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>
27 #include <asm/ptrace.h>
28 #include <asm/cputime.h>
30 #include <linux/smp.h>
31 #include <linux/sem.h>
32 #include <linux/signal.h>
33 #include <linux/compiler.h>
34 #include <linux/completion.h>
35 #include <linux/pid.h>
36 #include <linux/percpu.h>
37 #include <linux/topology.h>
38 #include <linux/proportions.h>
39 #include <linux/seccomp.h>
40 #include <linux/rcupdate.h>
41 #include <linux/rculist.h>
42 #include <linux/rtmutex.h>
44 #include <linux/time.h>
45 #include <linux/param.h>
46 #include <linux/resource.h>
47 #include <linux/timer.h>
48 #include <linux/hrtimer.h>
49 #include <linux/task_io_accounting.h>
50 #include <linux/latencytop.h>
51 #include <linux/cred.h>
52 #include <linux/llist.h>
53 #include <linux/uidgid.h>
54 #include <linux/gfp.h>
56 #include <asm/processor.h>
59 struct futex_pi_state;
60 struct robust_list_head;
63 struct perf_event_context;
67 * List of flags we want to share for kernel threads,
68 * if only because they are not used by them anyway.
70 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
73 * These are the constant used to fake the fixed-point load-average
74 * counting. Some notes:
75 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
76 * a load-average precision of 10 bits integer + 11 bits fractional
77 * - if you want to count load-averages more often, you need more
78 * precision, or rounding will get you. With 2-second counting freq,
79 * the EXP_n values would be 1981, 2034 and 2043 if still using only
82 extern unsigned long avenrun[]; /* Load averages */
83 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
85 #define FSHIFT 11 /* nr of bits of precision */
86 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
87 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
88 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
89 #define EXP_5 2014 /* 1/exp(5sec/5min) */
90 #define EXP_15 2037 /* 1/exp(5sec/15min) */
92 #define CALC_LOAD(load,exp,n) \
94 load += n*(FIXED_1-exp); \
97 extern unsigned long total_forks;
98 extern int nr_threads;
99 DECLARE_PER_CPU(unsigned long, process_counts);
100 extern int nr_processes(void);
101 extern unsigned long nr_running(void);
102 extern unsigned long nr_iowait(void);
103 extern unsigned long nr_iowait_cpu(int cpu);
104 extern unsigned long this_cpu_load(void);
107 extern void calc_global_load(unsigned long ticks);
108 extern void update_cpu_load_nohz(void);
110 /* Notifier for when a task gets migrated to a new CPU */
111 struct task_migration_notifier {
112 struct task_struct *task;
116 extern void register_task_migration_notifier(struct notifier_block *n);
118 extern unsigned long get_parent_ip(unsigned long addr);
120 extern void dump_cpu_task(int cpu);
125 #ifdef CONFIG_SCHED_DEBUG
126 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
127 extern void proc_sched_set_task(struct task_struct *p);
129 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
133 * Task state bitmask. NOTE! These bits are also
134 * encoded in fs/proc/array.c: get_task_state().
136 * We have two separate sets of flags: task->state
137 * is about runnability, while task->exit_state are
138 * about the task exiting. Confusing, but this way
139 * modifying one set can't modify the other one by
142 #define TASK_RUNNING 0
143 #define TASK_INTERRUPTIBLE 1
144 #define TASK_UNINTERRUPTIBLE 2
145 #define __TASK_STOPPED 4
146 #define __TASK_TRACED 8
147 /* in tsk->exit_state */
148 #define EXIT_ZOMBIE 16
150 /* in tsk->state again */
152 #define TASK_WAKEKILL 128
153 #define TASK_WAKING 256
154 #define TASK_PARKED 512
155 #define TASK_STATE_MAX 1024
157 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
159 extern char ___assert_task_state[1 - 2*!!(
160 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
162 /* Convenience macros for the sake of set_task_state */
163 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
164 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
165 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
167 /* Convenience macros for the sake of wake_up */
168 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
169 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
171 /* get_task_state() */
172 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
173 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
176 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
177 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
178 #define task_is_dead(task) ((task)->exit_state != 0)
179 #define task_is_stopped_or_traced(task) \
180 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
181 #define task_contributes_to_load(task) \
182 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
183 (task->flags & PF_FROZEN) == 0)
185 #define __set_task_state(tsk, state_value) \
186 do { (tsk)->state = (state_value); } while (0)
187 #define set_task_state(tsk, state_value) \
188 set_mb((tsk)->state, (state_value))
191 * set_current_state() includes a barrier so that the write of current->state
192 * is correctly serialised wrt the caller's subsequent test of whether to
195 * set_current_state(TASK_UNINTERRUPTIBLE);
196 * if (do_i_need_to_sleep())
199 * If the caller does not need such serialisation then use __set_current_state()
201 #define __set_current_state(state_value) \
202 do { current->state = (state_value); } while (0)
203 #define set_current_state(state_value) \
204 set_mb(current->state, (state_value))
206 /* Task command name length */
207 #define TASK_COMM_LEN 16
209 #include <linux/spinlock.h>
212 * This serializes "schedule()" and also protects
213 * the run-queue from deletions/modifications (but
214 * _adding_ to the beginning of the run-queue has
217 extern rwlock_t tasklist_lock;
218 extern spinlock_t mmlist_lock;
222 #ifdef CONFIG_PROVE_RCU
223 extern int lockdep_tasklist_lock_is_held(void);
224 #endif /* #ifdef CONFIG_PROVE_RCU */
226 extern void sched_init(void);
227 extern void sched_init_smp(void);
228 extern asmlinkage void schedule_tail(struct task_struct *prev);
229 extern void init_idle(struct task_struct *idle, int cpu);
230 extern void init_idle_bootup_task(struct task_struct *idle);
232 extern int runqueue_is_locked(int cpu);
234 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
235 extern void nohz_balance_enter_idle(int cpu);
236 extern void set_cpu_sd_state_idle(void);
237 extern int get_nohz_timer_target(void);
239 static inline void nohz_balance_enter_idle(int cpu) { }
240 static inline void set_cpu_sd_state_idle(void) { }
244 * Only dump TASK_* tasks. (0 for all tasks)
246 extern void show_state_filter(unsigned long state_filter);
248 static inline void show_state(void)
250 show_state_filter(0);
253 extern void show_regs(struct pt_regs *);
256 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
257 * task), SP is the stack pointer of the first frame that should be shown in the back
258 * trace (or NULL if the entire call-chain of the task should be shown).
