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);
132 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
135 static inline void proc_sched_set_task(struct task_struct *p)
139 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
145 * Task state bitmask. NOTE! These bits are also
146 * encoded in fs/proc/array.c: get_task_state().
148 * We have two separate sets of flags: task->state
149 * is about runnability, while task->exit_state are
150 * about the task exiting. Confusing, but this way
151 * modifying one set can't modify the other one by
154 #define TASK_RUNNING 0
155 #define TASK_INTERRUPTIBLE 1
156 #define TASK_UNINTERRUPTIBLE 2
157 #define __TASK_STOPPED 4
158 #define __TASK_TRACED 8
159 /* in tsk->exit_state */
160 #define EXIT_ZOMBIE 16
162 /* in tsk->state again */
164 #define TASK_WAKEKILL 128
165 #define TASK_WAKING 256
166 #define TASK_PARKED 512
167 #define TASK_STATE_MAX 1024
169 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
171 extern char ___assert_task_state[1 - 2*!!(
172 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
174 /* Convenience macros for the sake of set_task_state */
175 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
176 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
177 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
179 /* Convenience macros for the sake of wake_up */
180 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
181 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
183 /* get_task_state() */
184 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
185 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
188 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
189 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
190 #define task_is_dead(task) ((task)->exit_state != 0)
191 #define task_is_stopped_or_traced(task) \
192 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
193 #define task_contributes_to_load(task) \
194 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
195 (task->flags & PF_FROZEN) == 0)
197 #define __set_task_state(tsk, state_value) \
198 do { (tsk)->state = (state_value); } while (0)
199 #define set_task_state(tsk, state_value) \
200 set_mb((tsk)->state, (state_value))
203 * set_current_state() includes a barrier so that the write of current->state
204 * is correctly serialised wrt the caller's subsequent test of whether to
207 * set_current_state(TASK_UNINTERRUPTIBLE);
208 * if (do_i_need_to_sleep())
211 * If the caller does not need such serialisation then use __set_current_state()
213 #define __set_current_state(state_value) \
214 do { current->state = (state_value); } while (0)
215 #define set_current_state(state_value) \
216 set_mb(current->state, (state_value))
218 /* Task command name length */
219 #define TASK_COMM_LEN 16
221 #include <linux/spinlock.h>
224 * This serializes "schedule()" and also protects
225 * the run-queue from deletions/modifications (but
226 * _adding_ to the beginning of the run-queue has
229 extern rwlock_t tasklist_lock;
230 extern spinlock_t mmlist_lock;
234 #ifdef CONFIG_PROVE_RCU
235 extern int lockdep_tasklist_lock_is_held(void);
236 #endif /* #ifdef CONFIG_PROVE_RCU */
238 extern void sched_init(void);
239 extern void sched_init_smp(void);
240 extern asmlinkage void schedule_tail(struct task_struct *prev);
241 extern void init_idle(struct task_struct *idle, int cpu);
242 extern void init_idle_bootup_task(struct task_struct *idle);
244 extern int runqueue_is_locked(int cpu);
246 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
247 extern void nohz_balance_enter_idle(int cpu);
248 extern void set_cpu_sd_state_idle(void);
249 extern int get_nohz_timer_target(void);
251 static inline void nohz_balance_enter_idle(int cpu) { }
252 static inline void set_cpu_sd_state_idle(void) { }
256 * Only dump TASK_* tasks. (0 for all tasks)
258 extern void show_state_filter(unsigned long state_filter);
260 static inline void show_state(void)
262 show_state_filter(0);
265 extern void show_regs(struct pt_regs *);
268 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
269 * task), SP is the stack pointer of the first frame that should be shown in the back
270 * trace (or NULL if the entire call-chain of the task should be shown).
