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 | \
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) {}
395 extern void set_dumpable(struct mm_struct *mm, int value);
396 extern int get_dumpable(struct mm_struct *mm);
398 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
399 #define SUID_DUMP_USER 1 /* Dump as user of process */
400 #define SUID_DUMP_ROOT 2 /* Dump as root */
404 #define MMF_DUMPABLE 0 /* core dump is permitted */
405 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
407 #define MMF_DUMPABLE_BITS 2
408 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
410 /* coredump filter bits */
411 #define MMF_DUMP_ANON_PRIVATE 2
412 #define MMF_DUMP_ANON_SHARED 3
413 #define MMF_DUMP_MAPPED_PRIVATE 4
414 #define MMF_DUMP_MAPPED_SHARED 5
415 #define MMF_DUMP_ELF_HEADERS 6
416 #define MMF_DUMP_HUGETLB_PRIVATE 7
417 #define MMF_DUMP_HUGETLB_SHARED 8
419 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
420 #define MMF_DUMP_FILTER_BITS 7
421 #define MMF_DUMP_FILTER_MASK \
422 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
423 #define MMF_DUMP_FILTER_DEFAULT \
424 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
425 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
427 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
428 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
430 # define MMF_DUMP_MASK_DEFAULT_ELF 0
432 /* leave room for more dump flags */
433 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
434 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
435 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
437 #define MMF_HAS_UPROBES 19 /* has uprobes */
438 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
440 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
442 struct sighand_struct {
444 struct k_sigaction action[_NSIG];
446 wait_queue_head_t signalfd_wqh;
449 struct pacct_struct {
452 unsigned long ac_mem;
453 cputime_t ac_utime, ac_stime;
454 unsigned long ac_minflt, ac_majflt;
465 * struct cputime - snaphsot of system and user cputime
466 * @utime: time spent in user mode
467 * @stime: time spent in system mode
469 * Gathers a generic snapshot of user and system time.
477 * struct task_cputime - collected CPU time counts
478 * @utime: time spent in user mode, in &cputime_t units
479 * @stime: time spent in kernel mode, in &cputime_t units
480 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
482 * This is an extension of struct cputime that includes the total runtime
483 * spent by the task from the scheduler point of view.
485 * As a result, this structure groups together three kinds of CPU time
486 * that are tracked for threads and thread groups. Most things considering
487 * CPU time want to group these counts together and treat all three
488 * of them in parallel.
490 struct task_cputime {
493 unsigned long long sum_exec_runtime;
495 /* Alternate field names when used to cache expirations. */
496 #define prof_exp stime
497 #define virt_exp utime
498 #define sched_exp sum_exec_runtime
500 #define INIT_CPUTIME \
501 (struct task_cputime) { \
504 .sum_exec_runtime = 0, \
507 #ifdef CONFIG_PREEMPT_COUNT
508 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
510 #define PREEMPT_DISABLED PREEMPT_ENABLED
514 * Disable preemption until the scheduler is running.
515 * Reset by start_kernel()->sched_init()->init_idle().
517 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
518 * before the scheduler is active -- see should_resched().
520 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
523 * struct thread_group_cputimer - thread group interval timer counts
524 * @cputime: thread group interval timers.
525 * @running: non-zero when there are timers running and
526 * @cputime receives updates.
527 * @lock: lock for fields in this struct.
529 * This structure contains the version of task_cputime, above, that is
530 * used for thread group CPU timer calculations.
532 struct thread_group_cputimer {
533 struct task_cputime cputime;
538 #include <linux/rwsem.h>
542 * NOTE! "signal_struct" does not have its own
543 * locking, because a shared signal_struct always
544 * implies a shared sighand_struct, so locking
545 * sighand_struct is always a proper superset of
546 * the locking of signal_struct.
548 struct signal_struct {
553 wait_queue_head_t wait_chldexit; /* for wait4() */
555 /* current thread group signal load-balancing target: */
556 struct task_struct *curr_target;
558 /* shared signal handling: */
559 struct sigpending shared_pending;
561 /* thread group exit support */
564 * - notify group_exit_task when ->count is equal to notify_count
565 * - everyone except group_exit_task is stopped during signal delivery
566 * of fatal signals, group_exit_task processes the signal.
569 struct task_struct *group_exit_task;
571 /* thread group stop support, overloads group_exit_code too */
572 int group_stop_count;
573 unsigned int flags; /* see SIGNAL_* flags below */
576 * PR_SET_CHILD_SUBREAPER marks a process, like a service
577 * manager, to re-parent orphan (double-forking) child processes
578 * to this process instead of 'init'. The service manager is
579 * able to receive SIGCHLD signals and is able to investigate
580 * the process until it calls wait(). All children of this
581 * process will inherit a flag if they should look for a
582 * child_subreaper process at exit.
584 unsigned int is_child_subreaper:1;
585 unsigned int has_child_subreaper:1;
587 /* POSIX.1b Interval Timers */
589 struct list_head posix_timers;
591 /* ITIMER_REAL timer for the process */
592 struct hrtimer real_timer;
593 struct pid *leader_pid;
594 ktime_t it_real_incr;
597 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
598 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
599 * values are defined to 0 and 1 respectively
601 struct cpu_itimer it[2];
604 * Thread group totals for process CPU timers.
605 * See thread_group_cputimer(), et al, for details.
607 struct thread_group_cputimer cputimer;
609 /* Earliest-expiration cache. */
610 struct task_cputime cputime_expires;
612 struct list_head cpu_timers[3];
614 struct pid *tty_old_pgrp;
616 /* boolean value for session group leader */
619 struct tty_struct *tty; /* NULL if no tty */
621 #ifdef CONFIG_SCHED_AUTOGROUP
622 struct autogroup *autogroup;
625 * Cumulative resource counters for dead threads in the group,
626 * and for reaped dead child processes forked by this group.
627 * Live threads maintain their own counters and add to these
628 * in __exit_signal, except for the group leader.
630 cputime_t utime, stime, cutime, cstime;
633 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
634 struct cputime prev_cputime;
636 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
637 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
638 unsigned long inblock, oublock, cinblock, coublock;
639 unsigned long maxrss, cmaxrss;
640 struct task_io_accounting ioac;
643 * Cumulative ns of schedule CPU time fo dead threads in the
644 * group, not including a zombie group leader, (This only differs
645 * from jiffies_to_ns(utime + stime) if sched_clock uses something
646 * other than jiffies.)
648 unsigned long long sum_sched_runtime;
651 * We don't bother to synchronize most readers of this at all,
652 * because there is no reader checking a limit that actually needs
653 * to get both rlim_cur and rlim_max atomically, and either one
654 * alone is a single word that can safely be read normally.
655 * getrlimit/setrlimit use task_lock(current->group_leader) to
656 * protect this instead of the siglock, because they really
657 * have no need to disable irqs.
659 struct rlimit rlim[RLIM_NLIMITS];
661 #ifdef CONFIG_BSD_PROCESS_ACCT
662 struct pacct_struct pacct; /* per-process accounting information */
664 #ifdef CONFIG_TASKSTATS
665 struct taskstats *stats;
669 unsigned audit_tty_log_passwd;
670 struct tty_audit_buf *tty_audit_buf;
672 #ifdef CONFIG_CGROUPS
674 * group_rwsem prevents new tasks from entering the threadgroup and
675 * member tasks from exiting,a more specifically, setting of
676 * PF_EXITING. fork and exit paths are protected with this rwsem
677 * using threadgroup_change_begin/end(). Users which require
678 * threadgroup to remain stable should use threadgroup_[un]lock()
679 * which also takes care of exec path. Currently, cgroup is the
682 struct rw_semaphore group_rwsem;
685 oom_flags_t oom_flags;
686 short oom_score_adj; /* OOM kill score adjustment */
687 short oom_score_adj_min; /* OOM kill score adjustment min value.
