4 #include <uapi/linux/sched.h>
11 #include <asm/param.h> /* for HZ */
13 #include <linux/capability.h>
14 #include <linux/threads.h>
15 #include <linux/kernel.h>
16 #include <linux/types.h>
17 #include <linux/timex.h>
18 #include <linux/jiffies.h>
19 #include <linux/plist.h>
20 #include <linux/rbtree.h>
21 #include <linux/thread_info.h>
22 #include <linux/cpumask.h>
23 #include <linux/errno.h>
24 #include <linux/nodemask.h>
25 #include <linux/mm_types.h>
26 #include <linux/preempt_mask.h>
29 #include <asm/ptrace.h>
30 #include <asm/cputime.h>
32 #include <linux/smp.h>
33 #include <linux/sem.h>
34 #include <linux/signal.h>
35 #include <linux/compiler.h>
36 #include <linux/completion.h>
37 #include <linux/pid.h>
38 #include <linux/percpu.h>
39 #include <linux/topology.h>
40 #include <linux/proportions.h>
41 #include <linux/seccomp.h>
42 #include <linux/rcupdate.h>
43 #include <linux/rculist.h>
44 #include <linux/rtmutex.h>
46 #include <linux/time.h>
47 #include <linux/param.h>
48 #include <linux/resource.h>
49 #include <linux/timer.h>
50 #include <linux/hrtimer.h>
51 #include <linux/task_io_accounting.h>
52 #include <linux/latencytop.h>
53 #include <linux/cred.h>
54 #include <linux/llist.h>
55 #include <linux/uidgid.h>
56 #include <linux/gfp.h>
58 #include <asm/processor.h>
60 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
63 * Extended scheduling parameters data structure.
65 * This is needed because the original struct sched_param can not be
66 * altered without introducing ABI issues with legacy applications
67 * (e.g., in sched_getparam()).
69 * However, the possibility of specifying more than just a priority for
70 * the tasks may be useful for a wide variety of application fields, e.g.,
71 * multimedia, streaming, automation and control, and many others.
73 * This variant (sched_attr) is meant at describing a so-called
74 * sporadic time-constrained task. In such model a task is specified by:
75 * - the activation period or minimum instance inter-arrival time;
76 * - the maximum (or average, depending on the actual scheduling
77 * discipline) computation time of all instances, a.k.a. runtime;
78 * - the deadline (relative to the actual activation time) of each
80 * Very briefly, a periodic (sporadic) task asks for the execution of
81 * some specific computation --which is typically called an instance--
82 * (at most) every period. Moreover, each instance typically lasts no more
83 * than the runtime and must be completed by time instant t equal to
84 * the instance activation time + the deadline.
86 * This is reflected by the actual fields of the sched_attr structure:
88 * @size size of the structure, for fwd/bwd compat.
90 * @sched_policy task's scheduling policy
91 * @sched_flags for customizing the scheduler behaviour
92 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
93 * @sched_priority task's static priority (SCHED_FIFO/RR)
94 * @sched_deadline representative of the task's deadline
95 * @sched_runtime representative of the task's runtime
96 * @sched_period representative of the task's period
98 * Given this task model, there are a multiplicity of scheduling algorithms
99 * and policies, that can be used to ensure all the tasks will make their
100 * timing constraints.
102 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
103 * only user of this new interface. More information about the algorithm
104 * available in the scheduling class file or in Documentation/.
112 /* SCHED_NORMAL, SCHED_BATCH */
115 /* SCHED_FIFO, SCHED_RR */
125 struct futex_pi_state;
126 struct robust_list_head;
129 struct perf_event_context;
133 * List of flags we want to share for kernel threads,
134 * if only because they are not used by them anyway.
136 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
139 * These are the constant used to fake the fixed-point load-average
140 * counting. Some notes:
141 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
142 * a load-average precision of 10 bits integer + 11 bits fractional
143 * - if you want to count load-averages more often, you need more
144 * precision, or rounding will get you. With 2-second counting freq,
145 * the EXP_n values would be 1981, 2034 and 2043 if still using only
148 extern unsigned long avenrun[]; /* Load averages */
149 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
151 #define FSHIFT 11 /* nr of bits of precision */
152 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
153 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
154 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
155 #define EXP_5 2014 /* 1/exp(5sec/5min) */
156 #define EXP_15 2037 /* 1/exp(5sec/15min) */
158 #define CALC_LOAD(load,exp,n) \
160 load += n*(FIXED_1-exp); \
163 extern unsigned long total_forks;
164 extern int nr_threads;
165 DECLARE_PER_CPU(unsigned long, process_counts);
166 extern int nr_processes(void);
167 extern unsigned long nr_running(void);
168 extern unsigned long nr_iowait(void);
169 extern unsigned long nr_iowait_cpu(int cpu);
170 extern unsigned long this_cpu_load(void);
173 extern void calc_global_load(unsigned long ticks);
174 extern void update_cpu_load_nohz(void);
176 extern unsigned long get_parent_ip(unsigned long addr);
178 extern void dump_cpu_task(int cpu);
183 #ifdef CONFIG_SCHED_DEBUG
184 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
185 extern void proc_sched_set_task(struct task_struct *p);
187 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
191 * Task state bitmask. NOTE! These bits are also
192 * encoded in fs/proc/array.c: get_task_state().
194 * We have two separate sets of flags: task->state
195 * is about runnability, while task->exit_state are
196 * about the task exiting. Confusing, but this way
197 * modifying one set can't modify the other one by
200 #define TASK_RUNNING 0
201 #define TASK_INTERRUPTIBLE 1
202 #define TASK_UNINTERRUPTIBLE 2
203 #define __TASK_STOPPED 4
204 #define __TASK_TRACED 8
205 /* in tsk->exit_state */
206 #define EXIT_ZOMBIE 16
208 /* in tsk->state again */
210 #define TASK_WAKEKILL 128
211 #define TASK_WAKING 256
212 #define TASK_PARKED 512
213 #define TASK_STATE_MAX 1024
215 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
217 extern char ___assert_task_state[1 - 2*!!(
218 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
220 /* Convenience macros for the sake of set_task_state */
221 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
222 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
223 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
225 /* Convenience macros for the sake of wake_up */
226 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
227 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
229 /* get_task_state() */
230 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
231 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
232 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
234 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
235 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
236 #define task_is_stopped_or_traced(task) \
237 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
238 #define task_contributes_to_load(task) \
239 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
240 (task->flags & PF_FROZEN) == 0)
242 #define __set_task_state(tsk, state_value) \
243 do { (tsk)->state = (state_value); } while (0)
244 #define set_task_state(tsk, state_value) \
245 set_mb((tsk)->state, (state_value))
248 * set_current_state() includes a barrier so that the write of current->state
249 * is correctly serialised wrt the caller's subsequent test of whether to
252 * set_current_state(TASK_UNINTERRUPTIBLE);
253 * if (do_i_need_to_sleep())
256 * If the caller does not need such serialisation then use __set_current_state()
258 #define __set_current_state(state_value) \
259 do { current->state = (state_value); } while (0)
260 #define set_current_state(state_value) \
261 set_mb(current->state, (state_value))
263 /* Task command name length */
264 #define TASK_COMM_LEN 16
266 #include <linux/spinlock.h>
269 * This serializes "schedule()" and also protects
270 * the run-queue from deletions/modifications (but
271 * _adding_ to the beginning of the run-queue has
274 extern rwlock_t tasklist_lock;
275 extern spinlock_t mmlist_lock;
279 #ifdef CONFIG_PROVE_RCU
280 extern int lockdep_tasklist_lock_is_held(void);
281 #endif /* #ifdef CONFIG_PROVE_RCU */
283 extern void sched_init(void);
284 extern void sched_init_smp(void);
285 extern asmlinkage void schedule_tail(struct task_struct *prev);
286 extern void init_idle(struct task_struct *idle, int cpu);
287 extern void init_idle_bootup_task(struct task_struct *idle);
289 extern int runqueue_is_locked(int cpu);
291 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
292 extern void nohz_balance_enter_idle(int cpu);
293 extern void set_cpu_sd_state_idle(void);
294 extern int get_nohz_timer_target(void);
296 static inline void nohz_balance_enter_idle(int cpu) { }
297 static inline void set_cpu_sd_state_idle(void) { }
301 * Only dump TASK_* tasks. (0 for all tasks)
303 extern void show_state_filter(unsigned long state_filter);
305 static inline void show_state(void)
307 show_state_filter(0);
310 extern void show_regs(struct pt_regs *);
313 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
314 * task), SP is the stack pointer of the first frame that should be shown in the back
315 * trace (or NULL if the entire call-chain of the task should be shown).
