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_STATE_MAX 512
156 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
158 extern char ___assert_task_state[1 - 2*!!(
159 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
161 /* Convenience macros for the sake of set_task_state */
162 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
163 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
164 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
166 /* Convenience macros for the sake of wake_up */
167 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
168 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
170 /* get_task_state() */
171 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
172 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
175 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
176 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
177 #define task_is_dead(task) ((task)->exit_state != 0)
178 #define task_is_stopped_or_traced(task) \
179 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
180 #define task_contributes_to_load(task) \
181 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
182 (task->flags & PF_FROZEN) == 0)
184 #define __set_task_state(tsk, state_value) \
185 do { (tsk)->state = (state_value); } while (0)
186 #define set_task_state(tsk, state_value) \
187 set_mb((tsk)->state, (state_value))
190 * set_current_state() includes a barrier so that the write of current->state
191 * is correctly serialised wrt the caller's subsequent test of whether to
194 * set_current_state(TASK_UNINTERRUPTIBLE);
195 * if (do_i_need_to_sleep())
198 * If the caller does not need such serialisation then use __set_current_state()
200 #define __set_current_state(state_value) \
201 do { current->state = (state_value); } while (0)
202 #define set_current_state(state_value) \
203 set_mb(current->state, (state_value))
205 /* Task command name length */
206 #define TASK_COMM_LEN 16
208 #include <linux/spinlock.h>
211 * This serializes "schedule()" and also protects
212 * the run-queue from deletions/modifications (but
213 * _adding_ to the beginning of the run-queue has
216 extern rwlock_t tasklist_lock;
217 extern spinlock_t mmlist_lock;
221 #ifdef CONFIG_PROVE_RCU
222 extern int lockdep_tasklist_lock_is_held(void);
223 #endif /* #ifdef CONFIG_PROVE_RCU */
225 extern void sched_init(void);
226 extern void sched_init_smp(void);
227 extern asmlinkage void schedule_tail(struct task_struct *prev);
228 extern void init_idle(struct task_struct *idle, int cpu);
229 extern void init_idle_bootup_task(struct task_struct *idle);
231 extern int runqueue_is_locked(int cpu);
233 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
234 extern void nohz_balance_enter_idle(int cpu);
235 extern void set_cpu_sd_state_idle(void);
236 extern int get_nohz_timer_target(void);
238 static inline void nohz_balance_enter_idle(int cpu) { }
239 static inline void set_cpu_sd_state_idle(void) { }
243 * Only dump TASK_* tasks. (0 for all tasks)
245 extern void show_state_filter(unsigned long state_filter);
247 static inline void show_state(void)
249 show_state_filter(0);
252 extern void show_regs(struct pt_regs *);
255 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
256 * task), SP is the stack pointer of the first frame that should be shown in the back
257 * trace (or NULL if the entire call-chain of the task should be shown).
259 extern void show_stack(struct task_struct *task, unsigned long *sp);
261 void io_schedule(void);
262 long io_schedule_timeout(long timeout);
264 extern void cpu_init (void);
265 extern void trap_init(void);
266 extern void update_process_times(int user);
267 extern void scheduler_tick(void);
269 extern void sched_show_task(struct task_struct *p);
271 #ifdef CONFIG_LOCKUP_DETECTOR
272 extern void touch_softlockup_watchdog(void);
273 extern void touch_softlockup_watchdog_sync(void);
274 extern void touch_all_softlockup_watchdogs(void);
275 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
277 size_t *lenp, loff_t *ppos);
278 extern unsigned int softlockup_panic;
279 void lockup_detector_init(void);
281 static inline void touch_softlockup_watchdog(void)
284 static inline void touch_softlockup_watchdog_sync(void)
287 static inline void touch_all_softlockup_watchdogs(void)
290 static inline void lockup_detector_init(void)
295 /* Attach to any functions which should be ignored in wchan output. */
296 #define __sched __attribute__((__section__(".sched.text")))
298 /* Linker adds these: start and end of __sched functions */
299 extern char __sched_text_start[], __sched_text_end[];
301 /* Is this address in the __sched functions? */
302 extern int in_sched_functions(unsigned long addr);
304 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
305 extern signed long schedule_timeout(signed long timeout);
306 extern signed long schedule_timeout_interruptible(signed long timeout);
307 extern signed long schedule_timeout_killable(signed long timeout);
308 extern signed long schedule_timeout_uninterruptible(signed long timeout);
309 asmlinkage void schedule(void);
310 extern void schedule_preempt_disabled(void);
311 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
314 struct user_namespace;
316 #include <linux/aio.h>
319 extern void arch_pick_mmap_layout(struct mm_struct *mm);
321 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
322 unsigned long, unsigned long);
324 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
325 unsigned long len, unsigned long pgoff,
326 unsigned long flags);
327 extern void arch_unmap_area(struct mm_struct *, unsigned long);
328 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
330 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
334 extern void set_dumpable(struct mm_struct *mm, int value);
335 extern int get_dumpable(struct mm_struct *mm);
339 #define MMF_DUMPABLE 0 /* core dump is permitted */
340 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
342 #define MMF_DUMPABLE_BITS 2
343 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
345 /* coredump filter bits */
346 #define MMF_DUMP_ANON_PRIVATE 2
347 #define MMF_DUMP_ANON_SHARED 3
348 #define MMF_DUMP_MAPPED_PRIVATE 4
349 #define MMF_DUMP_MAPPED_SHARED 5
350 #define MMF_DUMP_ELF_HEADERS 6
351 #define MMF_DUMP_HUGETLB_PRIVATE 7
352 #define MMF_DUMP_HUGETLB_SHARED 8
354 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
355 #define MMF_DUMP_FILTER_BITS 7
356 #define MMF_DUMP_FILTER_MASK \
357 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
358 #define MMF_DUMP_FILTER_DEFAULT \
359 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
360 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
362 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
363 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
365 # define MMF_DUMP_MASK_DEFAULT_ELF 0
367 /* leave room for more dump flags */
368 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
369 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
370 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
372 #define MMF_HAS_UPROBES 19 /* has uprobes */
373 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
375 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
377 struct sighand_struct {
379 struct k_sigaction action[_NSIG];
381 wait_queue_head_t signalfd_wqh;
384 struct pacct_struct {
387 unsigned long ac_mem;
388 cputime_t ac_utime, ac_stime;
389 unsigned long ac_minflt, ac_majflt;
400 * struct cputime - snaphsot of system and user cputime
401 * @utime: time spent in user mode
402 * @stime: time spent in system mode
404 * Gathers a generic snapshot of user and system time.
412 * struct task_cputime - collected CPU time counts
413 * @utime: time spent in user mode, in &cputime_t units
414 * @stime: time spent in kernel mode, in &cputime_t units
415 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
417 * This is an extension of struct cputime that includes the total runtime
418 * spent by the task from the scheduler point of view.
420 * As a result, this structure groups together three kinds of CPU time
421 * that are tracked for threads and thread groups. Most things considering
422 * CPU time want to group these counts together and treat all three
423 * of them in parallel.
425 struct task_cputime {
428 unsigned long long sum_exec_runtime;
430 /* Alternate field names when used to cache expirations. */
431 #define prof_exp stime
432 #define virt_exp utime
433 #define sched_exp sum_exec_runtime
435 #define INIT_CPUTIME \
436 (struct task_cputime) { \
439 .sum_exec_runtime = 0, \
443 * Disable preemption until the scheduler is running.
444 * Reset by start_kernel()->sched_init()->init_idle().
446 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
447 * before the scheduler is active -- see should_resched().
449 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
452 * struct thread_group_cputimer - thread group interval timer counts
453 * @cputime: thread group interval timers.
454 * @running: non-zero when there are timers running and
455 * @cputime receives updates.
456 * @lock: lock for fields in this struct.
458 * This structure contains the version of task_cputime, above, that is
459 * used for thread group CPU timer calculations.
