4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
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>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/signal.h>
37 #include <linux/compiler.h>
38 #include <linux/completion.h>
39 #include <linux/pid.h>
40 #include <linux/percpu.h>
41 #include <linux/topology.h>
42 #include <linux/proportions.h>
43 #include <linux/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/task_io_accounting.h>
54 #include <linux/latencytop.h>
55 #include <linux/cred.h>
56 #include <linux/llist.h>
57 #include <linux/uidgid.h>
58 #include <linux/gfp.h>
60 #include <asm/processor.h>
62 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
65 * Extended scheduling parameters data structure.
67 * This is needed because the original struct sched_param can not be
68 * altered without introducing ABI issues with legacy applications
69 * (e.g., in sched_getparam()).
71 * However, the possibility of specifying more than just a priority for
72 * the tasks may be useful for a wide variety of application fields, e.g.,
73 * multimedia, streaming, automation and control, and many others.
75 * This variant (sched_attr) is meant at describing a so-called
76 * sporadic time-constrained task. In such model a task is specified by:
77 * - the activation period or minimum instance inter-arrival time;
78 * - the maximum (or average, depending on the actual scheduling
79 * discipline) computation time of all instances, a.k.a. runtime;
80 * - the deadline (relative to the actual activation time) of each
82 * Very briefly, a periodic (sporadic) task asks for the execution of
83 * some specific computation --which is typically called an instance--
84 * (at most) every period. Moreover, each instance typically lasts no more
85 * than the runtime and must be completed by time instant t equal to
86 * the instance activation time + the deadline.
88 * This is reflected by the actual fields of the sched_attr structure:
90 * @size size of the structure, for fwd/bwd compat.
92 * @sched_policy task's scheduling policy
93 * @sched_flags for customizing the scheduler behaviour
94 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
95 * @sched_priority task's static priority (SCHED_FIFO/RR)
96 * @sched_deadline representative of the task's deadline
97 * @sched_runtime representative of the task's runtime
98 * @sched_period representative of the task's period
100 * Given this task model, there are a multiplicity of scheduling algorithms
101 * and policies, that can be used to ensure all the tasks will make their
102 * timing constraints.
104 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
105 * only user of this new interface. More information about the algorithm
106 * available in the scheduling class file or in Documentation/.
114 /* SCHED_NORMAL, SCHED_BATCH */
117 /* SCHED_FIFO, SCHED_RR */
127 struct futex_pi_state;
128 struct robust_list_head;
131 struct perf_event_context;
136 * List of flags we want to share for kernel threads,
137 * if only because they are not used by them anyway.
139 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
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
151 extern unsigned long avenrun[]; /* Load averages */
152 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
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) */
161 #define CALC_LOAD(load,exp,n) \
163 load += n*(FIXED_1-exp); \
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 unsigned long this_cpu_load(void);
176 extern void calc_global_load(unsigned long ticks);
177 extern void update_cpu_load_nohz(void);
179 extern unsigned long get_parent_ip(unsigned long addr);
181 extern void dump_cpu_task(int cpu);
186 #ifdef CONFIG_SCHED_DEBUG
187 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
188 extern void proc_sched_set_task(struct task_struct *p);
190 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
194 * Task state bitmask. NOTE! These bits are also
195 * encoded in fs/proc/array.c: get_task_state().
197 * We have two separate sets of flags: task->state
198 * is about runnability, while task->exit_state are
199 * about the task exiting. Confusing, but this way
200 * modifying one set can't modify the other one by
203 #define TASK_RUNNING 0
204 #define TASK_INTERRUPTIBLE 1
205 #define TASK_UNINTERRUPTIBLE 2
206 #define __TASK_STOPPED 4
207 #define __TASK_TRACED 8
208 /* in tsk->exit_state */
209 #define EXIT_ZOMBIE 16
211 /* in tsk->state again */
213 #define TASK_WAKEKILL 128
214 #define TASK_WAKING 256
215 #define TASK_PARKED 512
216 #define TASK_STATE_MAX 1024
218 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
220 extern char ___assert_task_state[1 - 2*!!(
221 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
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)
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)
232 /* get_task_state() */
233 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
234 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
235 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
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)
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))
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
255 * set_current_state(TASK_UNINTERRUPTIBLE);
256 * if (do_i_need_to_sleep())
259 * If the caller does not need such serialisation then use __set_current_state()
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))
266 /* Task command name length */
267 #define TASK_COMM_LEN 16
269 #include <linux/spinlock.h>
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
277 extern rwlock_t tasklist_lock;
278 extern spinlock_t mmlist_lock;
282 #ifdef CONFIG_PROVE_RCU
283 extern int lockdep_tasklist_lock_is_held(void);
284 #endif /* #ifdef CONFIG_PROVE_RCU */
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);
292 extern int runqueue_is_locked(int cpu);
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(void);
299 static inline void nohz_balance_enter_idle(int cpu) { }
300 static inline void set_cpu_sd_state_idle(void) { }
304 * Only dump TASK_* tasks. (0 for all tasks)
306 extern void show_state_filter(unsigned long state_filter);
308 static inline void show_state(void)
310 show_state_filter(0);
313 extern void show_regs(struct pt_regs *);
316 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
317 * task), SP is the stack pointer of the first frame that should be shown in the back
318 * trace (or NULL if the entire call-chain of the task should be shown).
320 extern void show_stack(struct task_struct *task, unsigned long *sp);
322 void io_schedule(void);
323 long io_schedule_timeout(long timeout);
325 extern void cpu_init (void);
326 extern void trap_init(void);
327 extern void update_process_times(int user);
328 extern void scheduler_tick(void);
330 extern void sched_show_task(struct task_struct *p);
332 #ifdef CONFIG_LOCKUP_DETECTOR
333 extern void touch_softlockup_watchdog(void);
334 extern void touch_softlockup_watchdog_sync(void);
335 extern void touch_all_softlockup_watchdogs(void);
336 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
338 size_t *lenp, loff_t *ppos);
339 extern unsigned int softlockup_panic;
340 void lockup_detector_init(void);
342 static inline void touch_softlockup_watchdog(void)
345 static inline void touch_softlockup_watchdog_sync(void)
348 static inline void touch_all_softlockup_watchdogs(void)
351 static inline void lockup_detector_init(void)
356 #ifdef CONFIG_DETECT_HUNG_TASK
357 void reset_hung_task_detector(void);
359 static inline void reset_hung_task_detector(void)
364 /* Attach to any functions which should be ignored in wchan output. */
365 #define __sched __attribute__((__section__(".sched.text")))
367 /* Linker adds these: start and end of __sched functions */
368 extern char __sched_text_start[], __sched_text_end[];
370 /* Is this address in the __sched functions? */
371 extern int in_sched_functions(unsigned long addr);
373 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
374 extern signed long schedule_timeout(signed long timeout);
375 extern signed long schedule_timeout_interruptible(signed long timeout);
376 extern signed long schedule_timeout_killable(signed long timeout);
377 extern signed long schedule_timeout_uninterruptible(signed long timeout);
378 asmlinkage void schedule(void);
379 extern void schedule_preempt_disabled(void);
382 struct user_namespace;
385 extern void arch_pick_mmap_layout(struct mm_struct *mm);
387 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
388 unsigned long, unsigned long);
390 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
391 unsigned long len, unsigned long pgoff,
392 unsigned long flags);
394 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
397 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
398 #define SUID_DUMP_USER 1 /* Dump as user of process */
399 #define SUID_DUMP_ROOT 2 /* Dump as root */
403 /* for SUID_DUMP_* above */
404 #define MMF_DUMPABLE_BITS 2
405 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
407 extern void set_dumpable(struct mm_struct *mm, int value);
409 * This returns the actual value of the suid_dumpable flag. For things
410 * that are using this for checking for privilege transitions, it must
411 * test against SUID_DUMP_USER rather than treating it as a boolean
414 static inline int __get_dumpable(unsigned long mm_flags)
416 return mm_flags & MMF_DUMPABLE_MASK;
419 static inline int get_dumpable(struct mm_struct *mm)
421 return __get_dumpable(mm->flags);
424 /* coredump filter bits */
425 #define MMF_DUMP_ANON_PRIVATE 2
426 #define MMF_DUMP_ANON_SHARED 3
427 #define MMF_DUMP_MAPPED_PRIVATE 4
428 #define MMF_DUMP_MAPPED_SHARED 5
429 #define MMF_DUMP_ELF_HEADERS 6
430 #define MMF_DUMP_HUGETLB_PRIVATE 7
431 #define MMF_DUMP_HUGETLB_SHARED 8
433 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
434 #define MMF_DUMP_FILTER_BITS 7
435 #define MMF_DUMP_FILTER_MASK \
436 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
437 #define MMF_DUMP_FILTER_DEFAULT \
438 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
439 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
441 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
442 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
444 # define MMF_DUMP_MASK_DEFAULT_ELF 0
446 /* leave room for more dump flags */
447 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
448 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
449 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
451 #define MMF_HAS_UPROBES 19 /* has uprobes */
452 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
454 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
456 struct sighand_struct {
458 struct k_sigaction action[_NSIG];
460 wait_queue_head_t signalfd_wqh;
463 struct pacct_struct {
466 unsigned long ac_mem;
467 cputime_t ac_utime, ac_stime;
468 unsigned long ac_minflt, ac_majflt;
479 * struct cputime - snaphsot of system and user cputime
480 * @utime: time spent in user mode
481 * @stime: time spent in system mode
483 * Gathers a generic snapshot of user and system time.