317 extern void show_stack(struct task_struct *task, unsigned long *sp);
319 void io_schedule(void);
320 long io_schedule_timeout(long timeout);
322 extern void cpu_init (void);
323 extern void trap_init(void);
324 extern void update_process_times(int user);
325 extern void scheduler_tick(void);
327 extern void sched_show_task(struct task_struct *p);
329 #ifdef CONFIG_LOCKUP_DETECTOR
330 extern void touch_softlockup_watchdog(void);
331 extern void touch_softlockup_watchdog_sync(void);
332 extern void touch_all_softlockup_watchdogs(void);
333 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
335 size_t *lenp, loff_t *ppos);
336 extern unsigned int softlockup_panic;
337 void lockup_detector_init(void);
339 static inline void touch_softlockup_watchdog(void)
342 static inline void touch_softlockup_watchdog_sync(void)
345 static inline void touch_all_softlockup_watchdogs(void)
348 static inline void lockup_detector_init(void)
353 #ifdef CONFIG_DETECT_HUNG_TASK
354 void reset_hung_task_detector(void);
356 static inline void reset_hung_task_detector(void)
361 /* Attach to any functions which should be ignored in wchan output. */
362 #define __sched __attribute__((__section__(".sched.text")))
364 /* Linker adds these: start and end of __sched functions */
365 extern char __sched_text_start[], __sched_text_end[];
367 /* Is this address in the __sched functions? */
368 extern int in_sched_functions(unsigned long addr);
370 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
371 extern signed long schedule_timeout(signed long timeout);
372 extern signed long schedule_timeout_interruptible(signed long timeout);
373 extern signed long schedule_timeout_killable(signed long timeout);
374 extern signed long schedule_timeout_uninterruptible(signed long timeout);
375 asmlinkage void schedule(void);
376 extern void schedule_preempt_disabled(void);
379 struct user_namespace;
382 extern void arch_pick_mmap_layout(struct mm_struct *mm);
384 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
385 unsigned long, unsigned long);
387 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
388 unsigned long len, unsigned long pgoff,
389 unsigned long flags);
391 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
394 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
395 #define SUID_DUMP_USER 1 /* Dump as user of process */
396 #define SUID_DUMP_ROOT 2 /* Dump as root */
400 /* for SUID_DUMP_* above */
401 #define MMF_DUMPABLE_BITS 2
402 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
404 extern void set_dumpable(struct mm_struct *mm, int value);
406 * This returns the actual value of the suid_dumpable flag. For things
407 * that are using this for checking for privilege transitions, it must
408 * test against SUID_DUMP_USER rather than treating it as a boolean
411 static inline int __get_dumpable(unsigned long mm_flags)
413 return mm_flags & MMF_DUMPABLE_MASK;
416 static inline int get_dumpable(struct mm_struct *mm)
418 return __get_dumpable(mm->flags);
421 /* coredump filter bits */
422 #define MMF_DUMP_ANON_PRIVATE 2
423 #define MMF_DUMP_ANON_SHARED 3
424 #define MMF_DUMP_MAPPED_PRIVATE 4
425 #define MMF_DUMP_MAPPED_SHARED 5
426 #define MMF_DUMP_ELF_HEADERS 6
427 #define MMF_DUMP_HUGETLB_PRIVATE 7
428 #define MMF_DUMP_HUGETLB_SHARED 8
430 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
431 #define MMF_DUMP_FILTER_BITS 7
432 #define MMF_DUMP_FILTER_MASK \
433 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
434 #define MMF_DUMP_FILTER_DEFAULT \
435 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
436 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
438 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
439 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
441 # define MMF_DUMP_MASK_DEFAULT_ELF 0
443 /* leave room for more dump flags */
444 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
445 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
446 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
448 #define MMF_HAS_UPROBES 19 /* has uprobes */
449 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
451 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
453 struct sighand_struct {
455 struct k_sigaction action[_NSIG];
457 wait_queue_head_t signalfd_wqh;
460 struct pacct_struct {
463 unsigned long ac_mem;
464 cputime_t ac_utime, ac_stime;
465 unsigned long ac_minflt, ac_majflt;
476 * struct cputime - snaphsot of system and user cputime
477 * @utime: time spent in user mode
478 * @stime: time spent in system mode
480 * Gathers a generic snapshot of user and system time.
488 * struct task_cputime - collected CPU time counts
489 * @utime: time spent in user mode, in &cputime_t units
490 * @stime: time spent in kernel mode, in &cputime_t units
491 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
493 * This is an extension of struct cputime that includes the total runtime
494 * spent by the task from the scheduler point of view.
496 * As a result, this structure groups together three kinds of CPU time
497 * that are tracked for threads and thread groups. Most things considering
498 * CPU time want to group these counts together and treat all three
499 * of them in parallel.
501 struct task_cputime {
504 unsigned long long sum_exec_runtime;
506 /* Alternate field names when used to cache expirations. */
507 #define prof_exp stime
508 #define virt_exp utime
509 #define sched_exp sum_exec_runtime
511 #define INIT_CPUTIME \
512 (struct task_cputime) { \
515 .sum_exec_runtime = 0, \
518 #ifdef CONFIG_PREEMPT_COUNT
519 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
521 #define PREEMPT_DISABLED PREEMPT_ENABLED
525 * Disable preemption until the scheduler is running.
526 * Reset by start_kernel()->sched_init()->init_idle().
528 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
529 * before the scheduler is active -- see should_resched().
531 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
534 * struct thread_group_cputimer - thread group interval timer counts
535 * @cputime: thread group interval timers.
536 * @running: non-zero when there are timers running and
537 * @cputime receives updates.
538 * @lock: lock for fields in this struct.
540 * This structure contains the version of task_cputime, above, that is
541 * used for thread group CPU timer calculations.
543 struct thread_group_cputimer {
544 struct task_cputime cputime;
549 #include <linux/rwsem.h>
553 * NOTE! "signal_struct" does not have its own
554 * locking, because a shared signal_struct always
555 * implies a shared sighand_struct, so locking
556 * sighand_struct is always a proper superset of
557 * the locking of signal_struct.
559 struct signal_struct {
563 struct list_head thread_head;
565 wait_queue_head_t wait_chldexit; /* for wait4() */
567 /* current thread group signal load-balancing target: */
568 struct task_struct *curr_target;
570 /* shared signal handling: */
571 struct sigpending shared_pending;
573 /* thread group exit support */
576 * - notify group_exit_task when ->count is equal to notify_count
577 * - everyone except group_exit_task is stopped during signal delivery
578 * of fatal signals, group_exit_task processes the signal.
581 struct task_struct *group_exit_task;
583 /* thread group stop support, overloads group_exit_code too */
584 int group_stop_count;
585 unsigned int flags; /* see SIGNAL_* flags below */
588 * PR_SET_CHILD_SUBREAPER marks a process, like a service
589 * manager, to re-parent orphan (double-forking) child processes
590 * to this process instead of 'init'. The service manager is
591 * able to receive SIGCHLD signals and is able to investigate
592 * the process until it calls wait(). All children of this
593 * process will inherit a flag if they should look for a
594 * child_subreaper process at exit.
596 unsigned int is_child_subreaper:1;
597 unsigned int has_child_subreaper:1;
599 /* POSIX.1b Interval Timers */
601 struct list_head posix_timers;
603 /* ITIMER_REAL timer for the process */
604 struct hrtimer real_timer;
605 struct pid *leader_pid;
606 ktime_t it_real_incr;
609 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
610 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
611 * values are defined to 0 and 1 respectively
613 struct cpu_itimer it[2];
616 * Thread group totals for process CPU timers.
617 * See thread_group_cputimer(), et al, for details.
619 struct thread_group_cputimer cputimer;
621 /* Earliest-expiration cache. */
622 struct task_cputime cputime_expires;
624 struct list_head cpu_timers[3];
626 struct pid *tty_old_pgrp;
628 /* boolean value for session group leader */
631 struct tty_struct *tty; /* NULL if no tty */
633 #ifdef CONFIG_SCHED_AUTOGROUP
634 struct autogroup *autogroup;
637 * Cumulative resource counters for dead threads in the group,
638 * and for reaped dead child processes forked by this group.
639 * Live threads maintain their own counters and add to these
640 * in __exit_signal, except for the group leader.
642 cputime_t utime, stime, cutime, cstime;
645 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
646 struct cputime prev_cputime;
648 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
649 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
650 unsigned long inblock, oublock, cinblock, coublock;
651 unsigned long maxrss, cmaxrss;
652 struct task_io_accounting ioac;
655 * Cumulative ns of schedule CPU time fo dead threads in the
656 * group, not including a zombie group leader, (This only differs
657 * from jiffies_to_ns(utime + stime) if sched_clock uses something
658 * other than jiffies.)
660 unsigned long long sum_sched_runtime;
663 * We don't bother to synchronize most readers of this at all,
664 * because there is no reader checking a limit that actually needs
665 * to get both rlim_cur and rlim_max atomically, and either one
666 * alone is a single word that can safely be read normally.
667 * getrlimit/setrlimit use task_lock(current->group_leader) to
668 * protect this instead of the siglock, because they really
669 * have no need to disable irqs.