461 struct thread_group_cputimer {
462 struct task_cputime cputime;
467 #include <linux/rwsem.h>
471 * NOTE! "signal_struct" does not have its own
472 * locking, because a shared signal_struct always
473 * implies a shared sighand_struct, so locking
474 * sighand_struct is always a proper superset of
475 * the locking of signal_struct.
477 struct signal_struct {
482 wait_queue_head_t wait_chldexit; /* for wait4() */
484 /* current thread group signal load-balancing target: */
485 struct task_struct *curr_target;
487 /* shared signal handling: */
488 struct sigpending shared_pending;
490 /* thread group exit support */
493 * - notify group_exit_task when ->count is equal to notify_count
494 * - everyone except group_exit_task is stopped during signal delivery
495 * of fatal signals, group_exit_task processes the signal.
498 struct task_struct *group_exit_task;
500 /* thread group stop support, overloads group_exit_code too */
501 int group_stop_count;
502 unsigned int flags; /* see SIGNAL_* flags below */
505 * PR_SET_CHILD_SUBREAPER marks a process, like a service
506 * manager, to re-parent orphan (double-forking) child processes
507 * to this process instead of 'init'. The service manager is
508 * able to receive SIGCHLD signals and is able to investigate
509 * the process until it calls wait(). All children of this
510 * process will inherit a flag if they should look for a
511 * child_subreaper process at exit.
513 unsigned int is_child_subreaper:1;
514 unsigned int has_child_subreaper:1;
516 /* POSIX.1b Interval Timers */
517 struct list_head posix_timers;
519 /* ITIMER_REAL timer for the process */
520 struct hrtimer real_timer;
521 struct pid *leader_pid;
522 ktime_t it_real_incr;
525 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
526 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
527 * values are defined to 0 and 1 respectively
529 struct cpu_itimer it[2];
532 * Thread group totals for process CPU timers.
533 * See thread_group_cputimer(), et al, for details.
535 struct thread_group_cputimer cputimer;
537 /* Earliest-expiration cache. */
538 struct task_cputime cputime_expires;
540 struct list_head cpu_timers[3];
542 struct pid *tty_old_pgrp;
544 /* boolean value for session group leader */
547 struct tty_struct *tty; /* NULL if no tty */
549 #ifdef CONFIG_SCHED_AUTOGROUP
550 struct autogroup *autogroup;
553 * Cumulative resource counters for dead threads in the group,
554 * and for reaped dead child processes forked by this group.
555 * Live threads maintain their own counters and add to these
556 * in __exit_signal, except for the group leader.
558 cputime_t utime, stime, cutime, cstime;
561 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
562 struct cputime prev_cputime;
564 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
565 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
566 unsigned long inblock, oublock, cinblock, coublock;
567 unsigned long maxrss, cmaxrss;
568 struct task_io_accounting ioac;
571 * Cumulative ns of schedule CPU time fo dead threads in the
572 * group, not including a zombie group leader, (This only differs
573 * from jiffies_to_ns(utime + stime) if sched_clock uses something
574 * other than jiffies.)
576 unsigned long long sum_sched_runtime;
579 * We don't bother to synchronize most readers of this at all,
580 * because there is no reader checking a limit that actually needs
581 * to get both rlim_cur and rlim_max atomically, and either one
582 * alone is a single word that can safely be read normally.
583 * getrlimit/setrlimit use task_lock(current->group_leader) to
584 * protect this instead of the siglock, because they really
585 * have no need to disable irqs.
587 struct rlimit rlim[RLIM_NLIMITS];
589 #ifdef CONFIG_BSD_PROCESS_ACCT
590 struct pacct_struct pacct; /* per-process accounting information */
592 #ifdef CONFIG_TASKSTATS
593 struct taskstats *stats;
597 struct tty_audit_buf *tty_audit_buf;
599 #ifdef CONFIG_CGROUPS
601 * group_rwsem prevents new tasks from entering the threadgroup and
602 * member tasks from exiting,a more specifically, setting of
603 * PF_EXITING. fork and exit paths are protected with this rwsem
604 * using threadgroup_change_begin/end(). Users which require
605 * threadgroup to remain stable should use threadgroup_[un]lock()
606 * which also takes care of exec path. Currently, cgroup is the
609 struct rw_semaphore group_rwsem;
612 oom_flags_t oom_flags;
613 short oom_score_adj; /* OOM kill score adjustment */
614 short oom_score_adj_min; /* OOM kill score adjustment min value.
615 * Only settable by CAP_SYS_RESOURCE. */
617 struct mutex cred_guard_mutex; /* guard against foreign influences on
618 * credential calculations
619 * (notably. ptrace) */
623 * Bits in flags field of signal_struct.
625 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
626 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
627 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
629 * Pending notifications to parent.
631 #define SIGNAL_CLD_STOPPED 0x00000010
632 #define SIGNAL_CLD_CONTINUED 0x00000020
633 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
635 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
637 /* If true, all threads except ->group_exit_task have pending SIGKILL */
638 static inline int signal_group_exit(const struct signal_struct *sig)
640 return (sig->flags & SIGNAL_GROUP_EXIT) ||
641 (sig->group_exit_task != NULL);
645 * Some day this will be a full-fledged user tracking system..
648 atomic_t __count; /* reference count */
649 atomic_t processes; /* How many processes does this user have? */
650 atomic_t files; /* How many open files does this user have? */
651 atomic_t sigpending; /* How many pending signals does this user have? */
652 #ifdef CONFIG_INOTIFY_USER
653 atomic_t inotify_watches; /* How many inotify watches does this user have? */
654 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
656 #ifdef CONFIG_FANOTIFY
657 atomic_t fanotify_listeners;
660 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
662 #ifdef CONFIG_POSIX_MQUEUE
663 /* protected by mq_lock */
664 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
666 unsigned long locked_shm; /* How many pages of mlocked shm ? */
669 struct key *uid_keyring; /* UID specific keyring */
670 struct key *session_keyring; /* UID's default session keyring */
673 /* Hash table maintenance information */
674 struct hlist_node uidhash_node;
677 #ifdef CONFIG_PERF_EVENTS
678 atomic_long_t locked_vm;
682 extern int uids_sysfs_init(void);
684 extern struct user_struct *find_user(kuid_t);
686 extern struct user_struct root_user;
687 #define INIT_USER (&root_user)
690 struct backing_dev_info;
691 struct reclaim_state;
693 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
695 /* cumulative counters */
696 unsigned long pcount; /* # of times run on this cpu */
697 unsigned long long run_delay; /* time spent waiting on a runqueue */
700 unsigned long long last_arrival,/* when we last ran on a cpu */
701 last_queued; /* when we were last queued to run */
703 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
705 #ifdef CONFIG_TASK_DELAY_ACCT
706 struct task_delay_info {
708 unsigned int flags; /* Private per-task flags */
710 /* For each stat XXX, add following, aligned appropriately
712 * struct timespec XXX_start, XXX_end;
716 * Atomicity of updates to XXX_delay, XXX_count protected by
717 * single lock above (split into XXX_lock if contention is an issue).
721 * XXX_count is incremented on every XXX operation, the delay
722 * associated with the operation is added to XXX_delay.
723 * XXX_delay contains the accumulated delay time in nanoseconds.
725 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
726 u64 blkio_delay; /* wait for sync block io completion */
727 u64 swapin_delay; /* wait for swapin block io completion */
728 u32 blkio_count; /* total count of the number of sync block */
729 /* io operations performed */
730 u32 swapin_count; /* total count of the number of swapin block */
731 /* io operations performed */
733 struct timespec freepages_start, freepages_end;
734 u64 freepages_delay; /* wait for memory reclaim */
735 u32 freepages_count; /* total count of memory reclaim */
737 #endif /* CONFIG_TASK_DELAY_ACCT */
739 static inline int sched_info_on(void)
741 #ifdef CONFIG_SCHEDSTATS
743 #elif defined(CONFIG_TASK_DELAY_ACCT)
744 extern int delayacct_on;
759 * Increase resolution of nice-level calculations for 64-bit architectures.
760 * The extra resolution improves shares distribution and load balancing of
761 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
762 * hierarchies, especially on larger systems. This is not a user-visible change
763 * and does not change the user-interface for setting shares/weights.
