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;
135 #define VMACACHE_BITS 2
136 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
137 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
140 * List of flags we want to share for kernel threads,
141 * if only because they are not used by them anyway.
143 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
146 * These are the constant used to fake the fixed-point load-average
147 * counting. Some notes:
148 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
149 * a load-average precision of 10 bits integer + 11 bits fractional
150 * - if you want to count load-averages more often, you need more
151 * precision, or rounding will get you. With 2-second counting freq,
152 * the EXP_n values would be 1981, 2034 and 2043 if still using only
155 extern unsigned long avenrun[]; /* Load averages */
156 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
158 #define FSHIFT 11 /* nr of bits of precision */
159 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
160 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
161 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
162 #define EXP_5 2014 /* 1/exp(5sec/5min) */
163 #define EXP_15 2037 /* 1/exp(5sec/15min) */
165 #define CALC_LOAD(load,exp,n) \
167 load += n*(FIXED_1-exp); \
170 extern unsigned long total_forks;
171 extern int nr_threads;
172 DECLARE_PER_CPU(unsigned long, process_counts);
173 extern int nr_processes(void);
174 extern unsigned long nr_running(void);
175 extern unsigned long nr_iowait(void);
176 extern unsigned long nr_iowait_cpu(int cpu);
177 extern unsigned long this_cpu_load(void);
180 extern void calc_global_load(unsigned long ticks);
181 extern void update_cpu_load_nohz(void);
183 extern unsigned long get_parent_ip(unsigned long addr);
185 extern void dump_cpu_task(int cpu);
190 #ifdef CONFIG_SCHED_DEBUG
191 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
192 extern void proc_sched_set_task(struct task_struct *p);
194 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
198 * Task state bitmask. NOTE! These bits are also
199 * encoded in fs/proc/array.c: get_task_state().
201 * We have two separate sets of flags: task->state
202 * is about runnability, while task->exit_state are
203 * about the task exiting. Confusing, but this way
204 * modifying one set can't modify the other one by
207 #define TASK_RUNNING 0
208 #define TASK_INTERRUPTIBLE 1
209 #define TASK_UNINTERRUPTIBLE 2
210 #define __TASK_STOPPED 4
211 #define __TASK_TRACED 8
212 /* in tsk->exit_state */
214 #define EXIT_ZOMBIE 32
215 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
216 /* in tsk->state again */
218 #define TASK_WAKEKILL 128
219 #define TASK_WAKING 256
220 #define TASK_PARKED 512
221 #define TASK_STATE_MAX 1024
223 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
225 extern char ___assert_task_state[1 - 2*!!(
226 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
228 /* Convenience macros for the sake of set_task_state */
229 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
230 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
231 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
233 /* Convenience macros for the sake of wake_up */
234 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
235 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
237 /* get_task_state() */
238 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
239 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
240 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
242 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
243 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
244 #define task_is_stopped_or_traced(task) \
245 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
246 #define task_contributes_to_load(task) \
247 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
248 (task->flags & PF_FROZEN) == 0)
250 #define __set_task_state(tsk, state_value) \
251 do { (tsk)->state = (state_value); } while (0)
252 #define set_task_state(tsk, state_value) \
253 set_mb((tsk)->state, (state_value))
256 * set_current_state() includes a barrier so that the write of current->state
257 * is correctly serialised wrt the caller's subsequent test of whether to
260 * set_current_state(TASK_UNINTERRUPTIBLE);
261 * if (do_i_need_to_sleep())
264 * If the caller does not need such serialisation then use __set_current_state()
266 #define __set_current_state(state_value) \
267 do { current->state = (state_value); } while (0)
268 #define set_current_state(state_value) \
269 set_mb(current->state, (state_value))
271 /* Task command name length */
272 #define TASK_COMM_LEN 16
274 #include <linux/spinlock.h>
277 * This serializes "schedule()" and also protects
278 * the run-queue from deletions/modifications (but
279 * _adding_ to the beginning of the run-queue has
282 extern rwlock_t tasklist_lock;
283 extern spinlock_t mmlist_lock;
287 #ifdef CONFIG_PROVE_RCU
288 extern int lockdep_tasklist_lock_is_held(void);
289 #endif /* #ifdef CONFIG_PROVE_RCU */
291 extern void sched_init(void);
292 extern void sched_init_smp(void);
293 extern asmlinkage void schedule_tail(struct task_struct *prev);
294 extern void init_idle(struct task_struct *idle, int cpu);
295 extern void init_idle_bootup_task(struct task_struct *idle);
297 extern int runqueue_is_locked(int cpu);
299 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
300 extern void nohz_balance_enter_idle(int cpu);
301 extern void set_cpu_sd_state_idle(void);
302 extern int get_nohz_timer_target(int pinned);
304 static inline void nohz_balance_enter_idle(int cpu) { }
305 static inline void set_cpu_sd_state_idle(void) { }
306 static inline int get_nohz_timer_target(int pinned)
308 return smp_processor_id();
313 * Only dump TASK_* tasks. (0 for all tasks)
315 extern void show_state_filter(unsigned long state_filter);
317 static inline void show_state(void)
319 show_state_filter(0);
322 extern void show_regs(struct pt_regs *);
325 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
326 * task), SP is the stack pointer of the first frame that should be shown in the back
327 * trace (or NULL if the entire call-chain of the task should be shown).
329 extern void show_stack(struct task_struct *task, unsigned long *sp);
331 void io_schedule(void);
332 long io_schedule_timeout(long timeout);
334 extern void cpu_init (void);
335 extern void trap_init(void);
336 extern void update_process_times(int user);
337 extern void scheduler_tick(void);
339 extern void sched_show_task(struct task_struct *p);
341 #ifdef CONFIG_LOCKUP_DETECTOR
342 extern void touch_softlockup_watchdog(void);
343 extern void touch_softlockup_watchdog_sync(void);
344 extern void touch_all_softlockup_watchdogs(void);
345 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
347 size_t *lenp, loff_t *ppos);
348 extern unsigned int softlockup_panic;
349 void lockup_detector_init(void);
351 static inline void touch_softlockup_watchdog(void)
354 static inline void touch_softlockup_watchdog_sync(void)
357 static inline void touch_all_softlockup_watchdogs(void)
360 static inline void lockup_detector_init(void)
365 #ifdef CONFIG_DETECT_HUNG_TASK
366 void reset_hung_task_detector(void);
368 static inline void reset_hung_task_detector(void)
373 /* Attach to any functions which should be ignored in wchan output. */
374 #define __sched __attribute__((__section__(".sched.text")))
376 /* Linker adds these: start and end of __sched functions */
377 extern char __sched_text_start[], __sched_text_end[];
379 /* Is this address in the __sched functions? */
380 extern int in_sched_functions(unsigned long addr);
382 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
383 extern signed long schedule_timeout(signed long timeout);
384 extern signed long schedule_timeout_interruptible(signed long timeout);
385 extern signed long schedule_timeout_killable(signed long timeout);
386 extern signed long schedule_timeout_uninterruptible(signed long timeout);
387 asmlinkage void schedule(void);
388 extern void schedule_preempt_disabled(void);
391 struct user_namespace;
394 extern void arch_pick_mmap_layout(struct mm_struct *mm);
396 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
397 unsigned long, unsigned long);
399 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
400 unsigned long len, unsigned long pgoff,
401 unsigned long flags);
403 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
406 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
407 #define SUID_DUMP_USER 1 /* Dump as user of process */
408 #define SUID_DUMP_ROOT 2 /* Dump as root */
412 /* for SUID_DUMP_* above */
413 #define MMF_DUMPABLE_BITS 2
414 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
416 extern void set_dumpable(struct mm_struct *mm, int value);
418 * This returns the actual value of the suid_dumpable flag. For things
419 * that are using this for checking for privilege transitions, it must
420 * test against SUID_DUMP_USER rather than treating it as a boolean
423 static inline int __get_dumpable(unsigned long mm_flags)
425 return mm_flags & MMF_DUMPABLE_MASK;
428 static inline int get_dumpable(struct mm_struct *mm)
430 return __get_dumpable(mm->flags);
433 /* coredump filter bits */
434 #define MMF_DUMP_ANON_PRIVATE 2
435 #define MMF_DUMP_ANON_SHARED 3
436 #define MMF_DUMP_MAPPED_PRIVATE 4
437 #define MMF_DUMP_MAPPED_SHARED 5
438 #define MMF_DUMP_ELF_HEADERS 6
439 #define MMF_DUMP_HUGETLB_PRIVATE 7
440 #define MMF_DUMP_HUGETLB_SHARED 8
442 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
443 #define MMF_DUMP_FILTER_BITS 7
444 #define MMF_DUMP_FILTER_MASK \
445 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
446 #define MMF_DUMP_FILTER_DEFAULT \
447 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
448 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
450 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
451 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
453 # define MMF_DUMP_MASK_DEFAULT_ELF 0
455 /* leave room for more dump flags */
456 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
457 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
458 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
460 #define MMF_HAS_UPROBES 19 /* has uprobes */
461 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
463 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
465 struct sighand_struct {
467 struct k_sigaction action[_NSIG];
469 wait_queue_head_t signalfd_wqh;
472 struct pacct_struct {
475 unsigned long ac_mem;
476 cputime_t ac_utime, ac_stime;
477 unsigned long ac_minflt, ac_majflt;
488 * struct cputime - snaphsot of system and user cputime
489 * @utime: time spent in user mode
490 * @stime: time spent in system mode
492 * Gathers a generic snapshot of user and system time.