688 * Only settable by CAP_SYS_RESOURCE. */
690 struct mutex cred_guard_mutex; /* guard against foreign influences on
691 * credential calculations
692 * (notably. ptrace) */
696 * Bits in flags field of signal_struct.
698 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
699 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
700 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
701 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
703 * Pending notifications to parent.
705 #define SIGNAL_CLD_STOPPED 0x00000010
706 #define SIGNAL_CLD_CONTINUED 0x00000020
707 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
709 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
711 /* If true, all threads except ->group_exit_task have pending SIGKILL */
712 static inline int signal_group_exit(const struct signal_struct *sig)
714 return (sig->flags & SIGNAL_GROUP_EXIT) ||
715 (sig->group_exit_task != NULL);
719 * Some day this will be a full-fledged user tracking system..
722 atomic_t __count; /* reference count */
723 atomic_t processes; /* How many processes does this user have? */
724 atomic_t files; /* How many open files does this user have? */
725 atomic_t sigpending; /* How many pending signals does this user have? */
726 #ifdef CONFIG_INOTIFY_USER
727 atomic_t inotify_watches; /* How many inotify watches does this user have? */
728 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
730 #ifdef CONFIG_FANOTIFY
731 atomic_t fanotify_listeners;
734 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
736 #ifdef CONFIG_POSIX_MQUEUE
737 /* protected by mq_lock */
738 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
740 unsigned long locked_shm; /* How many pages of mlocked shm ? */
743 struct key *uid_keyring; /* UID specific keyring */
744 struct key *session_keyring; /* UID's default session keyring */
747 /* Hash table maintenance information */
748 struct hlist_node uidhash_node;
751 #ifdef CONFIG_PERF_EVENTS
752 atomic_long_t locked_vm;
756 extern int uids_sysfs_init(void);
758 extern struct user_struct *find_user(kuid_t);
760 extern struct user_struct root_user;
761 #define INIT_USER (&root_user)
764 struct backing_dev_info;
765 struct reclaim_state;
767 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
769 /* cumulative counters */
770 unsigned long pcount; /* # of times run on this cpu */
771 unsigned long long run_delay; /* time spent waiting on a runqueue */
774 unsigned long long last_arrival,/* when we last ran on a cpu */
775 last_queued; /* when we were last queued to run */
777 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
779 #ifdef CONFIG_TASK_DELAY_ACCT
780 struct task_delay_info {
782 unsigned int flags; /* Private per-task flags */
784 /* For each stat XXX, add following, aligned appropriately
786 * struct timespec XXX_start, XXX_end;
790 * Atomicity of updates to XXX_delay, XXX_count protected by
791 * single lock above (split into XXX_lock if contention is an issue).
795 * XXX_count is incremented on every XXX operation, the delay
796 * associated with the operation is added to XXX_delay.
797 * XXX_delay contains the accumulated delay time in nanoseconds.
799 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
800 u64 blkio_delay; /* wait for sync block io completion */
801 u64 swapin_delay; /* wait for swapin block io completion */
802 u32 blkio_count; /* total count of the number of sync block */
803 /* io operations performed */
804 u32 swapin_count; /* total count of the number of swapin block */
805 /* io operations performed */
807 struct timespec freepages_start, freepages_end;
808 u64 freepages_delay; /* wait for memory reclaim */
809 u32 freepages_count; /* total count of memory reclaim */
811 #endif /* CONFIG_TASK_DELAY_ACCT */
813 static inline int sched_info_on(void)
815 #ifdef CONFIG_SCHEDSTATS
817 #elif defined(CONFIG_TASK_DELAY_ACCT)
818 extern int delayacct_on;
833 * Increase resolution of cpu_power calculations
835 #define SCHED_POWER_SHIFT 10
836 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
839 * sched-domains (multiprocessor balancing) declarations:
842 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
843 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
844 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
845 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
846 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
847 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
848 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
849 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
850 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
851 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
852 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
853 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
854 #define SD_NUMA 0x4000 /* cross-node balancing */
856 extern int __weak arch_sd_sibiling_asym_packing(void);
858 struct sched_domain_attr {
859 int relax_domain_level;
862 #define SD_ATTR_INIT (struct sched_domain_attr) { \
863 .relax_domain_level = -1, \
866 extern int sched_domain_level_max;
870 struct sched_domain {
871 /* These fields must be setup */
872 struct sched_domain *parent; /* top domain must be null terminated */
873 struct sched_domain *child; /* bottom domain must be null terminated */
874 struct sched_group *groups; /* the balancing groups of the domain */
875 unsigned long min_interval; /* Minimum balance interval ms */
876 unsigned long max_interval; /* Maximum balance interval ms */
877 unsigned int busy_factor; /* less balancing by factor if busy */
878 unsigned int imbalance_pct; /* No balance until over watermark */
879 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
880 unsigned int busy_idx;
881 unsigned int idle_idx;
882 unsigned int newidle_idx;
883 unsigned int wake_idx;
884 unsigned int forkexec_idx;
885 unsigned int smt_gain;
887 int nohz_idle; /* NOHZ IDLE status */
888 int flags; /* See SD_* */
891 /* Runtime fields. */
892 unsigned long last_balance; /* init to jiffies. units in jiffies */
893 unsigned int balance_interval; /* initialise to 1. units in ms. */
894 unsigned int nr_balance_failed; /* initialise to 0 */
896 /* idle_balance() stats */
897 u64 max_newidle_lb_cost;
898 unsigned long next_decay_max_lb_cost;
900 #ifdef CONFIG_SCHEDSTATS
901 /* load_balance() stats */
902 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
903 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
904 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
905 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
906 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
907 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
908 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
909 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
911 /* Active load balancing */
912 unsigned int alb_count;
913 unsigned int alb_failed;
914 unsigned int alb_pushed;
916 /* SD_BALANCE_EXEC stats */
917 unsigned int sbe_count;
918 unsigned int sbe_balanced;
919 unsigned int sbe_pushed;
921 /* SD_BALANCE_FORK stats */
922 unsigned int sbf_count;
923 unsigned int sbf_balanced;
924 unsigned int sbf_pushed;
926 /* try_to_wake_up() stats */
927 unsigned int ttwu_wake_remote;
928 unsigned int ttwu_move_affine;
929 unsigned int ttwu_move_balance;
931 #ifdef CONFIG_SCHED_DEBUG
935 void *private; /* used during construction */
936 struct rcu_head rcu; /* used during destruction */
939 unsigned int span_weight;
941 * Span of all CPUs in this domain.
943 * NOTE: this field is variable length. (Allocated dynamically
944 * by attaching extra space to the end of the structure,
945 * depending on how many CPUs the kernel has booted up with)
947 unsigned long span[0];
950 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
952 return to_cpumask(sd->span);
955 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
956 struct sched_domain_attr *dattr_new);
958 /* Allocate an array of sched domains, for partition_sched_domains(). */
959 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
960 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
962 bool cpus_share_cache(int this_cpu, int that_cpu);
964 #else /* CONFIG_SMP */
966 struct sched_domain_attr;
969 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
970 struct sched_domain_attr *dattr_new)
974 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
979 #endif /* !CONFIG_SMP */
982 struct io_context; /* See blkdev.h */
985 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
986 extern void prefetch_stack(struct task_struct *t);
988 static inline void prefetch_stack(struct task_struct *t) { }
991 struct audit_context; /* See audit.c */
993 struct pipe_inode_info;
994 struct uts_namespace;
997 unsigned long weight;
1003 * These sums represent an infinite geometric series and so are bound
1004 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1005 * choices of y < 1-2^(-32)*1024.
