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/rbtree.h>
20 #include <linux/thread_info.h>
21 #include <linux/cpumask.h>
22 #include <linux/errno.h>
23 #include <linux/nodemask.h>
24 #include <linux/mm_types.h>
25 #include <linux/preempt.h>
28 #include <asm/ptrace.h>
29 #include <asm/cputime.h>
31 #include <linux/smp.h>
32 #include <linux/sem.h>
33 #include <linux/signal.h>
34 #include <linux/compiler.h>
35 #include <linux/completion.h>
36 #include <linux/pid.h>
37 #include <linux/percpu.h>
38 #include <linux/topology.h>
39 #include <linux/proportions.h>
40 #include <linux/seccomp.h>
41 #include <linux/rcupdate.h>
42 #include <linux/rculist.h>
43 #include <linux/rtmutex.h>
45 #include <linux/time.h>
46 #include <linux/param.h>
47 #include <linux/resource.h>
48 #include <linux/timer.h>
49 #include <linux/hrtimer.h>
50 #include <linux/task_io_accounting.h>
51 #include <linux/latencytop.h>
52 #include <linux/cred.h>
53 #include <linux/llist.h>
54 #include <linux/uidgid.h>
55 #include <linux/gfp.h>
57 #include <asm/processor.h>
60 struct futex_pi_state;
61 struct robust_list_head;
64 struct perf_event_context;
68 * List of flags we want to share for kernel threads,
69 * if only because they are not used by them anyway.
71 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
74 * These are the constant used to fake the fixed-point load-average
75 * counting. Some notes:
76 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
77 * a load-average precision of 10 bits integer + 11 bits fractional
78 * - if you want to count load-averages more often, you need more
79 * precision, or rounding will get you. With 2-second counting freq,
80 * the EXP_n values would be 1981, 2034 and 2043 if still using only
83 extern unsigned long avenrun[]; /* Load averages */
84 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
86 #define FSHIFT 11 /* nr of bits of precision */
87 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
88 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
89 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
90 #define EXP_5 2014 /* 1/exp(5sec/5min) */
91 #define EXP_15 2037 /* 1/exp(5sec/15min) */
93 #define CALC_LOAD(load,exp,n) \
95 load += n*(FIXED_1-exp); \
98 extern unsigned long total_forks;
99 extern int nr_threads;
100 DECLARE_PER_CPU(unsigned long, process_counts);
101 extern int nr_processes(void);
102 extern unsigned long nr_running(void);
103 extern unsigned long nr_iowait(void);
104 extern unsigned long nr_iowait_cpu(int cpu);
105 extern unsigned long this_cpu_load(void);
108 extern void calc_global_load(unsigned long ticks);
109 extern void update_cpu_load_nohz(void);
111 extern unsigned long get_parent_ip(unsigned long addr);
113 extern void dump_cpu_task(int cpu);
118 #ifdef CONFIG_SCHED_DEBUG
119 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
120 extern void proc_sched_set_task(struct task_struct *p);
122 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
126 * Task state bitmask. NOTE! These bits are also
127 * encoded in fs/proc/array.c: get_task_state().
129 * We have two separate sets of flags: task->state
130 * is about runnability, while task->exit_state are
131 * about the task exiting. Confusing, but this way
132 * modifying one set can't modify the other one by
135 #define TASK_RUNNING 0
136 #define TASK_INTERRUPTIBLE 1
137 #define TASK_UNINTERRUPTIBLE 2
138 #define __TASK_STOPPED 4
139 #define __TASK_TRACED 8
140 /* in tsk->exit_state */
141 #define EXIT_ZOMBIE 16
143 /* in tsk->state again */
145 #define TASK_WAKEKILL 128
146 #define TASK_WAKING 256
147 #define TASK_PARKED 512
148 #define TASK_STATE_MAX 1024
150 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
152 extern char ___assert_task_state[1 - 2*!!(
153 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
155 /* Convenience macros for the sake of set_task_state */
156 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
157 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
158 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
160 /* Convenience macros for the sake of wake_up */
161 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
162 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
164 /* get_task_state() */
165 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
166 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
169 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
170 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
171 #define task_is_dead(task) ((task)->exit_state != 0)
172 #define task_is_stopped_or_traced(task) \
173 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
174 #define task_contributes_to_load(task) \
175 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
176 (task->flags & PF_FROZEN) == 0)
178 #define __set_task_state(tsk, state_value) \
179 do { (tsk)->state = (state_value); } while (0)
180 #define set_task_state(tsk, state_value) \
181 set_mb((tsk)->state, (state_value))
184 * set_current_state() includes a barrier so that the write of current->state
185 * is correctly serialised wrt the caller's subsequent test of whether to
188 * set_current_state(TASK_UNINTERRUPTIBLE);
189 * if (do_i_need_to_sleep())
192 * If the caller does not need such serialisation then use __set_current_state()
194 #define __set_current_state(state_value) \
195 do { current->state = (state_value); } while (0)
196 #define set_current_state(state_value) \
197 set_mb(current->state, (state_value))
199 /* Task command name length */
200 #define TASK_COMM_LEN 16
202 #include <linux/spinlock.h>
205 * This serializes "schedule()" and also protects
206 * the run-queue from deletions/modifications (but
207 * _adding_ to the beginning of the run-queue has
210 extern rwlock_t tasklist_lock;
211 extern spinlock_t mmlist_lock;
215 #ifdef CONFIG_PROVE_RCU
216 extern int lockdep_tasklist_lock_is_held(void);
217 #endif /* #ifdef CONFIG_PROVE_RCU */
219 extern void sched_init(void);
220 extern void sched_init_smp(void);
221 extern asmlinkage void schedule_tail(struct task_struct *prev);
222 extern void init_idle(struct task_struct *idle, int cpu);
223 extern void init_idle_bootup_task(struct task_struct *idle);
225 extern int runqueue_is_locked(int cpu);
227 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
228 extern void nohz_balance_enter_idle(int cpu);
229 extern void set_cpu_sd_state_idle(void);
230 extern int get_nohz_timer_target(void);
232 static inline void nohz_balance_enter_idle(int cpu) { }
233 static inline void set_cpu_sd_state_idle(void) { }
237 * Only dump TASK_* tasks. (0 for all tasks)
239 extern void show_state_filter(unsigned long state_filter);
241 static inline void show_state(void)
243 show_state_filter(0);
246 extern void show_regs(struct pt_regs *);
249 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
250 * task), SP is the stack pointer of the first frame that should be shown in the back
251 * trace (or NULL if the entire call-chain of the task should be shown).
253 extern void show_stack(struct task_struct *task, unsigned long *sp);
255 void io_schedule(void);
256 long io_schedule_timeout(long timeout);
258 extern void cpu_init (void);
259 extern void trap_init(void);
260 extern void update_process_times(int user);
261 extern void scheduler_tick(void);
263 extern void sched_show_task(struct task_struct *p);
265 #ifdef CONFIG_LOCKUP_DETECTOR
266 extern void touch_softlockup_watchdog(void);
267 extern void touch_softlockup_watchdog_sync(void);
268 extern void touch_all_softlockup_watchdogs(void);
269 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
271 size_t *lenp, loff_t *ppos);
272 extern unsigned int softlockup_panic;
273 void lockup_detector_init(void);
275 static inline void touch_softlockup_watchdog(void)
278 static inline void touch_softlockup_watchdog_sync(void)
281 static inline void touch_all_softlockup_watchdogs(void)
284 static inline void lockup_detector_init(void)
289 /* Attach to any functions which should be ignored in wchan output. */
290 #define __sched __attribute__((__section__(".sched.text")))
292 /* Linker adds these: start and end of __sched functions */
293 extern char __sched_text_start[], __sched_text_end[];
295 /* Is this address in the __sched functions? */
296 extern int in_sched_functions(unsigned long addr);
298 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
299 extern signed long schedule_timeout(signed long timeout);
300 extern signed long schedule_timeout_interruptible(signed long timeout);
301 extern signed long schedule_timeout_killable(signed long timeout);
302 extern signed long schedule_timeout_uninterruptible(signed long timeout);
303 asmlinkage void schedule(void);
304 extern void schedule_preempt_disabled(void);
307 struct user_namespace;
310 extern void arch_pick_mmap_layout(struct mm_struct *mm);
312 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
313 unsigned long, unsigned long);
315 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
316 unsigned long len, unsigned long pgoff,
317 unsigned long flags);
319 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
323 extern void set_dumpable(struct mm_struct *mm, int value);
324 extern int get_dumpable(struct mm_struct *mm);
326 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
327 #define SUID_DUMP_USER 1 /* Dump as user of process */
328 #define SUID_DUMP_ROOT 2 /* Dump as root */
332 #define MMF_DUMPABLE 0 /* core dump is permitted */
333 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
335 #define MMF_DUMPABLE_BITS 2
336 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
338 /* coredump filter bits */
339 #define MMF_DUMP_ANON_PRIVATE 2
340 #define MMF_DUMP_ANON_SHARED 3
341 #define MMF_DUMP_MAPPED_PRIVATE 4
342 #define MMF_DUMP_MAPPED_SHARED 5
343 #define MMF_DUMP_ELF_HEADERS 6
344 #define MMF_DUMP_HUGETLB_PRIVATE 7
345 #define MMF_DUMP_HUGETLB_SHARED 8
347 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
348 #define MMF_DUMP_FILTER_BITS 7
349 #define MMF_DUMP_FILTER_MASK \
350 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
351 #define MMF_DUMP_FILTER_DEFAULT \
352 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
353 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
355 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
356 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
358 # define MMF_DUMP_MASK_DEFAULT_ELF 0
360 /* leave room for more dump flags */
361 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
362 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
363 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
365 #define MMF_HAS_UPROBES 19 /* has uprobes */
366 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
368 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
370 struct sighand_struct {
372 struct k_sigaction action[_NSIG];
374 wait_queue_head_t signalfd_wqh;
377 struct pacct_struct {
380 unsigned long ac_mem;
381 cputime_t ac_utime, ac_stime;
382 unsigned long ac_minflt, ac_majflt;
393 * struct cputime - snaphsot of system and user cputime
394 * @utime: time spent in user mode
395 * @stime: time spent in system mode
397 * Gathers a generic snapshot of user and system time.