272 extern void show_stack(struct task_struct *task, unsigned long *sp);
274 void io_schedule(void);
275 long io_schedule_timeout(long timeout);
277 extern void cpu_init (void);
278 extern void trap_init(void);
279 extern void update_process_times(int user);
280 extern void scheduler_tick(void);
282 extern void sched_show_task(struct task_struct *p);
284 #ifdef CONFIG_LOCKUP_DETECTOR
285 extern void touch_softlockup_watchdog(void);
286 extern void touch_softlockup_watchdog_sync(void);
287 extern void touch_all_softlockup_watchdogs(void);
288 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
290 size_t *lenp, loff_t *ppos);
291 extern unsigned int softlockup_panic;
292 void lockup_detector_init(void);
294 static inline void touch_softlockup_watchdog(void)
297 static inline void touch_softlockup_watchdog_sync(void)
300 static inline void touch_all_softlockup_watchdogs(void)
303 static inline void lockup_detector_init(void)
308 /* Attach to any functions which should be ignored in wchan output. */
309 #define __sched __attribute__((__section__(".sched.text")))
311 /* Linker adds these: start and end of __sched functions */
312 extern char __sched_text_start[], __sched_text_end[];
314 /* Is this address in the __sched functions? */
315 extern int in_sched_functions(unsigned long addr);
317 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
318 extern signed long schedule_timeout(signed long timeout);
319 extern signed long schedule_timeout_interruptible(signed long timeout);
320 extern signed long schedule_timeout_killable(signed long timeout);
321 extern signed long schedule_timeout_uninterruptible(signed long timeout);
322 asmlinkage void schedule(void);
323 extern void schedule_preempt_disabled(void);
324 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
327 struct user_namespace;
329 #include <linux/aio.h>
332 extern void arch_pick_mmap_layout(struct mm_struct *mm);
334 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
335 unsigned long, unsigned long);
337 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
338 unsigned long len, unsigned long pgoff,
339 unsigned long flags);
340 extern void arch_unmap_area(struct mm_struct *, unsigned long);
341 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
343 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
347 extern void set_dumpable(struct mm_struct *mm, int value);
348 extern int get_dumpable(struct mm_struct *mm);
352 #define MMF_DUMPABLE 0 /* core dump is permitted */
353 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
355 #define MMF_DUMPABLE_BITS 2
356 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
358 /* coredump filter bits */
359 #define MMF_DUMP_ANON_PRIVATE 2
360 #define MMF_DUMP_ANON_SHARED 3
361 #define MMF_DUMP_MAPPED_PRIVATE 4
362 #define MMF_DUMP_MAPPED_SHARED 5
363 #define MMF_DUMP_ELF_HEADERS 6
364 #define MMF_DUMP_HUGETLB_PRIVATE 7
365 #define MMF_DUMP_HUGETLB_SHARED 8
367 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
368 #define MMF_DUMP_FILTER_BITS 7
369 #define MMF_DUMP_FILTER_MASK \
370 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
371 #define MMF_DUMP_FILTER_DEFAULT \
372 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
373 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
375 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
376 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
378 # define MMF_DUMP_MASK_DEFAULT_ELF 0
380 /* leave room for more dump flags */
381 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
382 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
383 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
385 #define MMF_HAS_UPROBES 19 /* has uprobes */
386 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
388 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
390 struct sighand_struct {
392 struct k_sigaction action[_NSIG];
394 wait_queue_head_t signalfd_wqh;
397 struct pacct_struct {
400 unsigned long ac_mem;
401 cputime_t ac_utime, ac_stime;
402 unsigned long ac_minflt, ac_majflt;
413 * struct cputime - snaphsot of system and user cputime
414 * @utime: time spent in user mode
415 * @stime: time spent in system mode
417 * Gathers a generic snapshot of user and system time.
425 * struct task_cputime - collected CPU time counts
426 * @utime: time spent in user mode, in &cputime_t units
427 * @stime: time spent in kernel mode, in &cputime_t units
428 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
430 * This is an extension of struct cputime that includes the total runtime
431 * spent by the task from the scheduler point of view.
433 * As a result, this structure groups together three kinds of CPU time
434 * that are tracked for threads and thread groups. Most things considering
435 * CPU time want to group these counts together and treat all three
436 * of them in parallel.
438 struct task_cputime {
441 unsigned long long sum_exec_runtime;
443 /* Alternate field names when used to cache expirations. */
444 #define prof_exp stime
445 #define virt_exp utime
446 #define sched_exp sum_exec_runtime
448 #define INIT_CPUTIME \
449 (struct task_cputime) { \
452 .sum_exec_runtime = 0, \
456 * Disable preemption until the scheduler is running.