1007 u32 runnable_avg_sum, runnable_avg_period;
1008 u64 last_runnable_update;
1010 unsigned long load_avg_contrib;
1013 #ifdef CONFIG_SCHEDSTATS
1014 struct sched_statistics {
1024 s64 sum_sleep_runtime;
1031 u64 nr_migrations_cold;
1032 u64 nr_failed_migrations_affine;
1033 u64 nr_failed_migrations_running;
1034 u64 nr_failed_migrations_hot;
1035 u64 nr_forced_migrations;
1038 u64 nr_wakeups_sync;
1039 u64 nr_wakeups_migrate;
1040 u64 nr_wakeups_local;
1041 u64 nr_wakeups_remote;
1042 u64 nr_wakeups_affine;
1043 u64 nr_wakeups_affine_attempts;
1044 u64 nr_wakeups_passive;
1045 u64 nr_wakeups_idle;
1049 struct sched_entity {
1050 struct load_weight load; /* for load-balancing */
1051 struct rb_node run_node;
1052 struct list_head group_node;
1056 u64 sum_exec_runtime;
1058 u64 prev_sum_exec_runtime;
1062 #ifdef CONFIG_SCHEDSTATS
1063 struct sched_statistics statistics;
1066 #ifdef CONFIG_FAIR_GROUP_SCHED
1067 struct sched_entity *parent;
1068 /* rq on which this entity is (to be) queued: */
1069 struct cfs_rq *cfs_rq;
1070 /* rq "owned" by this entity/group: */
1071 struct cfs_rq *my_q;
1075 /* Per-entity load-tracking */
1076 struct sched_avg avg;
1080 struct sched_rt_entity {
1081 struct list_head run_list;
1082 unsigned long timeout;
1083 unsigned long watchdog_stamp;
1084 unsigned int time_slice;
1086 struct sched_rt_entity *back;
1087 #ifdef CONFIG_RT_GROUP_SCHED
1088 struct sched_rt_entity *parent;
1089 /* rq on which this entity is (to be) queued: */
1090 struct rt_rq *rt_rq;
1091 /* rq "owned" by this entity/group: */
1096 struct sched_dl_entity {
1097 struct rb_node rb_node;
1100 * Original scheduling parameters. Copied here from sched_attr
1101 * during sched_setscheduler2(), they will remain the same until
1102 * the next sched_setscheduler2().
1104 u64 dl_runtime; /* maximum runtime for each instance */
1105 u64 dl_deadline; /* relative deadline of each instance */
1106 u64 dl_period; /* separation of two instances (period) */
1107 u64 dl_bw; /* dl_runtime / dl_deadline */
1110 * Actual scheduling parameters. Initialized with the values above,
1111 * they are continously updated during task execution. Note that
1112 * the remaining runtime could be < 0 in case we are in overrun.
1114 s64 runtime; /* remaining runtime for this instance */
1115 u64 deadline; /* absolute deadline for this instance */
1116 unsigned int flags; /* specifying the scheduler behaviour */
1121 * @dl_throttled tells if we exhausted the runtime. If so, the
1122 * task has to wait for a replenishment to be performed at the
1123 * next firing of dl_timer.
1125 * @dl_new tells if a new instance arrived. If so we must
1126 * start executing it with full runtime and reset its absolute
1129 * @dl_boosted tells if we are boosted due to DI. If so we are
1130 * outside bandwidth enforcement mechanism (but only until we
1131 * exit the critical section).
1133 int dl_throttled, dl_new, dl_boosted;
1136 * Bandwidth enforcement timer. Each -deadline task has its
1137 * own bandwidth to be enforced, thus we need one timer per task.
1139 struct hrtimer dl_timer;
1144 enum perf_event_task_context {
1145 perf_invalid_context = -1,
1146 perf_hw_context = 0,
1148 perf_nr_task_contexts,
1151 struct task_struct {
1152 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1155 unsigned int flags; /* per process flags, defined below */
1156 unsigned int ptrace;
1159 struct llist_node wake_entry;
1161 struct task_struct *last_wakee;
1162 unsigned long wakee_flips;
1163 unsigned long wakee_flip_decay_ts;
1169 int prio, static_prio, normal_prio;
1170 unsigned int rt_priority;
1171 const struct sched_class *sched_class;
1172 struct sched_entity se;
1173 struct sched_rt_entity rt;
1174 #ifdef CONFIG_CGROUP_SCHED
1175 struct task_group *sched_task_group;
1177 struct sched_dl_entity dl;
1179 #ifdef CONFIG_PREEMPT_NOTIFIERS
1180 /* list of struct preempt_notifier: */
1181 struct hlist_head preempt_notifiers;
1184 #ifdef CONFIG_BLK_DEV_IO_TRACE
1185 unsigned int btrace_seq;
1188 unsigned int policy;
1189 int nr_cpus_allowed;
1190 cpumask_t cpus_allowed;
1192 #ifdef CONFIG_PREEMPT_RCU
1193 int rcu_read_lock_nesting;
1194 char rcu_read_unlock_special;
1195 struct list_head rcu_node_entry;
1196 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1197 #ifdef CONFIG_TREE_PREEMPT_RCU
1198 struct rcu_node *rcu_blocked_node;
1199 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1200 #ifdef CONFIG_RCU_BOOST
1201 struct rt_mutex *rcu_boost_mutex;
1202 #endif /* #ifdef CONFIG_RCU_BOOST */
1204 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1205 struct sched_info sched_info;
1208 struct list_head tasks;
1210 struct plist_node pushable_tasks;
1211 struct rb_node pushable_dl_tasks;
1214 struct mm_struct *mm, *active_mm;
1215 #ifdef CONFIG_COMPAT_BRK
1216 unsigned brk_randomized:1;
1218 #if defined(SPLIT_RSS_COUNTING)
1219 struct task_rss_stat rss_stat;
1223 int exit_code, exit_signal;
1224 int pdeath_signal; /* The signal sent when the parent dies */
1225 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1227 /* Used for emulating ABI behavior of previous Linux versions */
1228 unsigned int personality;
1230 unsigned did_exec:1;
1231 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1233 unsigned in_iowait:1;
1235 /* task may not gain privileges */
1236 unsigned no_new_privs:1;
1238 /* Revert to default priority/policy when forking */
1239 unsigned sched_reset_on_fork:1;
1240 unsigned sched_contributes_to_load:1;
1245 #ifdef CONFIG_CC_STACKPROTECTOR
1246 /* Canary value for the -fstack-protector gcc feature */
1247 unsigned long stack_canary;
1250 * pointers to (original) parent process, youngest child, younger sibling,
1251 * older sibling, respectively. (p->father can be replaced with
1252 * p->real_parent->pid)
1254 struct task_struct __rcu *real_parent; /* real parent process */
1255 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1257 * children/sibling forms the list of my natural children
1259 struct list_head children; /* list of my children */
1260 struct list_head sibling; /* linkage in my parent's children list */
1261 struct task_struct *group_leader; /* threadgroup leader */
1264 * ptraced is the list of tasks this task is using ptrace on.
1265 * This includes both natural children and PTRACE_ATTACH targets.