491 * struct task_cputime - collected CPU time counts
492 * @utime: time spent in user mode, in &cputime_t units
493 * @stime: time spent in kernel mode, in &cputime_t units
494 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
496 * This is an extension of struct cputime that includes the total runtime
497 * spent by the task from the scheduler point of view.
499 * As a result, this structure groups together three kinds of CPU time
500 * that are tracked for threads and thread groups. Most things considering
501 * CPU time want to group these counts together and treat all three
502 * of them in parallel.
504 struct task_cputime {
507 unsigned long long sum_exec_runtime;
509 /* Alternate field names when used to cache expirations. */
510 #define prof_exp stime
511 #define virt_exp utime
512 #define sched_exp sum_exec_runtime
514 #define INIT_CPUTIME \
515 (struct task_cputime) { \
518 .sum_exec_runtime = 0, \
521 #ifdef CONFIG_PREEMPT_COUNT
522 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
524 #define PREEMPT_DISABLED PREEMPT_ENABLED
528 * Disable preemption until the scheduler is running.
529 * Reset by start_kernel()->sched_init()->init_idle().
531 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
532 * before the scheduler is active -- see should_resched().
534 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
537 * struct thread_group_cputimer - thread group interval timer counts
538 * @cputime: thread group interval timers.
539 * @running: non-zero when there are timers running and
540 * @cputime receives updates.
541 * @lock: lock for fields in this struct.
543 * This structure contains the version of task_cputime, above, that is
544 * used for thread group CPU timer calculations.
546 struct thread_group_cputimer {
547 struct task_cputime cputime;
552 #include <linux/rwsem.h>
556 * NOTE! "signal_struct" does not have its own
557 * locking, because a shared signal_struct always
558 * implies a shared sighand_struct, so locking
559 * sighand_struct is always a proper superset of
560 * the locking of signal_struct.
562 struct signal_struct {
566 struct list_head thread_head;
568 wait_queue_head_t wait_chldexit; /* for wait4() */
570 /* current thread group signal load-balancing target: */
571 struct task_struct *curr_target;
573 /* shared signal handling: */
574 struct sigpending shared_pending;
576 /* thread group exit support */
579 * - notify group_exit_task when ->count is equal to notify_count
580 * - everyone except group_exit_task is stopped during signal delivery
581 * of fatal signals, group_exit_task processes the signal.
584 struct task_struct *group_exit_task;
586 /* thread group stop support, overloads group_exit_code too */
587 int group_stop_count;
588 unsigned int flags; /* see SIGNAL_* flags below */
591 * PR_SET_CHILD_SUBREAPER marks a process, like a service
592 * manager, to re-parent orphan (double-forking) child processes
593 * to this process instead of 'init'. The service manager is
594 * able to receive SIGCHLD signals and is able to investigate
595 * the process until it calls wait(). All children of this
596 * process will inherit a flag if they should look for a
597 * child_subreaper process at exit.
599 unsigned int is_child_subreaper:1;
600 unsigned int has_child_subreaper:1;
602 /* POSIX.1b Interval Timers */
604 struct list_head posix_timers;
606 /* ITIMER_REAL timer for the process */
607 struct hrtimer real_timer;
608 struct pid *leader_pid;
609 ktime_t it_real_incr;
612 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
613 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
614 * values are defined to 0 and 1 respectively
616 struct cpu_itimer it[2];
619 * Thread group totals for process CPU timers.
620 * See thread_group_cputimer(), et al, for details.
622 struct thread_group_cputimer cputimer;
624 /* Earliest-expiration cache. */
625 struct task_cputime cputime_expires;
627 struct list_head cpu_timers[3];
629 struct pid *tty_old_pgrp;
631 /* boolean value for session group leader */
634 struct tty_struct *tty; /* NULL if no tty */
636 #ifdef CONFIG_SCHED_AUTOGROUP
637 struct autogroup *autogroup;
640 * Cumulative resource counters for dead threads in the group,
641 * and for reaped dead child processes forked by this group.
642 * Live threads maintain their own counters and add to these
643 * in __exit_signal, except for the group leader.
645 cputime_t utime, stime, cutime, cstime;
648 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
649 struct cputime prev_cputime;
651 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
652 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
653 unsigned long inblock, oublock, cinblock, coublock;
654 unsigned long maxrss, cmaxrss;
655 struct task_io_accounting ioac;
658 * Cumulative ns of schedule CPU time fo dead threads in the
659 * group, not including a zombie group leader, (This only differs
660 * from jiffies_to_ns(utime + stime) if sched_clock uses something
661 * other than jiffies.)
663 unsigned long long sum_sched_runtime;
666 * We don't bother to synchronize most readers of this at all,
667 * because there is no reader checking a limit that actually needs
668 * to get both rlim_cur and rlim_max atomically, and either one
669 * alone is a single word that can safely be read normally.
670 * getrlimit/setrlimit use task_lock(current->group_leader) to
671 * protect this instead of the siglock, because they really
672 * have no need to disable irqs.
674 struct rlimit rlim[RLIM_NLIMITS];
676 #ifdef CONFIG_BSD_PROCESS_ACCT
677 struct pacct_struct pacct; /* per-process accounting information */
679 #ifdef CONFIG_TASKSTATS
680 struct taskstats *stats;
684 unsigned audit_tty_log_passwd;
685 struct tty_audit_buf *tty_audit_buf;
687 #ifdef CONFIG_CGROUPS
689 * group_rwsem prevents new tasks from entering the threadgroup and
690 * member tasks from exiting,a more specifically, setting of
691 * PF_EXITING. fork and exit paths are protected with this rwsem
692 * using threadgroup_change_begin/end(). Users which require
693 * threadgroup to remain stable should use threadgroup_[un]lock()
694 * which also takes care of exec path. Currently, cgroup is the
697 struct rw_semaphore group_rwsem;
700 oom_flags_t oom_flags;
701 short oom_score_adj; /* OOM kill score adjustment */
702 short oom_score_adj_min; /* OOM kill score adjustment min value.
703 * Only settable by CAP_SYS_RESOURCE. */
705 struct mutex cred_guard_mutex; /* guard against foreign influences on
706 * credential calculations
707 * (notably. ptrace) */
711 * Bits in flags field of signal_struct.
713 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
714 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
715 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
716 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
718 * Pending notifications to parent.
720 #define SIGNAL_CLD_STOPPED 0x00000010
721 #define SIGNAL_CLD_CONTINUED 0x00000020
722 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
724 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
726 /* If true, all threads except ->group_exit_task have pending SIGKILL */
727 static inline int signal_group_exit(const struct signal_struct *sig)
729 return (sig->flags & SIGNAL_GROUP_EXIT) ||
730 (sig->group_exit_task != NULL);
734 * Some day this will be a full-fledged user tracking system..
737 atomic_t __count; /* reference count */
738 atomic_t processes; /* How many processes does this user have? */
739 atomic_t files; /* How many open files does this user have? */
740 atomic_t sigpending; /* How many pending signals does this user have? */
741 #ifdef CONFIG_INOTIFY_USER
742 atomic_t inotify_watches; /* How many inotify watches does this user have? */
743 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
745 #ifdef CONFIG_FANOTIFY
746 atomic_t fanotify_listeners;
749 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
751 #ifdef CONFIG_POSIX_MQUEUE
752 /* protected by mq_lock */
753 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
755 unsigned long locked_shm; /* How many pages of mlocked shm ? */
758 struct key *uid_keyring; /* UID specific keyring */
759 struct key *session_keyring; /* UID's default session keyring */
762 /* Hash table maintenance information */
763 struct hlist_node uidhash_node;
766 #ifdef CONFIG_PERF_EVENTS
767 atomic_long_t locked_vm;
771 extern int uids_sysfs_init(void);
773 extern struct user_struct *find_user(kuid_t);
775 extern struct user_struct root_user;
776 #define INIT_USER (&root_user)
779 struct backing_dev_info;
780 struct reclaim_state;
782 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
784 /* cumulative counters */
785 unsigned long pcount; /* # of times run on this cpu */
786 unsigned long long run_delay; /* time spent waiting on a runqueue */
789 unsigned long long last_arrival,/* when we last ran on a cpu */
790 last_queued; /* when we were last queued to run */
792 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
794 #ifdef CONFIG_TASK_DELAY_ACCT
795 struct task_delay_info {
797 unsigned int flags; /* Private per-task flags */
799 /* For each stat XXX, add following, aligned appropriately
801 * struct timespec XXX_start, XXX_end;
805 * Atomicity of updates to XXX_delay, XXX_count protected by
806 * single lock above (split into XXX_lock if contention is an issue).