405 * struct task_cputime - collected CPU time counts
406 * @utime: time spent in user mode, in &cputime_t units
407 * @stime: time spent in kernel mode, in &cputime_t units
408 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
410 * This is an extension of struct cputime that includes the total runtime
411 * spent by the task from the scheduler point of view.
413 * As a result, this structure groups together three kinds of CPU time
414 * that are tracked for threads and thread groups. Most things considering
415 * CPU time want to group these counts together and treat all three
416 * of them in parallel.
418 struct task_cputime {
421 unsigned long long sum_exec_runtime;
423 /* Alternate field names when used to cache expirations. */
424 #define prof_exp stime
425 #define virt_exp utime
426 #define sched_exp sum_exec_runtime
428 #define INIT_CPUTIME \
429 (struct task_cputime) { \
432 .sum_exec_runtime = 0, \
435 #define PREEMPT_ENABLED (PREEMPT_NEED_RESCHED)
437 #ifdef CONFIG_PREEMPT_COUNT
438 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
440 #define PREEMPT_DISABLED PREEMPT_ENABLED
444 * Disable preemption until the scheduler is running.
445 * Reset by start_kernel()->sched_init()->init_idle().
447 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
448 * before the scheduler is active -- see should_resched().
450 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
453 * struct thread_group_cputimer - thread group interval timer counts
454 * @cputime: thread group interval timers.
455 * @running: non-zero when there are timers running and
456 * @cputime receives updates.
457 * @lock: lock for fields in this struct.
459 * This structure contains the version of task_cputime, above, that is
460 * used for thread group CPU timer calculations.
462 struct thread_group_cputimer {
463 struct task_cputime cputime;
468 #include <linux/rwsem.h>
472 * NOTE! "signal_struct" does not have its own
473 * locking, because a shared signal_struct always
474 * implies a shared sighand_struct, so locking
475 * sighand_struct is always a proper superset of
476 * the locking of signal_struct.
478 struct signal_struct {
483 wait_queue_head_t wait_chldexit; /* for wait4() */
485 /* current thread group signal load-balancing target: */
486 struct task_struct *curr_target;
488 /* shared signal handling: */
489 struct sigpending shared_pending;
491 /* thread group exit support */
494 * - notify group_exit_task when ->count is equal to notify_count
495 * - everyone except group_exit_task is stopped during signal delivery
496 * of fatal signals, group_exit_task processes the signal.
499 struct task_struct *group_exit_task;
501 /* thread group stop support, overloads group_exit_code too */
502 int group_stop_count;
503 unsigned int flags; /* see SIGNAL_* flags below */
506 * PR_SET_CHILD_SUBREAPER marks a process, like a service
507 * manager, to re-parent orphan (double-forking) child processes
508 * to this process instead of 'init'. The service manager is
509 * able to receive SIGCHLD signals and is able to investigate
510 * the process until it calls wait(). All children of this
511 * process will inherit a flag if they should look for a
512 * child_subreaper process at exit.
514 unsigned int is_child_subreaper:1;
515 unsigned int has_child_subreaper:1;
517 /* POSIX.1b Interval Timers */
519 struct list_head posix_timers;
521 /* ITIMER_REAL timer for the process */
522 struct hrtimer real_timer;
523 struct pid *leader_pid;
524 ktime_t it_real_incr;
527 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
528 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
529 * values are defined to 0 and 1 respectively
531 struct cpu_itimer it[2];
534 * Thread group totals for process CPU timers.
535 * See thread_group_cputimer(), et al, for details.
537 struct thread_group_cputimer cputimer;
539 /* Earliest-expiration cache. */
540 struct task_cputime cputime_expires;
542 struct list_head cpu_timers[3];
544 struct pid *tty_old_pgrp;
546 /* boolean value for session group leader */
549 struct tty_struct *tty; /* NULL if no tty */
551 #ifdef CONFIG_SCHED_AUTOGROUP
552 struct autogroup *autogroup;
555 * Cumulative resource counters for dead threads in the group,
556 * and for reaped dead child processes forked by this group.
557 * Live threads maintain their own counters and add to these
558 * in __exit_signal, except for the group leader.
560 cputime_t utime, stime, cutime, cstime;
563 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
564 struct cputime prev_cputime;
566 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
567 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
568 unsigned long inblock, oublock, cinblock, coublock;
569 unsigned long maxrss, cmaxrss;
570 struct task_io_accounting ioac;
573 * Cumulative ns of schedule CPU time fo dead threads in the
574 * group, not including a zombie group leader, (This only differs
575 * from jiffies_to_ns(utime + stime) if sched_clock uses something
576 * other than jiffies.)
578 unsigned long long sum_sched_runtime;
581 * We don't bother to synchronize most readers of this at all,
582 * because there is no reader checking a limit that actually needs
583 * to get both rlim_cur and rlim_max atomically, and either one
584 * alone is a single word that can safely be read normally.
585 * getrlimit/setrlimit use task_lock(current->group_leader) to
586 * protect this instead of the siglock, because they really
587 * have no need to disable irqs.
589 struct rlimit rlim[RLIM_NLIMITS];
591 #ifdef CONFIG_BSD_PROCESS_ACCT
592 struct pacct_struct pacct; /* per-process accounting information */
594 #ifdef CONFIG_TASKSTATS
595 struct taskstats *stats;
599 unsigned audit_tty_log_passwd;
600 struct tty_audit_buf *tty_audit_buf;
602 #ifdef CONFIG_CGROUPS
604 * group_rwsem prevents new tasks from entering the threadgroup and
605 * member tasks from exiting,a more specifically, setting of
606 * PF_EXITING. fork and exit paths are protected with this rwsem
607 * using threadgroup_change_begin/end(). Users which require
608 * threadgroup to remain stable should use threadgroup_[un]lock()
609 * which also takes care of exec path. Currently, cgroup is the
612 struct rw_semaphore group_rwsem;
615 oom_flags_t oom_flags;
616 short oom_score_adj; /* OOM kill score adjustment */
617 short oom_score_adj_min; /* OOM kill score adjustment min value.
618 * Only settable by CAP_SYS_RESOURCE. */
620 struct mutex cred_guard_mutex; /* guard against foreign influences on
621 * credential calculations
622 * (notably. ptrace) */
626 * Bits in flags field of signal_struct.
628 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
629 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
630 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
631 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
633 * Pending notifications to parent.
635 #define SIGNAL_CLD_STOPPED 0x00000010
636 #define SIGNAL_CLD_CONTINUED 0x00000020
637 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
639 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
641 /* If true, all threads except ->group_exit_task have pending SIGKILL */
642 static inline int signal_group_exit(const struct signal_struct *sig)
644 return (sig->flags & SIGNAL_GROUP_EXIT) ||
645 (sig->group_exit_task != NULL);
649 * Some day this will be a full-fledged user tracking system..