260 extern void show_stack(struct task_struct *task, unsigned long *sp);
262 void io_schedule(void);
263 long io_schedule_timeout(long timeout);
265 extern void cpu_init (void);
266 extern void trap_init(void);
267 extern void update_process_times(int user);
268 extern void scheduler_tick(void);
270 extern void sched_show_task(struct task_struct *p);
272 #ifdef CONFIG_LOCKUP_DETECTOR
273 extern void touch_softlockup_watchdog(void);
274 extern void touch_softlockup_watchdog_sync(void);
275 extern void touch_all_softlockup_watchdogs(void);
276 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
278 size_t *lenp, loff_t *ppos);
279 extern unsigned int softlockup_panic;
280 void lockup_detector_init(void);
282 static inline void touch_softlockup_watchdog(void)
285 static inline void touch_softlockup_watchdog_sync(void)
288 static inline void touch_all_softlockup_watchdogs(void)
291 static inline void lockup_detector_init(void)
296 /* Attach to any functions which should be ignored in wchan output. */
297 #define __sched __attribute__((__section__(".sched.text")))
299 /* Linker adds these: start and end of __sched functions */
300 extern char __sched_text_start[], __sched_text_end[];
302 /* Is this address in the __sched functions? */
303 extern int in_sched_functions(unsigned long addr);
305 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
306 extern signed long schedule_timeout(signed long timeout);
307 extern signed long schedule_timeout_interruptible(signed long timeout);
308 extern signed long schedule_timeout_killable(signed long timeout);
309 extern signed long schedule_timeout_uninterruptible(signed long timeout);
310 asmlinkage void schedule(void);
311 extern void schedule_preempt_disabled(void);
314 struct user_namespace;
317 extern void arch_pick_mmap_layout(struct mm_struct *mm);
319 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
320 unsigned long, unsigned long);
322 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
323 unsigned long len, unsigned long pgoff,
324 unsigned long flags);
325 extern void arch_unmap_area(struct mm_struct *, unsigned long);
326 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
328 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
332 extern void set_dumpable(struct mm_struct *mm, int value);
333 extern int get_dumpable(struct mm_struct *mm);
335 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
336 #define SUID_DUMP_USER 1 /* Dump as user of process */
337 #define SUID_DUMP_ROOT 2 /* Dump as root */
341 #define MMF_DUMPABLE 0 /* core dump is permitted */
342 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
344 #define MMF_DUMPABLE_BITS 2
345 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
347 /* coredump filter bits */
348 #define MMF_DUMP_ANON_PRIVATE 2
349 #define MMF_DUMP_ANON_SHARED 3
350 #define MMF_DUMP_MAPPED_PRIVATE 4
351 #define MMF_DUMP_MAPPED_SHARED 5
352 #define MMF_DUMP_ELF_HEADERS 6
353 #define MMF_DUMP_HUGETLB_PRIVATE 7
354 #define MMF_DUMP_HUGETLB_SHARED 8
356 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
357 #define MMF_DUMP_FILTER_BITS 7
358 #define MMF_DUMP_FILTER_MASK \
359 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
360 #define MMF_DUMP_FILTER_DEFAULT \
361 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
362 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
364 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
365 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
367 # define MMF_DUMP_MASK_DEFAULT_ELF 0
369 /* leave room for more dump flags */
370 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
371 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
372 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
374 #define MMF_HAS_UPROBES 19 /* has uprobes */
375 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
377 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
379 struct sighand_struct {
381 struct k_sigaction action[_NSIG];
383 wait_queue_head_t signalfd_wqh;
386 struct pacct_struct {
389 unsigned long ac_mem;
390 cputime_t ac_utime, ac_stime;
391 unsigned long ac_minflt, ac_majflt;
402 * struct cputime - snaphsot of system and user cputime
403 * @utime: time spent in user mode
404 * @stime: time spent in system mode
406 * Gathers a generic snapshot of user and system time.
414 * struct task_cputime - collected CPU time counts
415 * @utime: time spent in user mode, in &cputime_t units
416 * @stime: time spent in kernel mode, in &cputime_t units
417 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
419 * This is an extension of struct cputime that includes the total runtime
420 * spent by the task from the scheduler point of view.
422 * As a result, this structure groups together three kinds of CPU time
423 * that are tracked for threads and thread groups. Most things considering
424 * CPU time want to group these counts together and treat all three
425 * of them in parallel.
427 struct task_cputime {
430 unsigned long long sum_exec_runtime;
432 /* Alternate field names when used to cache expirations. */
433 #define prof_exp stime
434 #define virt_exp utime
435 #define sched_exp sum_exec_runtime
437 #define INIT_CPUTIME \
438 (struct task_cputime) { \
441 .sum_exec_runtime = 0, \
445 * Disable preemption until the scheduler is running.
446 * Reset by start_kernel()->sched_init()->init_idle().
448 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
449 * before the scheduler is active -- see should_resched().
451 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
454 * struct thread_group_cputimer - thread group interval timer counts
455 * @cputime: thread group interval timers.
456 * @running: non-zero when there are timers running and
457 * @cputime receives updates.
458 * @lock: lock for fields in this struct.
460 * This structure contains the version of task_cputime, above, that is
461 * used for thread group CPU timer calculations.
463 struct thread_group_cputimer {
464 struct task_cputime cputime;
469 #include <linux/rwsem.h>
473 * NOTE! "signal_struct" does not have its own
474 * locking, because a shared signal_struct always
475 * implies a shared sighand_struct, so locking
476 * sighand_struct is always a proper superset of
477 * the locking of signal_struct.
479 struct signal_struct {
484 wait_queue_head_t wait_chldexit; /* for wait4() */
486 /* current thread group signal load-balancing target: */
487 struct task_struct *curr_target;
489 /* shared signal handling: */
490 struct sigpending shared_pending;
492 /* thread group exit support */
495 * - notify group_exit_task when ->count is equal to notify_count
496 * - everyone except group_exit_task is stopped during signal delivery
497 * of fatal signals, group_exit_task processes the signal.
500 struct task_struct *group_exit_task;
502 /* thread group stop support, overloads group_exit_code too */
503 int group_stop_count;
504 unsigned int flags; /* see SIGNAL_* flags below */
507 * PR_SET_CHILD_SUBREAPER marks a process, like a service
508 * manager, to re-parent orphan (double-forking) child processes
509 * to this process instead of 'init'. The service manager is
510 * able to receive SIGCHLD signals and is able to investigate
511 * the process until it calls wait(). All children of this
512 * process will inherit a flag if they should look for a
513 * child_subreaper process at exit.
515 unsigned int is_child_subreaper:1;
516 unsigned int has_child_subreaper:1;
518 /* POSIX.1b Interval Timers */
520 struct list_head posix_timers;
522 /* ITIMER_REAL timer for the process */
523 struct hrtimer real_timer;
524 struct pid *leader_pid;
525 ktime_t it_real_incr;
528 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
529 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
530 * values are defined to 0 and 1 respectively
532 struct cpu_itimer it[2];
535 * Thread group totals for process CPU timers.
536 * See thread_group_cputimer(), et al, for details.
538 struct thread_group_cputimer cputimer;
540 /* Earliest-expiration cache. */
541 struct task_cputime cputime_expires;
543 struct list_head cpu_timers[3];
545 struct pid *tty_old_pgrp;
547 /* boolean value for session group leader */
550 struct tty_struct *tty; /* NULL if no tty */
552 #ifdef CONFIG_SCHED_AUTOGROUP
553 struct autogroup *autogroup;
556 * Cumulative resource counters for dead threads in the group,
557 * and for reaped dead child processes forked by this group.
558 * Live threads maintain their own counters and add to these
559 * in __exit_signal, except for the group leader.
561 cputime_t utime, stime, cutime, cstime;
564 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
565 struct cputime prev_cputime;
567 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
568 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
569 unsigned long inblock, oublock, cinblock, coublock;
570 unsigned long maxrss, cmaxrss;
571 struct task_io_accounting ioac;
574 * Cumulative ns of schedule CPU time fo dead threads in the
575 * group, not including a zombie group leader, (This only differs
576 * from jiffies_to_ns(utime + stime) if sched_clock uses something
577 * other than jiffies.)
579 unsigned long long sum_sched_runtime;
582 * We don't bother to synchronize most readers of this at all,
583 * because there is no reader checking a limit that actually needs
584 * to get both rlim_cur and rlim_max atomically, and either one
585 * alone is a single word that can safely be read normally.
586 * getrlimit/setrlimit use task_lock(current->group_leader) to
587 * protect this instead of the siglock, because they really
588 * have no need to disable irqs.
590 struct rlimit rlim[RLIM_NLIMITS];
592 #ifdef CONFIG_BSD_PROCESS_ACCT
593 struct pacct_struct pacct; /* per-process accounting information */
595 #ifdef CONFIG_TASKSTATS
596 struct taskstats *stats;
600 unsigned audit_tty_log_passwd;
601 struct tty_audit_buf *tty_audit_buf;
603 #ifdef CONFIG_CGROUPS
605 * group_rwsem prevents new tasks from entering the threadgroup and
606 * member tasks from exiting,a more specifically, setting of
607 * PF_EXITING. fork and exit paths are protected with this rwsem
608 * using threadgroup_change_begin/end(). Users which require
609 * threadgroup to remain stable should use threadgroup_[un]lock()
610 * which also takes care of exec path. Currently, cgroup is the
613 struct rw_semaphore group_rwsem;
616 oom_flags_t oom_flags;
617 short oom_score_adj; /* OOM kill score adjustment */
618 short oom_score_adj_min; /* OOM kill score adjustment min value.