457 * Reset by start_kernel()->sched_init()->init_idle().
459 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
460 * before the scheduler is active -- see should_resched().
462 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
465 * struct thread_group_cputimer - thread group interval timer counts
466 * @cputime: thread group interval timers.
467 * @running: non-zero when there are timers running and
468 * @cputime receives updates.
469 * @lock: lock for fields in this struct.
471 * This structure contains the version of task_cputime, above, that is
472 * used for thread group CPU timer calculations.
474 struct thread_group_cputimer {
475 struct task_cputime cputime;
480 #include <linux/rwsem.h>
484 * NOTE! "signal_struct" does not have its own
485 * locking, because a shared signal_struct always
486 * implies a shared sighand_struct, so locking
487 * sighand_struct is always a proper superset of
488 * the locking of signal_struct.
490 struct signal_struct {
495 wait_queue_head_t wait_chldexit; /* for wait4() */
497 /* current thread group signal load-balancing target: */
498 struct task_struct *curr_target;
500 /* shared signal handling: */
501 struct sigpending shared_pending;
503 /* thread group exit support */
506 * - notify group_exit_task when ->count is equal to notify_count
507 * - everyone except group_exit_task is stopped during signal delivery
508 * of fatal signals, group_exit_task processes the signal.
511 struct task_struct *group_exit_task;
513 /* thread group stop support, overloads group_exit_code too */
514 int group_stop_count;
515 unsigned int flags; /* see SIGNAL_* flags below */
518 * PR_SET_CHILD_SUBREAPER marks a process, like a service
519 * manager, to re-parent orphan (double-forking) child processes
520 * to this process instead of 'init'. The service manager is
521 * able to receive SIGCHLD signals and is able to investigate
522 * the process until it calls wait(). All children of this
523 * process will inherit a flag if they should look for a
524 * child_subreaper process at exit.
526 unsigned int is_child_subreaper:1;
527 unsigned int has_child_subreaper:1;
529 /* POSIX.1b Interval Timers */
531 struct list_head posix_timers;
533 /* ITIMER_REAL timer for the process */
534 struct hrtimer real_timer;
535 struct pid *leader_pid;
536 ktime_t it_real_incr;
539 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
540 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
541 * values are defined to 0 and 1 respectively
543 struct cpu_itimer it[2];
546 * Thread group totals for process CPU timers.
547 * See thread_group_cputimer(), et al, for details.
549 struct thread_group_cputimer cputimer;
551 /* Earliest-expiration cache. */
552 struct task_cputime cputime_expires;
554 struct list_head cpu_timers[3];
556 struct pid *tty_old_pgrp;
558 /* boolean value for session group leader */
561 struct tty_struct *tty; /* NULL if no tty */
563 #ifdef CONFIG_SCHED_AUTOGROUP
564 struct autogroup *autogroup;
567 * Cumulative resource counters for dead threads in the group,
568 * and for reaped dead child processes forked by this group.
569 * Live threads maintain their own counters and add to these
570 * in __exit_signal, except for the group leader.
572 cputime_t utime, stime, cutime, cstime;
575 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
576 struct cputime prev_cputime;
578 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
579 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
580 unsigned long inblock, oublock, cinblock, coublock;
581 unsigned long maxrss, cmaxrss;
582 struct task_io_accounting ioac;
585 * Cumulative ns of schedule CPU time fo dead threads in the
586 * group, not including a zombie group leader, (This only differs
587 * from jiffies_to_ns(utime + stime) if sched_clock uses something
588 * other than jiffies.)
590 unsigned long long sum_sched_runtime;
593 * We don't bother to synchronize most readers of this at all,
594 * because there is no reader checking a limit that actually needs
595 * to get both rlim_cur and rlim_max atomically, and either one
596 * alone is a single word that can safely be read normally.
597 * getrlimit/setrlimit use task_lock(current->group_leader) to
598 * protect this instead of the siglock, because they really
599 * have no need to disable irqs.