500 * struct task_cputime - collected CPU time counts
501 * @utime: time spent in user mode, in &cputime_t units
502 * @stime: time spent in kernel mode, in &cputime_t units
503 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
505 * This is an extension of struct cputime that includes the total runtime
506 * spent by the task from the scheduler point of view.
508 * As a result, this structure groups together three kinds of CPU time
509 * that are tracked for threads and thread groups. Most things considering
510 * CPU time want to group these counts together and treat all three
511 * of them in parallel.
513 struct task_cputime {
516 unsigned long long sum_exec_runtime;
518 /* Alternate field names when used to cache expirations. */
519 #define prof_exp stime
520 #define virt_exp utime
521 #define sched_exp sum_exec_runtime
523 #define INIT_CPUTIME \
524 (struct task_cputime) { \
527 .sum_exec_runtime = 0, \
530 #ifdef CONFIG_PREEMPT_COUNT
531 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
533 #define PREEMPT_DISABLED PREEMPT_ENABLED
537 * Disable preemption until the scheduler is running.
538 * Reset by start_kernel()->sched_init()->init_idle().
540 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
541 * before the scheduler is active -- see should_resched().
543 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
546 * struct thread_group_cputimer - thread group interval timer counts
547 * @cputime: thread group interval timers.
548 * @running: non-zero when there are timers running and
549 * @cputime receives updates.
550 * @lock: lock for fields in this struct.
552 * This structure contains the version of task_cputime, above, that is
553 * used for thread group CPU timer calculations.
555 struct thread_group_cputimer {
556 struct task_cputime cputime;
561 #include <linux/rwsem.h>
565 * NOTE! "signal_struct" does not have its own
566 * locking, because a shared signal_struct always
567 * implies a shared sighand_struct, so locking
568 * sighand_struct is always a proper superset of
569 * the locking of signal_struct.
571 struct signal_struct {
575 struct list_head thread_head;
577 wait_queue_head_t wait_chldexit; /* for wait4() */
579 /* current thread group signal load-balancing target: */
580 struct task_struct *curr_target;
582 /* shared signal handling: */
583 struct sigpending shared_pending;
585 /* thread group exit support */
588 * - notify group_exit_task when ->count is equal to notify_count
589 * - everyone except group_exit_task is stopped during signal delivery
590 * of fatal signals, group_exit_task processes the signal.
593 struct task_struct *group_exit_task;
595 /* thread group stop support, overloads group_exit_code too */
596 int group_stop_count;
597 unsigned int flags; /* see SIGNAL_* flags below */
600 * PR_SET_CHILD_SUBREAPER marks a process, like a service
601 * manager, to re-parent orphan (double-forking) child processes
602 * to this process instead of 'init'. The service manager is
603 * able to receive SIGCHLD signals and is able to investigate
604 * the process until it calls wait(). All children of this
605 * process will inherit a flag if they should look for a
606 * child_subreaper process at exit.
608 unsigned int is_child_subreaper:1;
609 unsigned int has_child_subreaper:1;
611 /* POSIX.1b Interval Timers */
613 struct list_head posix_timers;
615 /* ITIMER_REAL timer for the process */
616 struct hrtimer real_timer;
617 struct pid *leader_pid;
618 ktime_t it_real_incr;
621 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
622 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
623 * values are defined to 0 and 1 respectively
625 struct cpu_itimer it[2];
628 * Thread group totals for process CPU timers.
629 * See thread_group_cputimer(), et al, for details.
631 struct thread_group_cputimer cputimer;
633 /* Earliest-expiration cache. */
634 struct task_cputime cputime_expires;
636 struct list_head cpu_timers[3];
638 struct pid *tty_old_pgrp;
640 /* boolean value for session group leader */
643 struct tty_struct *tty; /* NULL if no tty */
645 #ifdef CONFIG_SCHED_AUTOGROUP
646 struct autogroup *autogroup;
649 * Cumulative resource counters for dead threads in the group,
650 * and for reaped dead child processes forked by this group.
651 * Live threads maintain their own counters and add to these
652 * in __exit_signal, except for the group leader.
654 cputime_t utime, stime, cutime, cstime;
657 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
658 struct cputime prev_cputime;
660 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
661 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
662 unsigned long inblock, oublock, cinblock, coublock;
663 unsigned long maxrss, cmaxrss;
664 struct task_io_accounting ioac;
667 * Cumulative ns of schedule CPU time fo dead threads in the
668 * group, not including a zombie group leader, (This only differs
669 * from jiffies_to_ns(utime + stime) if sched_clock uses something
670 * other than jiffies.)
672 unsigned long long sum_sched_runtime;
675 * We don't bother to synchronize most readers of this at all,
676 * because there is no reader checking a limit that actually needs
677 * to get both rlim_cur and rlim_max atomically, and either one
678 * alone is a single word that can safely be read normally.
679 * getrlimit/setrlimit use task_lock(current->group_leader) to
680 * protect this instead of the siglock, because they really
681 * have no need to disable irqs.
683 struct rlimit rlim[RLIM_NLIMITS];
685 #ifdef CONFIG_BSD_PROCESS_ACCT
686 struct pacct_struct pacct; /* per-process accounting information */
688 #ifdef CONFIG_TASKSTATS
689 struct taskstats *stats;
693 unsigned audit_tty_log_passwd;
694 struct tty_audit_buf *tty_audit_buf;
696 #ifdef CONFIG_CGROUPS
698 * group_rwsem prevents new tasks from entering the threadgroup and
699 * member tasks from exiting,a more specifically, setting of
700 * PF_EXITING. fork and exit paths are protected with this rwsem
701 * using threadgroup_change_begin/end(). Users which require
702 * threadgroup to remain stable should use threadgroup_[un]lock()
703 * which also takes care of exec path. Currently, cgroup is the
706 struct rw_semaphore group_rwsem;
709 oom_flags_t oom_flags;
710 short oom_score_adj; /* OOM kill score adjustment */
711 short oom_score_adj_min; /* OOM kill score adjustment min value.
712 * Only settable by CAP_SYS_RESOURCE. */
714 struct mutex cred_guard_mutex; /* guard against foreign influences on
715 * credential calculations
716 * (notably. ptrace) */
720 * Bits in flags field of signal_struct.
722 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
723 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
724 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
725 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
727 * Pending notifications to parent.
729 #define SIGNAL_CLD_STOPPED 0x00000010
730 #define SIGNAL_CLD_CONTINUED 0x00000020
731 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
733 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
735 /* If true, all threads except ->group_exit_task have pending SIGKILL */
736 static inline int signal_group_exit(const struct signal_struct *sig)
738 return (sig->flags & SIGNAL_GROUP_EXIT) ||
739 (sig->group_exit_task != NULL);
743 * Some day this will be a full-fledged user tracking system..
746 atomic_t __count; /* reference count */
747 atomic_t processes; /* How many processes does this user have? */
748 atomic_t files; /* How many open files does this user have? */
749 atomic_t sigpending; /* How many pending signals does this user have? */
750 #ifdef CONFIG_INOTIFY_USER
751 atomic_t inotify_watches; /* How many inotify watches does this user have? */
752 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
754 #ifdef CONFIG_FANOTIFY
755 atomic_t fanotify_listeners;
758 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
760 #ifdef CONFIG_POSIX_MQUEUE
761 /* protected by mq_lock */
762 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
764 unsigned long locked_shm; /* How many pages of mlocked shm ? */
767 struct key *uid_keyring; /* UID specific keyring */
768 struct key *session_keyring; /* UID's default session keyring */
771 /* Hash table maintenance information */
772 struct hlist_node uidhash_node;
775 #ifdef CONFIG_PERF_EVENTS
776 atomic_long_t locked_vm;
780 extern int uids_sysfs_init(void);
782 extern struct user_struct *find_user(kuid_t);
784 extern struct user_struct root_user;
785 #define INIT_USER (&root_user)
788 struct backing_dev_info;
789 struct reclaim_state;
791 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
793 /* cumulative counters */
794 unsigned long pcount; /* # of times run on this cpu */
795 unsigned long long run_delay; /* time spent waiting on a runqueue */
798 unsigned long long last_arrival,/* when we last ran on a cpu */
799 last_queued; /* when we were last queued to run */
801 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
803 #ifdef CONFIG_TASK_DELAY_ACCT
804 struct task_delay_info {
806 unsigned int flags; /* Private per-task flags */
808 /* For each stat XXX, add following, aligned appropriately
810 * struct timespec XXX_start, XXX_end;
814 * Atomicity of updates to XXX_delay, XXX_count protected by
815 * single lock above (split into XXX_lock if contention is an issue).