619 * Only settable by CAP_SYS_RESOURCE. */
621 struct mutex cred_guard_mutex; /* guard against foreign influences on
622 * credential calculations
623 * (notably. ptrace) */
627 * Bits in flags field of signal_struct.
629 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
630 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
631 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
632 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
634 * Pending notifications to parent.
636 #define SIGNAL_CLD_STOPPED 0x00000010
637 #define SIGNAL_CLD_CONTINUED 0x00000020
638 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
640 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
642 /* If true, all threads except ->group_exit_task have pending SIGKILL */
643 static inline int signal_group_exit(const struct signal_struct *sig)
645 return (sig->flags & SIGNAL_GROUP_EXIT) ||
646 (sig->group_exit_task != NULL);
650 * Some day this will be a full-fledged user tracking system..
653 atomic_t __count; /* reference count */
654 atomic_t processes; /* How many processes does this user have? */
655 atomic_t files; /* How many open files does this user have? */
656 atomic_t sigpending; /* How many pending signals does this user have? */
657 #ifdef CONFIG_INOTIFY_USER
658 atomic_t inotify_watches; /* How many inotify watches does this user have? */
659 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
661 #ifdef CONFIG_FANOTIFY
662 atomic_t fanotify_listeners;
665 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
667 #ifdef CONFIG_POSIX_MQUEUE
668 /* protected by mq_lock */
669 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
671 unsigned long locked_shm; /* How many pages of mlocked shm ? */
674 struct key *uid_keyring; /* UID specific keyring */
675 struct key *session_keyring; /* UID's default session keyring */
678 /* Hash table maintenance information */
679 struct hlist_node uidhash_node;
682 #ifdef CONFIG_PERF_EVENTS
683 atomic_long_t locked_vm;
687 extern int uids_sysfs_init(void);
689 extern struct user_struct *find_user(kuid_t);
691 extern struct user_struct root_user;
692 #define INIT_USER (&root_user)
695 struct backing_dev_info;
696 struct reclaim_state;
698 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
700 /* cumulative counters */
701 unsigned long pcount; /* # of times run on this cpu */
702 unsigned long long run_delay; /* time spent waiting on a runqueue */
705 unsigned long long last_arrival,/* when we last ran on a cpu */
706 last_queued; /* when we were last queued to run */
708 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
710 #ifdef CONFIG_TASK_DELAY_ACCT
711 struct task_delay_info {
713 unsigned int flags; /* Private per-task flags */
715 /* For each stat XXX, add following, aligned appropriately
717 * struct timespec XXX_start, XXX_end;
721 * Atomicity of updates to XXX_delay, XXX_count protected by
722 * single lock above (split into XXX_lock if contention is an issue).
726 * XXX_count is incremented on every XXX operation, the delay
727 * associated with the operation is added to XXX_delay.
728 * XXX_delay contains the accumulated delay time in nanoseconds.
730 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
731 u64 blkio_delay; /* wait for sync block io completion */
732 u64 swapin_delay; /* wait for swapin block io completion */
733 u32 blkio_count; /* total count of the number of sync block */
734 /* io operations performed */
735 u32 swapin_count; /* total count of the number of swapin block */
736 /* io operations performed */
738 struct timespec freepages_start, freepages_end;
739 u64 freepages_delay; /* wait for memory reclaim */
740 u32 freepages_count; /* total count of memory reclaim */
742 #endif /* CONFIG_TASK_DELAY_ACCT */
744 static inline int sched_info_on(void)
746 #ifdef CONFIG_SCHEDSTATS
748 #elif defined(CONFIG_TASK_DELAY_ACCT)
749 extern int delayacct_on;
764 * Increase resolution of cpu_power calculations
766 #define SCHED_POWER_SHIFT 10
767 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
770 * sched-domains (multiprocessor balancing) declarations:
773 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
774 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
775 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
776 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
777 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
778 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
779 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
780 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
781 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
782 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
783 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
784 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
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 #ifdef CONFIG_SCHEDSTATS
829 /* load_balance() stats */
830 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
831 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
832 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
833 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
834 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
835 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
836 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
837 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
839 /* Active load balancing */
840 unsigned int alb_count;
841 unsigned int alb_failed;
842 unsigned int alb_pushed;
844 /* SD_BALANCE_EXEC stats */
845 unsigned int sbe_count;
846 unsigned int sbe_balanced;
847 unsigned int sbe_pushed;
849 /* SD_BALANCE_FORK stats */
850 unsigned int sbf_count;
851 unsigned int sbf_balanced;
852 unsigned int sbf_pushed;
854 /* try_to_wake_up() stats */
855 unsigned int ttwu_wake_remote;
856 unsigned int ttwu_move_affine;
857 unsigned int ttwu_move_balance;
859 #ifdef CONFIG_SCHED_DEBUG
863 void *private; /* used during construction */
864 struct rcu_head rcu; /* used during destruction */
867 unsigned int span_weight;
869 * Span of all CPUs in this domain.
871 * NOTE: this field is variable length. (Allocated dynamically
872 * by attaching extra space to the end of the structure,
873 * depending on how many CPUs the kernel has booted up with)
875 unsigned long span[0];
878 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
880 return to_cpumask(sd->span);
883 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
884 struct sched_domain_attr *dattr_new);
886 /* Allocate an array of sched domains, for partition_sched_domains(). */
887 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
888 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
890 bool cpus_share_cache(int this_cpu, int that_cpu);
892 #else /* CONFIG_SMP */
894 struct sched_domain_attr;
897 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
898 struct sched_domain_attr *dattr_new)
902 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
907 #endif /* !CONFIG_SMP */
910 struct io_context; /* See blkdev.h */
913 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
914 extern void prefetch_stack(struct task_struct *t);
916 static inline void prefetch_stack(struct task_struct *t) { }
919 struct audit_context; /* See audit.c */
921 struct pipe_inode_info;
922 struct uts_namespace;
925 unsigned long weight, inv_weight;
930 * These sums represent an infinite geometric series and so are bound
931 * above by 1024/(1-y). Thus we only need a u32 to store them for for all
932 * choices of y < 1-2^(-32)*1024.
934 u32 runnable_avg_sum, runnable_avg_period;
935 u64 last_runnable_update;
937 unsigned long load_avg_contrib;
940 #ifdef CONFIG_SCHEDSTATS
941 struct sched_statistics {
951 s64 sum_sleep_runtime;
958 u64 nr_migrations_cold;
959 u64 nr_failed_migrations_affine;
960 u64 nr_failed_migrations_running;
961 u64 nr_failed_migrations_hot;
962 u64 nr_forced_migrations;
966 u64 nr_wakeups_migrate;
967 u64 nr_wakeups_local;
968 u64 nr_wakeups_remote;
969 u64 nr_wakeups_affine;
970 u64 nr_wakeups_affine_attempts;
971 u64 nr_wakeups_passive;
976 struct sched_entity {
977 struct load_weight load; /* for load-balancing */
978 struct rb_node run_node;
979 struct list_head group_node;
983 u64 sum_exec_runtime;
985 u64 prev_sum_exec_runtime;
989 #ifdef CONFIG_SCHEDSTATS
990 struct sched_statistics statistics;
993 #ifdef CONFIG_FAIR_GROUP_SCHED
994 struct sched_entity *parent;
995 /* rq on which this entity is (to be) queued: */
996 struct cfs_rq *cfs_rq;
997 /* rq "owned" by this entity/group: */
1002 * Load-tracking only depends on SMP, FAIR_GROUP_SCHED dependency below may be
1003 * removed when useful for applications beyond shares distribution (e.g.