671 struct rlimit rlim[RLIM_NLIMITS];
673 #ifdef CONFIG_BSD_PROCESS_ACCT
674 struct pacct_struct pacct; /* per-process accounting information */
676 #ifdef CONFIG_TASKSTATS
677 struct taskstats *stats;
681 unsigned audit_tty_log_passwd;
682 struct tty_audit_buf *tty_audit_buf;
684 #ifdef CONFIG_CGROUPS
686 * group_rwsem prevents new tasks from entering the threadgroup and
687 * member tasks from exiting,a more specifically, setting of
688 * PF_EXITING. fork and exit paths are protected with this rwsem
689 * using threadgroup_change_begin/end(). Users which require
690 * threadgroup to remain stable should use threadgroup_[un]lock()
691 * which also takes care of exec path. Currently, cgroup is the
694 struct rw_semaphore group_rwsem;
697 oom_flags_t oom_flags;
698 short oom_score_adj; /* OOM kill score adjustment */
699 short oom_score_adj_min; /* OOM kill score adjustment min value.
700 * Only settable by CAP_SYS_RESOURCE. */
702 struct mutex cred_guard_mutex; /* guard against foreign influences on
703 * credential calculations
704 * (notably. ptrace) */
708 * Bits in flags field of signal_struct.
710 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
711 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
712 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
713 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
715 * Pending notifications to parent.
717 #define SIGNAL_CLD_STOPPED 0x00000010
718 #define SIGNAL_CLD_CONTINUED 0x00000020
719 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
721 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
723 /* If true, all threads except ->group_exit_task have pending SIGKILL */
724 static inline int signal_group_exit(const struct signal_struct *sig)
726 return (sig->flags & SIGNAL_GROUP_EXIT) ||
727 (sig->group_exit_task != NULL);
731 * Some day this will be a full-fledged user tracking system..
734 atomic_t __count; /* reference count */
735 atomic_t processes; /* How many processes does this user have? */
736 atomic_t files; /* How many open files does this user have? */
737 atomic_t sigpending; /* How many pending signals does this user have? */
738 #ifdef CONFIG_INOTIFY_USER
739 atomic_t inotify_watches; /* How many inotify watches does this user have? */
740 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
742 #ifdef CONFIG_FANOTIFY
743 atomic_t fanotify_listeners;
746 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
748 #ifdef CONFIG_POSIX_MQUEUE
749 /* protected by mq_lock */
750 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
752 unsigned long locked_shm; /* How many pages of mlocked shm ? */
755 struct key *uid_keyring; /* UID specific keyring */
756 struct key *session_keyring; /* UID's default session keyring */
759 /* Hash table maintenance information */
760 struct hlist_node uidhash_node;
763 #ifdef CONFIG_PERF_EVENTS
764 atomic_long_t locked_vm;
768 extern int uids_sysfs_init(void);
770 extern struct user_struct *find_user(kuid_t);
772 extern struct user_struct root_user;
773 #define INIT_USER (&root_user)
776 struct backing_dev_info;
777 struct reclaim_state;
779 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
781 /* cumulative counters */
782 unsigned long pcount; /* # of times run on this cpu */
783 unsigned long long run_delay; /* time spent waiting on a runqueue */
786 unsigned long long last_arrival,/* when we last ran on a cpu */
787 last_queued; /* when we were last queued to run */
789 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
791 #ifdef CONFIG_TASK_DELAY_ACCT
792 struct task_delay_info {
794 unsigned int flags; /* Private per-task flags */
796 /* For each stat XXX, add following, aligned appropriately
798 * struct timespec XXX_start, XXX_end;
802 * Atomicity of updates to XXX_delay, XXX_count protected by
803 * single lock above (split into XXX_lock if contention is an issue).
807 * XXX_count is incremented on every XXX operation, the delay
808 * associated with the operation is added to XXX_delay.
809 * XXX_delay contains the accumulated delay time in nanoseconds.
811 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
812 u64 blkio_delay; /* wait for sync block io completion */
813 u64 swapin_delay; /* wait for swapin block io completion */
814 u32 blkio_count; /* total count of the number of sync block */
815 /* io operations performed */
816 u32 swapin_count; /* total count of the number of swapin block */
817 /* io operations performed */
819 struct timespec freepages_start, freepages_end;
820 u64 freepages_delay; /* wait for memory reclaim */
821 u32 freepages_count; /* total count of memory reclaim */
823 #endif /* CONFIG_TASK_DELAY_ACCT */
825 static inline int sched_info_on(void)
827 #ifdef CONFIG_SCHEDSTATS
829 #elif defined(CONFIG_TASK_DELAY_ACCT)
830 extern int delayacct_on;
845 * Increase resolution of cpu_power calculations
847 #define SCHED_POWER_SHIFT 10
848 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
851 * sched-domains (multiprocessor balancing) declarations:
854 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
855 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
856 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
857 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
858 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
859 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
860 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
861 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
862 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
863 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
864 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
865 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
866 #define SD_NUMA 0x4000 /* cross-node balancing */
868 extern int __weak arch_sd_sibiling_asym_packing(void);
870 struct sched_domain_attr {
871 int relax_domain_level;
874 #define SD_ATTR_INIT (struct sched_domain_attr) { \
875 .relax_domain_level = -1, \
878 extern int sched_domain_level_max;
882 struct sched_domain {
883 /* These fields must be setup */
884 struct sched_domain *parent; /* top domain must be null terminated */
885 struct sched_domain *child; /* bottom domain must be null terminated */
886 struct sched_group *groups; /* the balancing groups of the domain */
887 unsigned long min_interval; /* Minimum balance interval ms */
888 unsigned long max_interval; /* Maximum balance interval ms */
889 unsigned int busy_factor; /* less balancing by factor if busy */
890 unsigned int imbalance_pct; /* No balance until over watermark */
891 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
892 unsigned int busy_idx;
893 unsigned int idle_idx;
894 unsigned int newidle_idx;
895 unsigned int wake_idx;
896 unsigned int forkexec_idx;
897 unsigned int smt_gain;
899 int nohz_idle; /* NOHZ IDLE status */
900 int flags; /* See SD_* */
903 /* Runtime fields. */
904 unsigned long last_balance; /* init to jiffies. units in jiffies */
905 unsigned int balance_interval; /* initialise to 1. units in ms. */
906 unsigned int nr_balance_failed; /* initialise to 0 */
908 /* idle_balance() stats */
909 u64 max_newidle_lb_cost;
910 unsigned long next_decay_max_lb_cost;
912 #ifdef CONFIG_SCHEDSTATS
913 /* load_balance() stats */
914 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
915 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
916 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
917 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
918 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
919 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
920 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
921 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
923 /* Active load balancing */
924 unsigned int alb_count;
925 unsigned int alb_failed;
926 unsigned int alb_pushed;
928 /* SD_BALANCE_EXEC stats */
929 unsigned int sbe_count;
930 unsigned int sbe_balanced;
931 unsigned int sbe_pushed;
933 /* SD_BALANCE_FORK stats */
934 unsigned int sbf_count;
935 unsigned int sbf_balanced;
936 unsigned int sbf_pushed;
938 /* try_to_wake_up() stats */
939 unsigned int ttwu_wake_remote;
940 unsigned int ttwu_move_affine;
941 unsigned int ttwu_move_balance;
943 #ifdef CONFIG_SCHED_DEBUG
947 void *private; /* used during construction */
948 struct rcu_head rcu; /* used during destruction */
951 unsigned int span_weight;
953 * Span of all CPUs in this domain.
955 * NOTE: this field is variable length. (Allocated dynamically
956 * by attaching extra space to the end of the structure,
957 * depending on how many CPUs the kernel has booted up with)
959 unsigned long span[0];
962 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
964 return to_cpumask(sd->span);
967 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
968 struct sched_domain_attr *dattr_new);
970 /* Allocate an array of sched domains, for partition_sched_domains(). */
971 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
972 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
974 bool cpus_share_cache(int this_cpu, int that_cpu);
976 #else /* CONFIG_SMP */
978 struct sched_domain_attr;
981 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
982 struct sched_domain_attr *dattr_new)
986 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
991 #endif /* !CONFIG_SMP */
994 struct io_context; /* See blkdev.h */
997 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
998 extern void prefetch_stack(struct task_struct *t);
1000 static inline void prefetch_stack(struct task_struct *t) { }
1003 struct audit_context; /* See audit.c */
1005 struct pipe_inode_info;
1006 struct uts_namespace;
1008 struct load_weight {
1009 unsigned long weight;
1015 * These sums represent an infinite geometric series and so are bound
1016 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1017 * choices of y < 1-2^(-32)*1024.