1266 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1268 struct list_head ptraced;
1269 struct list_head ptrace_entry;
1271 /* PID/PID hash table linkage. */
1272 struct pid_link pids[PIDTYPE_MAX];
1273 struct list_head thread_group;
1275 struct completion *vfork_done; /* for vfork() */
1276 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1277 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1279 cputime_t utime, stime, utimescaled, stimescaled;
1281 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1282 struct cputime prev_cputime;
1284 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1285 seqlock_t vtime_seqlock;
1286 unsigned long long vtime_snap;
1291 } vtime_snap_whence;
1293 unsigned long nvcsw, nivcsw; /* context switch counts */
1294 struct timespec start_time; /* monotonic time */
1295 struct timespec real_start_time; /* boot based time */
1296 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1297 unsigned long min_flt, maj_flt;
1299 struct task_cputime cputime_expires;
1300 struct list_head cpu_timers[3];
1302 /* process credentials */
1303 const struct cred __rcu *real_cred; /* objective and real subjective task
1304 * credentials (COW) */
1305 const struct cred __rcu *cred; /* effective (overridable) subjective task
1306 * credentials (COW) */
1307 char comm[TASK_COMM_LEN]; /* executable name excluding path
1308 - access with [gs]et_task_comm (which lock
1309 it with task_lock())
1310 - initialized normally by setup_new_exec */
1311 /* file system info */
1312 int link_count, total_link_count;
1313 #ifdef CONFIG_SYSVIPC
1315 struct sysv_sem sysvsem;
1317 #ifdef CONFIG_DETECT_HUNG_TASK
1318 /* hung task detection */
1319 unsigned long last_switch_count;
1321 /* CPU-specific state of this task */
1322 struct thread_struct thread;
1323 /* filesystem information */
1324 struct fs_struct *fs;
1325 /* open file information */
1326 struct files_struct *files;
1328 struct nsproxy *nsproxy;
1329 /* signal handlers */
1330 struct signal_struct *signal;
1331 struct sighand_struct *sighand;
1333 sigset_t blocked, real_blocked;
1334 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1335 struct sigpending pending;
1337 unsigned long sas_ss_sp;
1339 int (*notifier)(void *priv);
1340 void *notifier_data;
1341 sigset_t *notifier_mask;
1342 struct callback_head *task_works;
1344 struct audit_context *audit_context;
1345 #ifdef CONFIG_AUDITSYSCALL
1347 unsigned int sessionid;
1349 struct seccomp seccomp;
1351 /* Thread group tracking */
1354 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1356 spinlock_t alloc_lock;
1358 /* Protection of the PI data structures: */
1359 raw_spinlock_t pi_lock;
1361 #ifdef CONFIG_RT_MUTEXES
1362 /* PI waiters blocked on a rt_mutex held by this task */
1363 struct rb_root pi_waiters;
1364 struct rb_node *pi_waiters_leftmost;
1365 /* Deadlock detection and priority inheritance handling */
1366 struct rt_mutex_waiter *pi_blocked_on;
1367 /* Top pi_waiters task */
1368 struct task_struct *pi_top_task;
1371 #ifdef CONFIG_DEBUG_MUTEXES
1372 /* mutex deadlock detection */
1373 struct mutex_waiter *blocked_on;
1375 #ifdef CONFIG_TRACE_IRQFLAGS
1376 unsigned int irq_events;
1377 unsigned long hardirq_enable_ip;
1378 unsigned long hardirq_disable_ip;
1379 unsigned int hardirq_enable_event;
1380 unsigned int hardirq_disable_event;
1381 int hardirqs_enabled;
1382 int hardirq_context;
1383 unsigned long softirq_disable_ip;
1384 unsigned long softirq_enable_ip;
1385 unsigned int softirq_disable_event;
1386 unsigned int softirq_enable_event;
1387 int softirqs_enabled;
1388 int softirq_context;
1390 #ifdef CONFIG_LOCKDEP
1391 # define MAX_LOCK_DEPTH 48UL
1394 unsigned int lockdep_recursion;
1395 struct held_lock held_locks[MAX_LOCK_DEPTH];
1396 gfp_t lockdep_reclaim_gfp;
1399 /* journalling filesystem info */
1402 /* stacked block device info */
1403 struct bio_list *bio_list;
1406 /* stack plugging */
1407 struct blk_plug *plug;
1411 struct reclaim_state *reclaim_state;
1413 struct backing_dev_info *backing_dev_info;
1415 struct io_context *io_context;
1417 unsigned long ptrace_message;
1418 siginfo_t *last_siginfo; /* For ptrace use. */
1419 struct task_io_accounting ioac;
1420 #if defined(CONFIG_TASK_XACCT)
1421 u64 acct_rss_mem1; /* accumulated rss usage */
1422 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1423 cputime_t acct_timexpd; /* stime + utime since last update */
1425 #ifdef CONFIG_CPUSETS
1426 nodemask_t mems_allowed; /* Protected by alloc_lock */
1427 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1428 int cpuset_mem_spread_rotor;
1429 int cpuset_slab_spread_rotor;
1431 #ifdef CONFIG_CGROUPS
1432 /* Control Group info protected by css_set_lock */
1433 struct css_set __rcu *cgroups;
1434 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1435 struct list_head cg_list;
1438 struct robust_list_head __user *robust_list;
1439 #ifdef CONFIG_COMPAT
1440 struct compat_robust_list_head __user *compat_robust_list;
1442 struct list_head pi_state_list;
1443 struct futex_pi_state *pi_state_cache;
1445 #ifdef CONFIG_PERF_EVENTS
1446 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1447 struct mutex perf_event_mutex;
1448 struct list_head perf_event_list;
1451 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1453 short pref_node_fork;
1455 #ifdef CONFIG_NUMA_BALANCING
1457 unsigned int numa_scan_period;
1458 unsigned int numa_scan_period_max;
1459 int numa_preferred_nid;
1460 unsigned long numa_migrate_retry;
1461 u64 node_stamp; /* migration stamp */
1462 struct callback_head numa_work;
1464 struct list_head numa_entry;
1465 struct numa_group *numa_group;
1468 * Exponential decaying average of faults on a per-node basis.
1469 * Scheduling placement decisions are made based on the these counts.
1470 * The values remain static for the duration of a PTE scan
1472 unsigned long *numa_faults_memory;
1473 unsigned long total_numa_faults;
1476 * numa_faults_buffer records faults per node during the current
1477 * scan window. When the scan completes, the counts in
1478 * numa_faults_memory decay and these values are copied.
1480 unsigned long *numa_faults_buffer_memory;
1483 * Track the nodes the process was running on when a NUMA hinting
1484 * fault was incurred.
1486 unsigned long *numa_faults_cpu;
1487 unsigned long *numa_faults_buffer_cpu;
1490 * numa_faults_locality tracks if faults recorded during the last
1491 * scan window were remote/local. The task scan period is adapted
1492 * based on the locality of the faults with different weights
1493 * depending on whether they were shared or private faults
1495 unsigned long numa_faults_locality[2];
1497 unsigned long numa_pages_migrated;
1498 #endif /* CONFIG_NUMA_BALANCING */
1500 struct rcu_head rcu;
1503 * cache last used pipe for splice
1505 struct pipe_inode_info *splice_pipe;
1507 struct page_frag task_frag;
1509 #ifdef CONFIG_TASK_DELAY_ACCT
1510 struct task_delay_info *delays;
1512 #ifdef CONFIG_FAULT_INJECTION
1516 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1517 * balance_dirty_pages() for some dirty throttling pause
1520 int nr_dirtied_pause;
1521 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1523 #ifdef CONFIG_LATENCYTOP
1524 int latency_record_count;
1525 struct latency_record latency_record[LT_SAVECOUNT];
1528 * time slack values; these are used to round up poll() and
1529 * select() etc timeout values. These are in nanoseconds.
1531 unsigned long timer_slack_ns;
1532 unsigned long default_timer_slack_ns;
1534 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1535 /* Index of current stored address in ret_stack */
1537 /* Stack of return addresses for return function tracing */
1538 struct ftrace_ret_stack *ret_stack;
1539 /* time stamp for last schedule */
1540 unsigned long long ftrace_timestamp;
1542 * Number of functions that haven't been traced
1543 * because of depth overrun.