652 atomic_t __count; /* reference count */
653 atomic_t processes; /* How many processes does this user have? */
654 atomic_t files; /* How many open files does this user have? */
655 atomic_t sigpending; /* How many pending signals does this user have? */
656 #ifdef CONFIG_INOTIFY_USER
657 atomic_t inotify_watches; /* How many inotify watches does this user have? */
658 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
660 #ifdef CONFIG_FANOTIFY
661 atomic_t fanotify_listeners;
664 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
666 #ifdef CONFIG_POSIX_MQUEUE
667 /* protected by mq_lock */
668 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
670 unsigned long locked_shm; /* How many pages of mlocked shm ? */
673 struct key *uid_keyring; /* UID specific keyring */
674 struct key *session_keyring; /* UID's default session keyring */
677 /* Hash table maintenance information */
678 struct hlist_node uidhash_node;
681 #ifdef CONFIG_PERF_EVENTS
682 atomic_long_t locked_vm;
686 extern int uids_sysfs_init(void);
688 extern struct user_struct *find_user(kuid_t);
690 extern struct user_struct root_user;
691 #define INIT_USER (&root_user)
694 struct backing_dev_info;
695 struct reclaim_state;
697 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
699 /* cumulative counters */
700 unsigned long pcount; /* # of times run on this cpu */
701 unsigned long long run_delay; /* time spent waiting on a runqueue */
704 unsigned long long last_arrival,/* when we last ran on a cpu */
705 last_queued; /* when we were last queued to run */
707 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
709 #ifdef CONFIG_TASK_DELAY_ACCT
710 struct task_delay_info {
712 unsigned int flags; /* Private per-task flags */
714 /* For each stat XXX, add following, aligned appropriately
716 * struct timespec XXX_start, XXX_end;
720 * Atomicity of updates to XXX_delay, XXX_count protected by
721 * single lock above (split into XXX_lock if contention is an issue).
725 * XXX_count is incremented on every XXX operation, the delay
726 * associated with the operation is added to XXX_delay.
727 * XXX_delay contains the accumulated delay time in nanoseconds.
729 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
730 u64 blkio_delay; /* wait for sync block io completion */
731 u64 swapin_delay; /* wait for swapin block io completion */
732 u32 blkio_count; /* total count of the number of sync block */
733 /* io operations performed */
734 u32 swapin_count; /* total count of the number of swapin block */
735 /* io operations performed */
737 struct timespec freepages_start, freepages_end;
738 u64 freepages_delay; /* wait for memory reclaim */
739 u32 freepages_count; /* total count of memory reclaim */
741 #endif /* CONFIG_TASK_DELAY_ACCT */
743 static inline int sched_info_on(void)
745 #ifdef CONFIG_SCHEDSTATS
747 #elif defined(CONFIG_TASK_DELAY_ACCT)
748 extern int delayacct_on;
763 * Increase resolution of cpu_power calculations
765 #define SCHED_POWER_SHIFT 10
766 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
769 * sched-domains (multiprocessor balancing) declarations:
772 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
773 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
774 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
775 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
776 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
777 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
778 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
779 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
780 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
781 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
782 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
783 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
784 #define SD_NUMA 0x4000 /* cross-node balancing */
786 extern int __weak arch_sd_sibiling_asym_packing(void);
788 struct sched_domain_attr {
789 int relax_domain_level;
792 #define SD_ATTR_INIT (struct sched_domain_attr) { \
793 .relax_domain_level = -1, \
796 extern int sched_domain_level_max;
800 struct sched_domain {
801 /* These fields must be setup */
802 struct sched_domain *parent; /* top domain must be null terminated */
803 struct sched_domain *child; /* bottom domain must be null terminated */
804 struct sched_group *groups; /* the balancing groups of the domain */
805 unsigned long min_interval; /* Minimum balance interval ms */
806 unsigned long max_interval; /* Maximum balance interval ms */
807 unsigned int busy_factor; /* less balancing by factor if busy */
808 unsigned int imbalance_pct; /* No balance until over watermark */
809 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
810 unsigned int busy_idx;
811 unsigned int idle_idx;
812 unsigned int newidle_idx;
813 unsigned int wake_idx;
814 unsigned int forkexec_idx;
815 unsigned int smt_gain;
817 int nohz_idle; /* NOHZ IDLE status */
818 int flags; /* See SD_* */
821 /* Runtime fields. */
822 unsigned long last_balance; /* init to jiffies. units in jiffies */
823 unsigned int balance_interval; /* initialise to 1. units in ms. */
824 unsigned int nr_balance_failed; /* initialise to 0 */
828 /* idle_balance() stats */
829 u64 max_newidle_lb_cost;
830 unsigned long next_decay_max_lb_cost;
832 #ifdef CONFIG_SCHEDSTATS
833 /* load_balance() stats */
834 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
835 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
836 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
837 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
838 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
839 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
840 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
841 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
843 /* Active load balancing */
844 unsigned int alb_count;
845 unsigned int alb_failed;
846 unsigned int alb_pushed;
848 /* SD_BALANCE_EXEC stats */
849 unsigned int sbe_count;
850 unsigned int sbe_balanced;
851 unsigned int sbe_pushed;
853 /* SD_BALANCE_FORK stats */
854 unsigned int sbf_count;
855 unsigned int sbf_balanced;
856 unsigned int sbf_pushed;
858 /* try_to_wake_up() stats */
859 unsigned int ttwu_wake_remote;
860 unsigned int ttwu_move_affine;
861 unsigned int ttwu_move_balance;
863 #ifdef CONFIG_SCHED_DEBUG
867 void *private; /* used during construction */
868 struct rcu_head rcu; /* used during destruction */
871 unsigned int span_weight;
873 * Span of all CPUs in this domain.
875 * NOTE: this field is variable length. (Allocated dynamically
876 * by attaching extra space to the end of the structure,
877 * depending on how many CPUs the kernel has booted up with)
879 unsigned long span[0];
882 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
884 return to_cpumask(sd->span);
887 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
888 struct sched_domain_attr *dattr_new);
890 /* Allocate an array of sched domains, for partition_sched_domains(). */
891 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
892 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
894 bool cpus_share_cache(int this_cpu, int that_cpu);
896 #else /* CONFIG_SMP */
898 struct sched_domain_attr;
901 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
902 struct sched_domain_attr *dattr_new)
906 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
911 #endif /* !CONFIG_SMP */
914 struct io_context; /* See blkdev.h */
917 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
918 extern void prefetch_stack(struct task_struct *t);
920 static inline void prefetch_stack(struct task_struct *t) { }
923 struct audit_context; /* See audit.c */
925 struct pipe_inode_info;
926 struct uts_namespace;
929 unsigned long weight, inv_weight;
934 * These sums represent an infinite geometric series and so are bound
935 * above by 1024/(1-y). Thus we only need a u32 to store them for all
936 * choices of y < 1-2^(-32)*1024.