765 * We increase resolution only if we have enough bits to allow this increased
766 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
767 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
770 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
771 # define SCHED_LOAD_RESOLUTION 10
772 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
773 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
775 # define SCHED_LOAD_RESOLUTION 0
776 # define scale_load(w) (w)
777 # define scale_load_down(w) (w)
780 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
781 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
784 * Increase resolution of cpu_power calculations
786 #define SCHED_POWER_SHIFT 10
787 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
790 * sched-domains (multiprocessor balancing) declarations:
793 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
794 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
795 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
796 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
797 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
798 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
799 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
800 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
801 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
802 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
803 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
804 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
806 extern int __weak arch_sd_sibiling_asym_packing(void);
808 struct sched_group_power {
811 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
814 unsigned int power, power_orig;
815 unsigned long next_update;
817 * Number of busy cpus in this group.
819 atomic_t nr_busy_cpus;
821 unsigned long cpumask[0]; /* iteration mask */
825 struct sched_group *next; /* Must be a circular list */
828 unsigned int group_weight;
829 struct sched_group_power *sgp;
832 * The CPUs this group covers.
834 * NOTE: this field is variable length. (Allocated dynamically
835 * by attaching extra space to the end of the structure,
836 * depending on how many CPUs the kernel has booted up with)
838 unsigned long cpumask[0];
841 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
843 return to_cpumask(sg->cpumask);
847 * cpumask masking which cpus in the group are allowed to iterate up the domain
850 static inline struct cpumask *sched_group_mask(struct sched_group *sg)
852 return to_cpumask(sg->sgp->cpumask);
856 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
857 * @group: The group whose first cpu is to be returned.
859 static inline unsigned int group_first_cpu(struct sched_group *group)
861 return cpumask_first(sched_group_cpus(group));
864 struct sched_domain_attr {
865 int relax_domain_level;
868 #define SD_ATTR_INIT (struct sched_domain_attr) { \
869 .relax_domain_level = -1, \
872 extern int sched_domain_level_max;
874 struct sched_domain {
875 /* These fields must be setup */
876 struct sched_domain *parent; /* top domain must be null terminated */
877 struct sched_domain *child; /* bottom domain must be null terminated */
878 struct sched_group *groups; /* the balancing groups of the domain */
879 unsigned long min_interval; /* Minimum balance interval ms */
880 unsigned long max_interval; /* Maximum balance interval ms */
881 unsigned int busy_factor; /* less balancing by factor if busy */
882 unsigned int imbalance_pct; /* No balance until over watermark */
883 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
884 unsigned int busy_idx;
885 unsigned int idle_idx;
886 unsigned int newidle_idx;
887 unsigned int wake_idx;
888 unsigned int forkexec_idx;
889 unsigned int smt_gain;
890 int flags; /* See SD_* */
893 /* Runtime fields. */
894 unsigned long last_balance; /* init to jiffies. units in jiffies */
895 unsigned int balance_interval; /* initialise to 1. units in ms. */
896 unsigned int nr_balance_failed; /* initialise to 0 */
900 #ifdef CONFIG_SCHEDSTATS
901 /* load_balance() stats */
902 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
903 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
904 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
905 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
906 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
907 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
908 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
909 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
911 /* Active load balancing */
912 unsigned int alb_count;
913 unsigned int alb_failed;
914 unsigned int alb_pushed;
916 /* SD_BALANCE_EXEC stats */
917 unsigned int sbe_count;
918 unsigned int sbe_balanced;
919 unsigned int sbe_pushed;
921 /* SD_BALANCE_FORK stats */
922 unsigned int sbf_count;
923 unsigned int sbf_balanced;
924 unsigned int sbf_pushed;
926 /* try_to_wake_up() stats */
927 unsigned int ttwu_wake_remote;
928 unsigned int ttwu_move_affine;
929 unsigned int ttwu_move_balance;
931 #ifdef CONFIG_SCHED_DEBUG
935 void *private; /* used during construction */
936 struct rcu_head rcu; /* used during destruction */
939 unsigned int span_weight;
941 * Span of all CPUs in this domain.
943 * NOTE: this field is variable length. (Allocated dynamically
944 * by attaching extra space to the end of the structure,
945 * depending on how many CPUs the kernel has booted up with)
947 unsigned long span[0];
950 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
952 return to_cpumask(sd->span);
955 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
956 struct sched_domain_attr *dattr_new);
958 /* Allocate an array of sched domains, for partition_sched_domains(). */
959 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
960 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
962 /* Test a flag in parent sched domain */
963 static inline int test_sd_parent(struct sched_domain *sd, int flag)
965 if (sd->parent && (sd->parent->flags & flag))
971 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
972 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
974 bool cpus_share_cache(int this_cpu, int that_cpu);
976 #else /* CONFIG_SMP */
978 struct sched_domain_attr;
981 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
982 struct sched_domain_attr *dattr_new)
986 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
991 #endif /* !CONFIG_SMP */
994 struct io_context; /* See blkdev.h */
997 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
998 extern void prefetch_stack(struct task_struct *t);
1000 static inline void prefetch_stack(struct task_struct *t) { }
1003 struct audit_context; /* See audit.c */
1005 struct pipe_inode_info;
1006 struct uts_namespace;
1009 struct sched_domain;
1014 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1015 #define WF_FORK 0x02 /* child wakeup after fork */
1016 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1018 #define ENQUEUE_WAKEUP 1
1019 #define ENQUEUE_HEAD 2
1021 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1023 #define ENQUEUE_WAKING 0
1026 #define DEQUEUE_SLEEP 1
1028 struct sched_class {
1029 const struct sched_class *next;
1031 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1032 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1033 void (*yield_task) (struct rq *rq);
1034 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1036 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1038 struct task_struct * (*pick_next_task) (struct rq *rq);
1039 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1042 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1043 void (*migrate_task_rq)(struct task_struct *p, int next_cpu);
1045 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1046 void (*post_schedule) (struct rq *this_rq);
1047 void (*task_waking) (struct task_struct *task);
1048 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1050 void (*set_cpus_allowed)(struct task_struct *p,
1051 const struct cpumask *newmask);
1053 void (*rq_online)(struct rq *rq);
1054 void (*rq_offline)(struct rq *rq);
1057 void (*set_curr_task) (struct rq *rq);
1058 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1059 void (*task_fork) (struct task_struct *p);
1061 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1062 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1063 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1066 unsigned int (*get_rr_interval) (struct rq *rq,
1067 struct task_struct *task);
1069 #ifdef CONFIG_FAIR_GROUP_SCHED
1070 void (*task_move_group) (struct task_struct *p, int on_rq);
1074 struct load_weight {
1075 unsigned long weight, inv_weight;
1080 * These sums represent an infinite geometric series and so are bound
1081 * above by 1024/(1-y). Thus we only need a u32 to store them for for all
1082 * choices of y < 1-2^(-32)*1024.
1084 u32 runnable_avg_sum, runnable_avg_period;
1085 u64 last_runnable_update;
1087 unsigned long load_avg_contrib;
1090 #ifdef CONFIG_SCHEDSTATS
1091 struct sched_statistics {
1101 s64 sum_sleep_runtime;
1108 u64 nr_migrations_cold;
1109 u64 nr_failed_migrations_affine;
1110 u64 nr_failed_migrations_running;
1111 u64 nr_failed_migrations_hot;
1112 u64 nr_forced_migrations;
1115 u64 nr_wakeups_sync;
1116 u64 nr_wakeups_migrate;
1117 u64 nr_wakeups_local;
1118 u64 nr_wakeups_remote;
1119 u64 nr_wakeups_affine;
1120 u64 nr_wakeups_affine_attempts;
1121 u64 nr_wakeups_passive;
1122 u64 nr_wakeups_idle;
1126 struct sched_entity {
1127 struct load_weight load; /* for load-balancing */
1128 struct rb_node run_node;
1129 struct list_head group_node;
1133 u64 sum_exec_runtime;
1135 u64 prev_sum_exec_runtime;
1139 #ifdef CONFIG_SCHEDSTATS
1140 struct sched_statistics statistics;
1143 #ifdef CONFIG_FAIR_GROUP_SCHED
1144 struct sched_entity *parent;
1145 /* rq on which this entity is (to be) queued: */
1146 struct cfs_rq *cfs_rq;
1147 /* rq "owned" by this entity/group: */
1148 struct cfs_rq *my_q;
1152 * Load-tracking only depends on SMP, FAIR_GROUP_SCHED dependency below may be
1153 * removed when useful for applications beyond shares distribution (e.g.