1019 u32 runnable_avg_sum, runnable_avg_period;
1020 u64 last_runnable_update;
1022 unsigned long load_avg_contrib;
1025 #ifdef CONFIG_SCHEDSTATS
1026 struct sched_statistics {
1036 s64 sum_sleep_runtime;
1043 u64 nr_migrations_cold;
1044 u64 nr_failed_migrations_affine;
1045 u64 nr_failed_migrations_running;
1046 u64 nr_failed_migrations_hot;
1047 u64 nr_forced_migrations;
1050 u64 nr_wakeups_sync;
1051 u64 nr_wakeups_migrate;
1052 u64 nr_wakeups_local;
1053 u64 nr_wakeups_remote;
1054 u64 nr_wakeups_affine;
1055 u64 nr_wakeups_affine_attempts;
1056 u64 nr_wakeups_passive;
1057 u64 nr_wakeups_idle;
1061 struct sched_entity {
1062 struct load_weight load; /* for load-balancing */
1063 struct rb_node run_node;
1064 struct list_head group_node;
1068 u64 sum_exec_runtime;
1070 u64 prev_sum_exec_runtime;
1074 #ifdef CONFIG_SCHEDSTATS
1075 struct sched_statistics statistics;
1078 #ifdef CONFIG_FAIR_GROUP_SCHED
1079 struct sched_entity *parent;
1080 /* rq on which this entity is (to be) queued: */
1081 struct cfs_rq *cfs_rq;
1082 /* rq "owned" by this entity/group: */
1083 struct cfs_rq *my_q;
1087 /* Per-entity load-tracking */
1088 struct sched_avg avg;
1092 struct sched_rt_entity {
1093 struct list_head run_list;
1094 unsigned long timeout;
1095 unsigned long watchdog_stamp;
1096 unsigned int time_slice;
1098 struct sched_rt_entity *back;
1099 #ifdef CONFIG_RT_GROUP_SCHED
1100 struct sched_rt_entity *parent;
1101 /* rq on which this entity is (to be) queued: */
1102 struct rt_rq *rt_rq;
1103 /* rq "owned" by this entity/group: */
1108 struct sched_dl_entity {
1109 struct rb_node rb_node;
1112 * Original scheduling parameters. Copied here from sched_attr
1113 * during sched_setscheduler2(), they will remain the same until
1114 * the next sched_setscheduler2().
1116 u64 dl_runtime; /* maximum runtime for each instance */
1117 u64 dl_deadline; /* relative deadline of each instance */
1118 u64 dl_period; /* separation of two instances (period) */
1119 u64 dl_bw; /* dl_runtime / dl_deadline */
1122 * Actual scheduling parameters. Initialized with the values above,
1123 * they are continously updated during task execution. Note that
1124 * the remaining runtime could be < 0 in case we are in overrun.
1126 s64 runtime; /* remaining runtime for this instance */
1127 u64 deadline; /* absolute deadline for this instance */
1128 unsigned int flags; /* specifying the scheduler behaviour */
1133 * @dl_throttled tells if we exhausted the runtime. If so, the
1134 * task has to wait for a replenishment to be performed at the
1135 * next firing of dl_timer.
1137 * @dl_new tells if a new instance arrived. If so we must
1138 * start executing it with full runtime and reset its absolute
1141 * @dl_boosted tells if we are boosted due to DI. If so we are
1142 * outside bandwidth enforcement mechanism (but only until we
1143 * exit the critical section).
1145 int dl_throttled, dl_new, dl_boosted;
1148 * Bandwidth enforcement timer. Each -deadline task has its
1149 * own bandwidth to be enforced, thus we need one timer per task.
1151 struct hrtimer dl_timer;
1156 enum perf_event_task_context {
1157 perf_invalid_context = -1,
1158 perf_hw_context = 0,
1160 perf_nr_task_contexts,
1163 struct task_struct {
1164 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1167 unsigned int flags; /* per process flags, defined below */
1168 unsigned int ptrace;
1171 struct llist_node wake_entry;
1173 struct task_struct *last_wakee;
1174 unsigned long wakee_flips;
1175 unsigned long wakee_flip_decay_ts;
1181 int prio, static_prio, normal_prio;
1182 unsigned int rt_priority;
1183 const struct sched_class *sched_class;
1184 struct sched_entity se;
1185 struct sched_rt_entity rt;
1186 #ifdef CONFIG_CGROUP_SCHED
1187 struct task_group *sched_task_group;
1189 struct sched_dl_entity dl;
1191 #ifdef CONFIG_PREEMPT_NOTIFIERS
1192 /* list of struct preempt_notifier: */
1193 struct hlist_head preempt_notifiers;
1196 #ifdef CONFIG_BLK_DEV_IO_TRACE
1197 unsigned int btrace_seq;
1200 unsigned int policy;
1201 int nr_cpus_allowed;
1202 cpumask_t cpus_allowed;
1204 #ifdef CONFIG_PREEMPT_RCU
1205 int rcu_read_lock_nesting;
1206 char rcu_read_unlock_special;
1207 struct list_head rcu_node_entry;
1208 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1209 #ifdef CONFIG_TREE_PREEMPT_RCU
1210 struct rcu_node *rcu_blocked_node;
1211 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1212 #ifdef CONFIG_RCU_BOOST
1213 struct rt_mutex *rcu_boost_mutex;
1214 #endif /* #ifdef CONFIG_RCU_BOOST */
1216 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1217 struct sched_info sched_info;
1220 struct list_head tasks;
1222 struct plist_node pushable_tasks;
1223 struct rb_node pushable_dl_tasks;
1226 struct mm_struct *mm, *active_mm;
1227 #ifdef CONFIG_COMPAT_BRK
1228 unsigned brk_randomized:1;
1230 #if defined(SPLIT_RSS_COUNTING)
1231 struct task_rss_stat rss_stat;
1235 int exit_code, exit_signal;
1236 int pdeath_signal; /* The signal sent when the parent dies */
1237 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1239 /* Used for emulating ABI behavior of previous Linux versions */
1240 unsigned int personality;
1242 unsigned did_exec:1;
1243 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1245 unsigned in_iowait:1;
1247 /* task may not gain privileges */
1248 unsigned no_new_privs:1;
1250 /* Revert to default priority/policy when forking */
1251 unsigned sched_reset_on_fork:1;
1252 unsigned sched_contributes_to_load:1;
1257 #ifdef CONFIG_CC_STACKPROTECTOR
1258 /* Canary value for the -fstack-protector gcc feature */
1259 unsigned long stack_canary;
1262 * pointers to (original) parent process, youngest child, younger sibling,
1263 * older sibling, respectively. (p->father can be replaced with
1264 * p->real_parent->pid)
1266 struct task_struct __rcu *real_parent; /* real parent process */
1267 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1269 * children/sibling forms the list of my natural children
1271 struct list_head children; /* list of my children */
1272 struct list_head sibling; /* linkage in my parent's children list */
1273 struct task_struct *group_leader; /* threadgroup leader */
1276 * ptraced is the list of tasks this task is using ptrace on.
1277 * This includes both natural children and PTRACE_ATTACH targets.
1278 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1280 struct list_head ptraced;
1281 struct list_head ptrace_entry;
1283 /* PID/PID hash table linkage. */
1284 struct pid_link pids[PIDTYPE_MAX];
1285 struct list_head thread_group;
1286 struct list_head thread_node;
1288 struct completion *vfork_done; /* for vfork() */
1289 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1290 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1292 cputime_t utime, stime, utimescaled, stimescaled;
1294 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1295 struct cputime prev_cputime;
1297 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1298 seqlock_t vtime_seqlock;
1299 unsigned long long vtime_snap;
1304 } vtime_snap_whence;
1306 unsigned long nvcsw, nivcsw; /* context switch counts */
1307 struct timespec start_time; /* monotonic time */
1308 struct timespec real_start_time; /* boot based time */
1309 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1310 unsigned long min_flt, maj_flt;
1312 struct task_cputime cputime_expires;
1313 struct list_head cpu_timers[3];
1315 /* process credentials */
1316 const struct cred __rcu *real_cred; /* objective and real subjective task
1317 * credentials (COW) */
1318 const struct cred __rcu *cred; /* effective (overridable) subjective task
1319 * credentials (COW) */
1320 char comm[TASK_COMM_LEN]; /* executable name excluding path
1321 - access with [gs]et_task_comm (which lock
1322 it with task_lock())
1323 - initialized normally by setup_new_exec */
1324 /* file system info */
1325 int link_count, total_link_count;
1326 #ifdef CONFIG_SYSVIPC
1328 struct sysv_sem sysvsem;
1330 #ifdef CONFIG_DETECT_HUNG_TASK
1331 /* hung task detection */
1332 unsigned long last_switch_count;
1334 /* CPU-specific state of this task */
1335 struct thread_struct thread;
1336 /* filesystem information */
1337 struct fs_struct *fs;
1338 /* open file information */
1339 struct files_struct *files;
1341 struct nsproxy *nsproxy;
1342 /* signal handlers */
1343 struct signal_struct *signal;
1344 struct sighand_struct *sighand;
1346 sigset_t blocked, real_blocked;
1347 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1348 struct sigpending pending;
1350 unsigned long sas_ss_sp;
1352 int (*notifier)(void *priv);
1353 void *notifier_data;
1354 sigset_t *notifier_mask;
1355 struct callback_head *task_works;
1357 struct audit_context *audit_context;
1358 #ifdef CONFIG_AUDITSYSCALL
1360 unsigned int sessionid;
1362 struct seccomp seccomp;
1364 /* Thread group tracking */
1367 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1369 spinlock_t alloc_lock;
1371 /* Protection of the PI data structures: */
1372 raw_spinlock_t pi_lock;
1374 #ifdef CONFIG_RT_MUTEXES
1375 /* PI waiters blocked on a rt_mutex held by this task */
1376 struct rb_root pi_waiters;
1377 struct rb_node *pi_waiters_leftmost;
1378 /* Deadlock detection and priority inheritance handling */
1379 struct rt_mutex_waiter *pi_blocked_on;
1380 /* Top pi_waiters task */
1381 struct task_struct *pi_top_task;
1384 #ifdef CONFIG_DEBUG_MUTEXES