938 u32 runnable_avg_sum, runnable_avg_period;
939 u64 last_runnable_update;
941 unsigned long load_avg_contrib;
944 #ifdef CONFIG_SCHEDSTATS
945 struct sched_statistics {
955 s64 sum_sleep_runtime;
962 u64 nr_migrations_cold;
963 u64 nr_failed_migrations_affine;
964 u64 nr_failed_migrations_running;
965 u64 nr_failed_migrations_hot;
966 u64 nr_forced_migrations;
970 u64 nr_wakeups_migrate;
971 u64 nr_wakeups_local;
972 u64 nr_wakeups_remote;
973 u64 nr_wakeups_affine;
974 u64 nr_wakeups_affine_attempts;
975 u64 nr_wakeups_passive;
980 struct sched_entity {
981 struct load_weight load; /* for load-balancing */
982 struct rb_node run_node;
983 struct list_head group_node;
987 u64 sum_exec_runtime;
989 u64 prev_sum_exec_runtime;
993 #ifdef CONFIG_SCHEDSTATS
994 struct sched_statistics statistics;
997 #ifdef CONFIG_FAIR_GROUP_SCHED
998 struct sched_entity *parent;
999 /* rq on which this entity is (to be) queued: */
1000 struct cfs_rq *cfs_rq;
1001 /* rq "owned" by this entity/group: */
1002 struct cfs_rq *my_q;
1006 /* Per-entity load-tracking */
1007 struct sched_avg avg;
1011 struct sched_rt_entity {
1012 struct list_head run_list;
1013 unsigned long timeout;
1014 unsigned long watchdog_stamp;
1015 unsigned int time_slice;
1017 struct sched_rt_entity *back;
1018 #ifdef CONFIG_RT_GROUP_SCHED
1019 struct sched_rt_entity *parent;
1020 /* rq on which this entity is (to be) queued: */
1021 struct rt_rq *rt_rq;
1022 /* rq "owned" by this entity/group: */
1030 enum perf_event_task_context {
1031 perf_invalid_context = -1,
1032 perf_hw_context = 0,
1034 perf_nr_task_contexts,
1037 struct task_struct {
1038 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1041 unsigned int flags; /* per process flags, defined below */
1042 unsigned int ptrace;
1045 struct llist_node wake_entry;
1047 struct task_struct *last_wakee;
1048 unsigned long wakee_flips;
1049 unsigned long wakee_flip_decay_ts;
1055 int prio, static_prio, normal_prio;
1056 unsigned int rt_priority;
1057 const struct sched_class *sched_class;
1058 struct sched_entity se;
1059 struct sched_rt_entity rt;
1060 #ifdef CONFIG_CGROUP_SCHED
1061 struct task_group *sched_task_group;
1064 #ifdef CONFIG_PREEMPT_NOTIFIERS
1065 /* list of struct preempt_notifier: */
1066 struct hlist_head preempt_notifiers;
1069 #ifdef CONFIG_BLK_DEV_IO_TRACE
1070 unsigned int btrace_seq;
1073 unsigned int policy;
1074 int nr_cpus_allowed;
1075 cpumask_t cpus_allowed;
1077 #ifdef CONFIG_PREEMPT_RCU
1078 int rcu_read_lock_nesting;
1079 char rcu_read_unlock_special;
1080 struct list_head rcu_node_entry;
1081 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1082 #ifdef CONFIG_TREE_PREEMPT_RCU
1083 struct rcu_node *rcu_blocked_node;
1084 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1085 #ifdef CONFIG_RCU_BOOST
1086 struct rt_mutex *rcu_boost_mutex;
1087 #endif /* #ifdef CONFIG_RCU_BOOST */
1089 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1090 struct sched_info sched_info;
1093 struct list_head tasks;
1095 struct plist_node pushable_tasks;
1098 struct mm_struct *mm, *active_mm;
1099 #ifdef CONFIG_COMPAT_BRK
1100 unsigned brk_randomized:1;
1102 #if defined(SPLIT_RSS_COUNTING)
1103 struct task_rss_stat rss_stat;
1107 int exit_code, exit_signal;
1108 int pdeath_signal; /* The signal sent when the parent dies */
1109 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1111 /* Used for emulating ABI behavior of previous Linux versions */
1112 unsigned int personality;
1114 unsigned did_exec:1;
1115 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1117 unsigned in_iowait:1;
1119 /* task may not gain privileges */
1120 unsigned no_new_privs:1;
1122 /* Revert to default priority/policy when forking */
1123 unsigned sched_reset_on_fork:1;
1124 unsigned sched_contributes_to_load:1;
1129 #ifdef CONFIG_CC_STACKPROTECTOR
1130 /* Canary value for the -fstack-protector gcc feature */
1131 unsigned long stack_canary;
1134 * pointers to (original) parent process, youngest child, younger sibling,
1135 * older sibling, respectively. (p->father can be replaced with
1136 * p->real_parent->pid)
1138 struct task_struct __rcu *real_parent; /* real parent process */
1139 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1141 * children/sibling forms the list of my natural children
1143 struct list_head children; /* list of my children */
1144 struct list_head sibling; /* linkage in my parent's children list */
1145 struct task_struct *group_leader; /* threadgroup leader */
1148 * ptraced is the list of tasks this task is using ptrace on.
1149 * This includes both natural children and PTRACE_ATTACH targets.
1150 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1152 struct list_head ptraced;
1153 struct list_head ptrace_entry;
1155 /* PID/PID hash table linkage. */
1156 struct pid_link pids[PIDTYPE_MAX];
1157 struct list_head thread_group;
1159 struct completion *vfork_done; /* for vfork() */
1160 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1161 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1163 cputime_t utime, stime, utimescaled, stimescaled;
1165 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1166 struct cputime prev_cputime;
1168 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1169 seqlock_t vtime_seqlock;
1170 unsigned long long vtime_snap;
1175 } vtime_snap_whence;
1177 unsigned long nvcsw, nivcsw; /* context switch counts */
1178 struct timespec start_time; /* monotonic time */
1179 struct timespec real_start_time; /* boot based time */
1180 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1181 unsigned long min_flt, maj_flt;
1183 struct task_cputime cputime_expires;
1184 struct list_head cpu_timers[3];
1186 /* process credentials */
1187 const struct cred __rcu *real_cred; /* objective and real subjective task
1188 * credentials (COW) */
1189 const struct cred __rcu *cred; /* effective (overridable) subjective task
1190 * credentials (COW) */
1191 char comm[TASK_COMM_LEN]; /* executable name excluding path
1192 - access with [gs]et_task_comm (which lock
1193 it with task_lock())
1194 - initialized normally by setup_new_exec */
1195 /* file system info */
1196 int link_count, total_link_count;
1197 #ifdef CONFIG_SYSVIPC
1199 struct sysv_sem sysvsem;
1201 #ifdef CONFIG_DETECT_HUNG_TASK
1202 /* hung task detection */
1203 unsigned long last_switch_count;
1205 /* CPU-specific state of this task */
1206 struct thread_struct thread;
1207 /* filesystem information */
1208 struct fs_struct *fs;
1209 /* open file information */
1210 struct files_struct *files;
1212 struct nsproxy *nsproxy;
1213 /* signal handlers */
1214 struct signal_struct *signal;
1215 struct sighand_struct *sighand;
1217 sigset_t blocked, real_blocked;
1218 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1219 struct sigpending pending;
1221 unsigned long sas_ss_sp;
1223 int (*notifier)(void *priv);
1224 void *notifier_data;
1225 sigset_t *notifier_mask;
1226 struct callback_head *task_works;
1228 struct audit_context *audit_context;
1229 #ifdef CONFIG_AUDITSYSCALL
1231 unsigned int sessionid;
1233 struct seccomp seccomp;
1235 /* Thread group tracking */
1238 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1240 spinlock_t alloc_lock;
1242 /* Protection of the PI data structures: */
1243 raw_spinlock_t pi_lock;
1245 #ifdef CONFIG_RT_MUTEXES
1246 /* PI waiters blocked on a rt_mutex held by this task */
1247 struct plist_head pi_waiters;
1248 /* Deadlock detection and priority inheritance handling */
1249 struct rt_mutex_waiter *pi_blocked_on;
1252 #ifdef CONFIG_DEBUG_MUTEXES
1253 /* mutex deadlock detection */
1254 struct mutex_waiter *blocked_on;
1256 #ifdef CONFIG_TRACE_IRQFLAGS
1257 unsigned int irq_events;
1258 unsigned long hardirq_enable_ip;
1259 unsigned long hardirq_disable_ip;
1260 unsigned int hardirq_enable_event;
1261 unsigned int hardirq_disable_event;
1262 int hardirqs_enabled;
1263 int hardirq_context;
1264 unsigned long softirq_disable_ip;
1265 unsigned long softirq_enable_ip;
1266 unsigned int softirq_disable_event;
1267 unsigned int softirq_enable_event;
1268 int softirqs_enabled;
1269 int softirq_context;
1271 #ifdef CONFIG_LOCKDEP
1272 # define MAX_LOCK_DEPTH 48UL
1275 unsigned int lockdep_recursion;
1276 struct held_lock held_locks[MAX_LOCK_DEPTH];
1277 gfp_t lockdep_reclaim_gfp;
1280 /* journalling filesystem info */
1283 /* stacked block device info */
1284 struct bio_list *bio_list;
1287 /* stack plugging */
1288 struct blk_plug *plug;
1292 struct reclaim_state *reclaim_state;
1294 struct backing_dev_info *backing_dev_info;
1296 struct io_context *io_context;
1298 unsigned long ptrace_message;
1299 siginfo_t *last_siginfo; /* For ptrace use. */
1300 struct task_io_accounting ioac;
1301 #if defined(CONFIG_TASK_XACCT)
1302 u64 acct_rss_mem1; /* accumulated rss usage */
1303 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1304 cputime_t acct_timexpd; /* stime + utime since last update */
1306 #ifdef CONFIG_CPUSETS
1307 nodemask_t mems_allowed; /* Protected by alloc_lock */
1308 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1309 int cpuset_mem_spread_rotor;
1310 int cpuset_slab_spread_rotor;
1312 #ifdef CONFIG_CGROUPS
1313 /* Control Group info protected by css_set_lock */
1314 struct css_set __rcu *cgroups;
1315 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1316 struct list_head cg_list;
1319 struct robust_list_head __user *robust_list;
1320 #ifdef CONFIG_COMPAT
1321 struct compat_robust_list_head __user *compat_robust_list;
1323 struct list_head pi_state_list;
1324 struct futex_pi_state *pi_state_cache;
1326 #ifdef CONFIG_PERF_EVENTS
1327 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1328 struct mutex perf_event_mutex;
1329 struct list_head perf_event_list;
1332 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1334 short pref_node_fork;
1336 #ifdef CONFIG_NUMA_BALANCING
1338 unsigned int numa_scan_period;
1339 unsigned int numa_scan_period_max;
1340 int numa_preferred_nid;
1341 int numa_migrate_deferred;
1342 unsigned long numa_migrate_retry;
1343 u64 node_stamp; /* migration stamp */
1344 struct callback_head numa_work;
1346 struct list_head numa_entry;
1347 struct numa_group *numa_group;
1350 * Exponential decaying average of faults on a per-node basis.