1545 atomic_t trace_overrun;
1546 /* Pause for the tracing */
1547 atomic_t tracing_graph_pause;
1549 #ifdef CONFIG_TRACING
1550 /* state flags for use by tracers */
1551 unsigned long trace;
1552 /* bitmask and counter of trace recursion */
1553 unsigned long trace_recursion;
1554 #endif /* CONFIG_TRACING */
1555 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1556 struct memcg_batch_info {
1557 int do_batch; /* incremented when batch uncharge started */
1558 struct mem_cgroup *memcg; /* target memcg of uncharge */
1559 unsigned long nr_pages; /* uncharged usage */
1560 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1562 unsigned int memcg_kmem_skip_account;
1563 struct memcg_oom_info {
1564 struct mem_cgroup *memcg;
1567 unsigned int may_oom:1;
1570 #ifdef CONFIG_UPROBES
1571 struct uprobe_task *utask;
1573 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1574 unsigned int sequential_io;
1575 unsigned int sequential_io_avg;
1579 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1580 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1582 #define TNF_MIGRATED 0x01
1583 #define TNF_NO_GROUP 0x02
1584 #define TNF_SHARED 0x04
1585 #define TNF_FAULT_LOCAL 0x08
1587 #ifdef CONFIG_NUMA_BALANCING
1588 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1589 extern pid_t task_numa_group_id(struct task_struct *p);
1590 extern void set_numabalancing_state(bool enabled);
1591 extern void task_numa_free(struct task_struct *p);
1593 static inline void task_numa_fault(int last_node, int node, int pages,
1597 static inline pid_t task_numa_group_id(struct task_struct *p)
1601 static inline void set_numabalancing_state(bool enabled)
1604 static inline void task_numa_free(struct task_struct *p)
1609 static inline struct pid *task_pid(struct task_struct *task)
1611 return task->pids[PIDTYPE_PID].pid;
1614 static inline struct pid *task_tgid(struct task_struct *task)
1616 return task->group_leader->pids[PIDTYPE_PID].pid;
1620 * Without tasklist or rcu lock it is not safe to dereference
1621 * the result of task_pgrp/task_session even if task == current,
1622 * we can race with another thread doing sys_setsid/sys_setpgid.
1624 static inline struct pid *task_pgrp(struct task_struct *task)
1626 return task->group_leader->pids[PIDTYPE_PGID].pid;
1629 static inline struct pid *task_session(struct task_struct *task)
1631 return task->group_leader->pids[PIDTYPE_SID].pid;
1634 struct pid_namespace;
1637 * the helpers to get the task's different pids as they are seen
1638 * from various namespaces
1640 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1641 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1643 * task_xid_nr_ns() : id seen from the ns specified;
1645 * set_task_vxid() : assigns a virtual id to a task;
1647 * see also pid_nr() etc in include/linux/pid.h
1649 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1650 struct pid_namespace *ns);
1652 static inline pid_t task_pid_nr(struct task_struct *tsk)
1657 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1658 struct pid_namespace *ns)
1660 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1663 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1665 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1669 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1674 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1676 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1678 return pid_vnr(task_tgid(tsk));
1682 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1683 struct pid_namespace *ns)
1685 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1688 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1690 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1694 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1695 struct pid_namespace *ns)
1697 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1700 static inline pid_t task_session_vnr(struct task_struct *tsk)
1702 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1705 /* obsolete, do not use */
1706 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1708 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1712 * pid_alive - check that a task structure is not stale
1713 * @p: Task structure to be checked.
1715 * Test if a process is not yet dead (at most zombie state)
1716 * If pid_alive fails, then pointers within the task structure
1717 * can be stale and must not be dereferenced.
1719 * Return: 1 if the process is alive. 0 otherwise.
1721 static inline int pid_alive(struct task_struct *p)
1723 return p->pids[PIDTYPE_PID].pid != NULL;
1727 * is_global_init - check if a task structure is init
1728 * @tsk: Task structure to be checked.
1730 * Check if a task structure is the first user space task the kernel created.
1732 * Return: 1 if the task structure is init. 0 otherwise.
1734 static inline int is_global_init(struct task_struct *tsk)
1736 return tsk->pid == 1;
1739 extern struct pid *cad_pid;
1741 extern void free_task(struct task_struct *tsk);
1742 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1744 extern void __put_task_struct(struct task_struct *t);
1746 static inline void put_task_struct(struct task_struct *t)
1748 if (atomic_dec_and_test(&t->usage))
1749 __put_task_struct(t);
1752 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1753 extern void task_cputime(struct task_struct *t,
1754 cputime_t *utime, cputime_t *stime);
1755 extern void task_cputime_scaled(struct task_struct *t,
1756 cputime_t *utimescaled, cputime_t *stimescaled);
1757 extern cputime_t task_gtime(struct task_struct *t);
1759 static inline void task_cputime(struct task_struct *t,
1760 cputime_t *utime, cputime_t *stime)
1768 static inline void task_cputime_scaled(struct task_struct *t,
1769 cputime_t *utimescaled,
1770 cputime_t *stimescaled)
1773 *utimescaled = t->utimescaled;
1775 *stimescaled = t->stimescaled;
1778 static inline cputime_t task_gtime(struct task_struct *t)
1783 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1784 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1789 #define PF_EXITING 0x00000004 /* getting shut down */
1790 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1791 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1792 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1793 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1794 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1795 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1796 #define PF_DUMPCORE 0x00000200 /* dumped core */
1797 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1798 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1799 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1800 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1801 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1802 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1803 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1804 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1805 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1806 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1807 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1808 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1809 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1810 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1811 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1812 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1813 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1814 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1815 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1816 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1817 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1818 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1821 * Only the _current_ task can read/write to tsk->flags, but other
1822 * tasks can access tsk->flags in readonly mode for example
1823 * with tsk_used_math (like during threaded core dumping).
1824 * There is however an exception to this rule during ptrace
1825 * or during fork: the ptracer task is allowed to write to the
1826 * child->flags of its traced child (same goes for fork, the parent
1827 * can write to the child->flags), because we're guaranteed the
1828 * child is not running and in turn not changing child->flags
1829 * at the same time the parent does it.
1831 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1832 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1833 #define clear_used_math() clear_stopped_child_used_math(current)
1834 #define set_used_math() set_stopped_child_used_math(current)
1835 #define conditional_stopped_child_used_math(condition, child) \
1836 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1837 #define conditional_used_math(condition) \
1838 conditional_stopped_child_used_math(condition, current)
1839 #define copy_to_stopped_child_used_math(child) \
1840 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1841 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1842 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1843 #define used_math() tsk_used_math(current)
1845 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1846 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1848 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1853 static inline unsigned int memalloc_noio_save(void)
1855 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1856 current->flags |= PF_MEMALLOC_NOIO;
1860 static inline void memalloc_noio_restore(unsigned int flags)
1862 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1866 * task->jobctl flags
1868 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1870 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1871 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1872 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1873 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1874 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1875 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1876 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1878 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1879 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1880 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1881 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1882 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1883 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1884 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1886 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1887 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1889 extern bool task_set_jobctl_pending(struct task_struct *task,
1891 extern void task_clear_jobctl_trapping(struct task_struct *task);
1892 extern void task_clear_jobctl_pending(struct task_struct *task,
1895 #ifdef CONFIG_PREEMPT_RCU
1897 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1898 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1900 static inline void rcu_copy_process(struct task_struct *p)
1902 p->rcu_read_lock_nesting = 0;
1903 p->rcu_read_unlock_special = 0;
1904 #ifdef CONFIG_TREE_PREEMPT_RCU
1905 p->rcu_blocked_node = NULL;
1906 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1907 #ifdef CONFIG_RCU_BOOST
1908 p->rcu_boost_mutex = NULL;
1909 #endif /* #ifdef CONFIG_RCU_BOOST */
1910 INIT_LIST_HEAD(&p->rcu_node_entry);
1915 static inline void rcu_copy_process(struct task_struct *p)
1921 static inline void tsk_restore_flags(struct task_struct *task,
1922 unsigned long orig_flags, unsigned long flags)
1924 task->flags &= ~flags;
1925 task->flags |= orig_flags & flags;
1929 extern void do_set_cpus_allowed(struct task_struct *p,
1930 const struct cpumask *new_mask);
1932 extern int set_cpus_allowed_ptr(struct task_struct *p,
1933 const struct cpumask *new_mask);
1935 static inline void do_set_cpus_allowed(struct task_struct *p,
1936 const struct cpumask *new_mask)
1939 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1940 const struct cpumask *new_mask)
1942 if (!cpumask_test_cpu(0, new_mask))
1948 #ifdef CONFIG_NO_HZ_COMMON
1949 void calc_load_enter_idle(void);
1950 void calc_load_exit_idle(void);
1952 static inline void calc_load_enter_idle(void) { }
1953 static inline void calc_load_exit_idle(void) { }
1954 #endif /* CONFIG_NO_HZ_COMMON */
1956 #ifndef CONFIG_CPUMASK_OFFSTACK
1957 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1959 return set_cpus_allowed_ptr(p, &new_mask);
1964 * Do not use outside of architecture code which knows its limitations.