810 * XXX_count is incremented on every XXX operation, the delay
811 * associated with the operation is added to XXX_delay.
812 * XXX_delay contains the accumulated delay time in nanoseconds.
814 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
815 u64 blkio_delay; /* wait for sync block io completion */
816 u64 swapin_delay; /* wait for swapin block io completion */
817 u32 blkio_count; /* total count of the number of sync block */
818 /* io operations performed */
819 u32 swapin_count; /* total count of the number of swapin block */
820 /* io operations performed */
822 struct timespec freepages_start, freepages_end;
823 u64 freepages_delay; /* wait for memory reclaim */
824 u32 freepages_count; /* total count of memory reclaim */
826 #endif /* CONFIG_TASK_DELAY_ACCT */
828 static inline int sched_info_on(void)
830 #ifdef CONFIG_SCHEDSTATS
832 #elif defined(CONFIG_TASK_DELAY_ACCT)
833 extern int delayacct_on;
848 * Increase resolution of cpu_power calculations
850 #define SCHED_POWER_SHIFT 10
851 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
854 * sched-domains (multiprocessor balancing) declarations:
857 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
858 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
859 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
860 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
861 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
862 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
863 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
864 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
865 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
866 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
867 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
868 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
869 #define SD_NUMA 0x4000 /* cross-node balancing */
871 extern int __weak arch_sd_sibiling_asym_packing(void);
873 struct sched_domain_attr {
874 int relax_domain_level;
877 #define SD_ATTR_INIT (struct sched_domain_attr) { \
878 .relax_domain_level = -1, \
881 extern int sched_domain_level_max;
885 struct sched_domain {
886 /* These fields must be setup */
887 struct sched_domain *parent; /* top domain must be null terminated */
888 struct sched_domain *child; /* bottom domain must be null terminated */
889 struct sched_group *groups; /* the balancing groups of the domain */
890 unsigned long min_interval; /* Minimum balance interval ms */
891 unsigned long max_interval; /* Maximum balance interval ms */
892 unsigned int busy_factor; /* less balancing by factor if busy */
893 unsigned int imbalance_pct; /* No balance until over watermark */
894 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
895 unsigned int busy_idx;
896 unsigned int idle_idx;
897 unsigned int newidle_idx;
898 unsigned int wake_idx;
899 unsigned int forkexec_idx;
900 unsigned int smt_gain;
902 int nohz_idle; /* NOHZ IDLE status */
903 int flags; /* See SD_* */
906 /* Runtime fields. */
907 unsigned long last_balance; /* init to jiffies. units in jiffies */
908 unsigned int balance_interval; /* initialise to 1. units in ms. */
909 unsigned int nr_balance_failed; /* initialise to 0 */
911 /* idle_balance() stats */
912 u64 max_newidle_lb_cost;
913 unsigned long next_decay_max_lb_cost;
915 #ifdef CONFIG_SCHEDSTATS
916 /* load_balance() stats */
917 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
918 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
919 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
920 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
921 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
922 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
923 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
924 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
926 /* Active load balancing */
927 unsigned int alb_count;
928 unsigned int alb_failed;
929 unsigned int alb_pushed;
931 /* SD_BALANCE_EXEC stats */
932 unsigned int sbe_count;
933 unsigned int sbe_balanced;
934 unsigned int sbe_pushed;
936 /* SD_BALANCE_FORK stats */
937 unsigned int sbf_count;
938 unsigned int sbf_balanced;
939 unsigned int sbf_pushed;
941 /* try_to_wake_up() stats */
942 unsigned int ttwu_wake_remote;
943 unsigned int ttwu_move_affine;
944 unsigned int ttwu_move_balance;
946 #ifdef CONFIG_SCHED_DEBUG
950 void *private; /* used during construction */
951 struct rcu_head rcu; /* used during destruction */
954 unsigned int span_weight;
956 * Span of all CPUs in this domain.
958 * NOTE: this field is variable length. (Allocated dynamically
959 * by attaching extra space to the end of the structure,
960 * depending on how many CPUs the kernel has booted up with)
962 unsigned long span[0];
965 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
967 return to_cpumask(sd->span);
970 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
971 struct sched_domain_attr *dattr_new);
973 /* Allocate an array of sched domains, for partition_sched_domains(). */
974 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
975 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
977 bool cpus_share_cache(int this_cpu, int that_cpu);
979 #else /* CONFIG_SMP */
981 struct sched_domain_attr;
984 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
985 struct sched_domain_attr *dattr_new)
989 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
994 #endif /* !CONFIG_SMP */
997 struct io_context; /* See blkdev.h */
1000 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1001 extern void prefetch_stack(struct task_struct *t);
1003 static inline void prefetch_stack(struct task_struct *t) { }
1006 struct audit_context; /* See audit.c */
1008 struct pipe_inode_info;
1009 struct uts_namespace;
1011 struct load_weight {
1012 unsigned long weight;
1018 * These sums represent an infinite geometric series and so are bound
1019 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1020 * choices of y < 1-2^(-32)*1024.
1022 u32 runnable_avg_sum, runnable_avg_period;
1023 u64 last_runnable_update;
1025 unsigned long load_avg_contrib;
1028 #ifdef CONFIG_SCHEDSTATS
1029 struct sched_statistics {
1039 s64 sum_sleep_runtime;
1046 u64 nr_migrations_cold;
1047 u64 nr_failed_migrations_affine;
1048 u64 nr_failed_migrations_running;
1049 u64 nr_failed_migrations_hot;
1050 u64 nr_forced_migrations;
1053 u64 nr_wakeups_sync;
1054 u64 nr_wakeups_migrate;
1055 u64 nr_wakeups_local;
1056 u64 nr_wakeups_remote;
1057 u64 nr_wakeups_affine;
1058 u64 nr_wakeups_affine_attempts;
1059 u64 nr_wakeups_passive;
1060 u64 nr_wakeups_idle;
1064 struct sched_entity {
1065 struct load_weight load; /* for load-balancing */
1066 struct rb_node run_node;
1067 struct list_head group_node;
1071 u64 sum_exec_runtime;
1073 u64 prev_sum_exec_runtime;
1077 #ifdef CONFIG_SCHEDSTATS
1078 struct sched_statistics statistics;
1081 #ifdef CONFIG_FAIR_GROUP_SCHED
1083 struct sched_entity *parent;
1084 /* rq on which this entity is (to be) queued: */
1085 struct cfs_rq *cfs_rq;
1086 /* rq "owned" by this entity/group: */
1087 struct cfs_rq *my_q;
1091 /* Per-entity load-tracking */
1092 struct sched_avg avg;
1096 struct sched_rt_entity {
1097 struct list_head run_list;
1098 unsigned long timeout;
1099 unsigned long watchdog_stamp;
1100 unsigned int time_slice;
1102 struct sched_rt_entity *back;
1103 #ifdef CONFIG_RT_GROUP_SCHED
1104 struct sched_rt_entity *parent;
1105 /* rq on which this entity is (to be) queued: */
1106 struct rt_rq *rt_rq;
1107 /* rq "owned" by this entity/group: */
1112 struct sched_dl_entity {
1113 struct rb_node rb_node;
1116 * Original scheduling parameters. Copied here from sched_attr
1117 * during sched_setscheduler2(), they will remain the same until
1118 * the next sched_setscheduler2().
1120 u64 dl_runtime; /* maximum runtime for each instance */
1121 u64 dl_deadline; /* relative deadline of each instance */
1122 u64 dl_period; /* separation of two instances (period) */
1123 u64 dl_bw; /* dl_runtime / dl_deadline */
1126 * Actual scheduling parameters. Initialized with the values above,
1127 * they are continously updated during task execution. Note that
1128 * the remaining runtime could be < 0 in case we are in overrun.
1130 s64 runtime; /* remaining runtime for this instance */
1131 u64 deadline; /* absolute deadline for this instance */
1132 unsigned int flags; /* specifying the scheduler behaviour */
1137 * @dl_throttled tells if we exhausted the runtime. If so, the
1138 * task has to wait for a replenishment to be performed at the
1139 * next firing of dl_timer.