601 struct rlimit rlim[RLIM_NLIMITS];
603 #ifdef CONFIG_BSD_PROCESS_ACCT
604 struct pacct_struct pacct; /* per-process accounting information */
606 #ifdef CONFIG_TASKSTATS
607 struct taskstats *stats;
611 struct tty_audit_buf *tty_audit_buf;
613 #ifdef CONFIG_CGROUPS
615 * group_rwsem prevents new tasks from entering the threadgroup and
616 * member tasks from exiting,a more specifically, setting of
617 * PF_EXITING. fork and exit paths are protected with this rwsem
618 * using threadgroup_change_begin/end(). Users which require
619 * threadgroup to remain stable should use threadgroup_[un]lock()
620 * which also takes care of exec path. Currently, cgroup is the
623 struct rw_semaphore group_rwsem;
626 oom_flags_t oom_flags;
627 short oom_score_adj; /* OOM kill score adjustment */
628 short oom_score_adj_min; /* OOM kill score adjustment min value.
629 * Only settable by CAP_SYS_RESOURCE. */
631 struct mutex cred_guard_mutex; /* guard against foreign influences on
632 * credential calculations
633 * (notably. ptrace) */
637 * Bits in flags field of signal_struct.
639 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
640 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
641 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
643 * Pending notifications to parent.
645 #define SIGNAL_CLD_STOPPED 0x00000010
646 #define SIGNAL_CLD_CONTINUED 0x00000020
647 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
649 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
651 /* If true, all threads except ->group_exit_task have pending SIGKILL */
652 static inline int signal_group_exit(const struct signal_struct *sig)
654 return (sig->flags & SIGNAL_GROUP_EXIT) ||
655 (sig->group_exit_task != NULL);
659 * Some day this will be a full-fledged user tracking system..
662 atomic_t __count; /* reference count */
663 atomic_t processes; /* How many processes does this user have? */
664 atomic_t files; /* How many open files does this user have? */
665 atomic_t sigpending; /* How many pending signals does this user have? */
666 #ifdef CONFIG_INOTIFY_USER
667 atomic_t inotify_watches; /* How many inotify watches does this user have? */
668 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
670 #ifdef CONFIG_FANOTIFY
671 atomic_t fanotify_listeners;
674 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
676 #ifdef CONFIG_POSIX_MQUEUE
677 /* protected by mq_lock */
678 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
680 unsigned long locked_shm; /* How many pages of mlocked shm ? */
683 struct key *uid_keyring; /* UID specific keyring */
684 struct key *session_keyring; /* UID's default session keyring */
687 /* Hash table maintenance information */
688 struct hlist_node uidhash_node;
691 #ifdef CONFIG_PERF_EVENTS
692 atomic_long_t locked_vm;
696 extern int uids_sysfs_init(void);
698 extern struct user_struct *find_user(kuid_t);
700 extern struct user_struct root_user;
701 #define INIT_USER (&root_user)
704 struct backing_dev_info;
705 struct reclaim_state;
707 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
709 /* cumulative counters */
710 unsigned long pcount; /* # of times run on this cpu */
711 unsigned long long run_delay; /* time spent waiting on a runqueue */
714 unsigned long long last_arrival,/* when we last ran on a cpu */
715 last_queued; /* when we were last queued to run */
717 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
719 #ifdef CONFIG_TASK_DELAY_ACCT
720 struct task_delay_info {
722 unsigned int flags; /* Private per-task flags */
724 /* For each stat XXX, add following, aligned appropriately
726 * struct timespec XXX_start, XXX_end;
730 * Atomicity of updates to XXX_delay, XXX_count protected by
731 * single lock above (split into XXX_lock if contention is an issue).
735 * XXX_count is incremented on every XXX operation, the delay
736 * associated with the operation is added to XXX_delay.
737 * XXX_delay contains the accumulated delay time in nanoseconds.
739 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
740 u64 blkio_delay; /* wait for sync block io completion */
741 u64 swapin_delay; /* wait for swapin block io completion */
742 u32 blkio_count; /* total count of the number of sync block */
743 /* io operations performed */
744 u32 swapin_count; /* total count of the number of swapin block */
745 /* io operations performed */
747 struct timespec freepages_start, freepages_end;
748 u64 freepages_delay; /* wait for memory reclaim */
749 u32 freepages_count; /* total count of memory reclaim */
751 #endif /* CONFIG_TASK_DELAY_ACCT */
753 static inline int sched_info_on(void)
755 #ifdef CONFIG_SCHEDSTATS
757 #elif defined(CONFIG_TASK_DELAY_ACCT)
758 extern int delayacct_on;
773 * Increase resolution of nice-level calculations for 64-bit architectures.