1966 * sched_clock() has no promise of monotonicity or bounded drift between
1967 * CPUs, use (which you should not) requires disabling IRQs.
1969 * Please use one of the three interfaces below.
1971 extern unsigned long long notrace sched_clock(void);
1973 * See the comment in kernel/sched/clock.c
1975 extern u64 cpu_clock(int cpu);
1976 extern u64 local_clock(void);
1977 extern u64 sched_clock_cpu(int cpu);
1980 extern void sched_clock_init(void);
1982 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1983 static inline void sched_clock_tick(void)
1987 static inline void sched_clock_idle_sleep_event(void)
1991 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1996 * Architectures can set this to 1 if they have specified
1997 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1998 * but then during bootup it turns out that sched_clock()
1999 * is reliable after all:
2001 extern int sched_clock_stable(void);
2002 extern void set_sched_clock_stable(void);
2003 extern void clear_sched_clock_stable(void);
2005 extern void sched_clock_tick(void);
2006 extern void sched_clock_idle_sleep_event(void);
2007 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2010 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2012 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2013 * The reason for this explicit opt-in is not to have perf penalty with
2014 * slow sched_clocks.
2016 extern void enable_sched_clock_irqtime(void);
2017 extern void disable_sched_clock_irqtime(void);
2019 static inline void enable_sched_clock_irqtime(void) {}
2020 static inline void disable_sched_clock_irqtime(void) {}
2023 extern unsigned long long
2024 task_sched_runtime(struct task_struct *task);
2026 /* sched_exec is called by processes performing an exec */
2028 extern void sched_exec(void);
2030 #define sched_exec() {}
2033 extern void sched_clock_idle_sleep_event(void);
2034 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2036 #ifdef CONFIG_HOTPLUG_CPU
2037 extern void idle_task_exit(void);
2039 static inline void idle_task_exit(void) {}
2042 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2043 extern void wake_up_nohz_cpu(int cpu);
2045 static inline void wake_up_nohz_cpu(int cpu) { }
2048 #ifdef CONFIG_NO_HZ_FULL
2049 extern bool sched_can_stop_tick(void);
2050 extern u64 scheduler_tick_max_deferment(void);
2052 static inline bool sched_can_stop_tick(void) { return false; }
2055 #ifdef CONFIG_SCHED_AUTOGROUP
2056 extern void sched_autogroup_create_attach(struct task_struct *p);
2057 extern void sched_autogroup_detach(struct task_struct *p);
2058 extern void sched_autogroup_fork(struct signal_struct *sig);
2059 extern void sched_autogroup_exit(struct signal_struct *sig);
2060 #ifdef CONFIG_PROC_FS
2061 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2062 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2065 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2066 static inline void sched_autogroup_detach(struct task_struct *p) { }
2067 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2068 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2071 extern bool yield_to(struct task_struct *p, bool preempt);
2072 extern void set_user_nice(struct task_struct *p, long nice);
2073 extern int task_prio(const struct task_struct *p);
2074 extern int task_nice(const struct task_struct *p);
2075 extern int can_nice(const struct task_struct *p, const int nice);
2076 extern int task_curr(const struct task_struct *p);
2077 extern int idle_cpu(int cpu);
2078 extern int sched_setscheduler(struct task_struct *, int,
2079 const struct sched_param *);
2080 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2081 const struct sched_param *);
2082 extern int sched_setattr(struct task_struct *,
2083 const struct sched_attr *);
2084 extern struct task_struct *idle_task(int cpu);
2086 * is_idle_task - is the specified task an idle task?
2087 * @p: the task in question.
2089 * Return: 1 if @p is an idle task. 0 otherwise.
2091 static inline bool is_idle_task(const struct task_struct *p)
2095 extern struct task_struct *curr_task(int cpu);
2096 extern void set_curr_task(int cpu, struct task_struct *p);
2101 * The default (Linux) execution domain.
2103 extern struct exec_domain default_exec_domain;
2105 union thread_union {
2106 struct thread_info thread_info;
2107 unsigned long stack[THREAD_SIZE/sizeof(long)];
2110 #ifndef __HAVE_ARCH_KSTACK_END
2111 static inline int kstack_end(void *addr)
2113 /* Reliable end of stack detection:
2114 * Some APM bios versions misalign the stack
2116 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2120 extern union thread_union init_thread_union;
2121 extern struct task_struct init_task;
2123 extern struct mm_struct init_mm;
2125 extern struct pid_namespace init_pid_ns;
2128 * find a task by one of its numerical ids
2130 * find_task_by_pid_ns():
2131 * finds a task by its pid in the specified namespace
2132 * find_task_by_vpid():
2133 * finds a task by its virtual pid
2135 * see also find_vpid() etc in include/linux/pid.h
2138 extern struct task_struct *find_task_by_vpid(pid_t nr);
2139 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2140 struct pid_namespace *ns);
2142 /* per-UID process charging. */
2143 extern struct user_struct * alloc_uid(kuid_t);
2144 static inline struct user_struct *get_uid(struct user_struct *u)
2146 atomic_inc(&u->__count);
2149 extern void free_uid(struct user_struct *);
2151 #include <asm/current.h>
2153 extern void xtime_update(unsigned long ticks);
2155 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2156 extern int wake_up_process(struct task_struct *tsk);
2157 extern void wake_up_new_task(struct task_struct *tsk);
2159 extern void kick_process(struct task_struct *tsk);
2161 static inline void kick_process(struct task_struct *tsk) { }
2163 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2164 extern void sched_dead(struct task_struct *p);
2166 extern void proc_caches_init(void);
2167 extern void flush_signals(struct task_struct *);
2168 extern void __flush_signals(struct task_struct *);
2169 extern void ignore_signals(struct task_struct *);
2170 extern void flush_signal_handlers(struct task_struct *, int force_default);
2171 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2173 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2175 unsigned long flags;
2178 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2179 ret = dequeue_signal(tsk, mask, info);
2180 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2185 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2187 extern void unblock_all_signals(void);
2188 extern void release_task(struct task_struct * p);
2189 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2190 extern int force_sigsegv(int, struct task_struct *);
2191 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2192 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2193 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2194 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2195 const struct cred *, u32);
2196 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2197 extern int kill_pid(struct pid *pid, int sig, int priv);
2198 extern int kill_proc_info(int, struct siginfo *, pid_t);
2199 extern __must_check bool do_notify_parent(struct task_struct *, int);
2200 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2201 extern void force_sig(int, struct task_struct *);
2202 extern int send_sig(int, struct task_struct *, int);
2203 extern int zap_other_threads(struct task_struct *p);
2204 extern struct sigqueue *sigqueue_alloc(void);
2205 extern void sigqueue_free(struct sigqueue *);
2206 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2207 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2209 static inline void restore_saved_sigmask(void)
2211 if (test_and_clear_restore_sigmask())
2212 __set_current_blocked(¤t->saved_sigmask);
2215 static inline sigset_t *sigmask_to_save(void)
2217 sigset_t *res = ¤t->blocked;
2218 if (unlikely(test_restore_sigmask()))
2219 res = ¤t->saved_sigmask;
2223 static inline int kill_cad_pid(int sig, int priv)
2225 return kill_pid(cad_pid, sig, priv);
2228 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2229 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2230 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2231 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2234 * True if we are on the alternate signal stack.