1141 * @dl_new tells if a new instance arrived. If so we must
1142 * start executing it with full runtime and reset its absolute
1145 * @dl_boosted tells if we are boosted due to DI. If so we are
1146 * outside bandwidth enforcement mechanism (but only until we
1147 * exit the critical section).
1149 int dl_throttled, dl_new, dl_boosted;
1152 * Bandwidth enforcement timer. Each -deadline task has its
1153 * own bandwidth to be enforced, thus we need one timer per task.
1155 struct hrtimer dl_timer;
1160 enum perf_event_task_context {
1161 perf_invalid_context = -1,
1162 perf_hw_context = 0,
1164 perf_nr_task_contexts,
1167 struct task_struct {
1168 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1171 unsigned int flags; /* per process flags, defined below */
1172 unsigned int ptrace;
1175 struct llist_node wake_entry;
1177 struct task_struct *last_wakee;
1178 unsigned long wakee_flips;
1179 unsigned long wakee_flip_decay_ts;
1185 int prio, static_prio, normal_prio;
1186 unsigned int rt_priority;
1187 const struct sched_class *sched_class;
1188 struct sched_entity se;
1189 struct sched_rt_entity rt;
1190 #ifdef CONFIG_CGROUP_SCHED
1191 struct task_group *sched_task_group;
1193 struct sched_dl_entity dl;
1195 #ifdef CONFIG_PREEMPT_NOTIFIERS
1196 /* list of struct preempt_notifier: */
1197 struct hlist_head preempt_notifiers;
1200 #ifdef CONFIG_BLK_DEV_IO_TRACE
1201 unsigned int btrace_seq;
1204 unsigned int policy;
1205 int nr_cpus_allowed;
1206 cpumask_t cpus_allowed;
1208 #ifdef CONFIG_PREEMPT_RCU
1209 int rcu_read_lock_nesting;
1210 char rcu_read_unlock_special;
1211 struct list_head rcu_node_entry;
1212 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1213 #ifdef CONFIG_TREE_PREEMPT_RCU
1214 struct rcu_node *rcu_blocked_node;
1215 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1216 #ifdef CONFIG_RCU_BOOST
1217 struct rt_mutex *rcu_boost_mutex;
1218 #endif /* #ifdef CONFIG_RCU_BOOST */
1220 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1221 struct sched_info sched_info;
1224 struct list_head tasks;
1226 struct plist_node pushable_tasks;
1227 struct rb_node pushable_dl_tasks;
1230 struct mm_struct *mm, *active_mm;
1231 #ifdef CONFIG_COMPAT_BRK
1232 unsigned brk_randomized:1;
1234 #if defined(SPLIT_RSS_COUNTING)
1235 struct task_rss_stat rss_stat;
1239 int exit_code, exit_signal;
1240 int pdeath_signal; /* The signal sent when the parent dies */
1241 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1243 /* Used for emulating ABI behavior of previous Linux versions */
1244 unsigned int personality;
1246 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1248 unsigned in_iowait:1;
1250 /* task may not gain privileges */
1251 unsigned no_new_privs:1;
1253 /* Revert to default priority/policy when forking */
1254 unsigned sched_reset_on_fork:1;
1255 unsigned sched_contributes_to_load:1;
1260 #ifdef CONFIG_CC_STACKPROTECTOR
1261 /* Canary value for the -fstack-protector gcc feature */
1262 unsigned long stack_canary;
1265 * pointers to (original) parent process, youngest child, younger sibling,
1266 * older sibling, respectively. (p->father can be replaced with
1267 * p->real_parent->pid)
1269 struct task_struct __rcu *real_parent; /* real parent process */
1270 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1272 * children/sibling forms the list of my natural children
1274 struct list_head children; /* list of my children */
1275 struct list_head sibling; /* linkage in my parent's children list */
1276 struct task_struct *group_leader; /* threadgroup leader */
1279 * ptraced is the list of tasks this task is using ptrace on.
1280 * This includes both natural children and PTRACE_ATTACH targets.
1281 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1283 struct list_head ptraced;
1284 struct list_head ptrace_entry;
1286 /* PID/PID hash table linkage. */
1287 struct pid_link pids[PIDTYPE_MAX];
1288 struct list_head thread_group;
1289 struct list_head thread_node;
1291 struct completion *vfork_done; /* for vfork() */
1292 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1293 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1295 cputime_t utime, stime, utimescaled, stimescaled;
1297 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1298 struct cputime prev_cputime;
1300 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1301 seqlock_t vtime_seqlock;
1302 unsigned long long vtime_snap;
1307 } vtime_snap_whence;
1309 unsigned long nvcsw, nivcsw; /* context switch counts */
1310 struct timespec start_time; /* monotonic time */
1311 struct timespec real_start_time; /* boot based time */
1312 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1313 unsigned long min_flt, maj_flt;
1315 struct task_cputime cputime_expires;
1316 struct list_head cpu_timers[3];
1318 /* process credentials */
1319 const struct cred __rcu *real_cred; /* objective and real subjective task
1320 * credentials (COW) */
1321 const struct cred __rcu *cred; /* effective (overridable) subjective task
1322 * credentials (COW) */
1323 char comm[TASK_COMM_LEN]; /* executable name excluding path
1324 - access with [gs]et_task_comm (which lock
1325 it with task_lock())
1326 - initialized normally by setup_new_exec */
1327 /* file system info */
1328 int link_count, total_link_count;
1329 #ifdef CONFIG_SYSVIPC
1331 struct sysv_sem sysvsem;
1333 #ifdef CONFIG_DETECT_HUNG_TASK
1334 /* hung task detection */
1335 unsigned long last_switch_count;
1337 /* CPU-specific state of this task */
1338 struct thread_struct thread;
1339 /* filesystem information */
1340 struct fs_struct *fs;
1341 /* open file information */
1342 struct files_struct *files;
1344 struct nsproxy *nsproxy;
1345 /* signal handlers */
1346 struct signal_struct *signal;
1347 struct sighand_struct *sighand;
1349 sigset_t blocked, real_blocked;
1350 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1351 struct sigpending pending;
1353 unsigned long sas_ss_sp;
1355 int (*notifier)(void *priv);
1356 void *notifier_data;
1357 sigset_t *notifier_mask;
1358 struct callback_head *task_works;
1360 struct audit_context *audit_context;
1361 #ifdef CONFIG_AUDITSYSCALL
1363 unsigned int sessionid;
1365 struct seccomp seccomp;
1367 /* Thread group tracking */
1370 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1372 spinlock_t alloc_lock;
1374 /* Protection of the PI data structures: */
1375 raw_spinlock_t pi_lock;
1377 #ifdef CONFIG_RT_MUTEXES
1378 /* PI waiters blocked on a rt_mutex held by this task */
1379 struct rb_root pi_waiters;
1380 struct rb_node *pi_waiters_leftmost;
1381 /* Deadlock detection and priority inheritance handling */
1382 struct rt_mutex_waiter *pi_blocked_on;
1383 /* Top pi_waiters task */
1384 struct task_struct *pi_top_task;
1387 #ifdef CONFIG_DEBUG_MUTEXES
1388 /* mutex deadlock detection */
1389 struct mutex_waiter *blocked_on;
1391 #ifdef CONFIG_TRACE_IRQFLAGS
1392 unsigned int irq_events;
1393 unsigned long hardirq_enable_ip;
1394 unsigned long hardirq_disable_ip;
1395 unsigned int hardirq_enable_event;
1396 unsigned int hardirq_disable_event;
1397 int hardirqs_enabled;
1398 int hardirq_context;
1399 unsigned long softirq_disable_ip;
1400 unsigned long softirq_enable_ip;
1401 unsigned int softirq_disable_event;
1402 unsigned int softirq_enable_event;
1403 int softirqs_enabled;
1404 int softirq_context;
1406 #ifdef CONFIG_LOCKDEP
1407 # define MAX_LOCK_DEPTH 48UL
1410 unsigned int lockdep_recursion;
1411 struct held_lock held_locks[MAX_LOCK_DEPTH];
1412 gfp_t lockdep_reclaim_gfp;
1415 /* journalling filesystem info */
1418 /* stacked block device info */
1419 struct bio_list *bio_list;
1422 /* stack plugging */
1423 struct blk_plug *plug;
1427 struct reclaim_state *reclaim_state;
1429 struct backing_dev_info *backing_dev_info;
1431 struct io_context *io_context;
1433 unsigned long ptrace_message;
1434 siginfo_t *last_siginfo; /* For ptrace use. */
1435 struct task_io_accounting ioac;
1436 #if defined(CONFIG_TASK_XACCT)
1437 u64 acct_rss_mem1; /* accumulated rss usage */
1438 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1439 cputime_t acct_timexpd; /* stime + utime since last update */
1441 #ifdef CONFIG_CPUSETS
1442 nodemask_t mems_allowed; /* Protected by alloc_lock */
1443 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1444 int cpuset_mem_spread_rotor;
1445 int cpuset_slab_spread_rotor;
1447 #ifdef CONFIG_CGROUPS
1448 /* Control Group info protected by css_set_lock */
1449 struct css_set __rcu *cgroups;
1450 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1451 struct list_head cg_list;
1454 struct robust_list_head __user *robust_list;
1455 #ifdef CONFIG_COMPAT
1456 struct compat_robust_list_head __user *compat_robust_list;
1458 struct list_head pi_state_list;
1459 struct futex_pi_state *pi_state_cache;
1461 #ifdef CONFIG_PERF_EVENTS
1462 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1463 struct mutex perf_event_mutex;
1464 struct list_head perf_event_list;
1466 #ifdef CONFIG_DEBUG_PREEMPT
1467 unsigned long preempt_disable_ip;
1470 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1472 short pref_node_fork;
1474 #ifdef CONFIG_NUMA_BALANCING
1476 unsigned int numa_scan_period;
1477 unsigned int numa_scan_period_max;
1478 int numa_preferred_nid;
1479 unsigned long numa_migrate_retry;
1480 u64 node_stamp; /* migration stamp */
1481 u64 last_task_numa_placement;
1482 u64 last_sum_exec_runtime;
1483 struct callback_head numa_work;
1485 struct list_head numa_entry;
1486 struct numa_group *numa_group;
1489 * Exponential decaying average of faults on a per-node basis.