1156 #if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)
1157 /* Per-entity load-tracking */
1158 struct sched_avg avg;
1162 struct sched_rt_entity {
1163 struct list_head run_list;
1164 unsigned long timeout;
1165 unsigned long watchdog_stamp;
1166 unsigned int time_slice;
1168 struct sched_rt_entity *back;
1169 #ifdef CONFIG_RT_GROUP_SCHED
1170 struct sched_rt_entity *parent;
1171 /* rq on which this entity is (to be) queued: */
1172 struct rt_rq *rt_rq;
1173 /* rq "owned" by this entity/group: */
1181 enum perf_event_task_context {
1182 perf_invalid_context = -1,
1183 perf_hw_context = 0,
1185 perf_nr_task_contexts,
1188 struct task_struct {
1189 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1192 unsigned int flags; /* per process flags, defined below */
1193 unsigned int ptrace;
1196 struct llist_node wake_entry;
1201 int prio, static_prio, normal_prio;
1202 unsigned int rt_priority;
1203 const struct sched_class *sched_class;
1204 struct sched_entity se;
1205 struct sched_rt_entity rt;
1206 #ifdef CONFIG_CGROUP_SCHED
1207 struct task_group *sched_task_group;
1210 #ifdef CONFIG_PREEMPT_NOTIFIERS
1211 /* list of struct preempt_notifier: */
1212 struct hlist_head preempt_notifiers;
1216 * fpu_counter contains the number of consecutive context switches
1217 * that the FPU is used. If this is over a threshold, the lazy fpu
1218 * saving becomes unlazy to save the trap. This is an unsigned char
1219 * so that after 256 times the counter wraps and the behavior turns
1220 * lazy again; this to deal with bursty apps that only use FPU for
1223 unsigned char fpu_counter;
1224 #ifdef CONFIG_BLK_DEV_IO_TRACE
1225 unsigned int btrace_seq;
1228 unsigned int policy;
1229 int nr_cpus_allowed;
1230 cpumask_t cpus_allowed;
1232 #ifdef CONFIG_PREEMPT_RCU
1233 int rcu_read_lock_nesting;
1234 char rcu_read_unlock_special;
1235 struct list_head rcu_node_entry;
1236 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1237 #ifdef CONFIG_TREE_PREEMPT_RCU
1238 struct rcu_node *rcu_blocked_node;
1239 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1240 #ifdef CONFIG_RCU_BOOST
1241 struct rt_mutex *rcu_boost_mutex;
1242 #endif /* #ifdef CONFIG_RCU_BOOST */
1244 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1245 struct sched_info sched_info;
1248 struct list_head tasks;
1250 struct plist_node pushable_tasks;
1253 struct mm_struct *mm, *active_mm;
1254 #ifdef CONFIG_COMPAT_BRK
1255 unsigned brk_randomized:1;
1257 #if defined(SPLIT_RSS_COUNTING)
1258 struct task_rss_stat rss_stat;
1262 int exit_code, exit_signal;
1263 int pdeath_signal; /* The signal sent when the parent dies */
1264 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1266 unsigned int personality;
1267 unsigned did_exec:1;
1268 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1270 unsigned in_iowait:1;
1272 /* task may not gain privileges */
1273 unsigned no_new_privs:1;
1275 /* Revert to default priority/policy when forking */
1276 unsigned sched_reset_on_fork:1;
1277 unsigned sched_contributes_to_load:1;
1282 #ifdef CONFIG_CC_STACKPROTECTOR
1283 /* Canary value for the -fstack-protector gcc feature */
1284 unsigned long stack_canary;
1287 * pointers to (original) parent process, youngest child, younger sibling,
1288 * older sibling, respectively. (p->father can be replaced with
1289 * p->real_parent->pid)
1291 struct task_struct __rcu *real_parent; /* real parent process */
1292 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1294 * children/sibling forms the list of my natural children
1296 struct list_head children; /* list of my children */
1297 struct list_head sibling; /* linkage in my parent's children list */
1298 struct task_struct *group_leader; /* threadgroup leader */
1301 * ptraced is the list of tasks this task is using ptrace on.
1302 * This includes both natural children and PTRACE_ATTACH targets.
1303 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1305 struct list_head ptraced;
1306 struct list_head ptrace_entry;
1308 /* PID/PID hash table linkage. */
1309 struct pid_link pids[PIDTYPE_MAX];
1310 struct list_head thread_group;
1312 struct completion *vfork_done; /* for vfork() */
1313 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1314 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1316 cputime_t utime, stime, utimescaled, stimescaled;
1318 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1319 struct cputime prev_cputime;
1321 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1322 seqlock_t vtime_seqlock;
1323 unsigned long long vtime_snap;
1328 } vtime_snap_whence;
1330 unsigned long nvcsw, nivcsw; /* context switch counts */
1331 struct timespec start_time; /* monotonic time */
1332 struct timespec real_start_time; /* boot based time */
1333 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1334 unsigned long min_flt, maj_flt;
1336 struct task_cputime cputime_expires;
1337 struct list_head cpu_timers[3];
1339 /* process credentials */
1340 const struct cred __rcu *real_cred; /* objective and real subjective task
1341 * credentials (COW) */
1342 const struct cred __rcu *cred; /* effective (overridable) subjective task
1343 * credentials (COW) */
1344 char comm[TASK_COMM_LEN]; /* executable name excluding path
1345 - access with [gs]et_task_comm (which lock
1346 it with task_lock())
1347 - initialized normally by setup_new_exec */
1348 /* file system info */
1349 int link_count, total_link_count;
1350 #ifdef CONFIG_SYSVIPC
1352 struct sysv_sem sysvsem;
1354 #ifdef CONFIG_DETECT_HUNG_TASK
1355 /* hung task detection */
1356 unsigned long last_switch_count;
1358 /* CPU-specific state of this task */
1359 struct thread_struct thread;
1360 /* filesystem information */
1361 struct fs_struct *fs;
1362 /* open file information */
1363 struct files_struct *files;
1365 struct nsproxy *nsproxy;
1366 /* signal handlers */
1367 struct signal_struct *signal;
1368 struct sighand_struct *sighand;
1370 sigset_t blocked, real_blocked;
1371 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1372 struct sigpending pending;
1374 unsigned long sas_ss_sp;
1376 int (*notifier)(void *priv);
1377 void *notifier_data;
1378 sigset_t *notifier_mask;
1379 struct callback_head *task_works;
1381 struct audit_context *audit_context;
1382 #ifdef CONFIG_AUDITSYSCALL
1384 unsigned int sessionid;
1386 struct seccomp seccomp;
1388 /* Thread group tracking */
1391 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1393 spinlock_t alloc_lock;
1395 /* Protection of the PI data structures: */
1396 raw_spinlock_t pi_lock;
1398 #ifdef CONFIG_RT_MUTEXES
1399 /* PI waiters blocked on a rt_mutex held by this task */
1400 struct plist_head pi_waiters;
1401 /* Deadlock detection and priority inheritance handling */
1402 struct rt_mutex_waiter *pi_blocked_on;
1405 #ifdef CONFIG_DEBUG_MUTEXES
1406 /* mutex deadlock detection */
1407 struct mutex_waiter *blocked_on;
1409 #ifdef CONFIG_TRACE_IRQFLAGS
1410 unsigned int irq_events;
1411 unsigned long hardirq_enable_ip;
1412 unsigned long hardirq_disable_ip;
1413 unsigned int hardirq_enable_event;
1414 unsigned int hardirq_disable_event;
1415 int hardirqs_enabled;
1416 int hardirq_context;
1417 unsigned long softirq_disable_ip;
1418 unsigned long softirq_enable_ip;
1419 unsigned int softirq_disable_event;
1420 unsigned int softirq_enable_event;
1421 int softirqs_enabled;
1422 int softirq_context;
1424 #ifdef CONFIG_LOCKDEP
1425 # define MAX_LOCK_DEPTH 48UL
1428 unsigned int lockdep_recursion;
1429 struct held_lock held_locks[MAX_LOCK_DEPTH];
1430 gfp_t lockdep_reclaim_gfp;
1433 /* journalling filesystem info */
1436 /* stacked block device info */
1437 struct bio_list *bio_list;
1440 /* stack plugging */
1441 struct blk_plug *plug;
1445 struct reclaim_state *reclaim_state;
1447 struct backing_dev_info *backing_dev_info;
1449 struct io_context *io_context;
1451 unsigned long ptrace_message;
1452 siginfo_t *last_siginfo; /* For ptrace use. */
1453 struct task_io_accounting ioac;
1454 #if defined(CONFIG_TASK_XACCT)
1455 u64 acct_rss_mem1; /* accumulated rss usage */
1456 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1457 cputime_t acct_timexpd; /* stime + utime since last update */
1459 #ifdef CONFIG_CPUSETS
1460 nodemask_t mems_allowed; /* Protected by alloc_lock */
1461 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1462 int cpuset_mem_spread_rotor;
1463 int cpuset_slab_spread_rotor;
1465 #ifdef CONFIG_CGROUPS
1466 /* Control Group info protected by css_set_lock */
1467 struct css_set __rcu *cgroups;
1468 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1469 struct list_head cg_list;
1472 struct robust_list_head __user *robust_list;
1473 #ifdef CONFIG_COMPAT
1474 struct compat_robust_list_head __user *compat_robust_list;
1476 struct list_head pi_state_list;
1477 struct futex_pi_state *pi_state_cache;
1479 #ifdef CONFIG_PERF_EVENTS
