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 sigpending; /* How many pending signals does this user have? */
749 #ifdef CONFIG_INOTIFY_USER
750 atomic_t inotify_watches; /* How many inotify watches does this user have? */
751 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
753 #ifdef CONFIG_FANOTIFY
754 atomic_t fanotify_listeners;
757 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
759 #ifdef CONFIG_POSIX_MQUEUE
760 /* protected by mq_lock */
761 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
763 unsigned long locked_shm; /* How many pages of mlocked shm ? */
766 struct key *uid_keyring; /* UID specific keyring */
767 struct key *session_keyring; /* UID's default session keyring */
770 /* Hash table maintenance information */
771 struct hlist_node uidhash_node;
774 #ifdef CONFIG_PERF_EVENTS
775 atomic_long_t locked_vm;
779 extern int uids_sysfs_init(void);
781 extern struct user_struct *find_user(kuid_t);
783 extern struct user_struct root_user;
784 #define INIT_USER (&root_user)
787 struct backing_dev_info;
788 struct reclaim_state;
790 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
792 /* cumulative counters */
793 unsigned long pcount; /* # of times run on this cpu */
794 unsigned long long run_delay; /* time spent waiting on a runqueue */
797 unsigned long long last_arrival,/* when we last ran on a cpu */
798 last_queued; /* when we were last queued to run */
800 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
802 #ifdef CONFIG_TASK_DELAY_ACCT
803 struct task_delay_info {
805 unsigned int flags; /* Private per-task flags */
807 /* For each stat XXX, add following, aligned appropriately
809 * struct timespec XXX_start, XXX_end;
813 * Atomicity of updates to XXX_delay, XXX_count protected by
814 * single lock above (split into XXX_lock if contention is an issue).
818 * XXX_count is incremented on every XXX operation, the delay
819 * associated with the operation is added to XXX_delay.
820 * XXX_delay contains the accumulated delay time in nanoseconds.
822 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
823 u64 blkio_delay; /* wait for sync block io completion */
824 u64 swapin_delay; /* wait for swapin block io completion */
825 u32 blkio_count; /* total count of the number of sync block */
826 /* io operations performed */
827 u32 swapin_count; /* total count of the number of swapin block */
828 /* io operations performed */
830 struct timespec freepages_start, freepages_end;
831 u64 freepages_delay; /* wait for memory reclaim */
832 u32 freepages_count; /* total count of memory reclaim */
834 #endif /* CONFIG_TASK_DELAY_ACCT */
836 static inline int sched_info_on(void)
838 #ifdef CONFIG_SCHEDSTATS
840 #elif defined(CONFIG_TASK_DELAY_ACCT)
841 extern int delayacct_on;
856 * Increase resolution of cpu_power calculations
858 #define SCHED_POWER_SHIFT 10
859 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
862 * sched-domains (multiprocessor balancing) declarations:
865 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
866 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
867 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
868 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
869 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
870 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
871 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
872 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
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 #ifdef CONFIG_SCHED_SMT
881 static inline const int cpu_smt_flags(void)
883 return SD_SHARE_CPUPOWER | SD_SHARE_PKG_RESOURCES;
887 #ifdef CONFIG_SCHED_MC
888 static inline const int cpu_core_flags(void)
890 return SD_SHARE_PKG_RESOURCES;
895 static inline const int cpu_numa_flags(void)
901 struct sched_domain_attr {
902 int relax_domain_level;
905 #define SD_ATTR_INIT (struct sched_domain_attr) { \
906 .relax_domain_level = -1, \
909 extern int sched_domain_level_max;
913 struct sched_domain {
914 /* These fields must be setup */
915 struct sched_domain *parent; /* top domain must be null terminated */
916 struct sched_domain *child; /* bottom domain must be null terminated */
917 struct sched_group *groups; /* the balancing groups of the domain */
918 unsigned long min_interval; /* Minimum balance interval ms */
919 unsigned long max_interval; /* Maximum balance interval ms */
920 unsigned int busy_factor; /* less balancing by factor if busy */
921 unsigned int imbalance_pct; /* No balance until over watermark */
922 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
923 unsigned int busy_idx;
924 unsigned int idle_idx;
925 unsigned int newidle_idx;
926 unsigned int wake_idx;
927 unsigned int forkexec_idx;
928 unsigned int smt_gain;
930 int nohz_idle; /* NOHZ IDLE status */
931 int flags; /* See SD_* */
934 /* Runtime fields. */
935 unsigned long last_balance; /* init to jiffies. units in jiffies */
936 unsigned int balance_interval; /* initialise to 1. units in ms. */
937 unsigned int nr_balance_failed; /* initialise to 0 */
939 /* idle_balance() stats */
940 u64 max_newidle_lb_cost;
941 unsigned long next_decay_max_lb_cost;
943 #ifdef CONFIG_SCHEDSTATS
944 /* load_balance() stats */
945 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
947 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
948 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
949 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
950 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
951 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
952 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
954 /* Active load balancing */
955 unsigned int alb_count;
956 unsigned int alb_failed;
957 unsigned int alb_pushed;
959 /* SD_BALANCE_EXEC stats */
960 unsigned int sbe_count;
961 unsigned int sbe_balanced;
962 unsigned int sbe_pushed;
964 /* SD_BALANCE_FORK stats */
965 unsigned int sbf_count;
966 unsigned int sbf_balanced;
967 unsigned int sbf_pushed;
969 /* try_to_wake_up() stats */
970 unsigned int ttwu_wake_remote;
971 unsigned int ttwu_move_affine;
972 unsigned int ttwu_move_balance;
974 #ifdef CONFIG_SCHED_DEBUG
978 void *private; /* used during construction */
979 struct rcu_head rcu; /* used during destruction */
982 unsigned int span_weight;
984 * Span of all CPUs in this domain.
986 * NOTE: this field is variable length. (Allocated dynamically
987 * by attaching extra space to the end of the structure,
988 * depending on how many CPUs the kernel has booted up with)
990 unsigned long span[0];
993 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
995 return to_cpumask(sd->span);
998 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
999 struct sched_domain_attr *dattr_new);
1001 /* Allocate an array of sched domains, for partition_sched_domains(). */
1002 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1003 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1005 bool cpus_share_cache(int this_cpu, int that_cpu);
1007 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1008 typedef const int (*sched_domain_flags_f)(void);
1010 #define SDTL_OVERLAP 0x01
1013 struct sched_domain **__percpu sd;
1014 struct sched_group **__percpu sg;
1015 struct sched_group_power **__percpu sgp;
1018 struct sched_domain_topology_level {
1019 sched_domain_mask_f mask;
1020 sched_domain_flags_f sd_flags;
1023 struct sd_data data;
1024 #ifdef CONFIG_SCHED_DEBUG
1029 extern struct sched_domain_topology_level *sched_domain_topology;
1031 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1033 #ifdef CONFIG_SCHED_DEBUG
1034 # define SD_INIT_NAME(type) .name = #type
1036 # define SD_INIT_NAME(type)
1039 #else /* CONFIG_SMP */
1041 struct sched_domain_attr;
1044 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1045 struct sched_domain_attr *dattr_new)
1049 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1054 #endif /* !CONFIG_SMP */
1057 struct io_context; /* See blkdev.h */
1060 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1061 extern void prefetch_stack(struct task_struct *t);
1063 static inline void prefetch_stack(struct task_struct *t) { }
1066 struct audit_context; /* See audit.c */
1068 struct pipe_inode_info;
1069 struct uts_namespace;
1071 struct load_weight {
1072 unsigned long weight;
1078 * These sums represent an infinite geometric series and so are bound
1079 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1080 * choices of y < 1-2^(-32)*1024.