1385 /* mutex deadlock detection */
1386 struct mutex_waiter *blocked_on;
1388 #ifdef CONFIG_TRACE_IRQFLAGS
1389 unsigned int irq_events;
1390 unsigned long hardirq_enable_ip;
1391 unsigned long hardirq_disable_ip;
1392 unsigned int hardirq_enable_event;
1393 unsigned int hardirq_disable_event;
1394 int hardirqs_enabled;
1395 int hardirq_context;
1396 unsigned long softirq_disable_ip;
1397 unsigned long softirq_enable_ip;
1398 unsigned int softirq_disable_event;
1399 unsigned int softirq_enable_event;
1400 int softirqs_enabled;
1401 int softirq_context;
1403 #ifdef CONFIG_LOCKDEP
1404 # define MAX_LOCK_DEPTH 48UL
1407 unsigned int lockdep_recursion;
1408 struct held_lock held_locks[MAX_LOCK_DEPTH];
1409 gfp_t lockdep_reclaim_gfp;
1412 /* journalling filesystem info */
1415 /* stacked block device info */
1416 struct bio_list *bio_list;
1419 /* stack plugging */
1420 struct blk_plug *plug;
1424 struct reclaim_state *reclaim_state;
1426 struct backing_dev_info *backing_dev_info;
1428 struct io_context *io_context;
1430 unsigned long ptrace_message;
1431 siginfo_t *last_siginfo; /* For ptrace use. */
1432 struct task_io_accounting ioac;
1433 #if defined(CONFIG_TASK_XACCT)
1434 u64 acct_rss_mem1; /* accumulated rss usage */
1435 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1436 cputime_t acct_timexpd; /* stime + utime since last update */
1438 #ifdef CONFIG_CPUSETS
1439 nodemask_t mems_allowed; /* Protected by alloc_lock */
1440 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1441 int cpuset_mem_spread_rotor;
1442 int cpuset_slab_spread_rotor;
1444 #ifdef CONFIG_CGROUPS
1445 /* Control Group info protected by css_set_lock */
1446 struct css_set __rcu *cgroups;
1447 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1448 struct list_head cg_list;
1451 struct robust_list_head __user *robust_list;
1452 #ifdef CONFIG_COMPAT
1453 struct compat_robust_list_head __user *compat_robust_list;
1455 struct list_head pi_state_list;
1456 struct futex_pi_state *pi_state_cache;
1458 #ifdef CONFIG_PERF_EVENTS
1459 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1460 struct mutex perf_event_mutex;
1461 struct list_head perf_event_list;
1464 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1466 short pref_node_fork;
1468 #ifdef CONFIG_NUMA_BALANCING
1470 unsigned int numa_scan_period;
1471 unsigned int numa_scan_period_max;
1472 int numa_preferred_nid;
1473 int numa_migrate_deferred;
1474 unsigned long numa_migrate_retry;
1475 u64 node_stamp; /* migration stamp */
1476 struct callback_head numa_work;
1478 struct list_head numa_entry;
1479 struct numa_group *numa_group;
1482 * Exponential decaying average of faults on a per-node basis.
1483 * Scheduling placement decisions are made based on the these counts.
1484 * The values remain static for the duration of a PTE scan
1486 unsigned long *numa_faults;
1487 unsigned long total_numa_faults;
1490 * numa_faults_buffer records faults per node during the current
1491 * scan window. When the scan completes, the counts in numa_faults
1492 * decay and these values are copied.
1494 unsigned long *numa_faults_buffer;
1497 * numa_faults_locality tracks if faults recorded during the last
1498 * scan window were remote/local. The task scan period is adapted
1499 * based on the locality of the faults with different weights
1500 * depending on whether they were shared or private faults
1502 unsigned long numa_faults_locality[2];
1504 unsigned long numa_pages_migrated;
1505 #endif /* CONFIG_NUMA_BALANCING */
1507 struct rcu_head rcu;
1510 * cache last used pipe for splice
1512 struct pipe_inode_info *splice_pipe;
1514 struct page_frag task_frag;
1516 #ifdef CONFIG_TASK_DELAY_ACCT
1517 struct task_delay_info *delays;
1519 #ifdef CONFIG_FAULT_INJECTION
1523 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1524 * balance_dirty_pages() for some dirty throttling pause
1527 int nr_dirtied_pause;
1528 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1530 #ifdef CONFIG_LATENCYTOP
1531 int latency_record_count;
1532 struct latency_record latency_record[LT_SAVECOUNT];
1535 * time slack values; these are used to round up poll() and
1536 * select() etc timeout values. These are in nanoseconds.
1538 unsigned long timer_slack_ns;
1539 unsigned long default_timer_slack_ns;
1541 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1542 /* Index of current stored address in ret_stack */
1544 /* Stack of return addresses for return function tracing */
1545 struct ftrace_ret_stack *ret_stack;
1546 /* time stamp for last schedule */
1547 unsigned long long ftrace_timestamp;
1549 * Number of functions that haven't been traced
1550 * because of depth overrun.
1552 atomic_t trace_overrun;
1553 /* Pause for the tracing */
1554 atomic_t tracing_graph_pause;
1556 #ifdef CONFIG_TRACING
1557 /* state flags for use by tracers */
1558 unsigned long trace;
1559 /* bitmask and counter of trace recursion */
1560 unsigned long trace_recursion;
1561 #endif /* CONFIG_TRACING */
1562 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1563 struct memcg_batch_info {
1564 int do_batch; /* incremented when batch uncharge started */
1565 struct mem_cgroup *memcg; /* target memcg of uncharge */
1566 unsigned long nr_pages; /* uncharged usage */
1567 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1569 unsigned int memcg_kmem_skip_account;
1570 struct memcg_oom_info {
1571 struct mem_cgroup *memcg;
1574 unsigned int may_oom:1;
1577 #ifdef CONFIG_UPROBES
1578 struct uprobe_task *utask;
1580 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1581 unsigned int sequential_io;
1582 unsigned int sequential_io_avg;
1586 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1587 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1589 #define TNF_MIGRATED 0x01
1590 #define TNF_NO_GROUP 0x02
1591 #define TNF_SHARED 0x04
1592 #define TNF_FAULT_LOCAL 0x08
1594 #ifdef CONFIG_NUMA_BALANCING
1595 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1596 extern pid_t task_numa_group_id(struct task_struct *p);
1597 extern void set_numabalancing_state(bool enabled);
1598 extern void task_numa_free(struct task_struct *p);
1600 extern unsigned int sysctl_numa_balancing_migrate_deferred;
1602 static inline void task_numa_fault(int last_node, int node, int pages,
1606 static inline pid_t task_numa_group_id(struct task_struct *p)
1610 static inline void set_numabalancing_state(bool enabled)
1613 static inline void task_numa_free(struct task_struct *p)
1618 static inline struct pid *task_pid(struct task_struct *task)
1620 return task->pids[PIDTYPE_PID].pid;
1623 static inline struct pid *task_tgid(struct task_struct *task)
1625 return task->group_leader->pids[PIDTYPE_PID].pid;
1629 * Without tasklist or rcu lock it is not safe to dereference
1630 * the result of task_pgrp/task_session even if task == current,
1631 * we can race with another thread doing sys_setsid/sys_setpgid.
1633 static inline struct pid *task_pgrp(struct task_struct *task)
1635 return task->group_leader->pids[PIDTYPE_PGID].pid;
1638 static inline struct pid *task_session(struct task_struct *task)
1640 return task->group_leader->pids[PIDTYPE_SID].pid;
1643 struct pid_namespace;
1646 * the helpers to get the task's different pids as they are seen
1647 * from various namespaces
1649 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1650 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1652 * task_xid_nr_ns() : id seen from the ns specified;
1654 * set_task_vxid() : assigns a virtual id to a task;
1656 * see also pid_nr() etc in include/linux/pid.h
1658 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1659 struct pid_namespace *ns);
1661 static inline pid_t task_pid_nr(struct task_struct *tsk)
1666 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1667 struct pid_namespace *ns)
1669 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1672 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1674 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1678 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1683 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1685 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1687 return pid_vnr(task_tgid(tsk));
1691 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1692 struct pid_namespace *ns)
1694 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1697 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1699 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1703 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1704 struct pid_namespace *ns)
1706 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1709 static inline pid_t task_session_vnr(struct task_struct *tsk)
1711 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1714 /* obsolete, do not use */
1715 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1717 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1721 * pid_alive - check that a task structure is not stale
1722 * @p: Task structure to be checked.
1724 * Test if a process is not yet dead (at most zombie state)
1725 * If pid_alive fails, then pointers within the task structure
1726 * can be stale and must not be dereferenced.
1728 * Return: 1 if the process is alive. 0 otherwise.
1730 static inline int pid_alive(struct task_struct *p)
1732 return p->pids[PIDTYPE_PID].pid != NULL;
1736 * is_global_init - check if a task structure is init
1737 * @tsk: Task structure to be checked.