1480 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1481 struct mutex perf_event_mutex;
1482 struct list_head perf_event_list;
1485 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1487 short pref_node_fork;
1489 #ifdef CONFIG_NUMA_BALANCING
1491 int numa_migrate_seq;
1492 unsigned int numa_scan_period;
1493 u64 node_stamp; /* migration stamp */
1494 struct callback_head numa_work;
1495 #endif /* CONFIG_NUMA_BALANCING */
1497 struct rcu_head rcu;
1500 * cache last used pipe for splice
1502 struct pipe_inode_info *splice_pipe;
1504 struct page_frag task_frag;
1506 #ifdef CONFIG_TASK_DELAY_ACCT
1507 struct task_delay_info *delays;
1509 #ifdef CONFIG_FAULT_INJECTION
1513 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1514 * balance_dirty_pages() for some dirty throttling pause
1517 int nr_dirtied_pause;
1518 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1520 #ifdef CONFIG_LATENCYTOP
1521 int latency_record_count;
1522 struct latency_record latency_record[LT_SAVECOUNT];
1525 * time slack values; these are used to round up poll() and
1526 * select() etc timeout values. These are in nanoseconds.
1528 unsigned long timer_slack_ns;
1529 unsigned long default_timer_slack_ns;
1531 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1532 /* Index of current stored address in ret_stack */
1534 /* Stack of return addresses for return function tracing */
1535 struct ftrace_ret_stack *ret_stack;
1536 /* time stamp for last schedule */
1537 unsigned long long ftrace_timestamp;
1539 * Number of functions that haven't been traced
1540 * because of depth overrun.
1542 atomic_t trace_overrun;
1543 /* Pause for the tracing */
1544 atomic_t tracing_graph_pause;
1546 #ifdef CONFIG_TRACING
1547 /* state flags for use by tracers */
1548 unsigned long trace;
1549 /* bitmask and counter of trace recursion */
1550 unsigned long trace_recursion;
1551 #endif /* CONFIG_TRACING */
1552 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1553 struct memcg_batch_info {
1554 int do_batch; /* incremented when batch uncharge started */
1555 struct mem_cgroup *memcg; /* target memcg of uncharge */
1556 unsigned long nr_pages; /* uncharged usage */
1557 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1559 unsigned int memcg_kmem_skip_account;
1561 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1562 atomic_t ptrace_bp_refcnt;
1564 #ifdef CONFIG_UPROBES
1565 struct uprobe_task *utask;
1569 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1570 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1572 #ifdef CONFIG_NUMA_BALANCING
1573 extern void task_numa_fault(int node, int pages, bool migrated);
1574 extern void set_numabalancing_state(bool enabled);
1576 static inline void task_numa_fault(int node, int pages, bool migrated)
1579 static inline void set_numabalancing_state(bool enabled)
1584 static inline struct pid *task_pid(struct task_struct *task)
1586 return task->pids[PIDTYPE_PID].pid;
1589 static inline struct pid *task_tgid(struct task_struct *task)
1591 return task->group_leader->pids[PIDTYPE_PID].pid;
1595 * Without tasklist or rcu lock it is not safe to dereference
1596 * the result of task_pgrp/task_session even if task == current,
1597 * we can race with another thread doing sys_setsid/sys_setpgid.
1599 static inline struct pid *task_pgrp(struct task_struct *task)
1601 return task->group_leader->pids[PIDTYPE_PGID].pid;
1604 static inline struct pid *task_session(struct task_struct *task)
1606 return task->group_leader->pids[PIDTYPE_SID].pid;
1609 struct pid_namespace;
1612 * the helpers to get the task's different pids as they are seen
1613 * from various namespaces
1615 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1616 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1618 * task_xid_nr_ns() : id seen from the ns specified;
1620 * set_task_vxid() : assigns a virtual id to a task;
1622 * see also pid_nr() etc in include/linux/pid.h
1624 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1625 struct pid_namespace *ns);
1627 static inline pid_t task_pid_nr(struct task_struct *tsk)
1632 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1633 struct pid_namespace *ns)
1635 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1638 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1640 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1644 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1649 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1651 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1653 return pid_vnr(task_tgid(tsk));
1657 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1658 struct pid_namespace *ns)
1660 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1663 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1665 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1669 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1670 struct pid_namespace *ns)
1672 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1675 static inline pid_t task_session_vnr(struct task_struct *tsk)
1677 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1680 /* obsolete, do not use */
1681 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1683 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1687 * pid_alive - check that a task structure is not stale
1688 * @p: Task structure to be checked.
1690 * Test if a process is not yet dead (at most zombie state)
1691 * If pid_alive fails, then pointers within the task structure
1692 * can be stale and must not be dereferenced.
1694 static inline int pid_alive(struct task_struct *p)
1696 return p->pids[PIDTYPE_PID].pid != NULL;
1700 * is_global_init - check if a task structure is init
1701 * @tsk: Task structure to be checked.
1703 * Check if a task structure is the first user space task the kernel created.
1705 static inline int is_global_init(struct task_struct *tsk)
1707 return tsk->pid == 1;
1710 extern struct pid *cad_pid;
1712 extern void free_task(struct task_struct *tsk);
1713 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1715 extern void __put_task_struct(struct task_struct *t);
1717 static inline void put_task_struct(struct task_struct *t)
1719 if (atomic_dec_and_test(&t->usage))
1720 __put_task_struct(t);
1723 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1724 extern void task_cputime(struct task_struct *t,
1725 cputime_t *utime, cputime_t *stime);
1726 extern void task_cputime_scaled(struct task_struct *t,
1727 cputime_t *utimescaled, cputime_t *stimescaled);
1728 extern cputime_t task_gtime(struct task_struct *t);
1730 static inline void task_cputime(struct task_struct *t,
1731 cputime_t *utime, cputime_t *stime)
1739 static inline void task_cputime_scaled(struct task_struct *t,
1740 cputime_t *utimescaled,
1741 cputime_t *stimescaled)
1744 *utimescaled = t->utimescaled;
1746 *stimescaled = t->stimescaled;
1749 static inline cputime_t task_gtime(struct task_struct *t)
1754 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1755 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1760 #define PF_EXITING 0x00000004 /* getting shut down */
1761 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1762 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1763 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1764 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1765 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1766 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1767 #define PF_DUMPCORE 0x00000200 /* dumped core */
1768 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1769 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1770 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1771 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1772 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1773 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1774 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1775 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1776 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1777 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1778 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1779 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1780 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1781 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1782 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1783 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1784 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1785 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1786 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1787 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1788 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1791 * Only the _current_ task can read/write to tsk->flags, but other
1792 * tasks can access tsk->flags in readonly mode for example
1793 * with tsk_used_math (like during threaded core dumping).