1490 * Scheduling placement decisions are made based on the these counts.
1491 * The values remain static for the duration of a PTE scan
1493 unsigned long *numa_faults_memory;
1494 unsigned long total_numa_faults;
1497 * numa_faults_buffer records faults per node during the current
1498 * scan window. When the scan completes, the counts in
1499 * numa_faults_memory decay and these values are copied.
1501 unsigned long *numa_faults_buffer_memory;
1504 * Track the nodes the process was running on when a NUMA hinting
1505 * fault was incurred.
1507 unsigned long *numa_faults_cpu;
1508 unsigned long *numa_faults_buffer_cpu;
1511 * numa_faults_locality tracks if faults recorded during the last
1512 * scan window were remote/local. The task scan period is adapted
1513 * based on the locality of the faults with different weights
1514 * depending on whether they were shared or private faults
1516 unsigned long numa_faults_locality[2];
1518 unsigned long numa_pages_migrated;
1519 #endif /* CONFIG_NUMA_BALANCING */
1521 struct rcu_head rcu;
1524 * cache last used pipe for splice
1526 struct pipe_inode_info *splice_pipe;
1528 struct page_frag task_frag;
1530 #ifdef CONFIG_TASK_DELAY_ACCT
1531 struct task_delay_info *delays;
1533 #ifdef CONFIG_FAULT_INJECTION
1537 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1538 * balance_dirty_pages() for some dirty throttling pause
1541 int nr_dirtied_pause;
1542 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1544 #ifdef CONFIG_LATENCYTOP
1545 int latency_record_count;
1546 struct latency_record latency_record[LT_SAVECOUNT];
1549 * time slack values; these are used to round up poll() and
1550 * select() etc timeout values. These are in nanoseconds.
1552 unsigned long timer_slack_ns;
1553 unsigned long default_timer_slack_ns;
1555 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1556 /* Index of current stored address in ret_stack */
1558 /* Stack of return addresses for return function tracing */
1559 struct ftrace_ret_stack *ret_stack;
1560 /* time stamp for last schedule */
1561 unsigned long long ftrace_timestamp;
1563 * Number of functions that haven't been traced
1564 * because of depth overrun.
1566 atomic_t trace_overrun;
1567 /* Pause for the tracing */
1568 atomic_t tracing_graph_pause;
1570 #ifdef CONFIG_TRACING
1571 /* state flags for use by tracers */
1572 unsigned long trace;
1573 /* bitmask and counter of trace recursion */
1574 unsigned long trace_recursion;
1575 #endif /* CONFIG_TRACING */
1576 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1577 struct memcg_batch_info {
1578 int do_batch; /* incremented when batch uncharge started */
1579 struct mem_cgroup *memcg; /* target memcg of uncharge */
1580 unsigned long nr_pages; /* uncharged usage */
1581 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1583 unsigned int memcg_kmem_skip_account;
1584 struct memcg_oom_info {
1585 struct mem_cgroup *memcg;
1588 unsigned int may_oom:1;
1591 #ifdef CONFIG_UPROBES
1592 struct uprobe_task *utask;
1594 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1595 unsigned int sequential_io;
1596 unsigned int sequential_io_avg;
1600 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1601 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1603 #define TNF_MIGRATED 0x01
1604 #define TNF_NO_GROUP 0x02
1605 #define TNF_SHARED 0x04
1606 #define TNF_FAULT_LOCAL 0x08
1608 #ifdef CONFIG_NUMA_BALANCING
1609 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1610 extern pid_t task_numa_group_id(struct task_struct *p);
1611 extern void set_numabalancing_state(bool enabled);
1612 extern void task_numa_free(struct task_struct *p);
1613 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1614 int src_nid, int dst_cpu);
1616 static inline void task_numa_fault(int last_node, int node, int pages,
1620 static inline pid_t task_numa_group_id(struct task_struct *p)
1624 static inline void set_numabalancing_state(bool enabled)
1627 static inline void task_numa_free(struct task_struct *p)
1630 static inline bool should_numa_migrate_memory(struct task_struct *p,
1631 struct page *page, int src_nid, int dst_cpu)
1637 static inline struct pid *task_pid(struct task_struct *task)
1639 return task->pids[PIDTYPE_PID].pid;
1642 static inline struct pid *task_tgid(struct task_struct *task)
1644 return task->group_leader->pids[PIDTYPE_PID].pid;
1648 * Without tasklist or rcu lock it is not safe to dereference
1649 * the result of task_pgrp/task_session even if task == current,
1650 * we can race with another thread doing sys_setsid/sys_setpgid.
1652 static inline struct pid *task_pgrp(struct task_struct *task)
1654 return task->group_leader->pids[PIDTYPE_PGID].pid;
1657 static inline struct pid *task_session(struct task_struct *task)
1659 return task->group_leader->pids[PIDTYPE_SID].pid;
1662 struct pid_namespace;
1665 * the helpers to get the task's different pids as they are seen
1666 * from various namespaces
1668 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1669 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1671 * task_xid_nr_ns() : id seen from the ns specified;
1673 * set_task_vxid() : assigns a virtual id to a task;
1675 * see also pid_nr() etc in include/linux/pid.h
1677 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1678 struct pid_namespace *ns);
1680 static inline pid_t task_pid_nr(struct task_struct *tsk)
1685 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1686 struct pid_namespace *ns)
1688 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1691 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1693 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1697 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1702 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1704 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1706 return pid_vnr(task_tgid(tsk));
1710 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1711 struct pid_namespace *ns)
1713 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1716 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1718 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1722 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1723 struct pid_namespace *ns)
1725 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1728 static inline pid_t task_session_vnr(struct task_struct *tsk)
1730 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1733 /* obsolete, do not use */
1734 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1736 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1740 * pid_alive - check that a task structure is not stale
1741 * @p: Task structure to be checked.
1743 * Test if a process is not yet dead (at most zombie state)
1744 * If pid_alive fails, then pointers within the task structure
1745 * can be stale and must not be dereferenced.
1747 * Return: 1 if the process is alive. 0 otherwise.
1749 static inline int pid_alive(struct task_struct *p)
1751 return p->pids[PIDTYPE_PID].pid != NULL;
1755 * is_global_init - check if a task structure is init
1756 * @tsk: Task structure to be checked.
1758 * Check if a task structure is the first user space task the kernel created.
1760 * Return: 1 if the task structure is init. 0 otherwise.
1762 static inline int is_global_init(struct task_struct *tsk)
1764 return tsk->pid == 1;
1767 extern struct pid *cad_pid;
1769 extern void free_task(struct task_struct *tsk);
1770 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1772 extern void __put_task_struct(struct task_struct *t);
1774 static inline void put_task_struct(struct task_struct *t)
1776 if (atomic_dec_and_test(&t->usage))
1777 __put_task_struct(t);
1780 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1781 extern void task_cputime(struct task_struct *t,
1782 cputime_t *utime, cputime_t *stime);
1783 extern void task_cputime_scaled(struct task_struct *t,
1784 cputime_t *utimescaled, cputime_t *stimescaled);
1785 extern cputime_t task_gtime(struct task_struct *t);
1787 static inline void task_cputime(struct task_struct *t,
1788 cputime_t *utime, cputime_t *stime)
1796 static inline void task_cputime_scaled(struct task_struct *t,
1797 cputime_t *utimescaled,
1798 cputime_t *stimescaled)
1801 *utimescaled = t->utimescaled;
1803 *stimescaled = t->stimescaled;
1806 static inline cputime_t task_gtime(struct task_struct *t)
1811 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1812 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1817 #define PF_EXITING 0x00000004 /* getting shut down */
1818 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1819 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1820 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1821 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1822 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1823 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1824 #define PF_DUMPCORE 0x00000200 /* dumped core */
1825 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1826 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1827 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1828 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1829 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1830 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1831 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1832 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1833 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1834 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1835 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1836 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1837 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1838 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1839 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1840 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1841 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1842 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1843 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1844 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1845 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1846 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1849 * Only the _current_ task can read/write to tsk->flags, but other
1850 * tasks can access tsk->flags in readonly mode for example
1851 * with tsk_used_math (like during threaded core dumping).