1082 u32 runnable_avg_sum, runnable_avg_period;
1083 u64 last_runnable_update;
1085 unsigned long load_avg_contrib;
1088 #ifdef CONFIG_SCHEDSTATS
1089 struct sched_statistics {
1099 s64 sum_sleep_runtime;
1106 u64 nr_migrations_cold;
1107 u64 nr_failed_migrations_affine;
1108 u64 nr_failed_migrations_running;
1109 u64 nr_failed_migrations_hot;
1110 u64 nr_forced_migrations;
1113 u64 nr_wakeups_sync;
1114 u64 nr_wakeups_migrate;
1115 u64 nr_wakeups_local;
1116 u64 nr_wakeups_remote;
1117 u64 nr_wakeups_affine;
1118 u64 nr_wakeups_affine_attempts;
1119 u64 nr_wakeups_passive;
1120 u64 nr_wakeups_idle;
1124 struct sched_entity {
1125 struct load_weight load; /* for load-balancing */
1126 struct rb_node run_node;
1127 struct list_head group_node;
1131 u64 sum_exec_runtime;
1133 u64 prev_sum_exec_runtime;
1137 #ifdef CONFIG_SCHEDSTATS
1138 struct sched_statistics statistics;
1141 #ifdef CONFIG_FAIR_GROUP_SCHED
1143 struct sched_entity *parent;
1144 /* rq on which this entity is (to be) queued: */
1145 struct cfs_rq *cfs_rq;
1146 /* rq "owned" by this entity/group: */
1147 struct cfs_rq *my_q;
1151 /* Per-entity load-tracking */
1152 struct sched_avg avg;
1156 struct sched_rt_entity {
1157 struct list_head run_list;
1158 unsigned long timeout;
1159 unsigned long watchdog_stamp;
1160 unsigned int time_slice;
1162 struct sched_rt_entity *back;
1163 #ifdef CONFIG_RT_GROUP_SCHED
1164 struct sched_rt_entity *parent;
1165 /* rq on which this entity is (to be) queued: */
1166 struct rt_rq *rt_rq;
1167 /* rq "owned" by this entity/group: */
1172 struct sched_dl_entity {
1173 struct rb_node rb_node;
1176 * Original scheduling parameters. Copied here from sched_attr
1177 * during sched_setattr(), they will remain the same until
1178 * the next sched_setattr().
1180 u64 dl_runtime; /* maximum runtime for each instance */
1181 u64 dl_deadline; /* relative deadline of each instance */
1182 u64 dl_period; /* separation of two instances (period) */
1183 u64 dl_bw; /* dl_runtime / dl_deadline */
1186 * Actual scheduling parameters. Initialized with the values above,
1187 * they are continously updated during task execution. Note that
1188 * the remaining runtime could be < 0 in case we are in overrun.
1190 s64 runtime; /* remaining runtime for this instance */
1191 u64 deadline; /* absolute deadline for this instance */
1192 unsigned int flags; /* specifying the scheduler behaviour */
1197 * @dl_throttled tells if we exhausted the runtime. If so, the
1198 * task has to wait for a replenishment to be performed at the
1199 * next firing of dl_timer.
1201 * @dl_new tells if a new instance arrived. If so we must
1202 * start executing it with full runtime and reset its absolute
1205 * @dl_boosted tells if we are boosted due to DI. If so we are
1206 * outside bandwidth enforcement mechanism (but only until we
1207 * exit the critical section);
1209 * @dl_yielded tells if task gave up the cpu before consuming
1210 * all its available runtime during the last job.
1212 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1215 * Bandwidth enforcement timer. Each -deadline task has its
1216 * own bandwidth to be enforced, thus we need one timer per task.
1218 struct hrtimer dl_timer;
1223 enum perf_event_task_context {
1224 perf_invalid_context = -1,
1225 perf_hw_context = 0,
1227 perf_nr_task_contexts,
1230 struct task_struct {
1231 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1234 unsigned int flags; /* per process flags, defined below */
1235 unsigned int ptrace;
1238 struct llist_node wake_entry;
1240 struct task_struct *last_wakee;
1241 unsigned long wakee_flips;
1242 unsigned long wakee_flip_decay_ts;
1248 int prio, static_prio, normal_prio;
1249 unsigned int rt_priority;
1250 const struct sched_class *sched_class;
1251 struct sched_entity se;
1252 struct sched_rt_entity rt;
1253 #ifdef CONFIG_CGROUP_SCHED
1254 struct task_group *sched_task_group;
1256 struct sched_dl_entity dl;
1258 #ifdef CONFIG_PREEMPT_NOTIFIERS
1259 /* list of struct preempt_notifier: */
1260 struct hlist_head preempt_notifiers;
1263 #ifdef CONFIG_BLK_DEV_IO_TRACE
1264 unsigned int btrace_seq;
1267 unsigned int policy;
1268 int nr_cpus_allowed;
1269 cpumask_t cpus_allowed;
1271 #ifdef CONFIG_PREEMPT_RCU
1272 int rcu_read_lock_nesting;
1273 char rcu_read_unlock_special;
1274 struct list_head rcu_node_entry;
1275 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1276 #ifdef CONFIG_TREE_PREEMPT_RCU
1277 struct rcu_node *rcu_blocked_node;
1278 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1279 #ifdef CONFIG_RCU_BOOST
1280 struct rt_mutex *rcu_boost_mutex;
1281 #endif /* #ifdef CONFIG_RCU_BOOST */
1283 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1284 struct sched_info sched_info;
1287 struct list_head tasks;
1289 struct plist_node pushable_tasks;
1290 struct rb_node pushable_dl_tasks;
1293 struct mm_struct *mm, *active_mm;
1294 #ifdef CONFIG_COMPAT_BRK
1295 unsigned brk_randomized:1;
1297 /* per-thread vma caching */
1298 u32 vmacache_seqnum;
1299 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1300 #if defined(SPLIT_RSS_COUNTING)
1301 struct task_rss_stat rss_stat;
1305 int exit_code, exit_signal;
1306 int pdeath_signal; /* The signal sent when the parent dies */
1307 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1309 /* Used for emulating ABI behavior of previous Linux versions */
1310 unsigned int personality;
1312 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1314 unsigned in_iowait:1;
1316 /* task may not gain privileges */
1317 unsigned no_new_privs:1;
1319 /* Revert to default priority/policy when forking */
1320 unsigned sched_reset_on_fork:1;
1321 unsigned sched_contributes_to_load:1;
1326 #ifdef CONFIG_CC_STACKPROTECTOR
1327 /* Canary value for the -fstack-protector gcc feature */
1328 unsigned long stack_canary;
1331 * pointers to (original) parent process, youngest child, younger sibling,
1332 * older sibling, respectively. (p->father can be replaced with
1333 * p->real_parent->pid)
1335 struct task_struct __rcu *real_parent; /* real parent process */
1336 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1338 * children/sibling forms the list of my natural children
1340 struct list_head children; /* list of my children */
1341 struct list_head sibling; /* linkage in my parent's children list */
1342 struct task_struct *group_leader; /* threadgroup leader */
1345 * ptraced is the list of tasks this task is using ptrace on.