2236 static inline int on_sig_stack(unsigned long sp)
2238 #ifdef CONFIG_STACK_GROWSUP
2239 return sp >= current->sas_ss_sp &&
2240 sp - current->sas_ss_sp < current->sas_ss_size;
2242 return sp > current->sas_ss_sp &&
2243 sp - current->sas_ss_sp <= current->sas_ss_size;
2247 static inline int sas_ss_flags(unsigned long sp)
2249 return (current->sas_ss_size == 0 ? SS_DISABLE
2250 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2253 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2255 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2256 #ifdef CONFIG_STACK_GROWSUP
2257 return current->sas_ss_sp;
2259 return current->sas_ss_sp + current->sas_ss_size;
2265 * Routines for handling mm_structs
2267 extern struct mm_struct * mm_alloc(void);
2269 /* mmdrop drops the mm and the page tables */
2270 extern void __mmdrop(struct mm_struct *);
2271 static inline void mmdrop(struct mm_struct * mm)
2273 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2277 /* mmput gets rid of the mappings and all user-space */
2278 extern void mmput(struct mm_struct *);
2279 /* Grab a reference to a task's mm, if it is not already going away */
2280 extern struct mm_struct *get_task_mm(struct task_struct *task);
2282 * Grab a reference to a task's mm, if it is not already going away
2283 * and ptrace_may_access with the mode parameter passed to it
2286 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2287 /* Remove the current tasks stale references to the old mm_struct */
2288 extern void mm_release(struct task_struct *, struct mm_struct *);
2289 /* Allocate a new mm structure and copy contents from tsk->mm */
2290 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2292 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2293 struct task_struct *);
2294 extern void flush_thread(void);
2295 extern void exit_thread(void);
2297 extern void exit_files(struct task_struct *);
2298 extern void __cleanup_sighand(struct sighand_struct *);
2300 extern void exit_itimers(struct signal_struct *);
2301 extern void flush_itimer_signals(void);
2303 extern void do_group_exit(int);
2305 extern int allow_signal(int);
2306 extern int disallow_signal(int);
2308 extern int do_execve(const char *,
2309 const char __user * const __user *,
2310 const char __user * const __user *);
2311 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2312 struct task_struct *fork_idle(int);
2313 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2315 extern void set_task_comm(struct task_struct *tsk, char *from);
2316 extern char *get_task_comm(char *to, struct task_struct *tsk);
2319 void scheduler_ipi(void);
2320 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2322 static inline void scheduler_ipi(void) { }
2323 static inline unsigned long wait_task_inactive(struct task_struct *p,
2330 #define next_task(p) \
2331 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2333 #define for_each_process(p) \
2334 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2336 extern bool current_is_single_threaded(void);
2339 * Careful: do_each_thread/while_each_thread is a double loop so
2340 * 'break' will not work as expected - use goto instead.
2342 #define do_each_thread(g, t) \
2343 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2345 #define while_each_thread(g, t) \
2346 while ((t = next_thread(t)) != g)
2348 static inline int get_nr_threads(struct task_struct *tsk)
2350 return tsk->signal->nr_threads;
2353 static inline bool thread_group_leader(struct task_struct *p)
2355 return p->exit_signal >= 0;
2358 /* Do to the insanities of de_thread it is possible for a process
2359 * to have the pid of the thread group leader without actually being
2360 * the thread group leader. For iteration through the pids in proc
2361 * all we care about is that we have a task with the appropriate
2362 * pid, we don't actually care if we have the right task.
2364 static inline bool has_group_leader_pid(struct task_struct *p)
2366 return task_pid(p) == p->signal->leader_pid;
2370 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2372 return p1->signal == p2->signal;
2375 static inline struct task_struct *next_thread(const struct task_struct *p)
2377 return list_entry_rcu(p->thread_group.next,
2378 struct task_struct, thread_group);
2381 static inline int thread_group_empty(struct task_struct *p)
2383 return list_empty(&p->thread_group);
2386 #define delay_group_leader(p) \
2387 (thread_group_leader(p) && !thread_group_empty(p))
2390 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2391 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2392 * pins the final release of task.io_context. Also protects ->cpuset and
2393 * ->cgroup.subsys[]. And ->vfork_done.
2395 * Nests both inside and outside of read_lock(&tasklist_lock).
2396 * It must not be nested with write_lock_irq(&tasklist_lock),
2397 * neither inside nor outside.
2399 static inline void task_lock(struct task_struct *p)
2401 spin_lock(&p->alloc_lock);
2404 static inline void task_unlock(struct task_struct *p)
2406 spin_unlock(&p->alloc_lock);
2409 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2410 unsigned long *flags);
2412 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2413 unsigned long *flags)
2415 struct sighand_struct *ret;
2417 ret = __lock_task_sighand(tsk, flags);
2418 (void)__cond_lock(&tsk->sighand->siglock, ret);
2422 static inline void unlock_task_sighand(struct task_struct *tsk,
2423 unsigned long *flags)
2425 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2428 #ifdef CONFIG_CGROUPS
2429 static inline void threadgroup_change_begin(struct task_struct *tsk)
2431 down_read(&tsk->signal->group_rwsem);
2433 static inline void threadgroup_change_end(struct task_struct *tsk)
2435 up_read(&tsk->signal->group_rwsem);
2439 * threadgroup_lock - lock threadgroup
2440 * @tsk: member task of the threadgroup to lock
2442 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2443 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2444 * change ->group_leader/pid. This is useful for cases where the threadgroup
2445 * needs to stay stable across blockable operations.
2447 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2448 * synchronization. While held, no new task will be added to threadgroup
2449 * and no existing live task will have its PF_EXITING set.
2451 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2452 * sub-thread becomes a new leader.
2454 static inline void threadgroup_lock(struct task_struct *tsk)
2456 down_write(&tsk->signal->group_rwsem);
2460 * threadgroup_unlock - unlock threadgroup
2461 * @tsk: member task of the threadgroup to unlock
2463 * Reverse threadgroup_lock().