774 * The extra resolution improves shares distribution and load balancing of
775 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
776 * hierarchies, especially on larger systems. This is not a user-visible change
777 * and does not change the user-interface for setting shares/weights.
779 * We increase resolution only if we have enough bits to allow this increased
780 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
781 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
784 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
785 # define SCHED_LOAD_RESOLUTION 10
786 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
787 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
789 # define SCHED_LOAD_RESOLUTION 0
790 # define scale_load(w) (w)
791 # define scale_load_down(w) (w)
794 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
795 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
798 * Increase resolution of cpu_power calculations
800 #define SCHED_POWER_SHIFT 10
801 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
804 * sched-domains (multiprocessor balancing) declarations:
807 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
808 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
809 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
810 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
811 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
812 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
813 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
814 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
815 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
816 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
817 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
818 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
820 extern int __weak arch_sd_sibiling_asym_packing(void);
822 struct sched_group_power {
825 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
828 unsigned int power, power_orig;
829 unsigned long next_update;
831 * Number of busy cpus in this group.
833 atomic_t nr_busy_cpus;
835 unsigned long cpumask[0]; /* iteration mask */
839 struct sched_group *next; /* Must be a circular list */
842 unsigned int group_weight;
843 struct sched_group_power *sgp;
846 * The CPUs this group covers.
848 * NOTE: this field is variable length. (Allocated dynamically
849 * by attaching extra space to the end of the structure,
850 * depending on how many CPUs the kernel has booted up with)
852 unsigned long cpumask[0];
855 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
857 return to_cpumask(sg->cpumask);
861 * cpumask masking which cpus in the group are allowed to iterate up the domain
864 static inline struct cpumask *sched_group_mask(struct sched_group *sg)
866 return to_cpumask(sg->sgp->cpumask);
870 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
871 * @group: The group whose first cpu is to be returned.
873 static inline unsigned int group_first_cpu(struct sched_group *group)
875 return cpumask_first(sched_group_cpus(group));
878 struct sched_domain_attr {
879 int relax_domain_level;
882 #define SD_ATTR_INIT (struct sched_domain_attr) { \
883 .relax_domain_level = -1, \
886 extern int sched_domain_level_max;
888 struct sched_domain {
889 /* These fields must be setup */
890 struct sched_domain *parent; /* top domain must be null terminated */
891 struct sched_domain *child; /* bottom domain must be null terminated */
892 struct sched_group *groups; /* the balancing groups of the domain */
893 unsigned long min_interval; /* Minimum balance interval ms */
894 unsigned long max_interval; /* Maximum balance interval ms */
895 unsigned int busy_factor; /* less balancing by factor if busy */
896 unsigned int imbalance_pct; /* No balance until over watermark */
897 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
898 unsigned int busy_idx;
899 unsigned int idle_idx;
900 unsigned int newidle_idx;
901 unsigned int wake_idx;
902 unsigned int forkexec_idx;
903 unsigned int smt_gain;
904 int flags; /* See SD_* */
907 /* Runtime fields. */
908 unsigned long last_balance; /* init to jiffies. units in jiffies */
909 unsigned int balance_interval; /* initialise to 1. units in ms. */
910 unsigned int nr_balance_failed; /* initialise to 0 */
914 #ifdef CONFIG_SCHEDSTATS
915 /* load_balance() stats */
916 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
917 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
918 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
919 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
920 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
921 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
922 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
923 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
925 /* Active load balancing */
926 unsigned int alb_count;
927 unsigned int alb_failed;
928 unsigned int alb_pushed;
930 /* SD_BALANCE_EXEC stats */
931 unsigned int sbe_count;
932 unsigned int sbe_balanced;
933 unsigned int sbe_pushed;
935 /* SD_BALANCE_FORK stats */
936 unsigned int sbf_count;
937 unsigned int sbf_balanced;
938 unsigned int sbf_pushed;
940 /* try_to_wake_up() stats */
941 unsigned int ttwu_wake_remote;
942 unsigned int ttwu_move_affine;
943 unsigned int ttwu_move_balance;
945 #ifdef CONFIG_SCHED_DEBUG
949 void *private; /* used during construction */
950 struct rcu_head rcu; /* used during destruction */
953 unsigned int span_weight;
955 * Span of all CPUs in this domain.