1351 * Scheduling placement decisions are made based on the these counts.
1352 * The values remain static for the duration of a PTE scan
1354 unsigned long *numa_faults;
1355 unsigned long total_numa_faults;
1358 * numa_faults_buffer records faults per node during the current
1359 * scan window. When the scan completes, the counts in numa_faults
1360 * decay and these values are copied.
1362 unsigned long *numa_faults_buffer;
1365 * numa_faults_locality tracks if faults recorded during the last
1366 * scan window were remote/local. The task scan period is adapted
1367 * based on the locality of the faults with different weights
1368 * depending on whether they were shared or private faults
1370 unsigned long numa_faults_locality[2];
1372 unsigned long numa_pages_migrated;
1373 #endif /* CONFIG_NUMA_BALANCING */
1375 struct rcu_head rcu;
1378 * cache last used pipe for splice
1380 struct pipe_inode_info *splice_pipe;
1382 struct page_frag task_frag;
1384 #ifdef CONFIG_TASK_DELAY_ACCT
1385 struct task_delay_info *delays;
1387 #ifdef CONFIG_FAULT_INJECTION
1391 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1392 * balance_dirty_pages() for some dirty throttling pause
1395 int nr_dirtied_pause;
1396 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1398 #ifdef CONFIG_LATENCYTOP
1399 int latency_record_count;
1400 struct latency_record latency_record[LT_SAVECOUNT];
1403 * time slack values; these are used to round up poll() and
1404 * select() etc timeout values. These are in nanoseconds.
1406 unsigned long timer_slack_ns;
1407 unsigned long default_timer_slack_ns;
1409 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1410 /* Index of current stored address in ret_stack */
1412 /* Stack of return addresses for return function tracing */
1413 struct ftrace_ret_stack *ret_stack;
1414 /* time stamp for last schedule */
1415 unsigned long long ftrace_timestamp;
1417 * Number of functions that haven't been traced
1418 * because of depth overrun.
1420 atomic_t trace_overrun;
1421 /* Pause for the tracing */
1422 atomic_t tracing_graph_pause;
1424 #ifdef CONFIG_TRACING
1425 /* state flags for use by tracers */
1426 unsigned long trace;
1427 /* bitmask and counter of trace recursion */
1428 unsigned long trace_recursion;
1429 #endif /* CONFIG_TRACING */
1430 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1431 struct memcg_batch_info {
1432 int do_batch; /* incremented when batch uncharge started */
1433 struct mem_cgroup *memcg; /* target memcg of uncharge */
1434 unsigned long nr_pages; /* uncharged usage */
1435 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1437 unsigned int memcg_kmem_skip_account;
1438 struct memcg_oom_info {
1439 struct mem_cgroup *memcg;
1442 unsigned int may_oom:1;
1445 #ifdef CONFIG_UPROBES
1446 struct uprobe_task *utask;
1448 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1449 unsigned int sequential_io;
1450 unsigned int sequential_io_avg;
1454 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1455 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1457 #define TNF_MIGRATED 0x01
1458 #define TNF_NO_GROUP 0x02
1459 #define TNF_SHARED 0x04
1460 #define TNF_FAULT_LOCAL 0x08
1462 #ifdef CONFIG_NUMA_BALANCING
1463 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1464 extern pid_t task_numa_group_id(struct task_struct *p);
1465 extern void set_numabalancing_state(bool enabled);
1466 extern void task_numa_free(struct task_struct *p);
1468 extern unsigned int sysctl_numa_balancing_migrate_deferred;
1470 static inline void task_numa_fault(int last_node, int node, int pages,
1474 static inline pid_t task_numa_group_id(struct task_struct *p)
1478 static inline void set_numabalancing_state(bool enabled)
1481 static inline void task_numa_free(struct task_struct *p)
1486 static inline struct pid *task_pid(struct task_struct *task)
1488 return task->pids[PIDTYPE_PID].pid;
1491 static inline struct pid *task_tgid(struct task_struct *task)
1493 return task->group_leader->pids[PIDTYPE_PID].pid;
1497 * Without tasklist or rcu lock it is not safe to dereference
1498 * the result of task_pgrp/task_session even if task == current,
1499 * we can race with another thread doing sys_setsid/sys_setpgid.
1501 static inline struct pid *task_pgrp(struct task_struct *task)
1503 return task->group_leader->pids[PIDTYPE_PGID].pid;
1506 static inline struct pid *task_session(struct task_struct *task)
1508 return task->group_leader->pids[PIDTYPE_SID].pid;
1511 struct pid_namespace;
1514 * the helpers to get the task's different pids as they are seen
1515 * from various namespaces
1517 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1518 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1520 * task_xid_nr_ns() : id seen from the ns specified;
1522 * set_task_vxid() : assigns a virtual id to a task;
1524 * see also pid_nr() etc in include/linux/pid.h
1526 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1527 struct pid_namespace *ns);
1529 static inline pid_t task_pid_nr(struct task_struct *tsk)
1534 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1535 struct pid_namespace *ns)
1537 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1540 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1542 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1546 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1551 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1553 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1555 return pid_vnr(task_tgid(tsk));
1559 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1560 struct pid_namespace *ns)
1562 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1565 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1567 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1571 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1572 struct pid_namespace *ns)
1574 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1577 static inline pid_t task_session_vnr(struct task_struct *tsk)
1579 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1582 /* obsolete, do not use */
1583 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1585 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1589 * pid_alive - check that a task structure is not stale
1590 * @p: Task structure to be checked.
1592 * Test if a process is not yet dead (at most zombie state)
1593 * If pid_alive fails, then pointers within the task structure
1594 * can be stale and must not be dereferenced.
1596 * Return: 1 if the process is alive. 0 otherwise.
1598 static inline int pid_alive(struct task_struct *p)
1600 return p->pids[PIDTYPE_PID].pid != NULL;
1604 * is_global_init - check if a task structure is init
1605 * @tsk: Task structure to be checked.
1607 * Check if a task structure is the first user space task the kernel created.
1609 * Return: 1 if the task structure is init. 0 otherwise.
1611 static inline int is_global_init(struct task_struct *tsk)
1613 return tsk->pid == 1;
1616 extern struct pid *cad_pid;
1618 extern void free_task(struct task_struct *tsk);
1619 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1621 extern void __put_task_struct(struct task_struct *t);
1623 static inline void put_task_struct(struct task_struct *t)
1625 if (atomic_dec_and_test(&t->usage))
1626 __put_task_struct(t);
1629 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1630 extern void task_cputime(struct task_struct *t,
1631 cputime_t *utime, cputime_t *stime);
1632 extern void task_cputime_scaled(struct task_struct *t,
1633 cputime_t *utimescaled, cputime_t *stimescaled);
1634 extern cputime_t task_gtime(struct task_struct *t);
1636 static inline void task_cputime(struct task_struct *t,
1637 cputime_t *utime, cputime_t *stime)
1645 static inline void task_cputime_scaled(struct task_struct *t,
1646 cputime_t *utimescaled,
1647 cputime_t *stimescaled)
1650 *utimescaled = t->utimescaled;
1652 *stimescaled = t->stimescaled;
1655 static inline cputime_t task_gtime(struct task_struct *t)
1660 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1661 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1666 #define PF_EXITING 0x00000004 /* getting shut down */
1667 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1668 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1669 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1670 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1671 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1672 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1673 #define PF_DUMPCORE 0x00000200 /* dumped core */
1674 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1675 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1676 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1677 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1678 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1679 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1680 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1681 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1682 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1683 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1684 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1685 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1686 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1687 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1688 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1689 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1690 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1691 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1692 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1693 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1694 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1695 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1698 * Only the _current_ task can read/write to tsk->flags, but other
1699 * tasks can access tsk->flags in readonly mode for example
1700 * with tsk_used_math (like during threaded core dumping).