1794 * There is however an exception to this rule during ptrace
1795 * or during fork: the ptracer task is allowed to write to the
1796 * child->flags of its traced child (same goes for fork, the parent
1797 * can write to the child->flags), because we're guaranteed the
1798 * child is not running and in turn not changing child->flags
1799 * at the same time the parent does it.
1801 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1802 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1803 #define clear_used_math() clear_stopped_child_used_math(current)
1804 #define set_used_math() set_stopped_child_used_math(current)
1805 #define conditional_stopped_child_used_math(condition, child) \
1806 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1807 #define conditional_used_math(condition) \
1808 conditional_stopped_child_used_math(condition, current)
1809 #define copy_to_stopped_child_used_math(child) \
1810 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1811 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1812 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1813 #define used_math() tsk_used_math(current)
1815 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1816 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1818 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1823 static inline unsigned int memalloc_noio_save(void)
1825 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1826 current->flags |= PF_MEMALLOC_NOIO;
1830 static inline void memalloc_noio_restore(unsigned int flags)
1832 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1836 * task->jobctl flags
1838 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1840 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1841 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1842 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1843 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1844 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1845 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1846 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1848 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1849 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1850 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1851 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1852 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1853 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1854 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1856 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1857 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1859 extern bool task_set_jobctl_pending(struct task_struct *task,
1861 extern void task_clear_jobctl_trapping(struct task_struct *task);
1862 extern void task_clear_jobctl_pending(struct task_struct *task,
1865 #ifdef CONFIG_PREEMPT_RCU
1867 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1868 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1870 static inline void rcu_copy_process(struct task_struct *p)
1872 p->rcu_read_lock_nesting = 0;
1873 p->rcu_read_unlock_special = 0;
1874 #ifdef CONFIG_TREE_PREEMPT_RCU
1875 p->rcu_blocked_node = NULL;
1876 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1877 #ifdef CONFIG_RCU_BOOST
1878 p->rcu_boost_mutex = NULL;
1879 #endif /* #ifdef CONFIG_RCU_BOOST */
1880 INIT_LIST_HEAD(&p->rcu_node_entry);
1885 static inline void rcu_copy_process(struct task_struct *p)
1891 static inline void tsk_restore_flags(struct task_struct *task,
1892 unsigned long orig_flags, unsigned long flags)
1894 task->flags &= ~flags;
1895 task->flags |= orig_flags & flags;
1899 extern void do_set_cpus_allowed(struct task_struct *p,
1900 const struct cpumask *new_mask);
1902 extern int set_cpus_allowed_ptr(struct task_struct *p,
1903 const struct cpumask *new_mask);
1905 static inline void do_set_cpus_allowed(struct task_struct *p,
1906 const struct cpumask *new_mask)
1909 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1910 const struct cpumask *new_mask)
1912 if (!cpumask_test_cpu(0, new_mask))
1919 void calc_load_enter_idle(void);
1920 void calc_load_exit_idle(void);
1922 static inline void calc_load_enter_idle(void) { }
1923 static inline void calc_load_exit_idle(void) { }
1924 #endif /* CONFIG_NO_HZ */
1926 #ifndef CONFIG_CPUMASK_OFFSTACK
1927 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1929 return set_cpus_allowed_ptr(p, &new_mask);
1934 * Do not use outside of architecture code which knows its limitations.
1936 * sched_clock() has no promise of monotonicity or bounded drift between
1937 * CPUs, use (which you should not) requires disabling IRQs.
1939 * Please use one of the three interfaces below.
1941 extern unsigned long long notrace sched_clock(void);
1943 * See the comment in kernel/sched/clock.c
1945 extern u64 cpu_clock(int cpu);
1946 extern u64 local_clock(void);
1947 extern u64 sched_clock_cpu(int cpu);
1950 extern void sched_clock_init(void);
1952 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1953 static inline void sched_clock_tick(void)
1957 static inline void sched_clock_idle_sleep_event(void)
1961 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1966 * Architectures can set this to 1 if they have specified
1967 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1968 * but then during bootup it turns out that sched_clock()
1969 * is reliable after all:
1971 extern int sched_clock_stable;
1973 extern void sched_clock_tick(void);
1974 extern void sched_clock_idle_sleep_event(void);
1975 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1978 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1980 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1981 * The reason for this explicit opt-in is not to have perf penalty with
1982 * slow sched_clocks.
1984 extern void enable_sched_clock_irqtime(void);
1985 extern void disable_sched_clock_irqtime(void);
1987 static inline void enable_sched_clock_irqtime(void) {}
1988 static inline void disable_sched_clock_irqtime(void) {}
1991 extern unsigned long long
1992 task_sched_runtime(struct task_struct *task);
1994 /* sched_exec is called by processes performing an exec */
1996 extern void sched_exec(void);
1998 #define sched_exec() {}
2001 extern void sched_clock_idle_sleep_event(void);
2002 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2004 #ifdef CONFIG_HOTPLUG_CPU
2005 extern void idle_task_exit(void);
2007 static inline void idle_task_exit(void) {}
2010 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2011 extern void wake_up_idle_cpu(int cpu);
2013 static inline void wake_up_idle_cpu(int cpu) { }
2016 #ifdef CONFIG_SCHED_AUTOGROUP
2017 extern void sched_autogroup_create_attach(struct task_struct *p);
2018 extern void sched_autogroup_detach(struct task_struct *p);
2019 extern void sched_autogroup_fork(struct signal_struct *sig);
2020 extern void sched_autogroup_exit(struct signal_struct *sig);
2021 #ifdef CONFIG_PROC_FS
2022 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2023 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2026 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2027 static inline void sched_autogroup_detach(struct task_struct *p) { }
2028 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2029 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2032 extern bool yield_to(struct task_struct *p, bool preempt);
2033 extern void set_user_nice(struct task_struct *p, long nice);
2034 extern int task_prio(const struct task_struct *p);
2035 extern int task_nice(const struct task_struct *p);
2036 extern int can_nice(const struct task_struct *p, const int nice);
2037 extern int task_curr(const struct task_struct *p);
2038 extern int idle_cpu(int cpu);
2039 extern int sched_setscheduler(struct task_struct *, int,
2040 const struct sched_param *);
2041 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2042 const struct sched_param *);
2043 extern struct task_struct *idle_task(int cpu);
2045 * is_idle_task - is the specified task an idle task?
2046 * @p: the task in question.
2048 static inline bool is_idle_task(const struct task_struct *p)
2052 extern struct task_struct *curr_task(int cpu);
2053 extern void set_curr_task(int cpu, struct task_struct *p);
2058 * The default (Linux) execution domain.