2465 static inline void threadgroup_unlock(struct task_struct *tsk)
2467 up_write(&tsk->signal->group_rwsem);
2470 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2471 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2472 static inline void threadgroup_lock(struct task_struct *tsk) {}
2473 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2476 #ifndef __HAVE_THREAD_FUNCTIONS
2478 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2479 #define task_stack_page(task) ((task)->stack)
2481 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2483 *task_thread_info(p) = *task_thread_info(org);
2484 task_thread_info(p)->task = p;
2487 static inline unsigned long *end_of_stack(struct task_struct *p)
2489 return (unsigned long *)(task_thread_info(p) + 1);
2494 static inline int object_is_on_stack(void *obj)
2496 void *stack = task_stack_page(current);
2498 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2501 extern void thread_info_cache_init(void);
2503 #ifdef CONFIG_DEBUG_STACK_USAGE
2504 static inline unsigned long stack_not_used(struct task_struct *p)
2506 unsigned long *n = end_of_stack(p);
2508 do { /* Skip over canary */
2512 return (unsigned long)n - (unsigned long)end_of_stack(p);
2516 /* set thread flags in other task's structures
2517 * - see asm/thread_info.h for TIF_xxxx flags available
2519 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2521 set_ti_thread_flag(task_thread_info(tsk), flag);
2524 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2526 clear_ti_thread_flag(task_thread_info(tsk), flag);
2529 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2531 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2534 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2536 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2539 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2541 return test_ti_thread_flag(task_thread_info(tsk), flag);
2544 static inline void set_tsk_need_resched(struct task_struct *tsk)
2546 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2549 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2551 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2554 static inline int test_tsk_need_resched(struct task_struct *tsk)
2556 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2559 static inline int restart_syscall(void)
2561 set_tsk_thread_flag(current, TIF_SIGPENDING);
2562 return -ERESTARTNOINTR;
2565 static inline int signal_pending(struct task_struct *p)
2567 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2570 static inline int __fatal_signal_pending(struct task_struct *p)
2572 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2575 static inline int fatal_signal_pending(struct task_struct *p)
2577 return signal_pending(p) && __fatal_signal_pending(p);
2580 static inline int signal_pending_state(long state, struct task_struct *p)
2582 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2584 if (!signal_pending(p))
2587 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2591 * cond_resched() and cond_resched_lock(): latency reduction via
2592 * explicit rescheduling in places that are safe. The return
2593 * value indicates whether a reschedule was done in fact.
2594 * cond_resched_lock() will drop the spinlock before scheduling,
2595 * cond_resched_softirq() will enable bhs before scheduling.
2597 extern int _cond_resched(void);
2599 #define cond_resched() ({ \
2600 __might_sleep(__FILE__, __LINE__, 0); \
2604 extern int __cond_resched_lock(spinlock_t *lock);
2606 #ifdef CONFIG_PREEMPT_COUNT
2607 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2609 #define PREEMPT_LOCK_OFFSET 0
2612 #define cond_resched_lock(lock) ({ \
2613 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2614 __cond_resched_lock(lock); \
2617 extern int __cond_resched_softirq(void);
2619 #define cond_resched_softirq() ({ \
2620 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2621 __cond_resched_softirq(); \
2624 static inline void cond_resched_rcu(void)
2626 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2634 * Does a critical section need to be broken due to another
2635 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2636 * but a general need for low latency)
2638 static inline int spin_needbreak(spinlock_t *lock)
2640 #ifdef CONFIG_PREEMPT
2641 return spin_is_contended(lock);
2648 * Idle thread specific functions to determine the need_resched
2649 * polling state. We have two versions, one based on TS_POLLING in
2650 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2654 static inline int tsk_is_polling(struct task_struct *p)
2656 return task_thread_info(p)->status & TS_POLLING;
2658 static inline void __current_set_polling(void)
2660 current_thread_info()->status |= TS_POLLING;
2663 static inline bool __must_check current_set_polling_and_test(void)
2665 __current_set_polling();
2668 * Polling state must be visible before we test NEED_RESCHED,
2669 * paired by resched_task()
2673 return unlikely(tif_need_resched());
2676 static inline void __current_clr_polling(void)
2678 current_thread_info()->status &= ~TS_POLLING;
2681 static inline bool __must_check current_clr_polling_and_test(void)
2683 __current_clr_polling();
2686 * Polling state must be visible before we test NEED_RESCHED,
2687 * paired by resched_task()
2691 return unlikely(tif_need_resched());
2693 #elif defined(TIF_POLLING_NRFLAG)
2694 static inline int tsk_is_polling(struct task_struct *p)
2696 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2699 static inline void __current_set_polling(void)
2701 set_thread_flag(TIF_POLLING_NRFLAG);
2704 static inline bool __must_check current_set_polling_and_test(void)
2706 __current_set_polling();
2709 * Polling state must be visible before we test NEED_RESCHED,
2710 * paired by resched_task()
2712 * XXX: assumes set/clear bit are identical barrier wise.
2714 smp_mb__after_clear_bit();
2716 return unlikely(tif_need_resched());
2719 static inline void __current_clr_polling(void)
2721 clear_thread_flag(TIF_POLLING_NRFLAG);
2724 static inline bool __must_check current_clr_polling_and_test(void)
2726 __current_clr_polling();
2729 * Polling state must be visible before we test NEED_RESCHED,
2730 * paired by resched_task()
2732 smp_mb__after_clear_bit();
2734 return unlikely(tif_need_resched());
2738 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2739 static inline void __current_set_polling(void) { }
2740 static inline void __current_clr_polling(void) { }
2742 static inline bool __must_check current_set_polling_and_test(void)
2744 return unlikely(tif_need_resched());
2746 static inline bool __must_check current_clr_polling_and_test(void)
2748 return unlikely(tif_need_resched());
2752 static inline void current_clr_polling(void)
2754 __current_clr_polling();
2757 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2758 * Once the bit is cleared, we'll get IPIs with every new
2759 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2762 smp_mb(); /* paired with resched_task() */
2764 preempt_fold_need_resched();
2767 static __always_inline bool need_resched(void)
2769 return unlikely(tif_need_resched());
2773 * Thread group CPU time accounting.
2775 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2776 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2778 static inline void thread_group_cputime_init(struct signal_struct *sig)
2780 raw_spin_lock_init(&sig->cputimer.lock);
2784 * Reevaluate whether the task has signals pending delivery.
2785 * Wake the task if so.
2786 * This is required every time the blocked sigset_t changes.
2787 * callers must hold sighand->siglock.
2789 extern void recalc_sigpending_and_wake(struct task_struct *t);
2790 extern void recalc_sigpending(void);
2792 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2794 static inline void signal_wake_up(struct task_struct *t, bool resume)
2796 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2798 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2800 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2804 * Wrappers for p->thread_info->cpu access. No-op on UP.
2808 static inline unsigned int task_cpu(const struct task_struct *p)
2810 return task_thread_info(p)->cpu;
2813 static inline int task_node(const struct task_struct *p)
2815 return cpu_to_node(task_cpu(p));
2818 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2822 static inline unsigned int task_cpu(const struct task_struct *p)
2827 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2831 #endif /* CONFIG_SMP */
2833 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2834 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2836 #ifdef CONFIG_CGROUP_SCHED
2837 extern struct task_group root_task_group;
2838 #endif /* CONFIG_CGROUP_SCHED */
2840 extern int task_can_switch_user(struct user_struct *up,
2841 struct task_struct *tsk);
2843 #ifdef CONFIG_TASK_XACCT
2844 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2846 tsk->ioac.rchar += amt;
2849 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2851 tsk->ioac.wchar += amt;
2854 static inline void inc_syscr(struct task_struct *tsk)
2859 static inline void inc_syscw(struct task_struct *tsk)
2864 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2868 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2872 static inline void inc_syscr(struct task_struct *tsk)
2876 static inline void inc_syscw(struct task_struct *tsk)
2881 #ifndef TASK_SIZE_OF
2882 #define TASK_SIZE_OF(tsk) TASK_SIZE
2885 #ifdef CONFIG_MM_OWNER
2886 extern void mm_update_next_owner(struct mm_struct *mm);
2887 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2889 static inline void mm_update_next_owner(struct mm_struct *mm)
2893 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2896 #endif /* CONFIG_MM_OWNER */
2898 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2901 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2904 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2907 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2910 static inline unsigned long rlimit(unsigned int limit)
2912 return task_rlimit(current, limit);
2915 static inline unsigned long rlimit_max(unsigned int limit)
2917 return task_rlimit_max(current, limit);