819 * XXX_count is incremented on every XXX operation, the delay
820 * associated with the operation is added to XXX_delay.
821 * XXX_delay contains the accumulated delay time in nanoseconds.
823 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
824 u64 blkio_delay; /* wait for sync block io completion */
825 u64 swapin_delay; /* wait for swapin block io completion */
826 u32 blkio_count; /* total count of the number of sync block */
827 /* io operations performed */
828 u32 swapin_count; /* total count of the number of swapin block */
829 /* io operations performed */
831 struct timespec freepages_start, freepages_end;
832 u64 freepages_delay; /* wait for memory reclaim */
833 u32 freepages_count; /* total count of memory reclaim */
835 #endif /* CONFIG_TASK_DELAY_ACCT */
837 static inline int sched_info_on(void)
839 #ifdef CONFIG_SCHEDSTATS
841 #elif defined(CONFIG_TASK_DELAY_ACCT)
842 extern int delayacct_on;
857 * Increase resolution of cpu_power calculations
859 #define SCHED_POWER_SHIFT 10
860 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
863 * sched-domains (multiprocessor balancing) declarations:
866 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
867 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
868 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
869 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
870 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
871 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
872 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
873 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
874 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
875 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
876 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
877 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
878 #define SD_NUMA 0x4000 /* cross-node balancing */
880 extern int __weak arch_sd_sibiling_asym_packing(void);
882 struct sched_domain_attr {
883 int relax_domain_level;
886 #define SD_ATTR_INIT (struct sched_domain_attr) { \
887 .relax_domain_level = -1, \
890 extern int sched_domain_level_max;
894 struct sched_domain {
895 /* These fields must be setup */
896 struct sched_domain *parent; /* top domain must be null terminated */
897 struct sched_domain *child; /* bottom domain must be null terminated */
898 struct sched_group *groups; /* the balancing groups of the domain */
899 unsigned long min_interval; /* Minimum balance interval ms */
900 unsigned long max_interval; /* Maximum balance interval ms */
901 unsigned int busy_factor; /* less balancing by factor if busy */
902 unsigned int imbalance_pct; /* No balance until over watermark */
903 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
904 unsigned int busy_idx;
905 unsigned int idle_idx;
906 unsigned int newidle_idx;
907 unsigned int wake_idx;
908 unsigned int forkexec_idx;
909 unsigned int smt_gain;
911 int nohz_idle; /* NOHZ IDLE status */
912 int flags; /* See SD_* */
915 /* Runtime fields. */
916 unsigned long last_balance; /* init to jiffies. units in jiffies */
917 unsigned int balance_interval; /* initialise to 1. units in ms. */
918 unsigned int nr_balance_failed; /* initialise to 0 */
920 /* idle_balance() stats */
921 u64 max_newidle_lb_cost;
922 unsigned long next_decay_max_lb_cost;
924 #ifdef CONFIG_SCHEDSTATS
925 /* load_balance() stats */
926 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
927 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
928 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
929 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
931 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
932 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
933 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
935 /* Active load balancing */
936 unsigned int alb_count;
937 unsigned int alb_failed;
938 unsigned int alb_pushed;
940 /* SD_BALANCE_EXEC stats */
941 unsigned int sbe_count;
942 unsigned int sbe_balanced;
943 unsigned int sbe_pushed;
945 /* SD_BALANCE_FORK stats */
946 unsigned int sbf_count;
947 unsigned int sbf_balanced;
948 unsigned int sbf_pushed;
950 /* try_to_wake_up() stats */
951 unsigned int ttwu_wake_remote;
952 unsigned int ttwu_move_affine;
953 unsigned int ttwu_move_balance;
955 #ifdef CONFIG_SCHED_DEBUG
959 void *private; /* used during construction */
960 struct rcu_head rcu; /* used during destruction */
963 unsigned int span_weight;
965 * Span of all CPUs in this domain.
967 * NOTE: this field is variable length. (Allocated dynamically
968 * by attaching extra space to the end of the structure,
969 * depending on how many CPUs the kernel has booted up with)
971 unsigned long span[0];
974 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
976 return to_cpumask(sd->span);
979 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
980 struct sched_domain_attr *dattr_new);
982 /* Allocate an array of sched domains, for partition_sched_domains(). */
983 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
984 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
986 bool cpus_share_cache(int this_cpu, int that_cpu);
988 #else /* CONFIG_SMP */
990 struct sched_domain_attr;
993 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
994 struct sched_domain_attr *dattr_new)
998 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1003 #endif /* !CONFIG_SMP */
1006 struct io_context; /* See blkdev.h */
1009 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1010 extern void prefetch_stack(struct task_struct *t);
1012 static inline void prefetch_stack(struct task_struct *t) { }
1015 struct audit_context; /* See audit.c */
1017 struct pipe_inode_info;
1018 struct uts_namespace;
1020 struct load_weight {
1021 unsigned long weight;
1027 * These sums represent an infinite geometric series and so are bound
1028 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1029 * choices of y < 1-2^(-32)*1024.
1031 u32 runnable_avg_sum, runnable_avg_period;
1032 u64 last_runnable_update;
1034 unsigned long load_avg_contrib;
1037 #ifdef CONFIG_SCHEDSTATS
1038 struct sched_statistics {
1048 s64 sum_sleep_runtime;
1055 u64 nr_migrations_cold;
1056 u64 nr_failed_migrations_affine;
1057 u64 nr_failed_migrations_running;
1058 u64 nr_failed_migrations_hot;
1059 u64 nr_forced_migrations;
1062 u64 nr_wakeups_sync;
1063 u64 nr_wakeups_migrate;
1064 u64 nr_wakeups_local;
1065 u64 nr_wakeups_remote;
1066 u64 nr_wakeups_affine;
1067 u64 nr_wakeups_affine_attempts;
1068 u64 nr_wakeups_passive;
1069 u64 nr_wakeups_idle;
1073 struct sched_entity {
1074 struct load_weight load; /* for load-balancing */
1075 struct rb_node run_node;
1076 struct list_head group_node;
1080 u64 sum_exec_runtime;
1082 u64 prev_sum_exec_runtime;
1086 #ifdef CONFIG_SCHEDSTATS
1087 struct sched_statistics statistics;
1090 #ifdef CONFIG_FAIR_GROUP_SCHED
1092 struct sched_entity *parent;
1093 /* rq on which this entity is (to be) queued: */
1094 struct cfs_rq *cfs_rq;
1095 /* rq "owned" by this entity/group: */
1096 struct cfs_rq *my_q;
1100 /* Per-entity load-tracking */
1101 struct sched_avg avg;
1105 struct sched_rt_entity {
1106 struct list_head run_list;
1107 unsigned long timeout;
1108 unsigned long watchdog_stamp;
1109 unsigned int time_slice;
1111 struct sched_rt_entity *back;
1112 #ifdef CONFIG_RT_GROUP_SCHED
1113 struct sched_rt_entity *parent;
1114 /* rq on which this entity is (to be) queued: */
1115 struct rt_rq *rt_rq;
1116 /* rq "owned" by this entity/group: */
1121 struct sched_dl_entity {
1122 struct rb_node rb_node;
1125 * Original scheduling parameters. Copied here from sched_attr
1126 * during sched_setscheduler2(), they will remain the same until
1127 * the next sched_setscheduler2().
1129 u64 dl_runtime; /* maximum runtime for each instance */
1130 u64 dl_deadline; /* relative deadline of each instance */
1131 u64 dl_period; /* separation of two instances (period) */
1132 u64 dl_bw; /* dl_runtime / dl_deadline */
1135 * Actual scheduling parameters. Initialized with the values above,
1136 * they are continously updated during task execution. Note that
1137 * the remaining runtime could be < 0 in case we are in overrun.
1139 s64 runtime; /* remaining runtime for this instance */
1140 u64 deadline; /* absolute deadline for this instance */
1141 unsigned int flags; /* specifying the scheduler behaviour */
1146 * @dl_throttled tells if we exhausted the runtime. If so, the
1147 * task has to wait for a replenishment to be performed at the
1148 * next firing of dl_timer.
1150 * @dl_new tells if a new instance arrived. If so we must
1151 * start executing it with full runtime and reset its absolute
1154 * @dl_boosted tells if we are boosted due to DI. If so we are
1155 * outside bandwidth enforcement mechanism (but only until we
1156 * exit the critical section);
1158 * @dl_yielded tells if task gave up the cpu before consuming
1159 * all its available runtime during the last job.