1346 * This includes both natural children and PTRACE_ATTACH targets.
1347 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1349 struct list_head ptraced;
1350 struct list_head ptrace_entry;
1352 /* PID/PID hash table linkage. */
1353 struct pid_link pids[PIDTYPE_MAX];
1354 struct list_head thread_group;
1355 struct list_head thread_node;
1357 struct completion *vfork_done; /* for vfork() */
1358 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1359 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1361 cputime_t utime, stime, utimescaled, stimescaled;
1363 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1364 struct cputime prev_cputime;
1366 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1367 seqlock_t vtime_seqlock;
1368 unsigned long long vtime_snap;
1373 } vtime_snap_whence;
1375 unsigned long nvcsw, nivcsw; /* context switch counts */
1376 struct timespec start_time; /* monotonic time */
1377 struct timespec real_start_time; /* boot based time */
1378 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1379 unsigned long min_flt, maj_flt;
1381 struct task_cputime cputime_expires;
1382 struct list_head cpu_timers[3];
1384 /* process credentials */
1385 const struct cred __rcu *real_cred; /* objective and real subjective task
1386 * credentials (COW) */
1387 const struct cred __rcu *cred; /* effective (overridable) subjective task
1388 * credentials (COW) */
1389 char comm[TASK_COMM_LEN]; /* executable name excluding path
1390 - access with [gs]et_task_comm (which lock
1391 it with task_lock())
1392 - initialized normally by setup_new_exec */
1393 /* file system info */
1394 int link_count, total_link_count;
1395 #ifdef CONFIG_SYSVIPC
1397 struct sysv_sem sysvsem;
1399 #ifdef CONFIG_DETECT_HUNG_TASK
1400 /* hung task detection */
1401 unsigned long last_switch_count;
1403 /* CPU-specific state of this task */
1404 struct thread_struct thread;
1405 /* filesystem information */
1406 struct fs_struct *fs;
1407 /* open file information */
1408 struct files_struct *files;
1410 struct nsproxy *nsproxy;
1411 /* signal handlers */
1412 struct signal_struct *signal;
1413 struct sighand_struct *sighand;
1415 sigset_t blocked, real_blocked;
1416 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1417 struct sigpending pending;
1419 unsigned long sas_ss_sp;
1421 int (*notifier)(void *priv);
1422 void *notifier_data;
1423 sigset_t *notifier_mask;
1424 struct callback_head *task_works;
1426 struct audit_context *audit_context;
1427 #ifdef CONFIG_AUDITSYSCALL
1429 unsigned int sessionid;
1431 struct seccomp seccomp;
1433 /* Thread group tracking */
1436 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1438 spinlock_t alloc_lock;
1440 /* Protection of the PI data structures: */
1441 raw_spinlock_t pi_lock;
1443 #ifdef CONFIG_RT_MUTEXES
1444 /* PI waiters blocked on a rt_mutex held by this task */
1445 struct rb_root pi_waiters;
1446 struct rb_node *pi_waiters_leftmost;
1447 /* Deadlock detection and priority inheritance handling */
1448 struct rt_mutex_waiter *pi_blocked_on;
1449 /* Top pi_waiters task */
1450 struct task_struct *pi_top_task;
1453 #ifdef CONFIG_DEBUG_MUTEXES
1454 /* mutex deadlock detection */
1455 struct mutex_waiter *blocked_on;
1457 #ifdef CONFIG_TRACE_IRQFLAGS
1458 unsigned int irq_events;
1459 unsigned long hardirq_enable_ip;
1460 unsigned long hardirq_disable_ip;
1461 unsigned int hardirq_enable_event;
1462 unsigned int hardirq_disable_event;
1463 int hardirqs_enabled;
1464 int hardirq_context;
1465 unsigned long softirq_disable_ip;
1466 unsigned long softirq_enable_ip;
1467 unsigned int softirq_disable_event;
1468 unsigned int softirq_enable_event;
1469 int softirqs_enabled;
1470 int softirq_context;
1472 #ifdef CONFIG_LOCKDEP
1473 # define MAX_LOCK_DEPTH 48UL
1476 unsigned int lockdep_recursion;
1477 struct held_lock held_locks[MAX_LOCK_DEPTH];
1478 gfp_t lockdep_reclaim_gfp;
1481 /* journalling filesystem info */
1484 /* stacked block device info */
1485 struct bio_list *bio_list;
1488 /* stack plugging */
1489 struct blk_plug *plug;
1493 struct reclaim_state *reclaim_state;
1495 struct backing_dev_info *backing_dev_info;
1497 struct io_context *io_context;
1499 unsigned long ptrace_message;
1500 siginfo_t *last_siginfo; /* For ptrace use. */
1501 struct task_io_accounting ioac;
1502 #if defined(CONFIG_TASK_XACCT)
1503 u64 acct_rss_mem1; /* accumulated rss usage */
1504 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1505 cputime_t acct_timexpd; /* stime + utime since last update */
1507 #ifdef CONFIG_CPUSETS
1508 nodemask_t mems_allowed; /* Protected by alloc_lock */
1509 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1510 int cpuset_mem_spread_rotor;
1511 int cpuset_slab_spread_rotor;
1513 #ifdef CONFIG_CGROUPS
1514 /* Control Group info protected by css_set_lock */
1515 struct css_set __rcu *cgroups;
1516 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1517 struct list_head cg_list;
1520 struct robust_list_head __user *robust_list;
1521 #ifdef CONFIG_COMPAT
1522 struct compat_robust_list_head __user *compat_robust_list;
1524 struct list_head pi_state_list;
1525 struct futex_pi_state *pi_state_cache;
1527 #ifdef CONFIG_PERF_EVENTS
1528 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1529 struct mutex perf_event_mutex;
1530 struct list_head perf_event_list;
1532 #ifdef CONFIG_DEBUG_PREEMPT
1533 unsigned long preempt_disable_ip;
1536 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1538 short pref_node_fork;
1540 #ifdef CONFIG_NUMA_BALANCING
1542 unsigned int numa_scan_period;
1543 unsigned int numa_scan_period_max;
1544 int numa_preferred_nid;
1545 unsigned long numa_migrate_retry;
1546 u64 node_stamp; /* migration stamp */
1547 u64 last_task_numa_placement;
1548 u64 last_sum_exec_runtime;
1549 struct callback_head numa_work;
1551 struct list_head numa_entry;
1552 struct numa_group *numa_group;
1555 * Exponential decaying average of faults on a per-node basis.