1006 #if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)
1007 /* Per-entity load-tracking */
1008 struct sched_avg avg;
1012 struct sched_rt_entity {
1013 struct list_head run_list;
1014 unsigned long timeout;
1015 unsigned long watchdog_stamp;
1016 unsigned int time_slice;
1018 struct sched_rt_entity *back;
1019 #ifdef CONFIG_RT_GROUP_SCHED
1020 struct sched_rt_entity *parent;
1021 /* rq on which this entity is (to be) queued: */
1022 struct rt_rq *rt_rq;
1023 /* rq "owned" by this entity/group: */
1031 enum perf_event_task_context {
1032 perf_invalid_context = -1,
1033 perf_hw_context = 0,
1035 perf_nr_task_contexts,
1038 struct task_struct {
1039 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1042 unsigned int flags; /* per process flags, defined below */
1043 unsigned int ptrace;
1046 struct llist_node wake_entry;
1051 int prio, static_prio, normal_prio;
1052 unsigned int rt_priority;
1053 const struct sched_class *sched_class;
1054 struct sched_entity se;
1055 struct sched_rt_entity rt;
1056 #ifdef CONFIG_CGROUP_SCHED
1057 struct task_group *sched_task_group;
1060 #ifdef CONFIG_PREEMPT_NOTIFIERS
1061 /* list of struct preempt_notifier: */
1062 struct hlist_head preempt_notifiers;
1066 * fpu_counter contains the number of consecutive context switches
1067 * that the FPU is used. If this is over a threshold, the lazy fpu
1068 * saving becomes unlazy to save the trap. This is an unsigned char
1069 * so that after 256 times the counter wraps and the behavior turns
1070 * lazy again; this to deal with bursty apps that only use FPU for
1073 unsigned char fpu_counter;
1074 #ifdef CONFIG_BLK_DEV_IO_TRACE
1075 unsigned int btrace_seq;
1078 unsigned int policy;
1079 int nr_cpus_allowed;
1080 cpumask_t cpus_allowed;
1082 #ifdef CONFIG_PREEMPT_RCU
1083 int rcu_read_lock_nesting;
1084 char rcu_read_unlock_special;
1085 struct list_head rcu_node_entry;
1086 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1087 #ifdef CONFIG_TREE_PREEMPT_RCU
1088 struct rcu_node *rcu_blocked_node;
1089 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1090 #ifdef CONFIG_RCU_BOOST
1091 struct rt_mutex *rcu_boost_mutex;
1092 #endif /* #ifdef CONFIG_RCU_BOOST */
1094 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1095 struct sched_info sched_info;
1098 struct list_head tasks;
1100 struct plist_node pushable_tasks;
1103 struct mm_struct *mm, *active_mm;
1104 #ifdef CONFIG_COMPAT_BRK
1105 unsigned brk_randomized:1;
1107 #if defined(SPLIT_RSS_COUNTING)
1108 struct task_rss_stat rss_stat;
1112 int exit_code, exit_signal;
1113 int pdeath_signal; /* The signal sent when the parent dies */
1114 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1116 /* Used for emulating ABI behavior of previous Linux versions */
1117 unsigned int personality;
1119 unsigned did_exec:1;
1120 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1122 unsigned in_iowait:1;
1124 /* task may not gain privileges */
1125 unsigned no_new_privs:1;
1127 /* Revert to default priority/policy when forking */
1128 unsigned sched_reset_on_fork:1;
1129 unsigned sched_contributes_to_load:1;
1134 #ifdef CONFIG_CC_STACKPROTECTOR
1135 /* Canary value for the -fstack-protector gcc feature */
1136 unsigned long stack_canary;
1139 * pointers to (original) parent process, youngest child, younger sibling,
1140 * older sibling, respectively. (p->father can be replaced with
1141 * p->real_parent->pid)
1143 struct task_struct __rcu *real_parent; /* real parent process */
1144 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1146 * children/sibling forms the list of my natural children
1148 struct list_head children; /* list of my children */
1149 struct list_head sibling; /* linkage in my parent's children list */
1150 struct task_struct *group_leader; /* threadgroup leader */
1153 * ptraced is the list of tasks this task is using ptrace on.
1154 * This includes both natural children and PTRACE_ATTACH targets.
1155 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1157 struct list_head ptraced;
1158 struct list_head ptrace_entry;
1160 /* PID/PID hash table linkage. */
1161 struct pid_link pids[PIDTYPE_MAX];
1162 struct list_head thread_group;
1164 struct completion *vfork_done; /* for vfork() */
1165 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1166 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1168 cputime_t utime, stime, utimescaled, stimescaled;
1170 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1171 struct cputime prev_cputime;
1173 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1174 seqlock_t vtime_seqlock;
1175 unsigned long long vtime_snap;
1180 } vtime_snap_whence;
1182 unsigned long nvcsw, nivcsw; /* context switch counts */
1183 struct timespec start_time; /* monotonic time */
1184 struct timespec real_start_time; /* boot based time */
1185 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1186 unsigned long min_flt, maj_flt;
1188 struct task_cputime cputime_expires;
1189 struct list_head cpu_timers[3];
1191 /* process credentials */
1192 const struct cred __rcu *real_cred; /* objective and real subjective task
1193 * credentials (COW) */
1194 const struct cred __rcu *cred; /* effective (overridable) subjective task
1195 * credentials (COW) */
1196 char comm[TASK_COMM_LEN]; /* executable name excluding path
1197 - access with [gs]et_task_comm (which lock
1198 it with task_lock())
1199 - initialized normally by setup_new_exec */
1200 /* file system info */
1201 int link_count, total_link_count;
1202 #ifdef CONFIG_SYSVIPC
1204 struct sysv_sem sysvsem;
1206 #ifdef CONFIG_DETECT_HUNG_TASK
1207 /* hung task detection */
1208 unsigned long last_switch_count;
1210 /* CPU-specific state of this task */
1211 struct thread_struct thread;
1212 /* filesystem information */
1213 struct fs_struct *fs;
1214 /* open file information */
1215 struct files_struct *files;
1217 struct nsproxy *nsproxy;
1218 /* signal handlers */
1219 struct signal_struct *signal;
1220 struct sighand_struct *sighand;
1222 sigset_t blocked, real_blocked;
1223 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1224 struct sigpending pending;
1226 unsigned long sas_ss_sp;
1228 int (*notifier)(void *priv);
1229 void *notifier_data;
1230 sigset_t *notifier_mask;
1231 struct callback_head *task_works;
1233 struct audit_context *audit_context;
1234 #ifdef CONFIG_AUDITSYSCALL
1236 unsigned int sessionid;
1238 struct seccomp seccomp;
1240 /* Thread group tracking */
1243 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1245 spinlock_t alloc_lock;
1247 /* Protection of the PI data structures: */
1248 raw_spinlock_t pi_lock;
1250 #ifdef CONFIG_RT_MUTEXES
1251 /* PI waiters blocked on a rt_mutex held by this task */
1252 struct plist_head pi_waiters;
1253 /* Deadlock detection and priority inheritance handling */
1254 struct rt_mutex_waiter *pi_blocked_on;
1257 #ifdef CONFIG_DEBUG_MUTEXES
1258 /* mutex deadlock detection */
1259 struct mutex_waiter *blocked_on;
1261 #ifdef CONFIG_TRACE_IRQFLAGS
1262 unsigned int irq_events;
1263 unsigned long hardirq_enable_ip;
1264 unsigned long hardirq_disable_ip;
1265 unsigned int hardirq_enable_event;
1266 unsigned int hardirq_disable_event;
1267 int hardirqs_enabled;
1268 int hardirq_context;
1269 unsigned long softirq_disable_ip;
1270 unsigned long softirq_enable_ip;
1271 unsigned int softirq_disable_event;
1272 unsigned int softirq_enable_event;
1273 int softirqs_enabled;
1274 int softirq_context;
1276 #ifdef CONFIG_LOCKDEP
1277 # define MAX_LOCK_DEPTH 48UL
1280 unsigned int lockdep_recursion;
1281 struct held_lock held_locks[MAX_LOCK_DEPTH];
1282 gfp_t lockdep_reclaim_gfp;
1285 /* journalling filesystem info */
1288 /* stacked block device info */
1289 struct bio_list *bio_list;
1292 /* stack plugging */
1293 struct blk_plug *plug;
1297 struct reclaim_state *reclaim_state;
1299 struct backing_dev_info *backing_dev_info;
1301 struct io_context *io_context;
1303 unsigned long ptrace_message;
1304 siginfo_t *last_siginfo; /* For ptrace use. */
1305 struct task_io_accounting ioac;
1306 #if defined(CONFIG_TASK_XACCT)
1307 u64 acct_rss_mem1; /* accumulated rss usage */
1308 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1309 cputime_t acct_timexpd; /* stime + utime since last update */
1311 #ifdef CONFIG_CPUSETS
1312 nodemask_t mems_allowed; /* Protected by alloc_lock */
1313 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1314 int cpuset_mem_spread_rotor;
1315 int cpuset_slab_spread_rotor;
1317 #ifdef CONFIG_CGROUPS
1318 /* Control Group info protected by css_set_lock */
1319 struct css_set __rcu *cgroups;
1320 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1321 struct list_head cg_list;
1324 struct robust_list_head __user *robust_list;
1325 #ifdef CONFIG_COMPAT
1326 struct compat_robust_list_head __user *compat_robust_list;
1328 struct list_head pi_state_list;
1329 struct futex_pi_state *pi_state_cache;
1331 #ifdef CONFIG_PERF_EVENTS
1332 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1333 struct mutex perf_event_mutex;
1334 struct list_head perf_event_list;
1337 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1339 short pref_node_fork;
1341 #ifdef CONFIG_NUMA_BALANCING
1343 int numa_migrate_seq;
1344 unsigned int numa_scan_period;
1345 u64 node_stamp; /* migration stamp */
1346 struct callback_head numa_work;
1347 #endif /* CONFIG_NUMA_BALANCING */
1349 struct rcu_head rcu;
1352 * cache last used pipe for splice
1354 struct pipe_inode_info *splice_pipe;
1356 struct page_frag task_frag;
1358 #ifdef CONFIG_TASK_DELAY_ACCT
1359 struct task_delay_info *delays;
1361 #ifdef CONFIG_FAULT_INJECTION
1365 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1366 * balance_dirty_pages() for some dirty throttling pause
1369 int nr_dirtied_pause;
1370 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1372 #ifdef CONFIG_LATENCYTOP
1373 int latency_record_count;
1374 struct latency_record latency_record[LT_SAVECOUNT];
1377 * time slack values; these are used to round up poll() and
1378 * select() etc timeout values. These are in nanoseconds.