1852 * There is however an exception to this rule during ptrace
1853 * or during fork: the ptracer task is allowed to write to the
1854 * child->flags of its traced child (same goes for fork, the parent
1855 * can write to the child->flags), because we're guaranteed the
1856 * child is not running and in turn not changing child->flags
1857 * at the same time the parent does it.
1859 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1860 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1861 #define clear_used_math() clear_stopped_child_used_math(current)
1862 #define set_used_math() set_stopped_child_used_math(current)
1863 #define conditional_stopped_child_used_math(condition, child) \
1864 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1865 #define conditional_used_math(condition) \
1866 conditional_stopped_child_used_math(condition, current)
1867 #define copy_to_stopped_child_used_math(child) \
1868 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1869 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1870 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1871 #define used_math() tsk_used_math(current)
1873 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1874 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1876 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1881 static inline unsigned int memalloc_noio_save(void)
1883 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1884 current->flags |= PF_MEMALLOC_NOIO;
1888 static inline void memalloc_noio_restore(unsigned int flags)
1890 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1894 * task->jobctl flags
1896 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1898 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1899 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1900 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1901 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1902 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1903 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1904 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1906 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1907 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1908 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1909 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1910 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1911 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1912 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1914 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1915 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1917 extern bool task_set_jobctl_pending(struct task_struct *task,
1919 extern void task_clear_jobctl_trapping(struct task_struct *task);
1920 extern void task_clear_jobctl_pending(struct task_struct *task,
1923 #ifdef CONFIG_PREEMPT_RCU
1925 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1926 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1928 static inline void rcu_copy_process(struct task_struct *p)
1930 p->rcu_read_lock_nesting = 0;
1931 p->rcu_read_unlock_special = 0;
1932 #ifdef CONFIG_TREE_PREEMPT_RCU
1933 p->rcu_blocked_node = NULL;
1934 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1935 #ifdef CONFIG_RCU_BOOST
1936 p->rcu_boost_mutex = NULL;
1937 #endif /* #ifdef CONFIG_RCU_BOOST */
1938 INIT_LIST_HEAD(&p->rcu_node_entry);
1943 static inline void rcu_copy_process(struct task_struct *p)
1949 static inline void tsk_restore_flags(struct task_struct *task,
1950 unsigned long orig_flags, unsigned long flags)
1952 task->flags &= ~flags;
1953 task->flags |= orig_flags & flags;
1957 extern void do_set_cpus_allowed(struct task_struct *p,
1958 const struct cpumask *new_mask);
1960 extern int set_cpus_allowed_ptr(struct task_struct *p,
1961 const struct cpumask *new_mask);
1963 static inline void do_set_cpus_allowed(struct task_struct *p,
1964 const struct cpumask *new_mask)
1967 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1968 const struct cpumask *new_mask)
1970 if (!cpumask_test_cpu(0, new_mask))
1976 #ifdef CONFIG_NO_HZ_COMMON
1977 void calc_load_enter_idle(void);
1978 void calc_load_exit_idle(void);
1980 static inline void calc_load_enter_idle(void) { }
1981 static inline void calc_load_exit_idle(void) { }
1982 #endif /* CONFIG_NO_HZ_COMMON */
1984 #ifndef CONFIG_CPUMASK_OFFSTACK
1985 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1987 return set_cpus_allowed_ptr(p, &new_mask);
1992 * Do not use outside of architecture code which knows its limitations.
1994 * sched_clock() has no promise of monotonicity or bounded drift between
1995 * CPUs, use (which you should not) requires disabling IRQs.
1997 * Please use one of the three interfaces below.
1999 extern unsigned long long notrace sched_clock(void);
2001 * See the comment in kernel/sched/clock.c
2003 extern u64 cpu_clock(int cpu);
2004 extern u64 local_clock(void);
2005 extern u64 sched_clock_cpu(int cpu);
2008 extern void sched_clock_init(void);
2010 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2011 static inline void sched_clock_tick(void)
2015 static inline void sched_clock_idle_sleep_event(void)
2019 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2024 * Architectures can set this to 1 if they have specified
2025 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2026 * but then during bootup it turns out that sched_clock()
2027 * is reliable after all:
2029 extern int sched_clock_stable(void);
2030 extern void set_sched_clock_stable(void);
2031 extern void clear_sched_clock_stable(void);
2033 extern void sched_clock_tick(void);
2034 extern void sched_clock_idle_sleep_event(void);
2035 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2038 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2040 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2041 * The reason for this explicit opt-in is not to have perf penalty with
2042 * slow sched_clocks.
2044 extern void enable_sched_clock_irqtime(void);
2045 extern void disable_sched_clock_irqtime(void);
2047 static inline void enable_sched_clock_irqtime(void) {}
2048 static inline void disable_sched_clock_irqtime(void) {}
2051 extern unsigned long long
2052 task_sched_runtime(struct task_struct *task);
2054 /* sched_exec is called by processes performing an exec */
2056 extern void sched_exec(void);
2058 #define sched_exec() {}
2061 extern void sched_clock_idle_sleep_event(void);
2062 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2064 #ifdef CONFIG_HOTPLUG_CPU
2065 extern void idle_task_exit(void);
2067 static inline void idle_task_exit(void) {}
2070 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2071 extern void wake_up_nohz_cpu(int cpu);
2073 static inline void wake_up_nohz_cpu(int cpu) { }
2076 #ifdef CONFIG_NO_HZ_FULL
2077 extern bool sched_can_stop_tick(void);
2078 extern u64 scheduler_tick_max_deferment(void);
2080 static inline bool sched_can_stop_tick(void) { return false; }
2083 #ifdef CONFIG_SCHED_AUTOGROUP
2084 extern void sched_autogroup_create_attach(struct task_struct *p);
2085 extern void sched_autogroup_detach(struct task_struct *p);
2086 extern void sched_autogroup_fork(struct signal_struct *sig);
2087 extern void sched_autogroup_exit(struct signal_struct *sig);
2088 #ifdef CONFIG_PROC_FS
2089 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2090 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2093 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2094 static inline void sched_autogroup_detach(struct task_struct *p) { }
2095 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2096 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2099 extern bool yield_to(struct task_struct *p, bool preempt);
2100 extern void set_user_nice(struct task_struct *p, long nice);
2101 extern int task_prio(const struct task_struct *p);
2103 * task_nice - return the nice value of a given task.
2104 * @p: the task in question.
2106 * Return: The nice value [ -20 ... 0 ... 19 ].
2108 static inline int task_nice(const struct task_struct *p)
2110 return PRIO_TO_NICE((p)->static_prio);
2112 extern int can_nice(const struct task_struct *p, const int nice);
2113 extern int task_curr(const struct task_struct *p);
2114 extern int idle_cpu(int cpu);
2115 extern int sched_setscheduler(struct task_struct *, int,
2116 const struct sched_param *);
2117 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2118 const struct sched_param *);
2119 extern int sched_setattr(struct task_struct *,
2120 const struct sched_attr *);
2121 extern struct task_struct *idle_task(int cpu);
2123 * is_idle_task - is the specified task an idle task?
2124 * @p: the task in question.
2126 * Return: 1 if @p is an idle task. 0 otherwise.
2128 static inline bool is_idle_task(const struct task_struct *p)
2132 extern struct task_struct *curr_task(int cpu);
2133 extern void set_curr_task(int cpu, struct task_struct *p);
2138 * The default (Linux) execution domain.