1556 * Scheduling placement decisions are made based on the these counts.
1557 * The values remain static for the duration of a PTE scan
1559 unsigned long *numa_faults_memory;
1560 unsigned long total_numa_faults;
1563 * numa_faults_buffer records faults per node during the current
1564 * scan window. When the scan completes, the counts in
1565 * numa_faults_memory decay and these values are copied.
1567 unsigned long *numa_faults_buffer_memory;
1570 * Track the nodes the process was running on when a NUMA hinting
1571 * fault was incurred.
1573 unsigned long *numa_faults_cpu;
1574 unsigned long *numa_faults_buffer_cpu;
1577 * numa_faults_locality tracks if faults recorded during the last
1578 * scan window were remote/local. The task scan period is adapted
1579 * based on the locality of the faults with different weights
1580 * depending on whether they were shared or private faults
1582 unsigned long numa_faults_locality[2];
1584 unsigned long numa_pages_migrated;
1585 #endif /* CONFIG_NUMA_BALANCING */
1587 struct rcu_head rcu;
1590 * cache last used pipe for splice
1592 struct pipe_inode_info *splice_pipe;
1594 struct page_frag task_frag;
1596 #ifdef CONFIG_TASK_DELAY_ACCT
1597 struct task_delay_info *delays;
1599 #ifdef CONFIG_FAULT_INJECTION
1603 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1604 * balance_dirty_pages() for some dirty throttling pause
1607 int nr_dirtied_pause;
1608 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1610 #ifdef CONFIG_LATENCYTOP
1611 int latency_record_count;
1612 struct latency_record latency_record[LT_SAVECOUNT];
1615 * time slack values; these are used to round up poll() and
1616 * select() etc timeout values. These are in nanoseconds.
1618 unsigned long timer_slack_ns;
1619 unsigned long default_timer_slack_ns;
1621 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1622 /* Index of current stored address in ret_stack */
1624 /* Stack of return addresses for return function tracing */
1625 struct ftrace_ret_stack *ret_stack;
1626 /* time stamp for last schedule */
1627 unsigned long long ftrace_timestamp;
1629 * Number of functions that haven't been traced
1630 * because of depth overrun.
1632 atomic_t trace_overrun;
1633 /* Pause for the tracing */
1634 atomic_t tracing_graph_pause;
1636 #ifdef CONFIG_TRACING
1637 /* state flags for use by tracers */
1638 unsigned long trace;
1639 /* bitmask and counter of trace recursion */
1640 unsigned long trace_recursion;
1641 #endif /* CONFIG_TRACING */
1642 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1643 struct memcg_batch_info {
1644 int do_batch; /* incremented when batch uncharge started */
1645 struct mem_cgroup *memcg; /* target memcg of uncharge */
1646 unsigned long nr_pages; /* uncharged usage */
1647 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1649 unsigned int memcg_kmem_skip_account;
1650 struct memcg_oom_info {
1651 struct mem_cgroup *memcg;
1654 unsigned int may_oom:1;
1657 #ifdef CONFIG_UPROBES
1658 struct uprobe_task *utask;
1660 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1661 unsigned int sequential_io;
1662 unsigned int sequential_io_avg;
1666 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1667 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1669 #define TNF_MIGRATED 0x01
1670 #define TNF_NO_GROUP 0x02
1671 #define TNF_SHARED 0x04
1672 #define TNF_FAULT_LOCAL 0x08
1674 #ifdef CONFIG_NUMA_BALANCING
1675 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1676 extern pid_t task_numa_group_id(struct task_struct *p);
1677 extern void set_numabalancing_state(bool enabled);
1678 extern void task_numa_free(struct task_struct *p);
1679 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1680 int src_nid, int dst_cpu);
1682 static inline void task_numa_fault(int last_node, int node, int pages,
1686 static inline pid_t task_numa_group_id(struct task_struct *p)
1690 static inline void set_numabalancing_state(bool enabled)
1693 static inline void task_numa_free(struct task_struct *p)
1696 static inline bool should_numa_migrate_memory(struct task_struct *p,
1697 struct page *page, int src_nid, int dst_cpu)
1703 static inline struct pid *task_pid(struct task_struct *task)
1705 return task->pids[PIDTYPE_PID].pid;
1708 static inline struct pid *task_tgid(struct task_struct *task)
1710 return task->group_leader->pids[PIDTYPE_PID].pid;
1714 * Without tasklist or rcu lock it is not safe to dereference
1715 * the result of task_pgrp/task_session even if task == current,
1716 * we can race with another thread doing sys_setsid/sys_setpgid.
1718 static inline struct pid *task_pgrp(struct task_struct *task)
1720 return task->group_leader->pids[PIDTYPE_PGID].pid;
1723 static inline struct pid *task_session(struct task_struct *task)
1725 return task->group_leader->pids[PIDTYPE_SID].pid;
1728 struct pid_namespace;
1731 * the helpers to get the task's different pids as they are seen
1732 * from various namespaces
1734 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1735 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1737 * task_xid_nr_ns() : id seen from the ns specified;
1739 * set_task_vxid() : assigns a virtual id to a task;
1741 * see also pid_nr() etc in include/linux/pid.h
1743 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1744 struct pid_namespace *ns);
1746 static inline pid_t task_pid_nr(struct task_struct *tsk)
1751 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1752 struct pid_namespace *ns)
1754 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1757 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1759 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1763 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1768 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1770 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1772 return pid_vnr(task_tgid(tsk));
1776 static inline int pid_alive(const struct task_struct *p);
1777 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1783 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1789 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1791 return task_ppid_nr_ns(tsk, &init_pid_ns);
1794 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1795 struct pid_namespace *ns)
1797 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1800 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1802 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1806 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1807 struct pid_namespace *ns)
1809 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1812 static inline pid_t task_session_vnr(struct task_struct *tsk)
1814 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1817 /* obsolete, do not use */
1818 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1820 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1824 * pid_alive - check that a task structure is not stale
1825 * @p: Task structure to be checked.