1161 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1164 * Bandwidth enforcement timer. Each -deadline task has its
1165 * own bandwidth to be enforced, thus we need one timer per task.
1167 struct hrtimer dl_timer;
1172 enum perf_event_task_context {
1173 perf_invalid_context = -1,
1174 perf_hw_context = 0,
1176 perf_nr_task_contexts,
1179 struct task_struct {
1180 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1183 unsigned int flags; /* per process flags, defined below */
1184 unsigned int ptrace;
1187 struct llist_node wake_entry;
1189 struct task_struct *last_wakee;
1190 unsigned long wakee_flips;
1191 unsigned long wakee_flip_decay_ts;
1197 int prio, static_prio, normal_prio;
1198 unsigned int rt_priority;
1199 const struct sched_class *sched_class;
1200 struct sched_entity se;
1201 struct sched_rt_entity rt;
1202 #ifdef CONFIG_CGROUP_SCHED
1203 struct task_group *sched_task_group;
1205 struct sched_dl_entity dl;
1207 #ifdef CONFIG_PREEMPT_NOTIFIERS
1208 /* list of struct preempt_notifier: */
1209 struct hlist_head preempt_notifiers;
1212 #ifdef CONFIG_BLK_DEV_IO_TRACE
1213 unsigned int btrace_seq;
1216 unsigned int policy;
1217 int nr_cpus_allowed;
1218 cpumask_t cpus_allowed;
1220 #ifdef CONFIG_PREEMPT_RCU
1221 int rcu_read_lock_nesting;
1222 char rcu_read_unlock_special;
1223 struct list_head rcu_node_entry;
1224 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1225 #ifdef CONFIG_TREE_PREEMPT_RCU
1226 struct rcu_node *rcu_blocked_node;
1227 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1228 #ifdef CONFIG_RCU_BOOST
1229 struct rt_mutex *rcu_boost_mutex;
1230 #endif /* #ifdef CONFIG_RCU_BOOST */
1232 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1233 struct sched_info sched_info;
1236 struct list_head tasks;
1238 struct plist_node pushable_tasks;
1239 struct rb_node pushable_dl_tasks;
1242 struct mm_struct *mm, *active_mm;
1243 #ifdef CONFIG_COMPAT_BRK
1244 unsigned brk_randomized:1;
1246 /* per-thread vma caching */
1247 u32 vmacache_seqnum;
1248 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1249 #if defined(SPLIT_RSS_COUNTING)
1250 struct task_rss_stat rss_stat;
1254 int exit_code, exit_signal;
1255 int pdeath_signal; /* The signal sent when the parent dies */
1256 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1258 /* Used for emulating ABI behavior of previous Linux versions */
1259 unsigned int personality;
1261 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1263 unsigned in_iowait:1;
1265 /* task may not gain privileges */
1266 unsigned no_new_privs:1;
1268 /* Revert to default priority/policy when forking */
1269 unsigned sched_reset_on_fork:1;
1270 unsigned sched_contributes_to_load:1;
1275 #ifdef CONFIG_CC_STACKPROTECTOR
1276 /* Canary value for the -fstack-protector gcc feature */
1277 unsigned long stack_canary;
1280 * pointers to (original) parent process, youngest child, younger sibling,
1281 * older sibling, respectively. (p->father can be replaced with
1282 * p->real_parent->pid)
1284 struct task_struct __rcu *real_parent; /* real parent process */
1285 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1287 * children/sibling forms the list of my natural children
1289 struct list_head children; /* list of my children */
1290 struct list_head sibling; /* linkage in my parent's children list */
1291 struct task_struct *group_leader; /* threadgroup leader */
1294 * ptraced is the list of tasks this task is using ptrace on.
1295 * This includes both natural children and PTRACE_ATTACH targets.
1296 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1298 struct list_head ptraced;
1299 struct list_head ptrace_entry;
1301 /* PID/PID hash table linkage. */
1302 struct pid_link pids[PIDTYPE_MAX];
1303 struct list_head thread_group;
1304 struct list_head thread_node;
1306 struct completion *vfork_done; /* for vfork() */
1307 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1308 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1310 cputime_t utime, stime, utimescaled, stimescaled;
1312 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1313 struct cputime prev_cputime;
1315 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1316 seqlock_t vtime_seqlock;
1317 unsigned long long vtime_snap;
1322 } vtime_snap_whence;
1324 unsigned long nvcsw, nivcsw; /* context switch counts */
1325 struct timespec start_time; /* monotonic time */
1326 struct timespec real_start_time; /* boot based time */
1327 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1328 unsigned long min_flt, maj_flt;
1330 struct task_cputime cputime_expires;
1331 struct list_head cpu_timers[3];
1333 /* process credentials */
1334 const struct cred __rcu *real_cred; /* objective and real subjective task
1335 * credentials (COW) */
1336 const struct cred __rcu *cred; /* effective (overridable) subjective task
1337 * credentials (COW) */
1338 char comm[TASK_COMM_LEN]; /* executable name excluding path
1339 - access with [gs]et_task_comm (which lock
1340 it with task_lock())
1341 - initialized normally by setup_new_exec */
1342 /* file system info */
1343 int link_count, total_link_count;
1344 #ifdef CONFIG_SYSVIPC
1346 struct sysv_sem sysvsem;
1348 #ifdef CONFIG_DETECT_HUNG_TASK
1349 /* hung task detection */
1350 unsigned long last_switch_count;
1352 /* CPU-specific state of this task */
1353 struct thread_struct thread;
1354 /* filesystem information */
1355 struct fs_struct *fs;
1356 /* open file information */
1357 struct files_struct *files;
1359 struct nsproxy *nsproxy;
1360 /* signal handlers */
1361 struct signal_struct *signal;
1362 struct sighand_struct *sighand;
1364 sigset_t blocked, real_blocked;
1365 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1366 struct sigpending pending;
1368 unsigned long sas_ss_sp;
1370 int (*notifier)(void *priv);
1371 void *notifier_data;
1372 sigset_t *notifier_mask;
1373 struct callback_head *task_works;
1375 struct audit_context *audit_context;
1376 #ifdef CONFIG_AUDITSYSCALL
1378 unsigned int sessionid;
1380 struct seccomp seccomp;
1382 /* Thread group tracking */
1385 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1387 spinlock_t alloc_lock;
1389 /* Protection of the PI data structures: */
1390 raw_spinlock_t pi_lock;
1392 #ifdef CONFIG_RT_MUTEXES
1393 /* PI waiters blocked on a rt_mutex held by this task */
1394 struct rb_root pi_waiters;
1395 struct rb_node *pi_waiters_leftmost;
1396 /* Deadlock detection and priority inheritance handling */
1397 struct rt_mutex_waiter *pi_blocked_on;
1398 /* Top pi_waiters task */
1399 struct task_struct *pi_top_task;
1402 #ifdef CONFIG_DEBUG_MUTEXES
1403 /* mutex deadlock detection */
1404 struct mutex_waiter *blocked_on;
1406 #ifdef CONFIG_TRACE_IRQFLAGS
1407 unsigned int irq_events;
1408 unsigned long hardirq_enable_ip;
1409 unsigned long hardirq_disable_ip;
1410 unsigned int hardirq_enable_event;
1411 unsigned int hardirq_disable_event;
1412 int hardirqs_enabled;
1413 int hardirq_context;
1414 unsigned long softirq_disable_ip;
1415 unsigned long softirq_enable_ip;
1416 unsigned int softirq_disable_event;
1417 unsigned int softirq_enable_event;
1418 int softirqs_enabled;
1419 int softirq_context;
1421 #ifdef CONFIG_LOCKDEP
1422 # define MAX_LOCK_DEPTH 48UL
1425 unsigned int lockdep_recursion;
1426 struct held_lock held_locks[MAX_LOCK_DEPTH];
1427 gfp_t lockdep_reclaim_gfp;
1430 /* journalling filesystem info */
1433 /* stacked block device info */
1434 struct bio_list *bio_list;
1437 /* stack plugging */
1438 struct blk_plug *plug;
1442 struct reclaim_state *reclaim_state;
1444 struct backing_dev_info *backing_dev_info;
1446 struct io_context *io_context;
1448 unsigned long ptrace_message;
1449 siginfo_t *last_siginfo; /* For ptrace use. */
1450 struct task_io_accounting ioac;
1451 #if defined(CONFIG_TASK_XACCT)
1452 u64 acct_rss_mem1; /* accumulated rss usage */
1453 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1454 cputime_t acct_timexpd; /* stime + utime since last update */
1456 #ifdef CONFIG_CPUSETS
1457 nodemask_t mems_allowed; /* Protected by alloc_lock */
1458 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1459 int cpuset_mem_spread_rotor;
1460 int cpuset_slab_spread_rotor;
1462 #ifdef CONFIG_CGROUPS
1463 /* Control Group info protected by css_set_lock */
1464 struct css_set __rcu *cgroups;
1465 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1466 struct list_head cg_list;
1469 struct robust_list_head __user *robust_list;
1470 #ifdef CONFIG_COMPAT
1471 struct compat_robust_list_head __user *compat_robust_list;
1473 struct list_head pi_state_list;
1474 struct futex_pi_state *pi_state_cache;
1476 #ifdef CONFIG_PERF_EVENTS
1477 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1478 struct mutex perf_event_mutex;
1479 struct list_head perf_event_list;
1481 #ifdef CONFIG_DEBUG_PREEMPT
1482 unsigned long preempt_disable_ip;
1485 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1487 short pref_node_fork;
1489 #ifdef CONFIG_NUMA_BALANCING
1491 unsigned int numa_scan_period;
1492 unsigned int numa_scan_period_max;
1493 int numa_preferred_nid;
1494 unsigned long numa_migrate_retry;
1495 u64 node_stamp; /* migration stamp */
1496 u64 last_task_numa_placement;
1497 u64 last_sum_exec_runtime;
1498 struct callback_head numa_work;
1500 struct list_head numa_entry;
1501 struct numa_group *numa_group;
1504 * Exponential decaying average of faults on a per-node basis.