2140 extern struct exec_domain default_exec_domain;
2142 union thread_union {
2143 struct thread_info thread_info;
2144 unsigned long stack[THREAD_SIZE/sizeof(long)];
2147 #ifndef __HAVE_ARCH_KSTACK_END
2148 static inline int kstack_end(void *addr)
2150 /* Reliable end of stack detection:
2151 * Some APM bios versions misalign the stack
2153 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2157 extern union thread_union init_thread_union;
2158 extern struct task_struct init_task;
2160 extern struct mm_struct init_mm;
2162 extern struct pid_namespace init_pid_ns;
2165 * find a task by one of its numerical ids
2167 * find_task_by_pid_ns():
2168 * finds a task by its pid in the specified namespace
2169 * find_task_by_vpid():
2170 * finds a task by its virtual pid
2172 * see also find_vpid() etc in include/linux/pid.h
2175 extern struct task_struct *find_task_by_vpid(pid_t nr);
2176 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2177 struct pid_namespace *ns);
2179 /* per-UID process charging. */
2180 extern struct user_struct * alloc_uid(kuid_t);
2181 static inline struct user_struct *get_uid(struct user_struct *u)
2183 atomic_inc(&u->__count);
2186 extern void free_uid(struct user_struct *);
2188 #include <asm/current.h>
2190 extern void xtime_update(unsigned long ticks);
2192 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2193 extern int wake_up_process(struct task_struct *tsk);
2194 extern void wake_up_new_task(struct task_struct *tsk);
2196 extern void kick_process(struct task_struct *tsk);
2198 static inline void kick_process(struct task_struct *tsk) { }
2200 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2201 extern void sched_dead(struct task_struct *p);
2203 extern void proc_caches_init(void);
2204 extern void flush_signals(struct task_struct *);
2205 extern void __flush_signals(struct task_struct *);
2206 extern void ignore_signals(struct task_struct *);
2207 extern void flush_signal_handlers(struct task_struct *, int force_default);
2208 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2210 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2212 unsigned long flags;
2215 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2216 ret = dequeue_signal(tsk, mask, info);
2217 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2222 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2224 extern void unblock_all_signals(void);
2225 extern void release_task(struct task_struct * p);
2226 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2227 extern int force_sigsegv(int, struct task_struct *);
2228 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2229 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2230 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2231 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2232 const struct cred *, u32);
2233 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2234 extern int kill_pid(struct pid *pid, int sig, int priv);
2235 extern int kill_proc_info(int, struct siginfo *, pid_t);
2236 extern __must_check bool do_notify_parent(struct task_struct *, int);
2237 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2238 extern void force_sig(int, struct task_struct *);
2239 extern int send_sig(int, struct task_struct *, int);
2240 extern int zap_other_threads(struct task_struct *p);
2241 extern struct sigqueue *sigqueue_alloc(void);
2242 extern void sigqueue_free(struct sigqueue *);
2243 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2244 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2246 static inline void restore_saved_sigmask(void)
2248 if (test_and_clear_restore_sigmask())
2249 __set_current_blocked(¤t->saved_sigmask);
2252 static inline sigset_t *sigmask_to_save(void)
2254 sigset_t *res = ¤t->blocked;
2255 if (unlikely(test_restore_sigmask()))
2256 res = ¤t->saved_sigmask;
2260 static inline int kill_cad_pid(int sig, int priv)
2262 return kill_pid(cad_pid, sig, priv);
2265 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2266 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2267 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2268 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2271 * True if we are on the alternate signal stack.
2273 static inline int on_sig_stack(unsigned long sp)
2275 #ifdef CONFIG_STACK_GROWSUP
2276 return sp >= current->sas_ss_sp &&
2277 sp - current->sas_ss_sp < current->sas_ss_size;
2279 return sp > current->sas_ss_sp &&
2280 sp - current->sas_ss_sp <= current->sas_ss_size;
2284 static inline int sas_ss_flags(unsigned long sp)
2286 return (current->sas_ss_size == 0 ? SS_DISABLE
2287 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2290 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2292 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2293 #ifdef CONFIG_STACK_GROWSUP
2294 return current->sas_ss_sp;
2296 return current->sas_ss_sp + current->sas_ss_size;
2302 * Routines for handling mm_structs
2304 extern struct mm_struct * mm_alloc(void);
2306 /* mmdrop drops the mm and the page tables */
2307 extern void __mmdrop(struct mm_struct *);
2308 static inline void mmdrop(struct mm_struct * mm)
2310 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2314 /* mmput gets rid of the mappings and all user-space */
2315 extern void mmput(struct mm_struct *);
2316 /* Grab a reference to a task's mm, if it is not already going away */
2317 extern struct mm_struct *get_task_mm(struct task_struct *task);
2319 * Grab a reference to a task's mm, if it is not already going away
2320 * and ptrace_may_access with the mode parameter passed to it
2323 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2324 /* Remove the current tasks stale references to the old mm_struct */
2325 extern void mm_release(struct task_struct *, struct mm_struct *);
2327 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2328 struct task_struct *);
2329 extern void flush_thread(void);
2330 extern void exit_thread(void);
2332 extern void exit_files(struct task_struct *);
2333 extern void __cleanup_sighand(struct sighand_struct *);
2335 extern void exit_itimers(struct signal_struct *);
2336 extern void flush_itimer_signals(void);
2338 extern void do_group_exit(int);
2340 extern int allow_signal(int);
2341 extern int disallow_signal(int);
2343 extern int do_execve(struct filename *,
2344 const char __user * const __user *,
2345 const char __user * const __user *);
2346 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2347 struct task_struct *fork_idle(int);
2348 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2350 extern void set_task_comm(struct task_struct *tsk, char *from);
2351 extern char *get_task_comm(char *to, struct task_struct *tsk);
2354 void scheduler_ipi(void);
2355 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2357 static inline void scheduler_ipi(void) { }
2358 static inline unsigned long wait_task_inactive(struct task_struct *p,
2365 #define next_task(p) \
2366 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2368 #define for_each_process(p) \
2369 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2371 extern bool current_is_single_threaded(void);
2374 * Careful: do_each_thread/while_each_thread is a double loop so
2375 * 'break' will not work as expected - use goto instead.
2377 #define do_each_thread(g, t) \
2378 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2380 #define while_each_thread(g, t) \
2381 while ((t = next_thread(t)) != g)
2383 #define __for_each_thread(signal, t) \
2384 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2386 #define for_each_thread(p, t) \
2387 __for_each_thread((p)->signal, t)
2389 /* Careful: this is a double loop, 'break' won't work as expected. */
2390 #define for_each_process_thread(p, t) \
2391 for_each_process(p) for_each_thread(p, t)
2393 static inline int get_nr_threads(struct task_struct *tsk)
2395 return tsk->signal->nr_threads;
2398 static inline bool thread_group_leader(struct task_struct *p)
2400 return p->exit_signal >= 0;
2403 /* Do to the insanities of de_thread it is possible for a process
2404 * to have the pid of the thread group leader without actually being
2405 * the thread group leader. For iteration through the pids in proc
2406 * all we care about is that we have a task with the appropriate
2407 * pid, we don't actually care if we have the right task.
2409 static inline bool has_group_leader_pid(struct task_struct *p)
2411 return task_pid(p) == p->signal->leader_pid;
2415 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2417 return p1->signal == p2->signal;
2420 static inline struct task_struct *next_thread(const struct task_struct *p)
2422 return list_entry_rcu(p->thread_group.next,
2423 struct task_struct, thread_group);
2426 static inline int thread_group_empty(struct task_struct *p)
2428 return list_empty(&p->thread_group);
2431 #define delay_group_leader(p) \
2432 (thread_group_leader(p) && !thread_group_empty(p))
2435 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2436 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2437 * pins the final release of task.io_context. Also protects ->cpuset and
2438 * ->cgroup.subsys[]. And ->vfork_done.
2440 * Nests both inside and outside of read_lock(&tasklist_lock).
2441 * It must not be nested with write_lock_irq(&tasklist_lock),
2442 * neither inside nor outside.
2444 static inline void task_lock(struct task_struct *p)
2446 spin_lock(&p->alloc_lock);
2449 static inline void task_unlock(struct task_struct *p)
2451 spin_unlock(&p->alloc_lock);
2454 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2455 unsigned long *flags);
2457 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2458 unsigned long *flags)
2460 struct sighand_struct *ret;
2462 ret = __lock_task_sighand(tsk, flags);
2463 (void)__cond_lock(&tsk->sighand->siglock, ret);
2467 static inline void unlock_task_sighand(struct task_struct *tsk,
2468 unsigned long *flags)
2470 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2473 #ifdef CONFIG_CGROUPS
2474 static inline void threadgroup_change_begin(struct task_struct *tsk)
2476 down_read(&tsk->signal->group_rwsem);
2478 static inline void threadgroup_change_end(struct task_struct *tsk)
2480 up_read(&tsk->signal->group_rwsem);
2484 * threadgroup_lock - lock threadgroup
2485 * @tsk: member task of the threadgroup to lock
2487 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2488 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2489 * change ->group_leader/pid. This is useful for cases where the threadgroup
2490 * needs to stay stable across blockable operations.
2492 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2493 * synchronization. While held, no new task will be added to threadgroup
2494 * and no existing live task will have its PF_EXITING set.
2496 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2497 * sub-thread becomes a new leader.