957 * NOTE: this field is variable length. (Allocated dynamically
958 * by attaching extra space to the end of the structure,
959 * depending on how many CPUs the kernel has booted up with)
961 unsigned long span[0];
964 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
966 return to_cpumask(sd->span);
969 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
970 struct sched_domain_attr *dattr_new);
972 /* Allocate an array of sched domains, for partition_sched_domains(). */
973 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
974 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
976 /* Test a flag in parent sched domain */
977 static inline int test_sd_parent(struct sched_domain *sd, int flag)
979 if (sd->parent && (sd->parent->flags & flag))
985 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
986 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
988 bool cpus_share_cache(int this_cpu, int that_cpu);
990 #else /* CONFIG_SMP */
992 struct sched_domain_attr;
995 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
996 struct sched_domain_attr *dattr_new)
1000 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1005 #endif /* !CONFIG_SMP */
1008 struct io_context; /* See blkdev.h */
1011 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1012 extern void prefetch_stack(struct task_struct *t);
1014 static inline void prefetch_stack(struct task_struct *t) { }
1017 struct audit_context; /* See audit.c */
1019 struct pipe_inode_info;
1020 struct uts_namespace;
1023 struct sched_domain;
1028 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1029 #define WF_FORK 0x02 /* child wakeup after fork */
1030 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1032 #define ENQUEUE_WAKEUP 1
1033 #define ENQUEUE_HEAD 2
1035 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1037 #define ENQUEUE_WAKING 0
1040 #define DEQUEUE_SLEEP 1
1042 struct sched_class {
1043 const struct sched_class *next;
1045 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1046 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1047 void (*yield_task) (struct rq *rq);
1048 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1050 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1052 struct task_struct * (*pick_next_task) (struct rq *rq);
1053 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1056 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1057 void (*migrate_task_rq)(struct task_struct *p, int next_cpu);
1059 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1060 void (*post_schedule) (struct rq *this_rq);
1061 void (*task_waking) (struct task_struct *task);
1062 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1064 void (*set_cpus_allowed)(struct task_struct *p,
1065 const struct cpumask *newmask);
1067 void (*rq_online)(struct rq *rq);
1068 void (*rq_offline)(struct rq *rq);
1071 void (*set_curr_task) (struct rq *rq);
1072 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1073 void (*task_fork) (struct task_struct *p);
1075 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1076 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1077 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1080 unsigned int (*get_rr_interval) (struct rq *rq,
1081 struct task_struct *task);
1083 #ifdef CONFIG_FAIR_GROUP_SCHED
1084 void (*task_move_group) (struct task_struct *p, int on_rq);
1088 struct load_weight {
1089 unsigned long weight, inv_weight;
1094 * These sums represent an infinite geometric series and so are bound
1095 * above by 1024/(1-y). Thus we only need a u32 to store them for for all
1096 * choices of y < 1-2^(-32)*1024.