1701 * There is however an exception to this rule during ptrace
1702 * or during fork: the ptracer task is allowed to write to the
1703 * child->flags of its traced child (same goes for fork, the parent
1704 * can write to the child->flags), because we're guaranteed the
1705 * child is not running and in turn not changing child->flags
1706 * at the same time the parent does it.
1708 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1709 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1710 #define clear_used_math() clear_stopped_child_used_math(current)
1711 #define set_used_math() set_stopped_child_used_math(current)
1712 #define conditional_stopped_child_used_math(condition, child) \
1713 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1714 #define conditional_used_math(condition) \
1715 conditional_stopped_child_used_math(condition, current)
1716 #define copy_to_stopped_child_used_math(child) \
1717 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1718 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1719 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1720 #define used_math() tsk_used_math(current)
1722 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1723 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1725 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1730 static inline unsigned int memalloc_noio_save(void)
1732 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1733 current->flags |= PF_MEMALLOC_NOIO;
1737 static inline void memalloc_noio_restore(unsigned int flags)
1739 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1743 * task->jobctl flags
1745 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1747 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1748 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1749 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1750 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1751 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1752 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1753 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1755 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1756 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1757 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1758 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1759 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1760 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1761 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1763 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1764 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1766 extern bool task_set_jobctl_pending(struct task_struct *task,
1768 extern void task_clear_jobctl_trapping(struct task_struct *task);
1769 extern void task_clear_jobctl_pending(struct task_struct *task,
1772 #ifdef CONFIG_PREEMPT_RCU
1774 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1775 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1777 static inline void rcu_copy_process(struct task_struct *p)
1779 p->rcu_read_lock_nesting = 0;
1780 p->rcu_read_unlock_special = 0;
1781 #ifdef CONFIG_TREE_PREEMPT_RCU
1782 p->rcu_blocked_node = NULL;
1783 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1784 #ifdef CONFIG_RCU_BOOST
1785 p->rcu_boost_mutex = NULL;
1786 #endif /* #ifdef CONFIG_RCU_BOOST */
1787 INIT_LIST_HEAD(&p->rcu_node_entry);
1792 static inline void rcu_copy_process(struct task_struct *p)
1798 static inline void tsk_restore_flags(struct task_struct *task,
1799 unsigned long orig_flags, unsigned long flags)
1801 task->flags &= ~flags;
1802 task->flags |= orig_flags & flags;
1806 extern void do_set_cpus_allowed(struct task_struct *p,
1807 const struct cpumask *new_mask);
1809 extern int set_cpus_allowed_ptr(struct task_struct *p,
1810 const struct cpumask *new_mask);
1812 static inline void do_set_cpus_allowed(struct task_struct *p,
1813 const struct cpumask *new_mask)
1816 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1817 const struct cpumask *new_mask)
1819 if (!cpumask_test_cpu(0, new_mask))
1825 #ifdef CONFIG_NO_HZ_COMMON
1826 void calc_load_enter_idle(void);
1827 void calc_load_exit_idle(void);
1829 static inline void calc_load_enter_idle(void) { }
1830 static inline void calc_load_exit_idle(void) { }
1831 #endif /* CONFIG_NO_HZ_COMMON */
1833 #ifndef CONFIG_CPUMASK_OFFSTACK
1834 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1836 return set_cpus_allowed_ptr(p, &new_mask);
1841 * Do not use outside of architecture code which knows its limitations.
1843 * sched_clock() has no promise of monotonicity or bounded drift between
1844 * CPUs, use (which you should not) requires disabling IRQs.
1846 * Please use one of the three interfaces below.
1848 extern unsigned long long notrace sched_clock(void);
1850 * See the comment in kernel/sched/clock.c
1852 extern u64 cpu_clock(int cpu);
1853 extern u64 local_clock(void);
1854 extern u64 sched_clock_cpu(int cpu);
1857 extern void sched_clock_init(void);
1859 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1860 static inline void sched_clock_tick(void)
1864 static inline void sched_clock_idle_sleep_event(void)
1868 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1873 * Architectures can set this to 1 if they have specified
1874 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1875 * but then during bootup it turns out that sched_clock()
1876 * is reliable after all:
1878 extern int sched_clock_stable;
1880 extern void sched_clock_tick(void);
1881 extern void sched_clock_idle_sleep_event(void);
1882 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1885 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1887 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1888 * The reason for this explicit opt-in is not to have perf penalty with
1889 * slow sched_clocks.
1891 extern void enable_sched_clock_irqtime(void);
1892 extern void disable_sched_clock_irqtime(void);
1894 static inline void enable_sched_clock_irqtime(void) {}
1895 static inline void disable_sched_clock_irqtime(void) {}
1898 extern unsigned long long
1899 task_sched_runtime(struct task_struct *task);
1901 /* sched_exec is called by processes performing an exec */
1903 extern void sched_exec(void);
1905 #define sched_exec() {}
1908 extern void sched_clock_idle_sleep_event(void);
1909 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1911 #ifdef CONFIG_HOTPLUG_CPU
1912 extern void idle_task_exit(void);
1914 static inline void idle_task_exit(void) {}
1917 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1918 extern void wake_up_nohz_cpu(int cpu);
1920 static inline void wake_up_nohz_cpu(int cpu) { }
1923 #ifdef CONFIG_NO_HZ_FULL
1924 extern bool sched_can_stop_tick(void);
1925 extern u64 scheduler_tick_max_deferment(void);
1927 static inline bool sched_can_stop_tick(void) { return false; }
1930 #ifdef CONFIG_SCHED_AUTOGROUP
1931 extern void sched_autogroup_create_attach(struct task_struct *p);
1932 extern void sched_autogroup_detach(struct task_struct *p);
1933 extern void sched_autogroup_fork(struct signal_struct *sig);
1934 extern void sched_autogroup_exit(struct signal_struct *sig);
1935 #ifdef CONFIG_PROC_FS
1936 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1937 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
1940 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1941 static inline void sched_autogroup_detach(struct task_struct *p) { }
1942 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1943 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1946 extern bool yield_to(struct task_struct *p, bool preempt);
1947 extern void set_user_nice(struct task_struct *p, long nice);
1948 extern int task_prio(const struct task_struct *p);
1949 extern int task_nice(const struct task_struct *p);
1950 extern int can_nice(const struct task_struct *p, const int nice);
1951 extern int task_curr(const struct task_struct *p);
1952 extern int idle_cpu(int cpu);
1953 extern int sched_setscheduler(struct task_struct *, int,
1954 const struct sched_param *);
1955 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1956 const struct sched_param *);
1957 extern struct task_struct *idle_task(int cpu);
1959 * is_idle_task - is the specified task an idle task?
1960 * @p: the task in question.
1962 * Return: 1 if @p is an idle task. 0 otherwise.
1964 static inline bool is_idle_task(const struct task_struct *p)
1968 extern struct task_struct *curr_task(int cpu);
1969 extern void set_curr_task(int cpu, struct task_struct *p);
1974 * The default (Linux) execution domain.