1505 * Scheduling placement decisions are made based on the these counts.
1506 * The values remain static for the duration of a PTE scan
1508 unsigned long *numa_faults_memory;
1509 unsigned long total_numa_faults;
1512 * numa_faults_buffer records faults per node during the current
1513 * scan window. When the scan completes, the counts in
1514 * numa_faults_memory decay and these values are copied.
1516 unsigned long *numa_faults_buffer_memory;
1519 * Track the nodes the process was running on when a NUMA hinting
1520 * fault was incurred.
1522 unsigned long *numa_faults_cpu;
1523 unsigned long *numa_faults_buffer_cpu;
1526 * numa_faults_locality tracks if faults recorded during the last
1527 * scan window were remote/local. The task scan period is adapted
1528 * based on the locality of the faults with different weights
1529 * depending on whether they were shared or private faults
1531 unsigned long numa_faults_locality[2];
1533 unsigned long numa_pages_migrated;
1534 #endif /* CONFIG_NUMA_BALANCING */
1536 struct rcu_head rcu;
1539 * cache last used pipe for splice
1541 struct pipe_inode_info *splice_pipe;
1543 struct page_frag task_frag;
1545 #ifdef CONFIG_TASK_DELAY_ACCT
1546 struct task_delay_info *delays;
1548 #ifdef CONFIG_FAULT_INJECTION
1552 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1553 * balance_dirty_pages() for some dirty throttling pause
1556 int nr_dirtied_pause;
1557 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1559 #ifdef CONFIG_LATENCYTOP
1560 int latency_record_count;
1561 struct latency_record latency_record[LT_SAVECOUNT];
1564 * time slack values; these are used to round up poll() and
1565 * select() etc timeout values. These are in nanoseconds.
1567 unsigned long timer_slack_ns;
1568 unsigned long default_timer_slack_ns;
1570 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1571 /* Index of current stored address in ret_stack */
1573 /* Stack of return addresses for return function tracing */
1574 struct ftrace_ret_stack *ret_stack;
1575 /* time stamp for last schedule */
1576 unsigned long long ftrace_timestamp;
1578 * Number of functions that haven't been traced
1579 * because of depth overrun.
1581 atomic_t trace_overrun;
1582 /* Pause for the tracing */
1583 atomic_t tracing_graph_pause;
1585 #ifdef CONFIG_TRACING
1586 /* state flags for use by tracers */
1587 unsigned long trace;
1588 /* bitmask and counter of trace recursion */
1589 unsigned long trace_recursion;
1590 #endif /* CONFIG_TRACING */
1591 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1592 struct memcg_batch_info {
1593 int do_batch; /* incremented when batch uncharge started */
1594 struct mem_cgroup *memcg; /* target memcg of uncharge */
1595 unsigned long nr_pages; /* uncharged usage */
1596 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1598 unsigned int memcg_kmem_skip_account;
1599 struct memcg_oom_info {
1600 struct mem_cgroup *memcg;
1603 unsigned int may_oom:1;
1606 #ifdef CONFIG_UPROBES
1607 struct uprobe_task *utask;
1609 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1610 unsigned int sequential_io;
1611 unsigned int sequential_io_avg;
1615 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1616 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1618 #define TNF_MIGRATED 0x01
1619 #define TNF_NO_GROUP 0x02
1620 #define TNF_SHARED 0x04
1621 #define TNF_FAULT_LOCAL 0x08
1623 #ifdef CONFIG_NUMA_BALANCING
1624 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1625 extern pid_t task_numa_group_id(struct task_struct *p);
1626 extern void set_numabalancing_state(bool enabled);
1627 extern void task_numa_free(struct task_struct *p);
1628 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1629 int src_nid, int dst_cpu);
1631 static inline void task_numa_fault(int last_node, int node, int pages,
1635 static inline pid_t task_numa_group_id(struct task_struct *p)
1639 static inline void set_numabalancing_state(bool enabled)
1642 static inline void task_numa_free(struct task_struct *p)
1645 static inline bool should_numa_migrate_memory(struct task_struct *p,
1646 struct page *page, int src_nid, int dst_cpu)
1652 static inline struct pid *task_pid(struct task_struct *task)
1654 return task->pids[PIDTYPE_PID].pid;
1657 static inline struct pid *task_tgid(struct task_struct *task)
1659 return task->group_leader->pids[PIDTYPE_PID].pid;
1663 * Without tasklist or rcu lock it is not safe to dereference
1664 * the result of task_pgrp/task_session even if task == current,
1665 * we can race with another thread doing sys_setsid/sys_setpgid.
1667 static inline struct pid *task_pgrp(struct task_struct *task)
1669 return task->group_leader->pids[PIDTYPE_PGID].pid;
1672 static inline struct pid *task_session(struct task_struct *task)
1674 return task->group_leader->pids[PIDTYPE_SID].pid;
1677 struct pid_namespace;
1680 * the helpers to get the task's different pids as they are seen
1681 * from various namespaces
1683 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1684 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1686 * task_xid_nr_ns() : id seen from the ns specified;
1688 * set_task_vxid() : assigns a virtual id to a task;
1690 * see also pid_nr() etc in include/linux/pid.h
1692 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1693 struct pid_namespace *ns);
1695 static inline pid_t task_pid_nr(struct task_struct *tsk)
1700 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1701 struct pid_namespace *ns)
1703 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1706 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1708 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1712 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1717 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1719 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1721 return pid_vnr(task_tgid(tsk));
1725 static inline int pid_alive(const struct task_struct *p);
1726 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1732 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1738 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1740 return task_ppid_nr_ns(tsk, &init_pid_ns);
1743 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1744 struct pid_namespace *ns)
1746 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1749 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1751 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1755 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1756 struct pid_namespace *ns)
1758 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1761 static inline pid_t task_session_vnr(struct task_struct *tsk)
1763 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1766 /* obsolete, do not use */
1767 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1769 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1773 * pid_alive - check that a task structure is not stale
1774 * @p: Task structure to be checked.
1776 * Test if a process is not yet dead (at most zombie state)
1777 * If pid_alive fails, then pointers within the task structure
1778 * can be stale and must not be dereferenced.
1780 * Return: 1 if the process is alive. 0 otherwise.
1782 static inline int pid_alive(const struct task_struct *p)
1784 return p->pids[PIDTYPE_PID].pid != NULL;
1788 * is_global_init - check if a task structure is init
1789 * @tsk: Task structure to be checked.
1791 * Check if a task structure is the first user space task the kernel created.
1793 * Return: 1 if the task structure is init. 0 otherwise.
1795 static inline int is_global_init(struct task_struct *tsk)
1797 return tsk->pid == 1;
1800 extern struct pid *cad_pid;
1802 extern void free_task(struct task_struct *tsk);
1803 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1805 extern void __put_task_struct(struct task_struct *t);
1807 static inline void put_task_struct(struct task_struct *t)
1809 if (atomic_dec_and_test(&t->usage))
1810 __put_task_struct(t);
1813 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1814 extern void task_cputime(struct task_struct *t,
1815 cputime_t *utime, cputime_t *stime);
1816 extern void task_cputime_scaled(struct task_struct *t,
1817 cputime_t *utimescaled, cputime_t *stimescaled);
1818 extern cputime_t task_gtime(struct task_struct *t);
1820 static inline void task_cputime(struct task_struct *t,
1821 cputime_t *utime, cputime_t *stime)
1829 static inline void task_cputime_scaled(struct task_struct *t,
1830 cputime_t *utimescaled,
1831 cputime_t *stimescaled)
1834 *utimescaled = t->utimescaled;
1836 *stimescaled = t->stimescaled;
1839 static inline cputime_t task_gtime(struct task_struct *t)
1844 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1845 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1850 #define PF_EXITING 0x00000004 /* getting shut down */
1851 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1852 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1853 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1854 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1855 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1856 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1857 #define PF_DUMPCORE 0x00000200 /* dumped core */
1858 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1859 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1860 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1861 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1862 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1863 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1864 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1865 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1866 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1867 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1868 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1869 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1870 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1871 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1872 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1873 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1874 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1875 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1876 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1877 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1878 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1881 * Only the _current_ task can read/write to tsk->flags, but other
1882 * tasks can access tsk->flags in readonly mode for example
1883 * with tsk_used_math (like during threaded core dumping).