1827 * Test if a process is not yet dead (at most zombie state)
1828 * If pid_alive fails, then pointers within the task structure
1829 * can be stale and must not be dereferenced.
1831 * Return: 1 if the process is alive. 0 otherwise.
1833 static inline int pid_alive(const struct task_struct *p)
1835 return p->pids[PIDTYPE_PID].pid != NULL;
1839 * is_global_init - check if a task structure is init
1840 * @tsk: Task structure to be checked.
1842 * Check if a task structure is the first user space task the kernel created.
1844 * Return: 1 if the task structure is init. 0 otherwise.
1846 static inline int is_global_init(struct task_struct *tsk)
1848 return tsk->pid == 1;
1851 extern struct pid *cad_pid;
1853 extern void free_task(struct task_struct *tsk);
1854 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1856 extern void __put_task_struct(struct task_struct *t);
1858 static inline void put_task_struct(struct task_struct *t)
1860 if (atomic_dec_and_test(&t->usage))
1861 __put_task_struct(t);
1864 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1865 extern void task_cputime(struct task_struct *t,
1866 cputime_t *utime, cputime_t *stime);
1867 extern void task_cputime_scaled(struct task_struct *t,
1868 cputime_t *utimescaled, cputime_t *stimescaled);
1869 extern cputime_t task_gtime(struct task_struct *t);
1871 static inline void task_cputime(struct task_struct *t,
1872 cputime_t *utime, cputime_t *stime)
1880 static inline void task_cputime_scaled(struct task_struct *t,
1881 cputime_t *utimescaled,
1882 cputime_t *stimescaled)
1885 *utimescaled = t->utimescaled;
1887 *stimescaled = t->stimescaled;
1890 static inline cputime_t task_gtime(struct task_struct *t)
1895 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1896 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1901 #define PF_EXITING 0x00000004 /* getting shut down */
1902 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1903 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1904 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1905 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1906 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1907 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1908 #define PF_DUMPCORE 0x00000200 /* dumped core */
1909 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1910 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1911 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1912 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1913 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1914 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1915 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1916 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1917 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1918 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1919 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1920 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1921 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1922 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1923 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1924 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1925 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1926 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1927 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1928 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1929 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1932 * Only the _current_ task can read/write to tsk->flags, but other
1933 * tasks can access tsk->flags in readonly mode for example
1934 * with tsk_used_math (like during threaded core dumping).
1935 * There is however an exception to this rule during ptrace
1936 * or during fork: the ptracer task is allowed to write to the
1937 * child->flags of its traced child (same goes for fork, the parent
1938 * can write to the child->flags), because we're guaranteed the
1939 * child is not running and in turn not changing child->flags
1940 * at the same time the parent does it.
1942 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1943 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1944 #define clear_used_math() clear_stopped_child_used_math(current)
1945 #define set_used_math() set_stopped_child_used_math(current)
1946 #define conditional_stopped_child_used_math(condition, child) \
1947 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1948 #define conditional_used_math(condition) \
1949 conditional_stopped_child_used_math(condition, current)
1950 #define copy_to_stopped_child_used_math(child) \
1951 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1952 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1953 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1954 #define used_math() tsk_used_math(current)
1956 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1957 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1959 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1964 static inline unsigned int memalloc_noio_save(void)
1966 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1967 current->flags |= PF_MEMALLOC_NOIO;
1971 static inline void memalloc_noio_restore(unsigned int flags)
1973 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1977 * task->jobctl flags
1979 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1981 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1982 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1983 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1984 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1985 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1986 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1987 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1989 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1990 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1991 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1992 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1993 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1994 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1995 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1997 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1998 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2000 extern bool task_set_jobctl_pending(struct task_struct *task,
2002 extern void task_clear_jobctl_trapping(struct task_struct *task);
2003 extern void task_clear_jobctl_pending(struct task_struct *task,
2006 #ifdef CONFIG_PREEMPT_RCU
2008 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
2009 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
2011 static inline void rcu_copy_process(struct task_struct *p)
2013 p->rcu_read_lock_nesting = 0;
2014 p->rcu_read_unlock_special = 0;
2015 #ifdef CONFIG_TREE_PREEMPT_RCU
2016 p->rcu_blocked_node = NULL;
2017 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
2018 #ifdef CONFIG_RCU_BOOST
2019 p->rcu_boost_mutex = NULL;
2020 #endif /* #ifdef CONFIG_RCU_BOOST */
2021 INIT_LIST_HEAD(&p->rcu_node_entry);
2026 static inline void rcu_copy_process(struct task_struct *p)
2032 static inline void tsk_restore_flags(struct task_struct *task,
2033 unsigned long orig_flags, unsigned long flags)
2035 task->flags &= ~flags;
2036 task->flags |= orig_flags & flags;
2040 extern void do_set_cpus_allowed(struct task_struct *p,
2041 const struct cpumask *new_mask);
2043 extern int set_cpus_allowed_ptr(struct task_struct *p,
2044 const struct cpumask *new_mask);
2046 static inline void do_set_cpus_allowed(struct task_struct *p,
2047 const struct cpumask *new_mask)
2050 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2051 const struct cpumask *new_mask)
2053 if (!cpumask_test_cpu(0, new_mask))
2059 #ifdef CONFIG_NO_HZ_COMMON
2060 void calc_load_enter_idle(void);
2061 void calc_load_exit_idle(void);
2063 static inline void calc_load_enter_idle(void) { }
2064 static inline void calc_load_exit_idle(void) { }
2065 #endif /* CONFIG_NO_HZ_COMMON */
2067 #ifndef CONFIG_CPUMASK_OFFSTACK
2068 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2070 return set_cpus_allowed_ptr(p, &new_mask);
2075 * Do not use outside of architecture code which knows its limitations.
2077 * sched_clock() has no promise of monotonicity or bounded drift between
2078 * CPUs, use (which you should not) requires disabling IRQs.
2080 * Please use one of the three interfaces below.
2082 extern unsigned long long notrace sched_clock(void);
2084 * See the comment in kernel/sched/clock.c
2086 extern u64 cpu_clock(int cpu);
2087 extern u64 local_clock(void);
2088 extern u64 sched_clock_cpu(int cpu);
2091 extern void sched_clock_init(void);
2093 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2094 static inline void sched_clock_tick(void)
2098 static inline void sched_clock_idle_sleep_event(void)
2102 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2107 * Architectures can set this to 1 if they have specified
2108 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2109 * but then during bootup it turns out that sched_clock()
2110 * is reliable after all:
2112 extern int sched_clock_stable(void);
2113 extern void set_sched_clock_stable(void);
2114 extern void clear_sched_clock_stable(void);
2116 extern void sched_clock_tick(void);
2117 extern void sched_clock_idle_sleep_event(void);
2118 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2121 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2123 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2124 * The reason for this explicit opt-in is not to have perf penalty with
2125 * slow sched_clocks.