1976 extern struct exec_domain default_exec_domain;
1978 union thread_union {
1979 struct thread_info thread_info;
1980 unsigned long stack[THREAD_SIZE/sizeof(long)];
1983 #ifndef __HAVE_ARCH_KSTACK_END
1984 static inline int kstack_end(void *addr)
1986 /* Reliable end of stack detection:
1987 * Some APM bios versions misalign the stack
1989 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1993 extern union thread_union init_thread_union;
1994 extern struct task_struct init_task;
1996 extern struct mm_struct init_mm;
1998 extern struct pid_namespace init_pid_ns;
2001 * find a task by one of its numerical ids
2003 * find_task_by_pid_ns():
2004 * finds a task by its pid in the specified namespace
2005 * find_task_by_vpid():
2006 * finds a task by its virtual pid
2008 * see also find_vpid() etc in include/linux/pid.h
2011 extern struct task_struct *find_task_by_vpid(pid_t nr);
2012 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2013 struct pid_namespace *ns);
2015 /* per-UID process charging. */
2016 extern struct user_struct * alloc_uid(kuid_t);
2017 static inline struct user_struct *get_uid(struct user_struct *u)
2019 atomic_inc(&u->__count);
2022 extern void free_uid(struct user_struct *);
2024 #include <asm/current.h>
2026 extern void xtime_update(unsigned long ticks);
2028 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2029 extern int wake_up_process(struct task_struct *tsk);
2030 extern void wake_up_new_task(struct task_struct *tsk);
2032 extern void kick_process(struct task_struct *tsk);
2034 static inline void kick_process(struct task_struct *tsk) { }
2036 extern void sched_fork(unsigned long clone_flags, struct task_struct *p);
2037 extern void sched_dead(struct task_struct *p);
2039 extern void proc_caches_init(void);
2040 extern void flush_signals(struct task_struct *);
2041 extern void __flush_signals(struct task_struct *);
2042 extern void ignore_signals(struct task_struct *);
2043 extern void flush_signal_handlers(struct task_struct *, int force_default);
2044 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2046 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2048 unsigned long flags;
2051 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2052 ret = dequeue_signal(tsk, mask, info);
2053 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2058 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2060 extern void unblock_all_signals(void);
2061 extern void release_task(struct task_struct * p);
2062 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2063 extern int force_sigsegv(int, struct task_struct *);
2064 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2065 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2066 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2067 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2068 const struct cred *, u32);
2069 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2070 extern int kill_pid(struct pid *pid, int sig, int priv);
2071 extern int kill_proc_info(int, struct siginfo *, pid_t);
2072 extern __must_check bool do_notify_parent(struct task_struct *, int);
2073 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2074 extern void force_sig(int, struct task_struct *);
2075 extern int send_sig(int, struct task_struct *, int);
2076 extern int zap_other_threads(struct task_struct *p);
2077 extern struct sigqueue *sigqueue_alloc(void);
2078 extern void sigqueue_free(struct sigqueue *);
2079 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2080 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2082 static inline void restore_saved_sigmask(void)
2084 if (test_and_clear_restore_sigmask())
2085 __set_current_blocked(¤t->saved_sigmask);
2088 static inline sigset_t *sigmask_to_save(void)
2090 sigset_t *res = ¤t->blocked;
2091 if (unlikely(test_restore_sigmask()))
2092 res = ¤t->saved_sigmask;
2096 static inline int kill_cad_pid(int sig, int priv)
2098 return kill_pid(cad_pid, sig, priv);
2101 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2102 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2103 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2104 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2107 * True if we are on the alternate signal stack.
2109 static inline int on_sig_stack(unsigned long sp)
2111 #ifdef CONFIG_STACK_GROWSUP
2112 return sp >= current->sas_ss_sp &&
2113 sp - current->sas_ss_sp < current->sas_ss_size;
2115 return sp > current->sas_ss_sp &&
2116 sp - current->sas_ss_sp <= current->sas_ss_size;
2120 static inline int sas_ss_flags(unsigned long sp)
2122 return (current->sas_ss_size == 0 ? SS_DISABLE
2123 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2126 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2128 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2129 #ifdef CONFIG_STACK_GROWSUP
2130 return current->sas_ss_sp;
2132 return current->sas_ss_sp + current->sas_ss_size;
2138 * Routines for handling mm_structs
2140 extern struct mm_struct * mm_alloc(void);
2142 /* mmdrop drops the mm and the page tables */
2143 extern void __mmdrop(struct mm_struct *);
2144 static inline void mmdrop(struct mm_struct * mm)
2146 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2150 /* mmput gets rid of the mappings and all user-space */
2151 extern void mmput(struct mm_struct *);
2152 /* Grab a reference to a task's mm, if it is not already going away */
2153 extern struct mm_struct *get_task_mm(struct task_struct *task);
2155 * Grab a reference to a task's mm, if it is not already going away
2156 * and ptrace_may_access with the mode parameter passed to it
2159 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2160 /* Remove the current tasks stale references to the old mm_struct */
2161 extern void mm_release(struct task_struct *, struct mm_struct *);
2162 /* Allocate a new mm structure and copy contents from tsk->mm */
2163 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2165 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2166 struct task_struct *);
2167 extern void flush_thread(void);
2168 extern void exit_thread(void);
2170 extern void exit_files(struct task_struct *);
2171 extern void __cleanup_sighand(struct sighand_struct *);
2173 extern void exit_itimers(struct signal_struct *);
2174 extern void flush_itimer_signals(void);
2176 extern void do_group_exit(int);
2178 extern int allow_signal(int);
2179 extern int disallow_signal(int);
2181 extern int do_execve(const char *,
2182 const char __user * const __user *,
2183 const char __user * const __user *);
2184 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2185 struct task_struct *fork_idle(int);
2186 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2188 extern void set_task_comm(struct task_struct *tsk, char *from);
2189 extern char *get_task_comm(char *to, struct task_struct *tsk);
2192 void scheduler_ipi(void);
2193 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2195 static inline void scheduler_ipi(void) { }
2196 static inline unsigned long wait_task_inactive(struct task_struct *p,
2203 #define next_task(p) \
2204 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2206 #define for_each_process(p) \
2207 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2209 extern bool current_is_single_threaded(void);
2212 * Careful: do_each_thread/while_each_thread is a double loop so
2213 * 'break' will not work as expected - use goto instead.
2215 #define do_each_thread(g, t) \
2216 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2218 #define while_each_thread(g, t) \
2219 while ((t = next_thread(t)) != g)
2221 static inline int get_nr_threads(struct task_struct *tsk)
2223 return tsk->signal->nr_threads;
2226 static inline bool thread_group_leader(struct task_struct *p)
2228 return p->exit_signal >= 0;
2231 /* Do to the insanities of de_thread it is possible for a process
2232 * to have the pid of the thread group leader without actually being
2233 * the thread group leader. For iteration through the pids in proc
2234 * all we care about is that we have a task with the appropriate
2235 * pid, we don't actually care if we have the right task.
2237 static inline bool has_group_leader_pid(struct task_struct *p)
2239 return task_pid(p) == p->signal->leader_pid;
2243 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2245 return p1->signal == p2->signal;
2248 static inline struct task_struct *next_thread(const struct task_struct *p)
2250 return list_entry_rcu(p->thread_group.next,
2251 struct task_struct, thread_group);
2254 static inline int thread_group_empty(struct task_struct *p)
2256 return list_empty(&p->thread_group);
2259 #define delay_group_leader(p) \
2260 (thread_group_leader(p) && !thread_group_empty(p))
2263 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2264 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2265 * pins the final release of task.io_context. Also protects ->cpuset and
2266 * ->cgroup.subsys[]. And ->vfork_done.
2268 * Nests both inside and outside of read_lock(&tasklist_lock).
2269 * It must not be nested with write_lock_irq(&tasklist_lock),
2270 * neither inside nor outside.
2272 static inline void task_lock(struct task_struct *p)
2274 spin_lock(&p->alloc_lock);
2277 static inline void task_unlock(struct task_struct *p)
2279 spin_unlock(&p->alloc_lock);
2282 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2283 unsigned long *flags);
2285 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2286 unsigned long *flags)
2288 struct sighand_struct *ret;
2290 ret = __lock_task_sighand(tsk, flags);
2291 (void)__cond_lock(&tsk->sighand->siglock, ret);
2295 static inline void unlock_task_sighand(struct task_struct *tsk,
2296 unsigned long *flags)
2298 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2301 #ifdef CONFIG_CGROUPS
2302 static inline void threadgroup_change_begin(struct task_struct *tsk)
2304 down_read(&tsk->signal->group_rwsem);
2306 static inline void threadgroup_change_end(struct task_struct *tsk)
2308 up_read(&tsk->signal->group_rwsem);
2312 * threadgroup_lock - lock threadgroup
2313 * @tsk: member task of the threadgroup to lock
2315 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2316 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2317 * change ->group_leader/pid. This is useful for cases where the threadgroup
2318 * needs to stay stable across blockable operations.
2320 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2321 * synchronization. While held, no new task will be added to threadgroup
2322 * and no existing live task will have its PF_EXITING set.
2324 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2325 * sub-thread becomes a new leader.