1739 * Check if a task structure is the first user space task the kernel created.
1741 * Return: 1 if the task structure is init. 0 otherwise.
1743 static inline int is_global_init(struct task_struct *tsk)
1745 return tsk->pid == 1;
1748 extern struct pid *cad_pid;
1750 extern void free_task(struct task_struct *tsk);
1751 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1753 extern void __put_task_struct(struct task_struct *t);
1755 static inline void put_task_struct(struct task_struct *t)
1757 if (atomic_dec_and_test(&t->usage))
1758 __put_task_struct(t);
1761 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1762 extern void task_cputime(struct task_struct *t,
1763 cputime_t *utime, cputime_t *stime);
1764 extern void task_cputime_scaled(struct task_struct *t,
1765 cputime_t *utimescaled, cputime_t *stimescaled);
1766 extern cputime_t task_gtime(struct task_struct *t);
1768 static inline void task_cputime(struct task_struct *t,
1769 cputime_t *utime, cputime_t *stime)
1777 static inline void task_cputime_scaled(struct task_struct *t,
1778 cputime_t *utimescaled,
1779 cputime_t *stimescaled)
1782 *utimescaled = t->utimescaled;
1784 *stimescaled = t->stimescaled;
1787 static inline cputime_t task_gtime(struct task_struct *t)
1792 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1793 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1798 #define PF_EXITING 0x00000004 /* getting shut down */
1799 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1800 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1801 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1802 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1803 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1804 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1805 #define PF_DUMPCORE 0x00000200 /* dumped core */
1806 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1807 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1808 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1809 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1810 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1811 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1812 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1813 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1814 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1815 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1816 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1817 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1818 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1819 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1820 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1821 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1822 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1823 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1824 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1825 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1826 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1827 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1830 * Only the _current_ task can read/write to tsk->flags, but other
1831 * tasks can access tsk->flags in readonly mode for example
1832 * with tsk_used_math (like during threaded core dumping).
1833 * There is however an exception to this rule during ptrace
1834 * or during fork: the ptracer task is allowed to write to the
1835 * child->flags of its traced child (same goes for fork, the parent
1836 * can write to the child->flags), because we're guaranteed the
1837 * child is not running and in turn not changing child->flags
1838 * at the same time the parent does it.
1840 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1841 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1842 #define clear_used_math() clear_stopped_child_used_math(current)
1843 #define set_used_math() set_stopped_child_used_math(current)
1844 #define conditional_stopped_child_used_math(condition, child) \
1845 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1846 #define conditional_used_math(condition) \
1847 conditional_stopped_child_used_math(condition, current)
1848 #define copy_to_stopped_child_used_math(child) \
1849 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1850 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1851 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1852 #define used_math() tsk_used_math(current)
1854 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1855 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1857 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1862 static inline unsigned int memalloc_noio_save(void)
1864 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1865 current->flags |= PF_MEMALLOC_NOIO;
1869 static inline void memalloc_noio_restore(unsigned int flags)
1871 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1875 * task->jobctl flags
1877 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1879 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1880 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1881 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1882 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1883 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1884 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1885 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1887 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1888 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1889 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1890 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1891 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1892 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1893 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1895 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1896 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1898 extern bool task_set_jobctl_pending(struct task_struct *task,
1900 extern void task_clear_jobctl_trapping(struct task_struct *task);
1901 extern void task_clear_jobctl_pending(struct task_struct *task,
1904 #ifdef CONFIG_PREEMPT_RCU
1906 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1907 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1909 static inline void rcu_copy_process(struct task_struct *p)
1911 p->rcu_read_lock_nesting = 0;
1912 p->rcu_read_unlock_special = 0;
1913 #ifdef CONFIG_TREE_PREEMPT_RCU
1914 p->rcu_blocked_node = NULL;
1915 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1916 #ifdef CONFIG_RCU_BOOST
1917 p->rcu_boost_mutex = NULL;
1918 #endif /* #ifdef CONFIG_RCU_BOOST */
1919 INIT_LIST_HEAD(&p->rcu_node_entry);
1924 static inline void rcu_copy_process(struct task_struct *p)
1930 static inline void tsk_restore_flags(struct task_struct *task,
1931 unsigned long orig_flags, unsigned long flags)
1933 task->flags &= ~flags;
1934 task->flags |= orig_flags & flags;
1938 extern void do_set_cpus_allowed(struct task_struct *p,
1939 const struct cpumask *new_mask);
1941 extern int set_cpus_allowed_ptr(struct task_struct *p,
1942 const struct cpumask *new_mask);
1944 static inline void do_set_cpus_allowed(struct task_struct *p,
1945 const struct cpumask *new_mask)
1948 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1949 const struct cpumask *new_mask)
1951 if (!cpumask_test_cpu(0, new_mask))
1957 #ifdef CONFIG_NO_HZ_COMMON
1958 void calc_load_enter_idle(void);
1959 void calc_load_exit_idle(void);
1961 static inline void calc_load_enter_idle(void) { }
1962 static inline void calc_load_exit_idle(void) { }
1963 #endif /* CONFIG_NO_HZ_COMMON */
1965 #ifndef CONFIG_CPUMASK_OFFSTACK
1966 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1968 return set_cpus_allowed_ptr(p, &new_mask);
1973 * Do not use outside of architecture code which knows its limitations.
1975 * sched_clock() has no promise of monotonicity or bounded drift between
1976 * CPUs, use (which you should not) requires disabling IRQs.
1978 * Please use one of the three interfaces below.
1980 extern unsigned long long notrace sched_clock(void);
1982 * See the comment in kernel/sched/clock.c
1984 extern u64 cpu_clock(int cpu);
1985 extern u64 local_clock(void);
1986 extern u64 sched_clock_cpu(int cpu);
1989 extern void sched_clock_init(void);
1991 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1992 static inline void sched_clock_tick(void)
1996 static inline void sched_clock_idle_sleep_event(void)
2000 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2005 * Architectures can set this to 1 if they have specified
2006 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2007 * but then during bootup it turns out that sched_clock()
2008 * is reliable after all:
2010 extern int sched_clock_stable(void);
2011 extern void set_sched_clock_stable(void);
2012 extern void clear_sched_clock_stable(void);
2014 extern void sched_clock_tick(void);
2015 extern void sched_clock_idle_sleep_event(void);
2016 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2019 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2021 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2022 * The reason for this explicit opt-in is not to have perf penalty with
2023 * slow sched_clocks.
2025 extern void enable_sched_clock_irqtime(void);
2026 extern void disable_sched_clock_irqtime(void);
2028 static inline void enable_sched_clock_irqtime(void) {}
2029 static inline void disable_sched_clock_irqtime(void) {}
2032 extern unsigned long long
2033 task_sched_runtime(struct task_struct *task);
2035 /* sched_exec is called by processes performing an exec */
2037 extern void sched_exec(void);
2039 #define sched_exec() {}
2042 extern void sched_clock_idle_sleep_event(void);
2043 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2045 #ifdef CONFIG_HOTPLUG_CPU
2046 extern void idle_task_exit(void);
2048 static inline void idle_task_exit(void) {}
2051 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2052 extern void wake_up_nohz_cpu(int cpu);
2054 static inline void wake_up_nohz_cpu(int cpu) { }
2057 #ifdef CONFIG_NO_HZ_FULL
2058 extern bool sched_can_stop_tick(void);
2059 extern u64 scheduler_tick_max_deferment(void);
2061 static inline bool sched_can_stop_tick(void) { return false; }
2064 #ifdef CONFIG_SCHED_AUTOGROUP
2065 extern void sched_autogroup_create_attach(struct task_struct *p);
2066 extern void sched_autogroup_detach(struct task_struct *p);
2067 extern void sched_autogroup_fork(struct signal_struct *sig);
2068 extern void sched_autogroup_exit(struct signal_struct *sig);
2069 #ifdef CONFIG_PROC_FS
2070 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2071 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2074 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2075 static inline void sched_autogroup_detach(struct task_struct *p) { }
2076 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2077 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2080 extern bool yield_to(struct task_struct *p, bool preempt);
2081 extern void set_user_nice(struct task_struct *p, long nice);
2082 extern int task_prio(const struct task_struct *p);
2083 extern int task_nice(const struct task_struct *p);
2084 extern int can_nice(const struct task_struct *p, const int nice);
2085 extern int task_curr(const struct task_struct *p);
2086 extern int idle_cpu(int cpu);
2087 extern int sched_setscheduler(struct task_struct *, int,
2088 const struct sched_param *);
2089 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2090 const struct sched_param *);
2091 extern int sched_setattr(struct task_struct *,
2092 const struct sched_attr *);
2093 extern struct task_struct *idle_task(int cpu);
2095 * is_idle_task - is the specified task an idle task?
2096 * @p: the task in question.
2098 * Return: 1 if @p is an idle task. 0 otherwise.
2100 static inline bool is_idle_task(const struct task_struct *p)
2104 extern struct task_struct *curr_task(int cpu);
2105 extern void set_curr_task(int cpu, struct task_struct *p);
2110 * The default (Linux) execution domain.