1884 * There is however an exception to this rule during ptrace
1885 * or during fork: the ptracer task is allowed to write to the
1886 * child->flags of its traced child (same goes for fork, the parent
1887 * can write to the child->flags), because we're guaranteed the
1888 * child is not running and in turn not changing child->flags
1889 * at the same time the parent does it.
1891 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1892 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1893 #define clear_used_math() clear_stopped_child_used_math(current)
1894 #define set_used_math() set_stopped_child_used_math(current)
1895 #define conditional_stopped_child_used_math(condition, child) \
1896 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1897 #define conditional_used_math(condition) \
1898 conditional_stopped_child_used_math(condition, current)
1899 #define copy_to_stopped_child_used_math(child) \
1900 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1901 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1902 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1903 #define used_math() tsk_used_math(current)
1905 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1906 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1908 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1913 static inline unsigned int memalloc_noio_save(void)
1915 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1916 current->flags |= PF_MEMALLOC_NOIO;
1920 static inline void memalloc_noio_restore(unsigned int flags)
1922 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1926 * task->jobctl flags
1928 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1930 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1931 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1932 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1933 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1934 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1935 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1936 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1938 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1939 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1940 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1941 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1942 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1943 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1944 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1946 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1947 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1949 extern bool task_set_jobctl_pending(struct task_struct *task,
1951 extern void task_clear_jobctl_trapping(struct task_struct *task);
1952 extern void task_clear_jobctl_pending(struct task_struct *task,
1955 #ifdef CONFIG_PREEMPT_RCU
1957 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1958 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1960 static inline void rcu_copy_process(struct task_struct *p)
1962 p->rcu_read_lock_nesting = 0;
1963 p->rcu_read_unlock_special = 0;
1964 #ifdef CONFIG_TREE_PREEMPT_RCU
1965 p->rcu_blocked_node = NULL;
1966 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1967 #ifdef CONFIG_RCU_BOOST
1968 p->rcu_boost_mutex = NULL;
1969 #endif /* #ifdef CONFIG_RCU_BOOST */
1970 INIT_LIST_HEAD(&p->rcu_node_entry);
1975 static inline void rcu_copy_process(struct task_struct *p)
1981 static inline void tsk_restore_flags(struct task_struct *task,
1982 unsigned long orig_flags, unsigned long flags)
1984 task->flags &= ~flags;
1985 task->flags |= orig_flags & flags;
1989 extern void do_set_cpus_allowed(struct task_struct *p,
1990 const struct cpumask *new_mask);
1992 extern int set_cpus_allowed_ptr(struct task_struct *p,
1993 const struct cpumask *new_mask);
1995 static inline void do_set_cpus_allowed(struct task_struct *p,
1996 const struct cpumask *new_mask)
1999 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2000 const struct cpumask *new_mask)
2002 if (!cpumask_test_cpu(0, new_mask))
2008 #ifdef CONFIG_NO_HZ_COMMON
2009 void calc_load_enter_idle(void);
2010 void calc_load_exit_idle(void);
2012 static inline void calc_load_enter_idle(void) { }
2013 static inline void calc_load_exit_idle(void) { }
2014 #endif /* CONFIG_NO_HZ_COMMON */
2016 #ifndef CONFIG_CPUMASK_OFFSTACK
2017 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2019 return set_cpus_allowed_ptr(p, &new_mask);
2024 * Do not use outside of architecture code which knows its limitations.
2026 * sched_clock() has no promise of monotonicity or bounded drift between
2027 * CPUs, use (which you should not) requires disabling IRQs.
2029 * Please use one of the three interfaces below.
2031 extern unsigned long long notrace sched_clock(void);
2033 * See the comment in kernel/sched/clock.c
2035 extern u64 cpu_clock(int cpu);
2036 extern u64 local_clock(void);
2037 extern u64 sched_clock_cpu(int cpu);
2040 extern void sched_clock_init(void);
2042 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2043 static inline void sched_clock_tick(void)
2047 static inline void sched_clock_idle_sleep_event(void)
2051 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2056 * Architectures can set this to 1 if they have specified
2057 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2058 * but then during bootup it turns out that sched_clock()
2059 * is reliable after all:
2061 extern int sched_clock_stable(void);
2062 extern void set_sched_clock_stable(void);
2063 extern void clear_sched_clock_stable(void);
2065 extern void sched_clock_tick(void);
2066 extern void sched_clock_idle_sleep_event(void);
2067 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2070 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2072 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2073 * The reason for this explicit opt-in is not to have perf penalty with
2074 * slow sched_clocks.
2076 extern void enable_sched_clock_irqtime(void);
2077 extern void disable_sched_clock_irqtime(void);
2079 static inline void enable_sched_clock_irqtime(void) {}
2080 static inline void disable_sched_clock_irqtime(void) {}
2083 extern unsigned long long
2084 task_sched_runtime(struct task_struct *task);
2086 /* sched_exec is called by processes performing an exec */
2088 extern void sched_exec(void);
2090 #define sched_exec() {}
2093 extern void sched_clock_idle_sleep_event(void);
2094 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2096 #ifdef CONFIG_HOTPLUG_CPU
2097 extern void idle_task_exit(void);
2099 static inline void idle_task_exit(void) {}
2102 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2103 extern void wake_up_nohz_cpu(int cpu);
2105 static inline void wake_up_nohz_cpu(int cpu) { }
2108 #ifdef CONFIG_NO_HZ_FULL
2109 extern bool sched_can_stop_tick(void);
2110 extern u64 scheduler_tick_max_deferment(void);
2112 static inline bool sched_can_stop_tick(void) { return false; }
2115 #ifdef CONFIG_SCHED_AUTOGROUP
2116 extern void sched_autogroup_create_attach(struct task_struct *p);
2117 extern void sched_autogroup_detach(struct task_struct *p);
2118 extern void sched_autogroup_fork(struct signal_struct *sig);
2119 extern void sched_autogroup_exit(struct signal_struct *sig);
2120 #ifdef CONFIG_PROC_FS
2121 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2122 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2125 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2126 static inline void sched_autogroup_detach(struct task_struct *p) { }
2127 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2128 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2131 extern bool yield_to(struct task_struct *p, bool preempt);
2132 extern void set_user_nice(struct task_struct *p, long nice);
2133 extern int task_prio(const struct task_struct *p);
2135 * task_nice - return the nice value of a given task.
2136 * @p: the task in question.
2138 * Return: The nice value [ -20 ... 0 ... 19 ].
2140 static inline int task_nice(const struct task_struct *p)
2142 return PRIO_TO_NICE((p)->static_prio);
2144 extern int can_nice(const struct task_struct *p, const int nice);
2145 extern int task_curr(const struct task_struct *p);
2146 extern int idle_cpu(int cpu);
2147 extern int sched_setscheduler(struct task_struct *, int,
2148 const struct sched_param *);
2149 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2150 const struct sched_param *);
2151 extern int sched_setattr(struct task_struct *,
2152 const struct sched_attr *);
2153 extern struct task_struct *idle_task(int cpu);
2155 * is_idle_task - is the specified task an idle task?
2156 * @p: the task in question.
2158 * Return: 1 if @p is an idle task. 0 otherwise.
2160 static inline bool is_idle_task(const struct task_struct *p)
2164 extern struct task_struct *curr_task(int cpu);
2165 extern void set_curr_task(int cpu, struct task_struct *p);
2170 * The default (Linux) execution domain.