2127 extern void enable_sched_clock_irqtime(void);
2128 extern void disable_sched_clock_irqtime(void);
2130 static inline void enable_sched_clock_irqtime(void) {}
2131 static inline void disable_sched_clock_irqtime(void) {}
2134 extern unsigned long long
2135 task_sched_runtime(struct task_struct *task);
2137 /* sched_exec is called by processes performing an exec */
2139 extern void sched_exec(void);
2141 #define sched_exec() {}
2144 extern void sched_clock_idle_sleep_event(void);
2145 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2147 #ifdef CONFIG_HOTPLUG_CPU
2148 extern void idle_task_exit(void);
2150 static inline void idle_task_exit(void) {}
2153 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2154 extern void wake_up_nohz_cpu(int cpu);
2156 static inline void wake_up_nohz_cpu(int cpu) { }
2159 #ifdef CONFIG_NO_HZ_FULL
2160 extern bool sched_can_stop_tick(void);
2161 extern u64 scheduler_tick_max_deferment(void);
2163 static inline bool sched_can_stop_tick(void) { return false; }
2166 #ifdef CONFIG_SCHED_AUTOGROUP
2167 extern void sched_autogroup_create_attach(struct task_struct *p);
2168 extern void sched_autogroup_detach(struct task_struct *p);
2169 extern void sched_autogroup_fork(struct signal_struct *sig);
2170 extern void sched_autogroup_exit(struct signal_struct *sig);
2171 #ifdef CONFIG_PROC_FS
2172 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2173 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2176 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2177 static inline void sched_autogroup_detach(struct task_struct *p) { }
2178 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2179 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2182 extern bool yield_to(struct task_struct *p, bool preempt);
2183 extern void set_user_nice(struct task_struct *p, long nice);
2184 extern int task_prio(const struct task_struct *p);
2186 * task_nice - return the nice value of a given task.
2187 * @p: the task in question.
2189 * Return: The nice value [ -20 ... 0 ... 19 ].
2191 static inline int task_nice(const struct task_struct *p)
2193 return PRIO_TO_NICE((p)->static_prio);
2195 extern int can_nice(const struct task_struct *p, const int nice);
2196 extern int task_curr(const struct task_struct *p);
2197 extern int idle_cpu(int cpu);
2198 extern int sched_setscheduler(struct task_struct *, int,
2199 const struct sched_param *);
2200 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2201 const struct sched_param *);
2202 extern int sched_setattr(struct task_struct *,
2203 const struct sched_attr *);
2204 extern struct task_struct *idle_task(int cpu);
2206 * is_idle_task - is the specified task an idle task?
2207 * @p: the task in question.
2209 * Return: 1 if @p is an idle task. 0 otherwise.
2211 static inline bool is_idle_task(const struct task_struct *p)
2215 extern struct task_struct *curr_task(int cpu);
2216 extern void set_curr_task(int cpu, struct task_struct *p);
2221 * The default (Linux) execution domain.
2223 extern struct exec_domain default_exec_domain;
2225 union thread_union {
2226 struct thread_info thread_info;
2227 unsigned long stack[THREAD_SIZE/sizeof(long)];
2230 #ifndef __HAVE_ARCH_KSTACK_END
2231 static inline int kstack_end(void *addr)
2233 /* Reliable end of stack detection:
2234 * Some APM bios versions misalign the stack
2236 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2240 extern union thread_union init_thread_union;
2241 extern struct task_struct init_task;
2243 extern struct mm_struct init_mm;
2245 extern struct pid_namespace init_pid_ns;
2248 * find a task by one of its numerical ids
2250 * find_task_by_pid_ns():
2251 * finds a task by its pid in the specified namespace
2252 * find_task_by_vpid():
2253 * finds a task by its virtual pid
2255 * see also find_vpid() etc in include/linux/pid.h
2258 extern struct task_struct *find_task_by_vpid(pid_t nr);
2259 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2260 struct pid_namespace *ns);
2262 /* per-UID process charging. */
2263 extern struct user_struct * alloc_uid(kuid_t);
2264 static inline struct user_struct *get_uid(struct user_struct *u)
2266 atomic_inc(&u->__count);
2269 extern void free_uid(struct user_struct *);
2271 #include <asm/current.h>
2273 extern void xtime_update(unsigned long ticks);
2275 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2276 extern int wake_up_process(struct task_struct *tsk);
2277 extern void wake_up_new_task(struct task_struct *tsk);
2279 extern void kick_process(struct task_struct *tsk);
2281 static inline void kick_process(struct task_struct *tsk) { }
2283 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2284 extern void sched_dead(struct task_struct *p);
2286 extern void proc_caches_init(void);
2287 extern void flush_signals(struct task_struct *);
2288 extern void __flush_signals(struct task_struct *);
2289 extern void ignore_signals(struct task_struct *);
2290 extern void flush_signal_handlers(struct task_struct *, int force_default);
2291 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2293 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2295 unsigned long flags;
2298 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2299 ret = dequeue_signal(tsk, mask, info);
2300 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2305 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2307 extern void unblock_all_signals(void);
2308 extern void release_task(struct task_struct * p);
2309 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2310 extern int force_sigsegv(int, struct task_struct *);
2311 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2312 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2313 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2314 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2315 const struct cred *, u32);
2316 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2317 extern int kill_pid(struct pid *pid, int sig, int priv);
2318 extern int kill_proc_info(int, struct siginfo *, pid_t);
2319 extern __must_check bool do_notify_parent(struct task_struct *, int);
2320 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2321 extern void force_sig(int, struct task_struct *);
2322 extern int send_sig(int, struct task_struct *, int);
2323 extern int zap_other_threads(struct task_struct *p);
2324 extern struct sigqueue *sigqueue_alloc(void);
2325 extern void sigqueue_free(struct sigqueue *);
2326 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2327 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2329 static inline void restore_saved_sigmask(void)
2331 if (test_and_clear_restore_sigmask())
2332 __set_current_blocked(¤t->saved_sigmask);
2335 static inline sigset_t *sigmask_to_save(void)
2337 sigset_t *res = ¤t->blocked;
2338 if (unlikely(test_restore_sigmask()))
2339 res = ¤t->saved_sigmask;
2343 static inline int kill_cad_pid(int sig, int priv)
2345 return kill_pid(cad_pid, sig, priv);
2348 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2349 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2350 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2351 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2354 * True if we are on the alternate signal stack.