1098 u32 runnable_avg_sum, runnable_avg_period;
1099 u64 last_runnable_update;
1101 unsigned long load_avg_contrib;
1104 #ifdef CONFIG_SCHEDSTATS
1105 struct sched_statistics {
1115 s64 sum_sleep_runtime;
1122 u64 nr_migrations_cold;
1123 u64 nr_failed_migrations_affine;
1124 u64 nr_failed_migrations_running;
1125 u64 nr_failed_migrations_hot;
1126 u64 nr_forced_migrations;
1129 u64 nr_wakeups_sync;
1130 u64 nr_wakeups_migrate;
1131 u64 nr_wakeups_local;
1132 u64 nr_wakeups_remote;
1133 u64 nr_wakeups_affine;
1134 u64 nr_wakeups_affine_attempts;
1135 u64 nr_wakeups_passive;
1136 u64 nr_wakeups_idle;
1140 struct sched_entity {
1141 struct load_weight load; /* for load-balancing */
1142 struct rb_node run_node;
1143 struct list_head group_node;
1147 u64 sum_exec_runtime;
1149 u64 prev_sum_exec_runtime;
1153 #ifdef CONFIG_SCHEDSTATS
1154 struct sched_statistics statistics;
1157 #ifdef CONFIG_FAIR_GROUP_SCHED
1158 struct sched_entity *parent;
1159 /* rq on which this entity is (to be) queued: */
1160 struct cfs_rq *cfs_rq;
1161 /* rq "owned" by this entity/group: */
1162 struct cfs_rq *my_q;
1166 * Load-tracking only depends on SMP, FAIR_GROUP_SCHED dependency below may be
1167 * removed when useful for applications beyond shares distribution (e.g.
1170 #if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)
1171 /* Per-entity load-tracking */
1172 struct sched_avg avg;
1176 struct sched_rt_entity {
1177 struct list_head run_list;
1178 unsigned long timeout;
1179 unsigned long watchdog_stamp;
1180 unsigned int time_slice;
1182 struct sched_rt_entity *back;
1183 #ifdef CONFIG_RT_GROUP_SCHED
1184 struct sched_rt_entity *parent;
1185 /* rq on which this entity is (to be) queued: */
1186 struct rt_rq *rt_rq;
1187 /* rq "owned" by this entity/group: */
1195 enum perf_event_task_context {
1196 perf_invalid_context = -1,
1197 perf_hw_context = 0,
1199 perf_nr_task_contexts,
1202 struct task_struct {
1203 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1206 unsigned int flags; /* per process flags, defined below */
1207 unsigned int ptrace;
1210 struct llist_node wake_entry;
1215 int prio, static_prio, normal_prio;
1216 unsigned int rt_priority;
1217 const struct sched_class *sched_class;
1218 struct sched_entity se;
1219 struct sched_rt_entity rt;
1220 #ifdef CONFIG_CGROUP_SCHED
1221 struct task_group *sched_task_group;
1224 #ifdef CONFIG_PREEMPT_NOTIFIERS
1225 /* list of struct preempt_notifier: */
1226 struct hlist_head preempt_notifiers;
1230 * fpu_counter contains the number of consecutive context switches
1231 * that the FPU is used. If this is over a threshold, the lazy fpu
1232 * saving becomes unlazy to save the trap. This is an unsigned char
1233 * so that after 256 times the counter wraps and the behavior turns
1234 * lazy again; this to deal with bursty apps that only use FPU for
1237 unsigned char fpu_counter;
1238 #ifdef CONFIG_BLK_DEV_IO_TRACE
1239 unsigned int btrace_seq;
1242 unsigned int policy;
1243 int nr_cpus_allowed;
1244 cpumask_t cpus_allowed;
1246 #ifdef CONFIG_PREEMPT_RCU
1247 int rcu_read_lock_nesting;
1248 char rcu_read_unlock_special;
1249 struct list_head rcu_node_entry;
1250 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1251 #ifdef CONFIG_TREE_PREEMPT_RCU
1252 struct rcu_node *rcu_blocked_node;
1253 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1254 #ifdef CONFIG_RCU_BOOST
1255 struct rt_mutex *rcu_boost_mutex;
1256 #endif /* #ifdef CONFIG_RCU_BOOST */
1258 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1259 struct sched_info sched_info;
1262 struct list_head tasks;
1264 struct plist_node pushable_tasks;
1267 struct mm_struct *mm, *active_mm;
1268 #ifdef CONFIG_COMPAT_BRK
1269 unsigned brk_randomized:1;
1271 #if defined(SPLIT_RSS_COUNTING)
1272 struct task_rss_stat rss_stat;
1276 int exit_code, exit_signal;
1277 int pdeath_signal; /* The signal sent when the parent dies */
1278 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1280 unsigned int personality;
1281 unsigned did_exec:1;
1282 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1284 unsigned in_iowait:1;
1286 /* task may not gain privileges */
1287 unsigned no_new_privs:1;
1289 /* Revert to default priority/policy when forking */
1290 unsigned sched_reset_on_fork:1;
1291 unsigned sched_contributes_to_load:1;
1296 #ifdef CONFIG_CC_STACKPROTECTOR
1297 /* Canary value for the -fstack-protector gcc feature */
1298 unsigned long stack_canary;
1301 * pointers to (original) parent process, youngest child, younger sibling,
1302 * older sibling, respectively. (p->father can be replaced with
1303 * p->real_parent->pid)
1305 struct task_struct __rcu *real_parent; /* real parent process */
1306 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1308 * children/sibling forms the list of my natural children
1310 struct list_head children; /* list of my children */
1311 struct list_head sibling; /* linkage in my parent's children list */
1312 struct task_struct *group_leader; /* threadgroup leader */
1315 * ptraced is the list of tasks this task is using ptrace on.