2172 extern struct exec_domain default_exec_domain;
2174 union thread_union {
2175 struct thread_info thread_info;
2176 unsigned long stack[THREAD_SIZE/sizeof(long)];
2179 #ifndef __HAVE_ARCH_KSTACK_END
2180 static inline int kstack_end(void *addr)
2182 /* Reliable end of stack detection:
2183 * Some APM bios versions misalign the stack
2185 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2189 extern union thread_union init_thread_union;
2190 extern struct task_struct init_task;
2192 extern struct mm_struct init_mm;
2194 extern struct pid_namespace init_pid_ns;
2197 * find a task by one of its numerical ids
2199 * find_task_by_pid_ns():
2200 * finds a task by its pid in the specified namespace
2201 * find_task_by_vpid():
2202 * finds a task by its virtual pid
2204 * see also find_vpid() etc in include/linux/pid.h
2207 extern struct task_struct *find_task_by_vpid(pid_t nr);
2208 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2209 struct pid_namespace *ns);
2211 /* per-UID process charging. */
2212 extern struct user_struct * alloc_uid(kuid_t);
2213 static inline struct user_struct *get_uid(struct user_struct *u)
2215 atomic_inc(&u->__count);
2218 extern void free_uid(struct user_struct *);
2220 #include <asm/current.h>
2222 extern void xtime_update(unsigned long ticks);
2224 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2225 extern int wake_up_process(struct task_struct *tsk);
2226 extern void wake_up_new_task(struct task_struct *tsk);
2228 extern void kick_process(struct task_struct *tsk);
2230 static inline void kick_process(struct task_struct *tsk) { }
2232 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2233 extern void sched_dead(struct task_struct *p);
2235 extern void proc_caches_init(void);
2236 extern void flush_signals(struct task_struct *);
2237 extern void __flush_signals(struct task_struct *);
2238 extern void ignore_signals(struct task_struct *);
2239 extern void flush_signal_handlers(struct task_struct *, int force_default);
2240 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2242 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2244 unsigned long flags;
2247 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2248 ret = dequeue_signal(tsk, mask, info);
2249 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2254 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2256 extern void unblock_all_signals(void);
2257 extern void release_task(struct task_struct * p);
2258 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2259 extern int force_sigsegv(int, struct task_struct *);
2260 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2261 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2262 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2263 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2264 const struct cred *, u32);
2265 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2266 extern int kill_pid(struct pid *pid, int sig, int priv);
2267 extern int kill_proc_info(int, struct siginfo *, pid_t);
2268 extern __must_check bool do_notify_parent(struct task_struct *, int);
2269 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2270 extern void force_sig(int, struct task_struct *);
2271 extern int send_sig(int, struct task_struct *, int);
2272 extern int zap_other_threads(struct task_struct *p);
2273 extern struct sigqueue *sigqueue_alloc(void);
2274 extern void sigqueue_free(struct sigqueue *);
2275 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2276 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2278 static inline void restore_saved_sigmask(void)
2280 if (test_and_clear_restore_sigmask())
2281 __set_current_blocked(¤t->saved_sigmask);
2284 static inline sigset_t *sigmask_to_save(void)
2286 sigset_t *res = ¤t->blocked;
2287 if (unlikely(test_restore_sigmask()))
2288 res = ¤t->saved_sigmask;
2292 static inline int kill_cad_pid(int sig, int priv)
2294 return kill_pid(cad_pid, sig, priv);
2297 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2298 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2299 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2300 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2303 * True if we are on the alternate signal stack.
2305 static inline int on_sig_stack(unsigned long sp)
2307 #ifdef CONFIG_STACK_GROWSUP
2308 return sp >= current->sas_ss_sp &&
2309 sp - current->sas_ss_sp < current->sas_ss_size;
2311 return sp > current->sas_ss_sp &&
2312 sp - current->sas_ss_sp <= current->sas_ss_size;
2316 static inline int sas_ss_flags(unsigned long sp)
2318 return (current->sas_ss_size == 0 ? SS_DISABLE
2319 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2322 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2324 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2325 #ifdef CONFIG_STACK_GROWSUP
2326 return current->sas_ss_sp;
2328 return current->sas_ss_sp + current->sas_ss_size;
2334 * Routines for handling mm_structs
2336 extern struct mm_struct * mm_alloc(void);
2338 /* mmdrop drops the mm and the page tables */
2339 extern void __mmdrop(struct mm_struct *);
2340 static inline void mmdrop(struct mm_struct * mm)
2342 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2346 /* mmput gets rid of the mappings and all user-space */
2347 extern void mmput(struct mm_struct *);
2348 /* Grab a reference to a task's mm, if it is not already going away */
2349 extern struct mm_struct *get_task_mm(struct task_struct *task);
2351 * Grab a reference to a task's mm, if it is not already going away
2352 * and ptrace_may_access with the mode parameter passed to it
2355 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2356 /* Remove the current tasks stale references to the old mm_struct */
2357 extern void mm_release(struct task_struct *, struct mm_struct *);
2359 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2360 struct task_struct *);
2361 extern void flush_thread(void);
2362 extern void exit_thread(void);
2364 extern void exit_files(struct task_struct *);
2365 extern void __cleanup_sighand(struct sighand_struct *);
2367 extern void exit_itimers(struct signal_struct *);
2368 extern void flush_itimer_signals(void);
2370 extern void do_group_exit(int);
2372 extern int allow_signal(int);
2373 extern int disallow_signal(int);
2375 extern int do_execve(struct filename *,
2376 const char __user * const __user *,
2377 const char __user * const __user *);
2378 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2379 struct task_struct *fork_idle(int);
2380 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2382 extern void set_task_comm(struct task_struct *tsk, const char *from);
2383 extern char *get_task_comm(char *to, struct task_struct *tsk);
2386 void scheduler_ipi(void);
2387 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2389 static inline void scheduler_ipi(void) { }
2390 static inline unsigned long wait_task_inactive(struct task_struct *p,
2397 #define next_task(p) \
2398 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2400 #define for_each_process(p) \
2401 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2403 extern bool current_is_single_threaded(void);
2406 * Careful: do_each_thread/while_each_thread is a double loop so
2407 * 'break' will not work as expected - use goto instead.
2409 #define do_each_thread(g, t) \
2410 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2412 #define while_each_thread(g, t) \
2413 while ((t = next_thread(t)) != g)
2415 #define __for_each_thread(signal, t) \
2416 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2418 #define for_each_thread(p, t) \
2419 __for_each_thread((p)->signal, t)
2421 /* Careful: this is a double loop, 'break' won't work as expected. */
2422 #define for_each_process_thread(p, t) \
2423 for_each_process(p) for_each_thread(p, t)
2425 static inline int get_nr_threads(struct task_struct *tsk)
2427 return tsk->signal->nr_threads;
2430 static inline bool thread_group_leader(struct task_struct *p)
2432 return p->exit_signal >= 0;
2435 /* Do to the insanities of de_thread it is possible for a process
2436 * to have the pid of the thread group leader without actually being
2437 * the thread group leader. For iteration through the pids in proc
2438 * all we care about is that we have a task with the appropriate
2439 * pid, we don't actually care if we have the right task.
2441 static inline bool has_group_leader_pid(struct task_struct *p)
2443 return task_pid(p) == p->signal->leader_pid;
2447 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2449 return p1->signal == p2->signal;
2452 static inline struct task_struct *next_thread(const struct task_struct *p)
2454 return list_entry_rcu(p->thread_group.next,
2455 struct task_struct, thread_group);
2458 static inline int thread_group_empty(struct task_struct *p)
2460 return list_empty(&p->thread_group);
2463 #define delay_group_leader(p) \
2464 (thread_group_leader(p) && !thread_group_empty(p))
2467 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2468 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2469 * pins the final release of task.io_context. Also protects ->cpuset and
2470 * ->cgroup.subsys[]. And ->vfork_done.
2472 * Nests both inside and outside of read_lock(&tasklist_lock).
2473 * It must not be nested with write_lock_irq(&tasklist_lock),
2474 * neither inside nor outside.
2476 static inline void task_lock(struct task_struct *p)
2478 spin_lock(&p->alloc_lock);
2481 static inline void task_unlock(struct task_struct *p)
2483 spin_unlock(&p->alloc_lock);
2486 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2487 unsigned long *flags);
2489 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2490 unsigned long *flags)
2492 struct sighand_struct *ret;
2494 ret = __lock_task_sighand(tsk, flags);
2495 (void)__cond_lock(&tsk->sighand->siglock, ret);
2499 static inline void unlock_task_sighand(struct task_struct *tsk,
2500 unsigned long *flags)
2502 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2505 #ifdef CONFIG_CGROUPS
2506 static inline void threadgroup_change_begin(struct task_struct *tsk)
2508 down_read(&tsk->signal->group_rwsem);
2510 static inline void threadgroup_change_end(struct task_struct *tsk)
2512 up_read(&tsk->signal->group_rwsem);
2516 * threadgroup_lock - lock threadgroup
2517 * @tsk: member task of the threadgroup to lock
2519 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2520 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2521 * change ->group_leader/pid. This is useful for cases where the threadgroup
2522 * needs to stay stable across blockable operations.
2524 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2525 * synchronization. While held, no new task will be added to threadgroup
2526 * and no existing live task will have its PF_EXITING set.
2528 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2529 * sub-thread becomes a new leader.