1380 unsigned long timer_slack_ns;
1381 unsigned long default_timer_slack_ns;
1383 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1384 /* Index of current stored address in ret_stack */
1386 /* Stack of return addresses for return function tracing */
1387 struct ftrace_ret_stack *ret_stack;
1388 /* time stamp for last schedule */
1389 unsigned long long ftrace_timestamp;
1391 * Number of functions that haven't been traced
1392 * because of depth overrun.
1394 atomic_t trace_overrun;
1395 /* Pause for the tracing */
1396 atomic_t tracing_graph_pause;
1398 #ifdef CONFIG_TRACING
1399 /* state flags for use by tracers */
1400 unsigned long trace;
1401 /* bitmask and counter of trace recursion */
1402 unsigned long trace_recursion;
1403 #endif /* CONFIG_TRACING */
1404 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1405 struct memcg_batch_info {
1406 int do_batch; /* incremented when batch uncharge started */
1407 struct mem_cgroup *memcg; /* target memcg of uncharge */
1408 unsigned long nr_pages; /* uncharged usage */
1409 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1411 unsigned int memcg_kmem_skip_account;
1413 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1414 atomic_t ptrace_bp_refcnt;
1416 #ifdef CONFIG_UPROBES
1417 struct uprobe_task *utask;
1419 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1420 unsigned int sequential_io;
1421 unsigned int sequential_io_avg;
1425 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1426 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1428 #ifdef CONFIG_NUMA_BALANCING
1429 extern void task_numa_fault(int node, int pages, bool migrated);
1430 extern void set_numabalancing_state(bool enabled);
1432 static inline void task_numa_fault(int node, int pages, bool migrated)
1435 static inline void set_numabalancing_state(bool enabled)
1440 static inline struct pid *task_pid(struct task_struct *task)
1442 return task->pids[PIDTYPE_PID].pid;
1445 static inline struct pid *task_tgid(struct task_struct *task)
1447 return task->group_leader->pids[PIDTYPE_PID].pid;
1451 * Without tasklist or rcu lock it is not safe to dereference
1452 * the result of task_pgrp/task_session even if task == current,
1453 * we can race with another thread doing sys_setsid/sys_setpgid.
1455 static inline struct pid *task_pgrp(struct task_struct *task)
1457 return task->group_leader->pids[PIDTYPE_PGID].pid;
1460 static inline struct pid *task_session(struct task_struct *task)
1462 return task->group_leader->pids[PIDTYPE_SID].pid;
1465 struct pid_namespace;
1468 * the helpers to get the task's different pids as they are seen
1469 * from various namespaces
1471 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1472 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1474 * task_xid_nr_ns() : id seen from the ns specified;
1476 * set_task_vxid() : assigns a virtual id to a task;
1478 * see also pid_nr() etc in include/linux/pid.h
1480 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1481 struct pid_namespace *ns);
1483 static inline pid_t task_pid_nr(struct task_struct *tsk)
1488 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1489 struct pid_namespace *ns)
1491 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1494 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1496 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1500 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1505 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1507 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1509 return pid_vnr(task_tgid(tsk));
1513 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1514 struct pid_namespace *ns)
1516 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1519 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1521 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1525 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1526 struct pid_namespace *ns)
1528 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1531 static inline pid_t task_session_vnr(struct task_struct *tsk)
1533 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1536 /* obsolete, do not use */
1537 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1539 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1543 * pid_alive - check that a task structure is not stale
1544 * @p: Task structure to be checked.
1546 * Test if a process is not yet dead (at most zombie state)
1547 * If pid_alive fails, then pointers within the task structure
1548 * can be stale and must not be dereferenced.
1550 static inline int pid_alive(struct task_struct *p)
1552 return p->pids[PIDTYPE_PID].pid != NULL;
1556 * is_global_init - check if a task structure is init
1557 * @tsk: Task structure to be checked.
1559 * Check if a task structure is the first user space task the kernel created.
1561 static inline int is_global_init(struct task_struct *tsk)
1563 return tsk->pid == 1;
1566 extern struct pid *cad_pid;
1568 extern void free_task(struct task_struct *tsk);
1569 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1571 extern void __put_task_struct(struct task_struct *t);
1573 static inline void put_task_struct(struct task_struct *t)
1575 if (atomic_dec_and_test(&t->usage))
1576 __put_task_struct(t);
1579 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1580 extern void task_cputime(struct task_struct *t,
1581 cputime_t *utime, cputime_t *stime);
1582 extern void task_cputime_scaled(struct task_struct *t,
1583 cputime_t *utimescaled, cputime_t *stimescaled);
1584 extern cputime_t task_gtime(struct task_struct *t);
1586 static inline void task_cputime(struct task_struct *t,
1587 cputime_t *utime, cputime_t *stime)
1595 static inline void task_cputime_scaled(struct task_struct *t,
1596 cputime_t *utimescaled,
1597 cputime_t *stimescaled)
1600 *utimescaled = t->utimescaled;
1602 *stimescaled = t->stimescaled;
1605 static inline cputime_t task_gtime(struct task_struct *t)
1610 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1611 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1616 #define PF_EXITING 0x00000004 /* getting shut down */
1617 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1618 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1619 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1620 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1621 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1622 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1623 #define PF_DUMPCORE 0x00000200 /* dumped core */
1624 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1625 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1626 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1627 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1628 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1629 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1630 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1631 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1632 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1633 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1634 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1635 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1636 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1637 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1638 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1639 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1640 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1641 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1642 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1643 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1644 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1647 * Only the _current_ task can read/write to tsk->flags, but other
1648 * tasks can access tsk->flags in readonly mode for example
1649 * with tsk_used_math (like during threaded core dumping).
1650 * There is however an exception to this rule during ptrace
1651 * or during fork: the ptracer task is allowed to write to the
1652 * child->flags of its traced child (same goes for fork, the parent
1653 * can write to the child->flags), because we're guaranteed the
1654 * child is not running and in turn not changing child->flags
1655 * at the same time the parent does it.