2060 extern struct exec_domain default_exec_domain;
2062 union thread_union {
2063 struct thread_info thread_info;
2064 unsigned long stack[THREAD_SIZE/sizeof(long)];
2067 #ifndef __HAVE_ARCH_KSTACK_END
2068 static inline int kstack_end(void *addr)
2070 /* Reliable end of stack detection:
2071 * Some APM bios versions misalign the stack
2073 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2077 extern union thread_union init_thread_union;
2078 extern struct task_struct init_task;
2080 extern struct mm_struct init_mm;
2082 extern struct pid_namespace init_pid_ns;
2085 * find a task by one of its numerical ids
2087 * find_task_by_pid_ns():
2088 * finds a task by its pid in the specified namespace
2089 * find_task_by_vpid():
2090 * finds a task by its virtual pid
2092 * see also find_vpid() etc in include/linux/pid.h
2095 extern struct task_struct *find_task_by_vpid(pid_t nr);
2096 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2097 struct pid_namespace *ns);
2099 extern void __set_special_pids(struct pid *pid);
2101 /* per-UID process charging. */
2102 extern struct user_struct * alloc_uid(kuid_t);
2103 static inline struct user_struct *get_uid(struct user_struct *u)
2105 atomic_inc(&u->__count);
2108 extern void free_uid(struct user_struct *);
2110 #include <asm/current.h>
2112 extern void xtime_update(unsigned long ticks);
2114 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2115 extern int wake_up_process(struct task_struct *tsk);
2116 extern void wake_up_new_task(struct task_struct *tsk);
2118 extern void kick_process(struct task_struct *tsk);
2120 static inline void kick_process(struct task_struct *tsk) { }
2122 extern void sched_fork(struct task_struct *p);
2123 extern void sched_dead(struct task_struct *p);
2125 extern void proc_caches_init(void);
2126 extern void flush_signals(struct task_struct *);
2127 extern void __flush_signals(struct task_struct *);
2128 extern void ignore_signals(struct task_struct *);
2129 extern void flush_signal_handlers(struct task_struct *, int force_default);
2130 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2132 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2134 unsigned long flags;
2137 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2138 ret = dequeue_signal(tsk, mask, info);
2139 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2144 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2146 extern void unblock_all_signals(void);
2147 extern void release_task(struct task_struct * p);
2148 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2149 extern int force_sigsegv(int, struct task_struct *);
2150 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2151 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2152 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2153 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2154 const struct cred *, u32);
2155 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2156 extern int kill_pid(struct pid *pid, int sig, int priv);
2157 extern int kill_proc_info(int, struct siginfo *, pid_t);
2158 extern __must_check bool do_notify_parent(struct task_struct *, int);
2159 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2160 extern void force_sig(int, struct task_struct *);
2161 extern int send_sig(int, struct task_struct *, int);
2162 extern int zap_other_threads(struct task_struct *p);
2163 extern struct sigqueue *sigqueue_alloc(void);
2164 extern void sigqueue_free(struct sigqueue *);
2165 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2166 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2168 static inline void restore_saved_sigmask(void)
2170 if (test_and_clear_restore_sigmask())
2171 __set_current_blocked(¤t->saved_sigmask);
2174 static inline sigset_t *sigmask_to_save(void)
2176 sigset_t *res = ¤t->blocked;
2177 if (unlikely(test_restore_sigmask()))
2178 res = ¤t->saved_sigmask;
2182 static inline int kill_cad_pid(int sig, int priv)
2184 return kill_pid(cad_pid, sig, priv);
2187 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2188 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2189 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2190 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2193 * True if we are on the alternate signal stack.
2195 static inline int on_sig_stack(unsigned long sp)
2197 #ifdef CONFIG_STACK_GROWSUP
2198 return sp >= current->sas_ss_sp &&
2199 sp - current->sas_ss_sp < current->sas_ss_size;
2201 return sp > current->sas_ss_sp &&
2202 sp - current->sas_ss_sp <= current->sas_ss_size;
2206 static inline int sas_ss_flags(unsigned long sp)
2208 return (current->sas_ss_size == 0 ? SS_DISABLE
2209 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2212 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2214 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2215 #ifdef CONFIG_STACK_GROWSUP
2216 return current->sas_ss_sp;
2218 return current->sas_ss_sp + current->sas_ss_size;
2224 * Routines for handling mm_structs
2226 extern struct mm_struct * mm_alloc(void);
2228 /* mmdrop drops the mm and the page tables */
2229 extern void __mmdrop(struct mm_struct *);
2230 static inline void mmdrop(struct mm_struct * mm)
2232 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2236 /* mmput gets rid of the mappings and all user-space */
2237 extern void mmput(struct mm_struct *);
2238 /* Grab a reference to a task's mm, if it is not already going away */
2239 extern struct mm_struct *get_task_mm(struct task_struct *task);
2241 * Grab a reference to a task's mm, if it is not already going away
2242 * and ptrace_may_access with the mode parameter passed to it
2245 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2246 /* Remove the current tasks stale references to the old mm_struct */
2247 extern void mm_release(struct task_struct *, struct mm_struct *);
2248 /* Allocate a new mm structure and copy contents from tsk->mm */
2249 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2251 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2252 struct task_struct *);
2253 extern void flush_thread(void);
2254 extern void exit_thread(void);
2256 extern void exit_files(struct task_struct *);
2257 extern void __cleanup_sighand(struct sighand_struct *);
2259 extern void exit_itimers(struct signal_struct *);
2260 extern void flush_itimer_signals(void);
2262 extern void do_group_exit(int);
2264 extern int allow_signal(int);
2265 extern int disallow_signal(int);
2267 extern int do_execve(const char *,
2268 const char __user * const __user *,
2269 const char __user * const __user *);
2270 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2271 struct task_struct *fork_idle(int);
2272 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2274 extern void set_task_comm(struct task_struct *tsk, char *from);
2275 extern char *get_task_comm(char *to, struct task_struct *tsk);
2278 void scheduler_ipi(void);
2279 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2281 static inline void scheduler_ipi(void) { }
2282 static inline unsigned long wait_task_inactive(struct task_struct *p,
2289 #define next_task(p) \
2290 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2292 #define for_each_process(p) \
2293 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2295 extern bool current_is_single_threaded(void);
2298 * Careful: do_each_thread/while_each_thread is a double loop so
2299 * 'break' will not work as expected - use goto instead.
2301 #define do_each_thread(g, t) \
2302 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2304 #define while_each_thread(g, t) \
2305 while ((t = next_thread(t)) != g)
2307 static inline int get_nr_threads(struct task_struct *tsk)
2309 return tsk->signal->nr_threads;
2312 static inline bool thread_group_leader(struct task_struct *p)
2314 return p->exit_signal >= 0;
2317 /* Do to the insanities of de_thread it is possible for a process
2318 * to have the pid of the thread group leader without actually being
2319 * the thread group leader. For iteration through the pids in proc
2320 * all we care about is that we have a task with the appropriate
2321 * pid, we don't actually care if we have the right task.
2323 static inline int has_group_leader_pid(struct task_struct *p)
2325 return p->pid == p->tgid;
2329 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2331 return p1->tgid == p2->tgid;
2334 static inline struct task_struct *next_thread(const struct task_struct *p)
2336 return list_entry_rcu(p->thread_group.next,
2337 struct task_struct, thread_group);
2340 static inline int thread_group_empty(struct task_struct *p)
2342 return list_empty(&p->thread_group);
2345 #define delay_group_leader(p) \
2346 (thread_group_leader(p) && !thread_group_empty(p))
2349 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2350 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2351 * pins the final release of task.io_context. Also protects ->cpuset and
2352 * ->cgroup.subsys[]. And ->vfork_done.
2354 * Nests both inside and outside of read_lock(&tasklist_lock).
2355 * It must not be nested with write_lock_irq(&tasklist_lock),
2356 * neither inside nor outside.
2358 static inline void task_lock(struct task_struct *p)
2360 spin_lock(&p->alloc_lock);
2363 static inline void task_unlock(struct task_struct *p)
2365 spin_unlock(&p->alloc_lock);
2368 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2369 unsigned long *flags);
2371 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2372 unsigned long *flags)
2374 struct sighand_struct *ret;
2376 ret = __lock_task_sighand(tsk, flags);
2377 (void)__cond_lock(&tsk->sighand->siglock, ret);
2381 static inline void unlock_task_sighand(struct task_struct *tsk,
2382 unsigned long *flags)
2384 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2387 #ifdef CONFIG_CGROUPS
2388 static inline void threadgroup_change_begin(struct task_struct *tsk)
2390 down_read(&tsk->signal->group_rwsem);
2392 static inline void threadgroup_change_end(struct task_struct *tsk)
2394 up_read(&tsk->signal->group_rwsem);
2398 * threadgroup_lock - lock threadgroup
2399 * @tsk: member task of the threadgroup to lock
2401 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2402 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2403 * perform exec. This is useful for cases where the threadgroup needs to
2404 * stay stable across blockable operations.