2356 static inline int on_sig_stack(unsigned long sp)
2358 #ifdef CONFIG_STACK_GROWSUP
2359 return sp >= current->sas_ss_sp &&
2360 sp - current->sas_ss_sp < current->sas_ss_size;
2362 return sp > current->sas_ss_sp &&
2363 sp - current->sas_ss_sp <= current->sas_ss_size;
2367 static inline int sas_ss_flags(unsigned long sp)
2369 return (current->sas_ss_size == 0 ? SS_DISABLE
2370 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2373 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2375 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2376 #ifdef CONFIG_STACK_GROWSUP
2377 return current->sas_ss_sp;
2379 return current->sas_ss_sp + current->sas_ss_size;
2385 * Routines for handling mm_structs
2387 extern struct mm_struct * mm_alloc(void);
2389 /* mmdrop drops the mm and the page tables */
2390 extern void __mmdrop(struct mm_struct *);
2391 static inline void mmdrop(struct mm_struct * mm)
2393 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2397 /* mmput gets rid of the mappings and all user-space */
2398 extern void mmput(struct mm_struct *);
2399 /* Grab a reference to a task's mm, if it is not already going away */
2400 extern struct mm_struct *get_task_mm(struct task_struct *task);
2402 * Grab a reference to a task's mm, if it is not already going away
2403 * and ptrace_may_access with the mode parameter passed to it
2406 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2407 /* Remove the current tasks stale references to the old mm_struct */
2408 extern void mm_release(struct task_struct *, struct mm_struct *);
2410 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2411 struct task_struct *);
2412 extern void flush_thread(void);
2413 extern void exit_thread(void);
2415 extern void exit_files(struct task_struct *);
2416 extern void __cleanup_sighand(struct sighand_struct *);
2418 extern void exit_itimers(struct signal_struct *);
2419 extern void flush_itimer_signals(void);
2421 extern void do_group_exit(int);
2423 extern int allow_signal(int);
2424 extern int disallow_signal(int);
2426 extern int do_execve(struct filename *,
2427 const char __user * const __user *,
2428 const char __user * const __user *);
2429 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2430 struct task_struct *fork_idle(int);
2431 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2433 extern void set_task_comm(struct task_struct *tsk, const char *from);
2434 extern char *get_task_comm(char *to, struct task_struct *tsk);
2437 void scheduler_ipi(void);
2438 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2440 static inline void scheduler_ipi(void) { }
2441 static inline unsigned long wait_task_inactive(struct task_struct *p,
2448 #define next_task(p) \
2449 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2451 #define for_each_process(p) \
2452 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2454 extern bool current_is_single_threaded(void);
2457 * Careful: do_each_thread/while_each_thread is a double loop so
2458 * 'break' will not work as expected - use goto instead.
2460 #define do_each_thread(g, t) \
2461 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2463 #define while_each_thread(g, t) \
2464 while ((t = next_thread(t)) != g)
2466 #define __for_each_thread(signal, t) \
2467 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2469 #define for_each_thread(p, t) \
2470 __for_each_thread((p)->signal, t)
2472 /* Careful: this is a double loop, 'break' won't work as expected. */
2473 #define for_each_process_thread(p, t) \
2474 for_each_process(p) for_each_thread(p, t)
2476 static inline int get_nr_threads(struct task_struct *tsk)
2478 return tsk->signal->nr_threads;
2481 static inline bool thread_group_leader(struct task_struct *p)
2483 return p->exit_signal >= 0;
2486 /* Do to the insanities of de_thread it is possible for a process
2487 * to have the pid of the thread group leader without actually being
2488 * the thread group leader. For iteration through the pids in proc
2489 * all we care about is that we have a task with the appropriate
2490 * pid, we don't actually care if we have the right task.
2492 static inline bool has_group_leader_pid(struct task_struct *p)
2494 return task_pid(p) == p->signal->leader_pid;
2498 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2500 return p1->signal == p2->signal;
2503 static inline struct task_struct *next_thread(const struct task_struct *p)
2505 return list_entry_rcu(p->thread_group.next,
2506 struct task_struct, thread_group);
2509 static inline int thread_group_empty(struct task_struct *p)
2511 return list_empty(&p->thread_group);
2514 #define delay_group_leader(p) \
2515 (thread_group_leader(p) && !thread_group_empty(p))
2518 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2519 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2520 * pins the final release of task.io_context. Also protects ->cpuset and
2521 * ->cgroup.subsys[]. And ->vfork_done.
2523 * Nests both inside and outside of read_lock(&tasklist_lock).
2524 * It must not be nested with write_lock_irq(&tasklist_lock),
2525 * neither inside nor outside.
2527 static inline void task_lock(struct task_struct *p)
2529 spin_lock(&p->alloc_lock);
2532 static inline void task_unlock(struct task_struct *p)
2534 spin_unlock(&p->alloc_lock);
2537 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2538 unsigned long *flags);
2540 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2541 unsigned long *flags)
2543 struct sighand_struct *ret;
2545 ret = __lock_task_sighand(tsk, flags);
2546 (void)__cond_lock(&tsk->sighand->siglock, ret);
2550 static inline void unlock_task_sighand(struct task_struct *tsk,
2551 unsigned long *flags)
2553 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2556 #ifdef CONFIG_CGROUPS
2557 static inline void threadgroup_change_begin(struct task_struct *tsk)
2559 down_read(&tsk->signal->group_rwsem);
2561 static inline void threadgroup_change_end(struct task_struct *tsk)
2563 up_read(&tsk->signal->group_rwsem);
2567 * threadgroup_lock - lock threadgroup
2568 * @tsk: member task of the threadgroup to lock
2570 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2571 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2572 * change ->group_leader/pid. This is useful for cases where the threadgroup
2573 * needs to stay stable across blockable operations.
2575 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2576 * synchronization. While held, no new task will be added to threadgroup
2577 * and no existing live task will have its PF_EXITING set.
2579 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2580 * sub-thread becomes a new leader.
2582 static inline void threadgroup_lock(struct task_struct *tsk)
2584 down_write(&tsk->signal->group_rwsem);
2588 * threadgroup_unlock - unlock threadgroup
2589 * @tsk: member task of the threadgroup to unlock
2591 * Reverse threadgroup_lock().