1657 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1658 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1659 #define clear_used_math() clear_stopped_child_used_math(current)
1660 #define set_used_math() set_stopped_child_used_math(current)
1661 #define conditional_stopped_child_used_math(condition, child) \
1662 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1663 #define conditional_used_math(condition) \
1664 conditional_stopped_child_used_math(condition, current)
1665 #define copy_to_stopped_child_used_math(child) \
1666 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1667 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1668 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1669 #define used_math() tsk_used_math(current)
1671 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1672 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1674 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1679 static inline unsigned int memalloc_noio_save(void)
1681 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1682 current->flags |= PF_MEMALLOC_NOIO;
1686 static inline void memalloc_noio_restore(unsigned int flags)
1688 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1692 * task->jobctl flags
1694 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1696 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1697 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1698 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1699 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1700 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1701 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1702 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1704 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1705 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1706 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1707 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1708 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1709 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1710 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1712 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1713 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1715 extern bool task_set_jobctl_pending(struct task_struct *task,
1717 extern void task_clear_jobctl_trapping(struct task_struct *task);
1718 extern void task_clear_jobctl_pending(struct task_struct *task,
1721 #ifdef CONFIG_PREEMPT_RCU
1723 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1724 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1726 static inline void rcu_copy_process(struct task_struct *p)
1728 p->rcu_read_lock_nesting = 0;
1729 p->rcu_read_unlock_special = 0;
1730 #ifdef CONFIG_TREE_PREEMPT_RCU
1731 p->rcu_blocked_node = NULL;
1732 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1733 #ifdef CONFIG_RCU_BOOST
1734 p->rcu_boost_mutex = NULL;
1735 #endif /* #ifdef CONFIG_RCU_BOOST */
1736 INIT_LIST_HEAD(&p->rcu_node_entry);
1741 static inline void rcu_copy_process(struct task_struct *p)
1747 static inline void tsk_restore_flags(struct task_struct *task,
1748 unsigned long orig_flags, unsigned long flags)
1750 task->flags &= ~flags;
1751 task->flags |= orig_flags & flags;
1755 extern void do_set_cpus_allowed(struct task_struct *p,
1756 const struct cpumask *new_mask);
1758 extern int set_cpus_allowed_ptr(struct task_struct *p,
1759 const struct cpumask *new_mask);
1761 static inline void do_set_cpus_allowed(struct task_struct *p,
1762 const struct cpumask *new_mask)
1765 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1766 const struct cpumask *new_mask)
1768 if (!cpumask_test_cpu(0, new_mask))
1774 #ifdef CONFIG_NO_HZ_COMMON
1775 void calc_load_enter_idle(void);
1776 void calc_load_exit_idle(void);
1778 static inline void calc_load_enter_idle(void) { }
1779 static inline void calc_load_exit_idle(void) { }
1780 #endif /* CONFIG_NO_HZ_COMMON */
1782 #ifndef CONFIG_CPUMASK_OFFSTACK
1783 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1785 return set_cpus_allowed_ptr(p, &new_mask);
1790 * Do not use outside of architecture code which knows its limitations.
1792 * sched_clock() has no promise of monotonicity or bounded drift between
1793 * CPUs, use (which you should not) requires disabling IRQs.
1795 * Please use one of the three interfaces below.
1797 extern unsigned long long notrace sched_clock(void);
1799 * See the comment in kernel/sched/clock.c
1801 extern u64 cpu_clock(int cpu);
1802 extern u64 local_clock(void);
1803 extern u64 sched_clock_cpu(int cpu);
1806 extern void sched_clock_init(void);
1808 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1809 static inline void sched_clock_tick(void)
1813 static inline void sched_clock_idle_sleep_event(void)
1817 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1822 * Architectures can set this to 1 if they have specified
1823 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1824 * but then during bootup it turns out that sched_clock()
1825 * is reliable after all:
1827 extern int sched_clock_stable;
1829 extern void sched_clock_tick(void);
1830 extern void sched_clock_idle_sleep_event(void);
1831 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1834 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1836 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1837 * The reason for this explicit opt-in is not to have perf penalty with
1838 * slow sched_clocks.
1840 extern void enable_sched_clock_irqtime(void);
1841 extern void disable_sched_clock_irqtime(void);
1843 static inline void enable_sched_clock_irqtime(void) {}
1844 static inline void disable_sched_clock_irqtime(void) {}
1847 extern unsigned long long
1848 task_sched_runtime(struct task_struct *task);
1850 /* sched_exec is called by processes performing an exec */
1852 extern void sched_exec(void);
1854 #define sched_exec() {}
1857 extern void sched_clock_idle_sleep_event(void);
1858 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1860 #ifdef CONFIG_HOTPLUG_CPU
1861 extern void idle_task_exit(void);
1863 static inline void idle_task_exit(void) {}
1866 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1867 extern void wake_up_nohz_cpu(int cpu);
1869 static inline void wake_up_nohz_cpu(int cpu) { }
1872 #ifdef CONFIG_NO_HZ_FULL
1873 extern bool sched_can_stop_tick(void);
1874 extern u64 scheduler_tick_max_deferment(void);
1876 static inline bool sched_can_stop_tick(void) { return false; }
1879 #ifdef CONFIG_SCHED_AUTOGROUP
1880 extern void sched_autogroup_create_attach(struct task_struct *p);
1881 extern void sched_autogroup_detach(struct task_struct *p);
1882 extern void sched_autogroup_fork(struct signal_struct *sig);
1883 extern void sched_autogroup_exit(struct signal_struct *sig);
1884 #ifdef CONFIG_PROC_FS
1885 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1886 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
1889 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1890 static inline void sched_autogroup_detach(struct task_struct *p) { }
1891 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1892 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1895 extern bool yield_to(struct task_struct *p, bool preempt);
1896 extern void set_user_nice(struct task_struct *p, long nice);
1897 extern int task_prio(const struct task_struct *p);
1898 extern int task_nice(const struct task_struct *p);
1899 extern int can_nice(const struct task_struct *p, const int nice);
1900 extern int task_curr(const struct task_struct *p);
1901 extern int idle_cpu(int cpu);
1902 extern int sched_setscheduler(struct task_struct *, int,
1903 const struct sched_param *);
1904 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1905 const struct sched_param *);
1906 extern struct task_struct *idle_task(int cpu);
1908 * is_idle_task - is the specified task an idle task?
1909 * @p: the task in question.
1911 static inline bool is_idle_task(const struct task_struct *p)
1915 extern struct task_struct *curr_task(int cpu);
1916 extern void set_curr_task(int cpu, struct task_struct *p);
1921 * The default (Linux) execution domain.