2406 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2407 * synchronization. While held, no new task will be added to threadgroup
2408 * and no existing live task will have its PF_EXITING set.
2410 * During exec, a task goes and puts its thread group through unusual
2411 * changes. After de-threading, exclusive access is assumed to resources
2412 * which are usually shared by tasks in the same group - e.g. sighand may
2413 * be replaced with a new one. Also, the exec'ing task takes over group
2414 * leader role including its pid. Exclude these changes while locked by
2415 * grabbing cred_guard_mutex which is used to synchronize exec path.
2417 static inline void threadgroup_lock(struct task_struct *tsk)
2420 * exec uses exit for de-threading nesting group_rwsem inside
2421 * cred_guard_mutex. Grab cred_guard_mutex first.
2423 mutex_lock(&tsk->signal->cred_guard_mutex);
2424 down_write(&tsk->signal->group_rwsem);
2428 * threadgroup_unlock - unlock threadgroup
2429 * @tsk: member task of the threadgroup to unlock
2431 * Reverse threadgroup_lock().
2433 static inline void threadgroup_unlock(struct task_struct *tsk)
2435 up_write(&tsk->signal->group_rwsem);
2436 mutex_unlock(&tsk->signal->cred_guard_mutex);
2439 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2440 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2441 static inline void threadgroup_lock(struct task_struct *tsk) {}
2442 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2445 #ifndef __HAVE_THREAD_FUNCTIONS
2447 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2448 #define task_stack_page(task) ((task)->stack)
2450 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2452 *task_thread_info(p) = *task_thread_info(org);
2453 task_thread_info(p)->task = p;
2456 static inline unsigned long *end_of_stack(struct task_struct *p)
2458 return (unsigned long *)(task_thread_info(p) + 1);
2463 static inline int object_is_on_stack(void *obj)
2465 void *stack = task_stack_page(current);
2467 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2470 extern void thread_info_cache_init(void);
2472 #ifdef CONFIG_DEBUG_STACK_USAGE
2473 static inline unsigned long stack_not_used(struct task_struct *p)
2475 unsigned long *n = end_of_stack(p);
2477 do { /* Skip over canary */
2481 return (unsigned long)n - (unsigned long)end_of_stack(p);
2485 /* set thread flags in other task's structures
2486 * - see asm/thread_info.h for TIF_xxxx flags available
2488 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2490 set_ti_thread_flag(task_thread_info(tsk), flag);
2493 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2495 clear_ti_thread_flag(task_thread_info(tsk), flag);
2498 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2500 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2503 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2505 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2508 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2510 return test_ti_thread_flag(task_thread_info(tsk), flag);
2513 static inline void set_tsk_need_resched(struct task_struct *tsk)
2515 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2518 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2520 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2523 static inline int test_tsk_need_resched(struct task_struct *tsk)
2525 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2528 static inline int restart_syscall(void)
2530 set_tsk_thread_flag(current, TIF_SIGPENDING);
2531 return -ERESTARTNOINTR;
2534 static inline int signal_pending(struct task_struct *p)
2536 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2539 static inline int __fatal_signal_pending(struct task_struct *p)
2541 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2544 static inline int fatal_signal_pending(struct task_struct *p)
2546 return signal_pending(p) && __fatal_signal_pending(p);
2549 static inline int signal_pending_state(long state, struct task_struct *p)
2551 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2553 if (!signal_pending(p))
2556 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2559 static inline int need_resched(void)
2561 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2565 * cond_resched() and cond_resched_lock(): latency reduction via
2566 * explicit rescheduling in places that are safe. The return
2567 * value indicates whether a reschedule was done in fact.
2568 * cond_resched_lock() will drop the spinlock before scheduling,
2569 * cond_resched_softirq() will enable bhs before scheduling.
2571 extern int _cond_resched(void);
2573 #define cond_resched() ({ \
2574 __might_sleep(__FILE__, __LINE__, 0); \
2578 extern int __cond_resched_lock(spinlock_t *lock);
2580 #ifdef CONFIG_PREEMPT_COUNT
2581 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2583 #define PREEMPT_LOCK_OFFSET 0
2586 #define cond_resched_lock(lock) ({ \
2587 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2588 __cond_resched_lock(lock); \
2591 extern int __cond_resched_softirq(void);
2593 #define cond_resched_softirq() ({ \
2594 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2595 __cond_resched_softirq(); \
2599 * Does a critical section need to be broken due to another
2600 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2601 * but a general need for low latency)
2603 static inline int spin_needbreak(spinlock_t *lock)
2605 #ifdef CONFIG_PREEMPT
2606 return spin_is_contended(lock);
2613 * Thread group CPU time accounting.
2615 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2616 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2618 static inline void thread_group_cputime_init(struct signal_struct *sig)
2620 raw_spin_lock_init(&sig->cputimer.lock);
2624 * Reevaluate whether the task has signals pending delivery.
2625 * Wake the task if so.
2626 * This is required every time the blocked sigset_t changes.
2627 * callers must hold sighand->siglock.
2629 extern void recalc_sigpending_and_wake(struct task_struct *t);
2630 extern void recalc_sigpending(void);
2632 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2634 static inline void signal_wake_up(struct task_struct *t, bool resume)
2636 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2638 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2640 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2644 * Wrappers for p->thread_info->cpu access. No-op on UP.
2648 static inline unsigned int task_cpu(const struct task_struct *p)
2650 return task_thread_info(p)->cpu;
2653 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2657 static inline unsigned int task_cpu(const struct task_struct *p)
2662 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2666 #endif /* CONFIG_SMP */
2668 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2669 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2671 #ifdef CONFIG_CGROUP_SCHED
2673 extern struct task_group root_task_group;
2675 extern struct task_group *sched_create_group(struct task_group *parent);
2676 extern void sched_online_group(struct task_group *tg,
2677 struct task_group *parent);
2678 extern void sched_destroy_group(struct task_group *tg);
2679 extern void sched_offline_group(struct task_group *tg);
2680 extern void sched_move_task(struct task_struct *tsk);
2681 #ifdef CONFIG_FAIR_GROUP_SCHED
2682 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2683 extern unsigned long sched_group_shares(struct task_group *tg);
2685 #ifdef CONFIG_RT_GROUP_SCHED
2686 extern int sched_group_set_rt_runtime(struct task_group *tg,
2687 long rt_runtime_us);
2688 extern long sched_group_rt_runtime(struct task_group *tg);
2689 extern int sched_group_set_rt_period(struct task_group *tg,
2691 extern long sched_group_rt_period(struct task_group *tg);
2692 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2694 #endif /* CONFIG_CGROUP_SCHED */
2696 extern int task_can_switch_user(struct user_struct *up,
2697 struct task_struct *tsk);
2699 #ifdef CONFIG_TASK_XACCT
2700 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2702 tsk->ioac.rchar += amt;
2705 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2707 tsk->ioac.wchar += amt;
2710 static inline void inc_syscr(struct task_struct *tsk)
2715 static inline void inc_syscw(struct task_struct *tsk)
2720 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2724 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2728 static inline void inc_syscr(struct task_struct *tsk)
2732 static inline void inc_syscw(struct task_struct *tsk)
2737 #ifndef TASK_SIZE_OF
2738 #define TASK_SIZE_OF(tsk) TASK_SIZE
2741 #ifdef CONFIG_MM_OWNER
2742 extern void mm_update_next_owner(struct mm_struct *mm);
2743 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2745 static inline void mm_update_next_owner(struct mm_struct *mm)
2749 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2752 #endif /* CONFIG_MM_OWNER */
2754 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2757 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2760 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2763 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2766 static inline unsigned long rlimit(unsigned int limit)
2768 return task_rlimit(current, limit);
2771 static inline unsigned long rlimit_max(unsigned int limit)
2773 return task_rlimit_max(current, limit);