2499 static inline void threadgroup_lock(struct task_struct *tsk)
2501 down_write(&tsk->signal->group_rwsem);
2505 * threadgroup_unlock - unlock threadgroup
2506 * @tsk: member task of the threadgroup to unlock
2508 * Reverse threadgroup_lock().
2510 static inline void threadgroup_unlock(struct task_struct *tsk)
2512 up_write(&tsk->signal->group_rwsem);
2515 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2516 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2517 static inline void threadgroup_lock(struct task_struct *tsk) {}
2518 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2521 #ifndef __HAVE_THREAD_FUNCTIONS
2523 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2524 #define task_stack_page(task) ((task)->stack)
2526 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2528 *task_thread_info(p) = *task_thread_info(org);
2529 task_thread_info(p)->task = p;
2532 static inline unsigned long *end_of_stack(struct task_struct *p)
2534 return (unsigned long *)(task_thread_info(p) + 1);
2539 static inline int object_is_on_stack(void *obj)
2541 void *stack = task_stack_page(current);
2543 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2546 extern void thread_info_cache_init(void);
2548 #ifdef CONFIG_DEBUG_STACK_USAGE
2549 static inline unsigned long stack_not_used(struct task_struct *p)
2551 unsigned long *n = end_of_stack(p);
2553 do { /* Skip over canary */
2557 return (unsigned long)n - (unsigned long)end_of_stack(p);
2561 /* set thread flags in other task's structures
2562 * - see asm/thread_info.h for TIF_xxxx flags available
2564 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2566 set_ti_thread_flag(task_thread_info(tsk), flag);
2569 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2571 clear_ti_thread_flag(task_thread_info(tsk), flag);
2574 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2576 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2579 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2581 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2584 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2586 return test_ti_thread_flag(task_thread_info(tsk), flag);
2589 static inline void set_tsk_need_resched(struct task_struct *tsk)
2591 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2594 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2596 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2599 static inline int test_tsk_need_resched(struct task_struct *tsk)
2601 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2604 static inline int restart_syscall(void)
2606 set_tsk_thread_flag(current, TIF_SIGPENDING);
2607 return -ERESTARTNOINTR;
2610 static inline int signal_pending(struct task_struct *p)
2612 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2615 static inline int __fatal_signal_pending(struct task_struct *p)
2617 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2620 static inline int fatal_signal_pending(struct task_struct *p)
2622 return signal_pending(p) && __fatal_signal_pending(p);
2625 static inline int signal_pending_state(long state, struct task_struct *p)
2627 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2629 if (!signal_pending(p))
2632 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2636 * cond_resched() and cond_resched_lock(): latency reduction via
2637 * explicit rescheduling in places that are safe. The return
2638 * value indicates whether a reschedule was done in fact.
2639 * cond_resched_lock() will drop the spinlock before scheduling,
2640 * cond_resched_softirq() will enable bhs before scheduling.
2642 extern int _cond_resched(void);
2644 #define cond_resched() ({ \
2645 __might_sleep(__FILE__, __LINE__, 0); \
2649 extern int __cond_resched_lock(spinlock_t *lock);
2651 #ifdef CONFIG_PREEMPT_COUNT
2652 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2654 #define PREEMPT_LOCK_OFFSET 0
2657 #define cond_resched_lock(lock) ({ \
2658 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2659 __cond_resched_lock(lock); \
2662 extern int __cond_resched_softirq(void);
2664 #define cond_resched_softirq() ({ \
2665 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2666 __cond_resched_softirq(); \
2669 static inline void cond_resched_rcu(void)
2671 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2679 * Does a critical section need to be broken due to another
2680 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2681 * but a general need for low latency)
2683 static inline int spin_needbreak(spinlock_t *lock)
2685 #ifdef CONFIG_PREEMPT
2686 return spin_is_contended(lock);
2693 * Idle thread specific functions to determine the need_resched
2694 * polling state. We have two versions, one based on TS_POLLING in
2695 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2699 static inline int tsk_is_polling(struct task_struct *p)
2701 return task_thread_info(p)->status & TS_POLLING;
2703 static inline void __current_set_polling(void)
2705 current_thread_info()->status |= TS_POLLING;
2708 static inline bool __must_check current_set_polling_and_test(void)
2710 __current_set_polling();
2713 * Polling state must be visible before we test NEED_RESCHED,
2714 * paired by resched_task()
2718 return unlikely(tif_need_resched());
2721 static inline void __current_clr_polling(void)
2723 current_thread_info()->status &= ~TS_POLLING;
2726 static inline bool __must_check current_clr_polling_and_test(void)
2728 __current_clr_polling();
2731 * Polling state must be visible before we test NEED_RESCHED,
2732 * paired by resched_task()
2736 return unlikely(tif_need_resched());
2738 #elif defined(TIF_POLLING_NRFLAG)
2739 static inline int tsk_is_polling(struct task_struct *p)
2741 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2744 static inline void __current_set_polling(void)
2746 set_thread_flag(TIF_POLLING_NRFLAG);
2749 static inline bool __must_check current_set_polling_and_test(void)
2751 __current_set_polling();
2754 * Polling state must be visible before we test NEED_RESCHED,
2755 * paired by resched_task()
2757 * XXX: assumes set/clear bit are identical barrier wise.
2759 smp_mb__after_clear_bit();
2761 return unlikely(tif_need_resched());
2764 static inline void __current_clr_polling(void)
2766 clear_thread_flag(TIF_POLLING_NRFLAG);
2769 static inline bool __must_check current_clr_polling_and_test(void)
2771 __current_clr_polling();
2774 * Polling state must be visible before we test NEED_RESCHED,
2775 * paired by resched_task()
2777 smp_mb__after_clear_bit();
2779 return unlikely(tif_need_resched());
2783 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2784 static inline void __current_set_polling(void) { }
2785 static inline void __current_clr_polling(void) { }
2787 static inline bool __must_check current_set_polling_and_test(void)
2789 return unlikely(tif_need_resched());
2791 static inline bool __must_check current_clr_polling_and_test(void)
2793 return unlikely(tif_need_resched());
2797 static inline void current_clr_polling(void)
2799 __current_clr_polling();
2802 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2803 * Once the bit is cleared, we'll get IPIs with every new
2804 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2807 smp_mb(); /* paired with resched_task() */
2809 preempt_fold_need_resched();
2812 static __always_inline bool need_resched(void)
2814 return unlikely(tif_need_resched());
2818 * Thread group CPU time accounting.
2820 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2821 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2823 static inline void thread_group_cputime_init(struct signal_struct *sig)
2825 raw_spin_lock_init(&sig->cputimer.lock);
2829 * Reevaluate whether the task has signals pending delivery.
2830 * Wake the task if so.
2831 * This is required every time the blocked sigset_t changes.
2832 * callers must hold sighand->siglock.
2834 extern void recalc_sigpending_and_wake(struct task_struct *t);
2835 extern void recalc_sigpending(void);
2837 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2839 static inline void signal_wake_up(struct task_struct *t, bool resume)
2841 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2843 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2845 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2849 * Wrappers for p->thread_info->cpu access. No-op on UP.
2853 static inline unsigned int task_cpu(const struct task_struct *p)
2855 return task_thread_info(p)->cpu;
2858 static inline int task_node(const struct task_struct *p)
2860 return cpu_to_node(task_cpu(p));
2863 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2867 static inline unsigned int task_cpu(const struct task_struct *p)
2872 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2876 #endif /* CONFIG_SMP */
2878 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2879 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2881 #ifdef CONFIG_CGROUP_SCHED
2882 extern struct task_group root_task_group;
2883 #endif /* CONFIG_CGROUP_SCHED */
2885 extern int task_can_switch_user(struct user_struct *up,
2886 struct task_struct *tsk);
2888 #ifdef CONFIG_TASK_XACCT
2889 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2891 tsk->ioac.rchar += amt;
2894 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2896 tsk->ioac.wchar += amt;
2899 static inline void inc_syscr(struct task_struct *tsk)
2904 static inline void inc_syscw(struct task_struct *tsk)
2909 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2913 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2917 static inline void inc_syscr(struct task_struct *tsk)
2921 static inline void inc_syscw(struct task_struct *tsk)
2926 #ifndef TASK_SIZE_OF
2927 #define TASK_SIZE_OF(tsk) TASK_SIZE
2930 #ifdef CONFIG_MM_OWNER
2931 extern void mm_update_next_owner(struct mm_struct *mm);
2932 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2934 static inline void mm_update_next_owner(struct mm_struct *mm)
2938 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2941 #endif /* CONFIG_MM_OWNER */
2943 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2946 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2949 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2952 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2955 static inline unsigned long rlimit(unsigned int limit)
2957 return task_rlimit(current, limit);
2960 static inline unsigned long rlimit_max(unsigned int limit)
2962 return task_rlimit_max(current, limit);