1923 extern struct exec_domain default_exec_domain;
1925 union thread_union {
1926 struct thread_info thread_info;
1927 unsigned long stack[THREAD_SIZE/sizeof(long)];
1930 #ifndef __HAVE_ARCH_KSTACK_END
1931 static inline int kstack_end(void *addr)
1933 /* Reliable end of stack detection:
1934 * Some APM bios versions misalign the stack
1936 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1940 extern union thread_union init_thread_union;
1941 extern struct task_struct init_task;
1943 extern struct mm_struct init_mm;
1945 extern struct pid_namespace init_pid_ns;
1948 * find a task by one of its numerical ids
1950 * find_task_by_pid_ns():
1951 * finds a task by its pid in the specified namespace
1952 * find_task_by_vpid():
1953 * finds a task by its virtual pid
1955 * see also find_vpid() etc in include/linux/pid.h
1958 extern struct task_struct *find_task_by_vpid(pid_t nr);
1959 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1960 struct pid_namespace *ns);
1962 extern void __set_special_pids(struct pid *pid);
1964 /* per-UID process charging. */
1965 extern struct user_struct * alloc_uid(kuid_t);
1966 static inline struct user_struct *get_uid(struct user_struct *u)
1968 atomic_inc(&u->__count);
1971 extern void free_uid(struct user_struct *);
1973 #include <asm/current.h>
1975 extern void xtime_update(unsigned long ticks);
1977 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1978 extern int wake_up_process(struct task_struct *tsk);
1979 extern void wake_up_new_task(struct task_struct *tsk);
1981 extern void kick_process(struct task_struct *tsk);
1983 static inline void kick_process(struct task_struct *tsk) { }
1985 extern void sched_fork(struct task_struct *p);
1986 extern void sched_dead(struct task_struct *p);
1988 extern void proc_caches_init(void);
1989 extern void flush_signals(struct task_struct *);
1990 extern void __flush_signals(struct task_struct *);
1991 extern void ignore_signals(struct task_struct *);
1992 extern void flush_signal_handlers(struct task_struct *, int force_default);
1993 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1995 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1997 unsigned long flags;
2000 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2001 ret = dequeue_signal(tsk, mask, info);
2002 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2007 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2009 extern void unblock_all_signals(void);
2010 extern void release_task(struct task_struct * p);
2011 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2012 extern int force_sigsegv(int, struct task_struct *);
2013 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2014 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2015 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2016 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2017 const struct cred *, u32);
2018 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2019 extern int kill_pid(struct pid *pid, int sig, int priv);
2020 extern int kill_proc_info(int, struct siginfo *, pid_t);
2021 extern __must_check bool do_notify_parent(struct task_struct *, int);
2022 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2023 extern void force_sig(int, struct task_struct *);
2024 extern int send_sig(int, struct task_struct *, int);
2025 extern int zap_other_threads(struct task_struct *p);
2026 extern struct sigqueue *sigqueue_alloc(void);
2027 extern void sigqueue_free(struct sigqueue *);
2028 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2029 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2031 static inline void restore_saved_sigmask(void)
2033 if (test_and_clear_restore_sigmask())
2034 __set_current_blocked(¤t->saved_sigmask);
2037 static inline sigset_t *sigmask_to_save(void)
2039 sigset_t *res = ¤t->blocked;
2040 if (unlikely(test_restore_sigmask()))
2041 res = ¤t->saved_sigmask;
2045 static inline int kill_cad_pid(int sig, int priv)
2047 return kill_pid(cad_pid, sig, priv);
2050 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2051 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2052 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2053 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2056 * True if we are on the alternate signal stack.
2058 static inline int on_sig_stack(unsigned long sp)
2060 #ifdef CONFIG_STACK_GROWSUP
2061 return sp >= current->sas_ss_sp &&
2062 sp - current->sas_ss_sp < current->sas_ss_size;
2064 return sp > current->sas_ss_sp &&
2065 sp - current->sas_ss_sp <= current->sas_ss_size;
2069 static inline int sas_ss_flags(unsigned long sp)
2071 return (current->sas_ss_size == 0 ? SS_DISABLE
2072 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2075 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2077 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2078 #ifdef CONFIG_STACK_GROWSUP
2079 return current->sas_ss_sp;
2081 return current->sas_ss_sp + current->sas_ss_size;
2087 * Routines for handling mm_structs
2089 extern struct mm_struct * mm_alloc(void);
2091 /* mmdrop drops the mm and the page tables */
2092 extern void __mmdrop(struct mm_struct *);
2093 static inline void mmdrop(struct mm_struct * mm)
2095 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2099 /* mmput gets rid of the mappings and all user-space */
2100 extern void mmput(struct mm_struct *);
2101 /* Grab a reference to a task's mm, if it is not already going away */
2102 extern struct mm_struct *get_task_mm(struct task_struct *task);
2104 * Grab a reference to a task's mm, if it is not already going away
2105 * and ptrace_may_access with the mode parameter passed to it
2108 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2109 /* Remove the current tasks stale references to the old mm_struct */
2110 extern void mm_release(struct task_struct *, struct mm_struct *);
2111 /* Allocate a new mm structure and copy contents from tsk->mm */
2112 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2114 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2115 struct task_struct *);
2116 extern void flush_thread(void);
2117 extern void exit_thread(void);
2119 extern void exit_files(struct task_struct *);
2120 extern void __cleanup_sighand(struct sighand_struct *);
2122 extern void exit_itimers(struct signal_struct *);
2123 extern void flush_itimer_signals(void);
2125 extern void do_group_exit(int);
2127 extern int allow_signal(int);
2128 extern int disallow_signal(int);
2130 extern int do_execve(const char *,
2131 const char __user * const __user *,
2132 const char __user * const __user *);
2133 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2134 struct task_struct *fork_idle(int);
2135 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2137 extern void set_task_comm(struct task_struct *tsk, char *from);
2138 extern char *get_task_comm(char *to, struct task_struct *tsk);
2141 void scheduler_ipi(void);
2142 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2144 static inline void scheduler_ipi(void) { }
2145 static inline unsigned long wait_task_inactive(struct task_struct *p,
2152 #define next_task(p) \
2153 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2155 #define for_each_process(p) \
2156 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2158 extern bool current_is_single_threaded(void);
2161 * Careful: do_each_thread/while_each_thread is a double loop so
2162 * 'break' will not work as expected - use goto instead.
2164 #define do_each_thread(g, t) \
2165 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2167 #define while_each_thread(g, t) \
2168 while ((t = next_thread(t)) != g)
2170 static inline int get_nr_threads(struct task_struct *tsk)
2172 return tsk->signal->nr_threads;
2175 static inline bool thread_group_leader(struct task_struct *p)
2177 return p->exit_signal >= 0;
2180 /* Do to the insanities of de_thread it is possible for a process
2181 * to have the pid of the thread group leader without actually being
2182 * the thread group leader. For iteration through the pids in proc
2183 * all we care about is that we have a task with the appropriate
2184 * pid, we don't actually care if we have the right task.
2186 static inline int has_group_leader_pid(struct task_struct *p)
2188 return p->pid == p->tgid;
2192 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2194 return p1->tgid == p2->tgid;
2197 static inline struct task_struct *next_thread(const struct task_struct *p)
2199 return list_entry_rcu(p->thread_group.next,
2200 struct task_struct, thread_group);
2203 static inline int thread_group_empty(struct task_struct *p)
2205 return list_empty(&p->thread_group);
2208 #define delay_group_leader(p) \
2209 (thread_group_leader(p) && !thread_group_empty(p))
2212 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2213 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2214 * pins the final release of task.io_context. Also protects ->cpuset and
2215 * ->cgroup.subsys[]. And ->vfork_done.
2217 * Nests both inside and outside of read_lock(&tasklist_lock).
2218 * It must not be nested with write_lock_irq(&tasklist_lock),
2219 * neither inside nor outside.
2221 static inline void task_lock(struct task_struct *p)
2223 spin_lock(&p->alloc_lock);
2226 static inline void task_unlock(struct task_struct *p)
2228 spin_unlock(&p->alloc_lock);
2231 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2232 unsigned long *flags);
2234 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2235 unsigned long *flags)
2237 struct sighand_struct *ret;
2239 ret = __lock_task_sighand(tsk, flags);
2240 (void)__cond_lock(&tsk->sighand->siglock, ret);
2244 static inline void unlock_task_sighand(struct task_struct *tsk,
2245 unsigned long *flags)
2247 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2250 #ifdef CONFIG_CGROUPS
2251 static inline void threadgroup_change_begin(struct task_struct *tsk)
2253 down_read(&tsk->signal->group_rwsem);
2255 static inline void threadgroup_change_end(struct task_struct *tsk)
2257 up_read(&tsk->signal->group_rwsem);
2261 * threadgroup_lock - lock threadgroup
2262 * @tsk: member task of the threadgroup to lock
2264 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2265 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2266 * change ->group_leader/pid. This is useful for cases where the threadgroup
2267 * needs to stay stable across blockable operations.
2269 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2270 * synchronization. While held, no new task will be added to threadgroup
2271 * and no existing live task will have its PF_EXITING set.
2273 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2274 * sub-thread becomes a new leader.
2276 static inline void threadgroup_lock(struct task_struct *tsk)
2278 down_write(&tsk->signal->group_rwsem);
2282 * threadgroup_unlock - unlock threadgroup
2283 * @tsk: member task of the threadgroup to unlock
2285 * Reverse threadgroup_lock().