2327 static inline void threadgroup_lock(struct task_struct *tsk)
2329 down_write(&tsk->signal->group_rwsem);
2333 * threadgroup_unlock - unlock threadgroup
2334 * @tsk: member task of the threadgroup to unlock
2336 * Reverse threadgroup_lock().
2338 static inline void threadgroup_unlock(struct task_struct *tsk)
2340 up_write(&tsk->signal->group_rwsem);
2343 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2344 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2345 static inline void threadgroup_lock(struct task_struct *tsk) {}
2346 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2349 #ifndef __HAVE_THREAD_FUNCTIONS
2351 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2352 #define task_stack_page(task) ((task)->stack)
2354 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2356 *task_thread_info(p) = *task_thread_info(org);
2357 task_thread_info(p)->task = p;
2360 static inline unsigned long *end_of_stack(struct task_struct *p)
2362 return (unsigned long *)(task_thread_info(p) + 1);
2367 static inline int object_is_on_stack(void *obj)
2369 void *stack = task_stack_page(current);
2371 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2374 extern void thread_info_cache_init(void);
2376 #ifdef CONFIG_DEBUG_STACK_USAGE
2377 static inline unsigned long stack_not_used(struct task_struct *p)
2379 unsigned long *n = end_of_stack(p);
2381 do { /* Skip over canary */
2385 return (unsigned long)n - (unsigned long)end_of_stack(p);
2389 /* set thread flags in other task's structures
2390 * - see asm/thread_info.h for TIF_xxxx flags available
2392 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2394 set_ti_thread_flag(task_thread_info(tsk), flag);
2397 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2399 clear_ti_thread_flag(task_thread_info(tsk), flag);
2402 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2404 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2407 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2409 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2412 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2414 return test_ti_thread_flag(task_thread_info(tsk), flag);
2417 static inline void set_tsk_need_resched(struct task_struct *tsk)
2419 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2422 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2424 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2427 static inline int test_tsk_need_resched(struct task_struct *tsk)
2429 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2432 static inline int restart_syscall(void)
2434 set_tsk_thread_flag(current, TIF_SIGPENDING);
2435 return -ERESTARTNOINTR;
2438 static inline int signal_pending(struct task_struct *p)
2440 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2443 static inline int __fatal_signal_pending(struct task_struct *p)
2445 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2448 static inline int fatal_signal_pending(struct task_struct *p)
2450 return signal_pending(p) && __fatal_signal_pending(p);
2453 static inline int signal_pending_state(long state, struct task_struct *p)
2455 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2457 if (!signal_pending(p))
2460 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2464 * cond_resched() and cond_resched_lock(): latency reduction via
2465 * explicit rescheduling in places that are safe. The return
2466 * value indicates whether a reschedule was done in fact.
2467 * cond_resched_lock() will drop the spinlock before scheduling,
2468 * cond_resched_softirq() will enable bhs before scheduling.
2470 extern int _cond_resched(void);
2472 #define cond_resched() ({ \
2473 __might_sleep(__FILE__, __LINE__, 0); \
2477 extern int __cond_resched_lock(spinlock_t *lock);
2479 #ifdef CONFIG_PREEMPT_COUNT
2480 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2482 #define PREEMPT_LOCK_OFFSET 0
2485 #define cond_resched_lock(lock) ({ \
2486 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2487 __cond_resched_lock(lock); \
2490 extern int __cond_resched_softirq(void);
2492 #define cond_resched_softirq() ({ \
2493 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2494 __cond_resched_softirq(); \
2497 static inline void cond_resched_rcu(void)
2499 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2507 * Does a critical section need to be broken due to another
2508 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2509 * but a general need for low latency)
2511 static inline int spin_needbreak(spinlock_t *lock)
2513 #ifdef CONFIG_PREEMPT
2514 return spin_is_contended(lock);
2521 * Idle thread specific functions to determine the need_resched
2522 * polling state. We have two versions, one based on TS_POLLING in
2523 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2527 static inline int tsk_is_polling(struct task_struct *p)
2529 return task_thread_info(p)->status & TS_POLLING;
2531 static inline void __current_set_polling(void)
2533 current_thread_info()->status |= TS_POLLING;
2536 static inline bool __must_check current_set_polling_and_test(void)
2538 __current_set_polling();
2541 * Polling state must be visible before we test NEED_RESCHED,
2542 * paired by resched_task()
2546 return unlikely(tif_need_resched());
2549 static inline void __current_clr_polling(void)
2551 current_thread_info()->status &= ~TS_POLLING;
2554 static inline bool __must_check current_clr_polling_and_test(void)
2556 __current_clr_polling();
2559 * Polling state must be visible before we test NEED_RESCHED,
2560 * paired by resched_task()
2564 return unlikely(tif_need_resched());
2566 #elif defined(TIF_POLLING_NRFLAG)
2567 static inline int tsk_is_polling(struct task_struct *p)
2569 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2572 static inline void __current_set_polling(void)
2574 set_thread_flag(TIF_POLLING_NRFLAG);
2577 static inline bool __must_check current_set_polling_and_test(void)
2579 __current_set_polling();
2582 * Polling state must be visible before we test NEED_RESCHED,
2583 * paired by resched_task()
2585 * XXX: assumes set/clear bit are identical barrier wise.
2587 smp_mb__after_clear_bit();
2589 return unlikely(tif_need_resched());
2592 static inline void __current_clr_polling(void)
2594 clear_thread_flag(TIF_POLLING_NRFLAG);
2597 static inline bool __must_check current_clr_polling_and_test(void)
2599 __current_clr_polling();
2602 * Polling state must be visible before we test NEED_RESCHED,
2603 * paired by resched_task()
2605 smp_mb__after_clear_bit();
2607 return unlikely(tif_need_resched());
2611 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2612 static inline void __current_set_polling(void) { }
2613 static inline void __current_clr_polling(void) { }
2615 static inline bool __must_check current_set_polling_and_test(void)
2617 return unlikely(tif_need_resched());
2619 static inline bool __must_check current_clr_polling_and_test(void)
2621 return unlikely(tif_need_resched());
2625 static __always_inline bool need_resched(void)
2627 return unlikely(tif_need_resched());
2631 * Thread group CPU time accounting.
2633 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2634 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2636 static inline void thread_group_cputime_init(struct signal_struct *sig)
2638 raw_spin_lock_init(&sig->cputimer.lock);
2642 * Reevaluate whether the task has signals pending delivery.
2643 * Wake the task if so.
2644 * This is required every time the blocked sigset_t changes.
2645 * callers must hold sighand->siglock.
2647 extern void recalc_sigpending_and_wake(struct task_struct *t);
2648 extern void recalc_sigpending(void);
2650 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2652 static inline void signal_wake_up(struct task_struct *t, bool resume)
2654 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2656 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2658 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2662 * Wrappers for p->thread_info->cpu access. No-op on UP.
2666 static inline unsigned int task_cpu(const struct task_struct *p)
2668 return task_thread_info(p)->cpu;
2671 static inline int task_node(const struct task_struct *p)
2673 return cpu_to_node(task_cpu(p));
2676 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2680 static inline unsigned int task_cpu(const struct task_struct *p)
2685 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2689 #endif /* CONFIG_SMP */
2691 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2692 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2694 #ifdef CONFIG_CGROUP_SCHED
2695 extern struct task_group root_task_group;
2696 #endif /* CONFIG_CGROUP_SCHED */
2698 extern int task_can_switch_user(struct user_struct *up,
2699 struct task_struct *tsk);
2701 #ifdef CONFIG_TASK_XACCT
2702 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2704 tsk->ioac.rchar += amt;
2707 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2709 tsk->ioac.wchar += amt;
2712 static inline void inc_syscr(struct task_struct *tsk)
2717 static inline void inc_syscw(struct task_struct *tsk)
2722 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2726 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2730 static inline void inc_syscr(struct task_struct *tsk)
2734 static inline void inc_syscw(struct task_struct *tsk)
2739 #ifndef TASK_SIZE_OF
2740 #define TASK_SIZE_OF(tsk) TASK_SIZE
2743 #ifdef CONFIG_MM_OWNER
2744 extern void mm_update_next_owner(struct mm_struct *mm);
2745 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2747 static inline void mm_update_next_owner(struct mm_struct *mm)
2751 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2754 #endif /* CONFIG_MM_OWNER */
2756 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2759 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2762 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2765 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2768 static inline unsigned long rlimit(unsigned int limit)
2770 return task_rlimit(current, limit);
2773 static inline unsigned long rlimit_max(unsigned int limit)
2775 return task_rlimit_max(current, limit);