2593 static inline void threadgroup_unlock(struct task_struct *tsk)
2595 up_write(&tsk->signal->group_rwsem);
2598 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2599 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2600 static inline void threadgroup_lock(struct task_struct *tsk) {}
2601 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2604 #ifndef __HAVE_THREAD_FUNCTIONS
2606 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2607 #define task_stack_page(task) ((task)->stack)
2609 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2611 *task_thread_info(p) = *task_thread_info(org);
2612 task_thread_info(p)->task = p;
2615 static inline unsigned long *end_of_stack(struct task_struct *p)
2617 return (unsigned long *)(task_thread_info(p) + 1);
2622 static inline int object_is_on_stack(void *obj)
2624 void *stack = task_stack_page(current);
2626 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2629 extern void thread_info_cache_init(void);
2631 #ifdef CONFIG_DEBUG_STACK_USAGE
2632 static inline unsigned long stack_not_used(struct task_struct *p)
2634 unsigned long *n = end_of_stack(p);
2636 do { /* Skip over canary */
2640 return (unsigned long)n - (unsigned long)end_of_stack(p);
2644 /* set thread flags in other task's structures
2645 * - see asm/thread_info.h for TIF_xxxx flags available
2647 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2649 set_ti_thread_flag(task_thread_info(tsk), flag);
2652 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2654 clear_ti_thread_flag(task_thread_info(tsk), flag);
2657 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2659 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2662 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2664 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2667 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2669 return test_ti_thread_flag(task_thread_info(tsk), flag);
2672 static inline void set_tsk_need_resched(struct task_struct *tsk)
2674 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2677 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2679 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2682 static inline int test_tsk_need_resched(struct task_struct *tsk)
2684 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2687 static inline int restart_syscall(void)
2689 set_tsk_thread_flag(current, TIF_SIGPENDING);
2690 return -ERESTARTNOINTR;
2693 static inline int signal_pending(struct task_struct *p)
2695 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2698 static inline int __fatal_signal_pending(struct task_struct *p)
2700 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2703 static inline int fatal_signal_pending(struct task_struct *p)
2705 return signal_pending(p) && __fatal_signal_pending(p);
2708 static inline int signal_pending_state(long state, struct task_struct *p)
2710 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2712 if (!signal_pending(p))
2715 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2719 * cond_resched() and cond_resched_lock(): latency reduction via
2720 * explicit rescheduling in places that are safe. The return
2721 * value indicates whether a reschedule was done in fact.
2722 * cond_resched_lock() will drop the spinlock before scheduling,
2723 * cond_resched_softirq() will enable bhs before scheduling.
2725 extern int _cond_resched(void);
2727 #define cond_resched() ({ \
2728 __might_sleep(__FILE__, __LINE__, 0); \
2732 extern int __cond_resched_lock(spinlock_t *lock);
2734 #ifdef CONFIG_PREEMPT_COUNT
2735 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2737 #define PREEMPT_LOCK_OFFSET 0
2740 #define cond_resched_lock(lock) ({ \
2741 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2742 __cond_resched_lock(lock); \
2745 extern int __cond_resched_softirq(void);
2747 #define cond_resched_softirq() ({ \
2748 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2749 __cond_resched_softirq(); \
2752 static inline void cond_resched_rcu(void)
2754 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2762 * Does a critical section need to be broken due to another
2763 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2764 * but a general need for low latency)
2766 static inline int spin_needbreak(spinlock_t *lock)
2768 #ifdef CONFIG_PREEMPT
2769 return spin_is_contended(lock);
2776 * Idle thread specific functions to determine the need_resched
2779 #ifdef TIF_POLLING_NRFLAG
2780 static inline int tsk_is_polling(struct task_struct *p)
2782 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2785 static inline void __current_set_polling(void)
2787 set_thread_flag(TIF_POLLING_NRFLAG);
2790 static inline bool __must_check current_set_polling_and_test(void)
2792 __current_set_polling();
2795 * Polling state must be visible before we test NEED_RESCHED,
2796 * paired by resched_task()
2798 smp_mb__after_atomic();
2800 return unlikely(tif_need_resched());
2803 static inline void __current_clr_polling(void)
2805 clear_thread_flag(TIF_POLLING_NRFLAG);
2808 static inline bool __must_check current_clr_polling_and_test(void)
2810 __current_clr_polling();
2813 * Polling state must be visible before we test NEED_RESCHED,
2814 * paired by resched_task()
2816 smp_mb__after_atomic();
2818 return unlikely(tif_need_resched());
2822 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2823 static inline void __current_set_polling(void) { }
2824 static inline void __current_clr_polling(void) { }
2826 static inline bool __must_check current_set_polling_and_test(void)
2828 return unlikely(tif_need_resched());
2830 static inline bool __must_check current_clr_polling_and_test(void)
2832 return unlikely(tif_need_resched());
2836 static inline void current_clr_polling(void)
2838 __current_clr_polling();
2841 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2842 * Once the bit is cleared, we'll get IPIs with every new
2843 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2846 smp_mb(); /* paired with resched_task() */
2848 preempt_fold_need_resched();
2851 static __always_inline bool need_resched(void)
2853 return unlikely(tif_need_resched());
2857 * Thread group CPU time accounting.
2859 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2860 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2862 static inline void thread_group_cputime_init(struct signal_struct *sig)
2864 raw_spin_lock_init(&sig->cputimer.lock);
2868 * Reevaluate whether the task has signals pending delivery.
2869 * Wake the task if so.
2870 * This is required every time the blocked sigset_t changes.
2871 * callers must hold sighand->siglock.
2873 extern void recalc_sigpending_and_wake(struct task_struct *t);
2874 extern void recalc_sigpending(void);
2876 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2878 static inline void signal_wake_up(struct task_struct *t, bool resume)
2880 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2882 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2884 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2888 * Wrappers for p->thread_info->cpu access. No-op on UP.
2892 static inline unsigned int task_cpu(const struct task_struct *p)
2894 return task_thread_info(p)->cpu;
2897 static inline int task_node(const struct task_struct *p)
2899 return cpu_to_node(task_cpu(p));
2902 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2906 static inline unsigned int task_cpu(const struct task_struct *p)
2911 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2915 #endif /* CONFIG_SMP */
2917 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2918 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2920 #ifdef CONFIG_CGROUP_SCHED
2921 extern struct task_group root_task_group;
2922 #endif /* CONFIG_CGROUP_SCHED */
2924 extern int task_can_switch_user(struct user_struct *up,
2925 struct task_struct *tsk);
2927 #ifdef CONFIG_TASK_XACCT
2928 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2930 tsk->ioac.rchar += amt;
2933 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2935 tsk->ioac.wchar += amt;
2938 static inline void inc_syscr(struct task_struct *tsk)
2943 static inline void inc_syscw(struct task_struct *tsk)
2948 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2952 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2956 static inline void inc_syscr(struct task_struct *tsk)
2960 static inline void inc_syscw(struct task_struct *tsk)
2965 #ifndef TASK_SIZE_OF
2966 #define TASK_SIZE_OF(tsk) TASK_SIZE
2970 extern void mm_update_next_owner(struct mm_struct *mm);
2971 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2973 static inline void mm_update_next_owner(struct mm_struct *mm)
2977 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2980 #endif /* CONFIG_MEMCG */
2982 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2985 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2988 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2991 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2994 static inline unsigned long rlimit(unsigned int limit)
2996 return task_rlimit(current, limit);
2999 static inline unsigned long rlimit_max(unsigned int limit)
3001 return task_rlimit_max(current, limit);