2112 extern struct exec_domain default_exec_domain;
2114 union thread_union {
2115 struct thread_info thread_info;
2116 unsigned long stack[THREAD_SIZE/sizeof(long)];
2119 #ifndef __HAVE_ARCH_KSTACK_END
2120 static inline int kstack_end(void *addr)
2122 /* Reliable end of stack detection:
2123 * Some APM bios versions misalign the stack
2125 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2129 extern union thread_union init_thread_union;
2130 extern struct task_struct init_task;
2132 extern struct mm_struct init_mm;
2134 extern struct pid_namespace init_pid_ns;
2137 * find a task by one of its numerical ids
2139 * find_task_by_pid_ns():
2140 * finds a task by its pid in the specified namespace
2141 * find_task_by_vpid():
2142 * finds a task by its virtual pid
2144 * see also find_vpid() etc in include/linux/pid.h
2147 extern struct task_struct *find_task_by_vpid(pid_t nr);
2148 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2149 struct pid_namespace *ns);
2151 /* per-UID process charging. */
2152 extern struct user_struct * alloc_uid(kuid_t);
2153 static inline struct user_struct *get_uid(struct user_struct *u)
2155 atomic_inc(&u->__count);
2158 extern void free_uid(struct user_struct *);
2160 #include <asm/current.h>
2162 extern void xtime_update(unsigned long ticks);
2164 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2165 extern int wake_up_process(struct task_struct *tsk);
2166 extern void wake_up_new_task(struct task_struct *tsk);
2168 extern void kick_process(struct task_struct *tsk);
2170 static inline void kick_process(struct task_struct *tsk) { }
2172 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2173 extern void sched_dead(struct task_struct *p);
2175 extern void proc_caches_init(void);
2176 extern void flush_signals(struct task_struct *);
2177 extern void __flush_signals(struct task_struct *);
2178 extern void ignore_signals(struct task_struct *);
2179 extern void flush_signal_handlers(struct task_struct *, int force_default);
2180 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2182 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2184 unsigned long flags;
2187 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2188 ret = dequeue_signal(tsk, mask, info);
2189 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2194 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2196 extern void unblock_all_signals(void);
2197 extern void release_task(struct task_struct * p);
2198 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2199 extern int force_sigsegv(int, struct task_struct *);
2200 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2201 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2202 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2203 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2204 const struct cred *, u32);
2205 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2206 extern int kill_pid(struct pid *pid, int sig, int priv);
2207 extern int kill_proc_info(int, struct siginfo *, pid_t);
2208 extern __must_check bool do_notify_parent(struct task_struct *, int);
2209 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2210 extern void force_sig(int, struct task_struct *);
2211 extern int send_sig(int, struct task_struct *, int);
2212 extern int zap_other_threads(struct task_struct *p);
2213 extern struct sigqueue *sigqueue_alloc(void);
2214 extern void sigqueue_free(struct sigqueue *);
2215 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2216 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2218 static inline void restore_saved_sigmask(void)
2220 if (test_and_clear_restore_sigmask())
2221 __set_current_blocked(¤t->saved_sigmask);
2224 static inline sigset_t *sigmask_to_save(void)
2226 sigset_t *res = ¤t->blocked;
2227 if (unlikely(test_restore_sigmask()))
2228 res = ¤t->saved_sigmask;
2232 static inline int kill_cad_pid(int sig, int priv)
2234 return kill_pid(cad_pid, sig, priv);
2237 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2238 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2239 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2240 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2243 * True if we are on the alternate signal stack.
2245 static inline int on_sig_stack(unsigned long sp)
2247 #ifdef CONFIG_STACK_GROWSUP
2248 return sp >= current->sas_ss_sp &&
2249 sp - current->sas_ss_sp < current->sas_ss_size;
2251 return sp > current->sas_ss_sp &&
2252 sp - current->sas_ss_sp <= current->sas_ss_size;
2256 static inline int sas_ss_flags(unsigned long sp)
2258 return (current->sas_ss_size == 0 ? SS_DISABLE
2259 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2262 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2264 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2265 #ifdef CONFIG_STACK_GROWSUP
2266 return current->sas_ss_sp;
2268 return current->sas_ss_sp + current->sas_ss_size;
2274 * Routines for handling mm_structs
2276 extern struct mm_struct * mm_alloc(void);
2278 /* mmdrop drops the mm and the page tables */
2279 extern void __mmdrop(struct mm_struct *);
2280 static inline void mmdrop(struct mm_struct * mm)
2282 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2286 /* mmput gets rid of the mappings and all user-space */
2287 extern void mmput(struct mm_struct *);
2288 /* Grab a reference to a task's mm, if it is not already going away */
2289 extern struct mm_struct *get_task_mm(struct task_struct *task);
2291 * Grab a reference to a task's mm, if it is not already going away
2292 * and ptrace_may_access with the mode parameter passed to it
2295 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2296 /* Remove the current tasks stale references to the old mm_struct */
2297 extern void mm_release(struct task_struct *, struct mm_struct *);
2298 /* Allocate a new mm structure and copy contents from tsk->mm */
2299 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2301 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2302 struct task_struct *);
2303 extern void flush_thread(void);
2304 extern void exit_thread(void);
2306 extern void exit_files(struct task_struct *);
2307 extern void __cleanup_sighand(struct sighand_struct *);
2309 extern void exit_itimers(struct signal_struct *);
2310 extern void flush_itimer_signals(void);
2312 extern void do_group_exit(int);
2314 extern int allow_signal(int);
2315 extern int disallow_signal(int);
2317 extern int do_execve(const char *,
2318 const char __user * const __user *,
2319 const char __user * const __user *);
2320 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2321 struct task_struct *fork_idle(int);
2322 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2324 extern void set_task_comm(struct task_struct *tsk, char *from);
2325 extern char *get_task_comm(char *to, struct task_struct *tsk);
2328 void scheduler_ipi(void);
2329 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2331 static inline void scheduler_ipi(void) { }
2332 static inline unsigned long wait_task_inactive(struct task_struct *p,
2339 #define next_task(p) \
2340 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2342 #define for_each_process(p) \
2343 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2345 extern bool current_is_single_threaded(void);
2348 * Careful: do_each_thread/while_each_thread is a double loop so
2349 * 'break' will not work as expected - use goto instead.
2351 #define do_each_thread(g, t) \
2352 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2354 #define while_each_thread(g, t) \
2355 while ((t = next_thread(t)) != g)
2357 #define __for_each_thread(signal, t) \
2358 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2360 #define for_each_thread(p, t) \
2361 __for_each_thread((p)->signal, t)
2363 /* Careful: this is a double loop, 'break' won't work as expected. */
2364 #define for_each_process_thread(p, t) \
2365 for_each_process(p) for_each_thread(p, t)
2367 static inline int get_nr_threads(struct task_struct *tsk)
2369 return tsk->signal->nr_threads;
2372 static inline bool thread_group_leader(struct task_struct *p)
2374 return p->exit_signal >= 0;
2377 /* Do to the insanities of de_thread it is possible for a process
2378 * to have the pid of the thread group leader without actually being
2379 * the thread group leader. For iteration through the pids in proc
2380 * all we care about is that we have a task with the appropriate
2381 * pid, we don't actually care if we have the right task.
2383 static inline bool has_group_leader_pid(struct task_struct *p)
2385 return task_pid(p) == p->signal->leader_pid;
2389 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2391 return p1->signal == p2->signal;
2394 static inline struct task_struct *next_thread(const struct task_struct *p)
2396 return list_entry_rcu(p->thread_group.next,
2397 struct task_struct, thread_group);
2400 static inline int thread_group_empty(struct task_struct *p)
2402 return list_empty(&p->thread_group);
2405 #define delay_group_leader(p) \
2406 (thread_group_leader(p) && !thread_group_empty(p))
2409 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2410 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2411 * pins the final release of task.io_context. Also protects ->cpuset and
2412 * ->cgroup.subsys[]. And ->vfork_done.
2414 * Nests both inside and outside of read_lock(&tasklist_lock).
2415 * It must not be nested with write_lock_irq(&tasklist_lock),
2416 * neither inside nor outside.
2418 static inline void task_lock(struct task_struct *p)
2420 spin_lock(&p->alloc_lock);
2423 static inline void task_unlock(struct task_struct *p)
2425 spin_unlock(&p->alloc_lock);
2428 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2429 unsigned long *flags);
2431 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2432 unsigned long *flags)
2434 struct sighand_struct *ret;
2436 ret = __lock_task_sighand(tsk, flags);
2437 (void)__cond_lock(&tsk->sighand->siglock, ret);
2441 static inline void unlock_task_sighand(struct task_struct *tsk,
2442 unsigned long *flags)
2444 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2447 #ifdef CONFIG_CGROUPS
2448 static inline void threadgroup_change_begin(struct task_struct *tsk)
2450 down_read(&tsk->signal->group_rwsem);
2452 static inline void threadgroup_change_end(struct task_struct *tsk)
2454 up_read(&tsk->signal->group_rwsem);
2458 * threadgroup_lock - lock threadgroup
2459 * @tsk: member task of the threadgroup to lock
2461 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2462 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2463 * change ->group_leader/pid. This is useful for cases where the threadgroup
2464 * needs to stay stable across blockable operations.