2531 static inline void threadgroup_lock(struct task_struct *tsk)
2533 down_write(&tsk->signal->group_rwsem);
2537 * threadgroup_unlock - unlock threadgroup
2538 * @tsk: member task of the threadgroup to unlock
2540 * Reverse threadgroup_lock().
2542 static inline void threadgroup_unlock(struct task_struct *tsk)
2544 up_write(&tsk->signal->group_rwsem);
2547 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2548 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2549 static inline void threadgroup_lock(struct task_struct *tsk) {}
2550 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2553 #ifndef __HAVE_THREAD_FUNCTIONS
2555 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2556 #define task_stack_page(task) ((task)->stack)
2558 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2560 *task_thread_info(p) = *task_thread_info(org);
2561 task_thread_info(p)->task = p;
2564 static inline unsigned long *end_of_stack(struct task_struct *p)
2566 return (unsigned long *)(task_thread_info(p) + 1);
2571 static inline int object_is_on_stack(void *obj)
2573 void *stack = task_stack_page(current);
2575 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2578 extern void thread_info_cache_init(void);
2580 #ifdef CONFIG_DEBUG_STACK_USAGE
2581 static inline unsigned long stack_not_used(struct task_struct *p)
2583 unsigned long *n = end_of_stack(p);
2585 do { /* Skip over canary */
2589 return (unsigned long)n - (unsigned long)end_of_stack(p);
2593 /* set thread flags in other task's structures
2594 * - see asm/thread_info.h for TIF_xxxx flags available
2596 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2598 set_ti_thread_flag(task_thread_info(tsk), flag);
2601 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2603 clear_ti_thread_flag(task_thread_info(tsk), flag);
2606 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2608 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2611 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2613 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2616 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2618 return test_ti_thread_flag(task_thread_info(tsk), flag);
2621 static inline void set_tsk_need_resched(struct task_struct *tsk)
2623 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2626 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2628 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2631 static inline int test_tsk_need_resched(struct task_struct *tsk)
2633 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2636 static inline int restart_syscall(void)
2638 set_tsk_thread_flag(current, TIF_SIGPENDING);
2639 return -ERESTARTNOINTR;
2642 static inline int signal_pending(struct task_struct *p)
2644 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2647 static inline int __fatal_signal_pending(struct task_struct *p)
2649 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2652 static inline int fatal_signal_pending(struct task_struct *p)
2654 return signal_pending(p) && __fatal_signal_pending(p);
2657 static inline int signal_pending_state(long state, struct task_struct *p)
2659 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2661 if (!signal_pending(p))
2664 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2668 * cond_resched() and cond_resched_lock(): latency reduction via
2669 * explicit rescheduling in places that are safe. The return
2670 * value indicates whether a reschedule was done in fact.
2671 * cond_resched_lock() will drop the spinlock before scheduling,
2672 * cond_resched_softirq() will enable bhs before scheduling.
2674 extern int _cond_resched(void);
2676 #define cond_resched() ({ \
2677 __might_sleep(__FILE__, __LINE__, 0); \
2681 extern int __cond_resched_lock(spinlock_t *lock);
2683 #ifdef CONFIG_PREEMPT_COUNT
2684 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2686 #define PREEMPT_LOCK_OFFSET 0
2689 #define cond_resched_lock(lock) ({ \
2690 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2691 __cond_resched_lock(lock); \
2694 extern int __cond_resched_softirq(void);
2696 #define cond_resched_softirq() ({ \
2697 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2698 __cond_resched_softirq(); \
2701 static inline void cond_resched_rcu(void)
2703 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2711 * Does a critical section need to be broken due to another
2712 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2713 * but a general need for low latency)
2715 static inline int spin_needbreak(spinlock_t *lock)
2717 #ifdef CONFIG_PREEMPT
2718 return spin_is_contended(lock);
2725 * Idle thread specific functions to determine the need_resched
2726 * polling state. We have two versions, one based on TS_POLLING in
2727 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2731 static inline int tsk_is_polling(struct task_struct *p)
2733 return task_thread_info(p)->status & TS_POLLING;
2735 static inline void __current_set_polling(void)
2737 current_thread_info()->status |= TS_POLLING;
2740 static inline bool __must_check current_set_polling_and_test(void)
2742 __current_set_polling();
2745 * Polling state must be visible before we test NEED_RESCHED,
2746 * paired by resched_task()
2750 return unlikely(tif_need_resched());
2753 static inline void __current_clr_polling(void)
2755 current_thread_info()->status &= ~TS_POLLING;
2758 static inline bool __must_check current_clr_polling_and_test(void)
2760 __current_clr_polling();
2763 * Polling state must be visible before we test NEED_RESCHED,
2764 * paired by resched_task()
2768 return unlikely(tif_need_resched());
2770 #elif defined(TIF_POLLING_NRFLAG)
2771 static inline int tsk_is_polling(struct task_struct *p)
2773 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2776 static inline void __current_set_polling(void)
2778 set_thread_flag(TIF_POLLING_NRFLAG);
2781 static inline bool __must_check current_set_polling_and_test(void)
2783 __current_set_polling();
2786 * Polling state must be visible before we test NEED_RESCHED,
2787 * paired by resched_task()
2789 * XXX: assumes set/clear bit are identical barrier wise.
2791 smp_mb__after_clear_bit();
2793 return unlikely(tif_need_resched());
2796 static inline void __current_clr_polling(void)
2798 clear_thread_flag(TIF_POLLING_NRFLAG);
2801 static inline bool __must_check current_clr_polling_and_test(void)
2803 __current_clr_polling();
2806 * Polling state must be visible before we test NEED_RESCHED,
2807 * paired by resched_task()
2809 smp_mb__after_clear_bit();
2811 return unlikely(tif_need_resched());
2815 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2816 static inline void __current_set_polling(void) { }
2817 static inline void __current_clr_polling(void) { }
2819 static inline bool __must_check current_set_polling_and_test(void)
2821 return unlikely(tif_need_resched());
2823 static inline bool __must_check current_clr_polling_and_test(void)
2825 return unlikely(tif_need_resched());
2829 static inline void current_clr_polling(void)
2831 __current_clr_polling();
2834 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2835 * Once the bit is cleared, we'll get IPIs with every new
2836 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2839 smp_mb(); /* paired with resched_task() */
2841 preempt_fold_need_resched();
2844 static __always_inline bool need_resched(void)
2846 return unlikely(tif_need_resched());
2850 * Thread group CPU time accounting.
2852 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2853 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2855 static inline void thread_group_cputime_init(struct signal_struct *sig)
2857 raw_spin_lock_init(&sig->cputimer.lock);
2861 * Reevaluate whether the task has signals pending delivery.
2862 * Wake the task if so.
2863 * This is required every time the blocked sigset_t changes.
2864 * callers must hold sighand->siglock.
2866 extern void recalc_sigpending_and_wake(struct task_struct *t);
2867 extern void recalc_sigpending(void);
2869 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2871 static inline void signal_wake_up(struct task_struct *t, bool resume)
2873 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2875 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2877 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2881 * Wrappers for p->thread_info->cpu access. No-op on UP.
2885 static inline unsigned int task_cpu(const struct task_struct *p)
2887 return task_thread_info(p)->cpu;
2890 static inline int task_node(const struct task_struct *p)
2892 return cpu_to_node(task_cpu(p));
2895 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2899 static inline unsigned int task_cpu(const struct task_struct *p)
2904 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2908 #endif /* CONFIG_SMP */
2910 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2911 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2913 #ifdef CONFIG_CGROUP_SCHED
2914 extern struct task_group root_task_group;
2915 #endif /* CONFIG_CGROUP_SCHED */
2917 extern int task_can_switch_user(struct user_struct *up,
2918 struct task_struct *tsk);
2920 #ifdef CONFIG_TASK_XACCT
2921 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2923 tsk->ioac.rchar += amt;
2926 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2928 tsk->ioac.wchar += amt;
2931 static inline void inc_syscr(struct task_struct *tsk)
2936 static inline void inc_syscw(struct task_struct *tsk)
2941 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2945 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2949 static inline void inc_syscr(struct task_struct *tsk)
2953 static inline void inc_syscw(struct task_struct *tsk)
2958 #ifndef TASK_SIZE_OF
2959 #define TASK_SIZE_OF(tsk) TASK_SIZE
2962 #ifdef CONFIG_MM_OWNER
2963 extern void mm_update_next_owner(struct mm_struct *mm);
2964 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2966 static inline void mm_update_next_owner(struct mm_struct *mm)
2970 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2973 #endif /* CONFIG_MM_OWNER */
2975 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2978 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2981 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2984 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2987 static inline unsigned long rlimit(unsigned int limit)
2989 return task_rlimit(current, limit);
2992 static inline unsigned long rlimit_max(unsigned int limit)
2994 return task_rlimit_max(current, limit);