2287 static inline void threadgroup_unlock(struct task_struct *tsk)
2289 up_write(&tsk->signal->group_rwsem);
2292 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2293 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2294 static inline void threadgroup_lock(struct task_struct *tsk) {}
2295 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2298 #ifndef __HAVE_THREAD_FUNCTIONS
2300 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2301 #define task_stack_page(task) ((task)->stack)
2303 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2305 *task_thread_info(p) = *task_thread_info(org);
2306 task_thread_info(p)->task = p;
2309 static inline unsigned long *end_of_stack(struct task_struct *p)
2311 return (unsigned long *)(task_thread_info(p) + 1);
2316 static inline int object_is_on_stack(void *obj)
2318 void *stack = task_stack_page(current);
2320 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2323 extern void thread_info_cache_init(void);
2325 #ifdef CONFIG_DEBUG_STACK_USAGE
2326 static inline unsigned long stack_not_used(struct task_struct *p)
2328 unsigned long *n = end_of_stack(p);
2330 do { /* Skip over canary */
2334 return (unsigned long)n - (unsigned long)end_of_stack(p);
2338 /* set thread flags in other task's structures
2339 * - see asm/thread_info.h for TIF_xxxx flags available
2341 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2343 set_ti_thread_flag(task_thread_info(tsk), flag);
2346 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2348 clear_ti_thread_flag(task_thread_info(tsk), flag);
2351 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2353 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2356 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2358 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2361 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2363 return test_ti_thread_flag(task_thread_info(tsk), flag);
2366 static inline void set_tsk_need_resched(struct task_struct *tsk)
2368 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2371 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2373 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2376 static inline int test_tsk_need_resched(struct task_struct *tsk)
2378 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2381 static inline int restart_syscall(void)
2383 set_tsk_thread_flag(current, TIF_SIGPENDING);
2384 return -ERESTARTNOINTR;
2387 static inline int signal_pending(struct task_struct *p)
2389 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2392 static inline int __fatal_signal_pending(struct task_struct *p)
2394 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2397 static inline int fatal_signal_pending(struct task_struct *p)
2399 return signal_pending(p) && __fatal_signal_pending(p);
2402 static inline int signal_pending_state(long state, struct task_struct *p)
2404 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2406 if (!signal_pending(p))
2409 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2412 static inline int need_resched(void)
2414 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2418 * cond_resched() and cond_resched_lock(): latency reduction via
2419 * explicit rescheduling in places that are safe. The return
2420 * value indicates whether a reschedule was done in fact.
2421 * cond_resched_lock() will drop the spinlock before scheduling,
2422 * cond_resched_softirq() will enable bhs before scheduling.
2424 extern int _cond_resched(void);
2426 #define cond_resched() ({ \
2427 __might_sleep(__FILE__, __LINE__, 0); \
2431 extern int __cond_resched_lock(spinlock_t *lock);
2433 #ifdef CONFIG_PREEMPT_COUNT
2434 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2436 #define PREEMPT_LOCK_OFFSET 0
2439 #define cond_resched_lock(lock) ({ \
2440 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2441 __cond_resched_lock(lock); \
2444 extern int __cond_resched_softirq(void);
2446 #define cond_resched_softirq() ({ \
2447 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2448 __cond_resched_softirq(); \
2452 * Does a critical section need to be broken due to another
2453 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2454 * but a general need for low latency)
2456 static inline int spin_needbreak(spinlock_t *lock)
2458 #ifdef CONFIG_PREEMPT
2459 return spin_is_contended(lock);
2466 * Idle thread specific functions to determine the need_resched
2467 * polling state. We have two versions, one based on TS_POLLING in
2468 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2472 static inline int tsk_is_polling(struct task_struct *p)
2474 return task_thread_info(p)->status & TS_POLLING;
2476 static inline void __current_set_polling(void)
2478 current_thread_info()->status |= TS_POLLING;
2481 static inline bool __must_check current_set_polling_and_test(void)
2483 __current_set_polling();
2486 * Polling state must be visible before we test NEED_RESCHED,
2487 * paired by resched_task()
2491 return unlikely(tif_need_resched());
2494 static inline void __current_clr_polling(void)
2496 current_thread_info()->status &= ~TS_POLLING;
2499 static inline bool __must_check current_clr_polling_and_test(void)
2501 __current_clr_polling();
2504 * Polling state must be visible before we test NEED_RESCHED,
2505 * paired by resched_task()
2509 return unlikely(tif_need_resched());
2511 #elif defined(TIF_POLLING_NRFLAG)
2512 static inline int tsk_is_polling(struct task_struct *p)
2514 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2517 static inline void __current_set_polling(void)
2519 set_thread_flag(TIF_POLLING_NRFLAG);
2522 static inline bool __must_check current_set_polling_and_test(void)
2524 __current_set_polling();
2527 * Polling state must be visible before we test NEED_RESCHED,
2528 * paired by resched_task()
2530 * XXX: assumes set/clear bit are identical barrier wise.
2532 smp_mb__after_clear_bit();
2534 return unlikely(tif_need_resched());
2537 static inline void __current_clr_polling(void)
2539 clear_thread_flag(TIF_POLLING_NRFLAG);
2542 static inline bool __must_check current_clr_polling_and_test(void)
2544 __current_clr_polling();
2547 * Polling state must be visible before we test NEED_RESCHED,
2548 * paired by resched_task()
2550 smp_mb__after_clear_bit();
2552 return unlikely(tif_need_resched());
2556 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2557 static inline void __current_set_polling(void) { }
2558 static inline void __current_clr_polling(void) { }
2560 static inline bool __must_check current_set_polling_and_test(void)
2562 return unlikely(tif_need_resched());
2564 static inline bool __must_check current_clr_polling_and_test(void)
2566 return unlikely(tif_need_resched());
2571 * Thread group CPU time accounting.
2573 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2574 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2576 static inline void thread_group_cputime_init(struct signal_struct *sig)
2578 raw_spin_lock_init(&sig->cputimer.lock);
2582 * Reevaluate whether the task has signals pending delivery.
2583 * Wake the task if so.
2584 * This is required every time the blocked sigset_t changes.
2585 * callers must hold sighand->siglock.
2587 extern void recalc_sigpending_and_wake(struct task_struct *t);
2588 extern void recalc_sigpending(void);
2590 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2592 static inline void signal_wake_up(struct task_struct *t, bool resume)
2594 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2596 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2598 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2602 * Wrappers for p->thread_info->cpu access. No-op on UP.
2606 static inline unsigned int task_cpu(const struct task_struct *p)
2608 return task_thread_info(p)->cpu;
2611 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2615 static inline unsigned int task_cpu(const struct task_struct *p)
2620 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2624 #endif /* CONFIG_SMP */
2626 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2627 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2629 #ifdef CONFIG_CGROUP_SCHED
2630 extern struct task_group root_task_group;
2631 #endif /* CONFIG_CGROUP_SCHED */
2633 extern int task_can_switch_user(struct user_struct *up,
2634 struct task_struct *tsk);
2636 #ifdef CONFIG_TASK_XACCT
2637 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2639 tsk->ioac.rchar += amt;
2642 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2644 tsk->ioac.wchar += amt;
2647 static inline void inc_syscr(struct task_struct *tsk)
2652 static inline void inc_syscw(struct task_struct *tsk)
2657 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2661 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2665 static inline void inc_syscr(struct task_struct *tsk)
2669 static inline void inc_syscw(struct task_struct *tsk)
2674 #ifndef TASK_SIZE_OF
2675 #define TASK_SIZE_OF(tsk) TASK_SIZE
2678 #ifdef CONFIG_MM_OWNER
2679 extern void mm_update_next_owner(struct mm_struct *mm);
2680 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2682 static inline void mm_update_next_owner(struct mm_struct *mm)
2686 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2689 #endif /* CONFIG_MM_OWNER */
2691 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2694 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2697 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2700 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2703 static inline unsigned long rlimit(unsigned int limit)
2705 return task_rlimit(current, limit);
2708 static inline unsigned long rlimit_max(unsigned int limit)
2710 return task_rlimit_max(current, limit);