2466 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2467 * synchronization. While held, no new task will be added to threadgroup
2468 * and no existing live task will have its PF_EXITING set.
2470 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2471 * sub-thread becomes a new leader.
2473 static inline void threadgroup_lock(struct task_struct *tsk)
2475 down_write(&tsk->signal->group_rwsem);
2479 * threadgroup_unlock - unlock threadgroup
2480 * @tsk: member task of the threadgroup to unlock
2482 * Reverse threadgroup_lock().
2484 static inline void threadgroup_unlock(struct task_struct *tsk)
2486 up_write(&tsk->signal->group_rwsem);
2489 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2490 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2491 static inline void threadgroup_lock(struct task_struct *tsk) {}
2492 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2495 #ifndef __HAVE_THREAD_FUNCTIONS
2497 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2498 #define task_stack_page(task) ((task)->stack)
2500 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2502 *task_thread_info(p) = *task_thread_info(org);
2503 task_thread_info(p)->task = p;
2506 static inline unsigned long *end_of_stack(struct task_struct *p)
2508 return (unsigned long *)(task_thread_info(p) + 1);
2513 static inline int object_is_on_stack(void *obj)
2515 void *stack = task_stack_page(current);
2517 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2520 extern void thread_info_cache_init(void);
2522 #ifdef CONFIG_DEBUG_STACK_USAGE
2523 static inline unsigned long stack_not_used(struct task_struct *p)
2525 unsigned long *n = end_of_stack(p);
2527 do { /* Skip over canary */
2531 return (unsigned long)n - (unsigned long)end_of_stack(p);
2535 /* set thread flags in other task's structures
2536 * - see asm/thread_info.h for TIF_xxxx flags available
2538 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2540 set_ti_thread_flag(task_thread_info(tsk), flag);
2543 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2545 clear_ti_thread_flag(task_thread_info(tsk), flag);
2548 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2550 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2553 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2555 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2558 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2560 return test_ti_thread_flag(task_thread_info(tsk), flag);
2563 static inline void set_tsk_need_resched(struct task_struct *tsk)
2565 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2568 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2570 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2573 static inline int test_tsk_need_resched(struct task_struct *tsk)
2575 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2578 static inline int restart_syscall(void)
2580 set_tsk_thread_flag(current, TIF_SIGPENDING);
2581 return -ERESTARTNOINTR;
2584 static inline int signal_pending(struct task_struct *p)
2586 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2589 static inline int __fatal_signal_pending(struct task_struct *p)
2591 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2594 static inline int fatal_signal_pending(struct task_struct *p)
2596 return signal_pending(p) && __fatal_signal_pending(p);
2599 static inline int signal_pending_state(long state, struct task_struct *p)
2601 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2603 if (!signal_pending(p))
2606 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2610 * cond_resched() and cond_resched_lock(): latency reduction via
2611 * explicit rescheduling in places that are safe. The return
2612 * value indicates whether a reschedule was done in fact.
2613 * cond_resched_lock() will drop the spinlock before scheduling,
2614 * cond_resched_softirq() will enable bhs before scheduling.
2616 extern int _cond_resched(void);
2618 #define cond_resched() ({ \
2619 __might_sleep(__FILE__, __LINE__, 0); \
2623 extern int __cond_resched_lock(spinlock_t *lock);
2625 #ifdef CONFIG_PREEMPT_COUNT
2626 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2628 #define PREEMPT_LOCK_OFFSET 0
2631 #define cond_resched_lock(lock) ({ \
2632 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2633 __cond_resched_lock(lock); \
2636 extern int __cond_resched_softirq(void);
2638 #define cond_resched_softirq() ({ \
2639 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2640 __cond_resched_softirq(); \
2643 static inline void cond_resched_rcu(void)
2645 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2653 * Does a critical section need to be broken due to another
2654 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2655 * but a general need for low latency)
2657 static inline int spin_needbreak(spinlock_t *lock)
2659 #ifdef CONFIG_PREEMPT
2660 return spin_is_contended(lock);
2667 * Idle thread specific functions to determine the need_resched
2668 * polling state. We have two versions, one based on TS_POLLING in
2669 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2673 static inline int tsk_is_polling(struct task_struct *p)
2675 return task_thread_info(p)->status & TS_POLLING;
2677 static inline void __current_set_polling(void)
2679 current_thread_info()->status |= TS_POLLING;
2682 static inline bool __must_check current_set_polling_and_test(void)
2684 __current_set_polling();
2687 * Polling state must be visible before we test NEED_RESCHED,
2688 * paired by resched_task()
2692 return unlikely(tif_need_resched());
2695 static inline void __current_clr_polling(void)
2697 current_thread_info()->status &= ~TS_POLLING;
2700 static inline bool __must_check current_clr_polling_and_test(void)
2702 __current_clr_polling();
2705 * Polling state must be visible before we test NEED_RESCHED,
2706 * paired by resched_task()
2710 return unlikely(tif_need_resched());
2712 #elif defined(TIF_POLLING_NRFLAG)
2713 static inline int tsk_is_polling(struct task_struct *p)
2715 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2718 static inline void __current_set_polling(void)
2720 set_thread_flag(TIF_POLLING_NRFLAG);
2723 static inline bool __must_check current_set_polling_and_test(void)
2725 __current_set_polling();
2728 * Polling state must be visible before we test NEED_RESCHED,
2729 * paired by resched_task()
2731 * XXX: assumes set/clear bit are identical barrier wise.
2733 smp_mb__after_clear_bit();
2735 return unlikely(tif_need_resched());
2738 static inline void __current_clr_polling(void)
2740 clear_thread_flag(TIF_POLLING_NRFLAG);
2743 static inline bool __must_check current_clr_polling_and_test(void)
2745 __current_clr_polling();
2748 * Polling state must be visible before we test NEED_RESCHED,
2749 * paired by resched_task()
2751 smp_mb__after_clear_bit();
2753 return unlikely(tif_need_resched());
2757 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2758 static inline void __current_set_polling(void) { }
2759 static inline void __current_clr_polling(void) { }
2761 static inline bool __must_check current_set_polling_and_test(void)
2763 return unlikely(tif_need_resched());
2765 static inline bool __must_check current_clr_polling_and_test(void)
2767 return unlikely(tif_need_resched());
2771 static inline void current_clr_polling(void)
2773 __current_clr_polling();
2776 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2777 * Once the bit is cleared, we'll get IPIs with every new
2778 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2781 smp_mb(); /* paired with resched_task() */
2783 preempt_fold_need_resched();
2786 static __always_inline bool need_resched(void)
2788 return unlikely(tif_need_resched());
2792 * Thread group CPU time accounting.
2794 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2795 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2797 static inline void thread_group_cputime_init(struct signal_struct *sig)
2799 raw_spin_lock_init(&sig->cputimer.lock);
2803 * Reevaluate whether the task has signals pending delivery.
2804 * Wake the task if so.
2805 * This is required every time the blocked sigset_t changes.
2806 * callers must hold sighand->siglock.
2808 extern void recalc_sigpending_and_wake(struct task_struct *t);
2809 extern void recalc_sigpending(void);
2811 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2813 static inline void signal_wake_up(struct task_struct *t, bool resume)
2815 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2817 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2819 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2823 * Wrappers for p->thread_info->cpu access. No-op on UP.
2827 static inline unsigned int task_cpu(const struct task_struct *p)
2829 return task_thread_info(p)->cpu;
2832 static inline int task_node(const struct task_struct *p)
2834 return cpu_to_node(task_cpu(p));
2837 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2841 static inline unsigned int task_cpu(const struct task_struct *p)
2846 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2850 #endif /* CONFIG_SMP */
2852 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2853 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2855 #ifdef CONFIG_CGROUP_SCHED
2856 extern struct task_group root_task_group;
2857 #endif /* CONFIG_CGROUP_SCHED */
2859 extern int task_can_switch_user(struct user_struct *up,
2860 struct task_struct *tsk);
2862 #ifdef CONFIG_TASK_XACCT
2863 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2865 tsk->ioac.rchar += amt;
2868 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2870 tsk->ioac.wchar += amt;
2873 static inline void inc_syscr(struct task_struct *tsk)
2878 static inline void inc_syscw(struct task_struct *tsk)
2883 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2887 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2891 static inline void inc_syscr(struct task_struct *tsk)
2895 static inline void inc_syscw(struct task_struct *tsk)
2900 #ifndef TASK_SIZE_OF
2901 #define TASK_SIZE_OF(tsk) TASK_SIZE
2904 #ifdef CONFIG_MM_OWNER
2905 extern void mm_update_next_owner(struct mm_struct *mm);
2906 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2908 static inline void mm_update_next_owner(struct mm_struct *mm)
2912 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2915 #endif /* CONFIG_MM_OWNER */
2917 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2920 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2923 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2926 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2929 static inline unsigned long rlimit(unsigned int limit)
2931 return task_rlimit(current, limit);
2934 static inline unsigned long rlimit_max(unsigned int limit)
2936 return task_rlimit_max(current, limit);