1316 * This includes both natural children and PTRACE_ATTACH targets.
1317 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1319 struct list_head ptraced;
1320 struct list_head ptrace_entry;
1322 /* PID/PID hash table linkage. */
1323 struct pid_link pids[PIDTYPE_MAX];
1324 struct list_head thread_group;
1326 struct completion *vfork_done; /* for vfork() */
1327 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1328 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1330 cputime_t utime, stime, utimescaled, stimescaled;
1332 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1333 struct cputime prev_cputime;
1335 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1336 seqlock_t vtime_seqlock;
1337 unsigned long long vtime_snap;
1342 } vtime_snap_whence;
1344 unsigned long nvcsw, nivcsw; /* context switch counts */
1345 struct timespec start_time; /* monotonic time */
1346 struct timespec real_start_time; /* boot based time */
1347 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1348 unsigned long min_flt, maj_flt;
1350 struct task_cputime cputime_expires;
1351 struct list_head cpu_timers[3];
1353 /* process credentials */
1354 const struct cred __rcu *real_cred; /* objective and real subjective task
1355 * credentials (COW) */
1356 const struct cred __rcu *cred; /* effective (overridable) subjective task
1357 * credentials (COW) */
1358 char comm[TASK_COMM_LEN]; /* executable name excluding path
1359 - access with [gs]et_task_comm (which lock
1360 it with task_lock())
1361 - initialized normally by setup_new_exec */
1362 /* file system info */
1363 int link_count, total_link_count;
1364 #ifdef CONFIG_SYSVIPC
1366 struct sysv_sem sysvsem;
1368 #ifdef CONFIG_DETECT_HUNG_TASK
1369 /* hung task detection */
1370 unsigned long last_switch_count;
1372 /* CPU-specific state of this task */
1373 struct thread_struct thread;
1374 /* filesystem information */
1375 struct fs_struct *fs;
1376 /* open file information */
1377 struct files_struct *files;
1379 struct nsproxy *nsproxy;
1380 /* signal handlers */
1381 struct signal_struct *signal;
1382 struct sighand_struct *sighand;
1384 sigset_t blocked, real_blocked;
1385 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1386 struct sigpending pending;
1388 unsigned long sas_ss_sp;
1390 int (*notifier)(void *priv);
1391 void *notifier_data;
1392 sigset_t *notifier_mask;
1393 struct callback_head *task_works;
1395 struct audit_context *audit_context;
1396 #ifdef CONFIG_AUDITSYSCALL
1398 unsigned int sessionid;
1400 struct seccomp seccomp;
1402 /* Thread group tracking */
1405 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1407 spinlock_t alloc_lock;
1409 /* Protection of the PI data structures: */
1410 raw_spinlock_t pi_lock;
1412 #ifdef CONFIG_RT_MUTEXES
1413 /* PI waiters blocked on a rt_mutex held by this task */
1414 struct plist_head pi_waiters;
1415 /* Deadlock detection and priority inheritance handling */
1416 struct rt_mutex_waiter *pi_blocked_on;
1419 #ifdef CONFIG_DEBUG_MUTEXES
1420 /* mutex deadlock detection */
1421 struct mutex_waiter *blocked_on;
1423 #ifdef CONFIG_TRACE_IRQFLAGS
1424 unsigned int irq_events;
1425 unsigned long hardirq_enable_ip;
1426 unsigned long hardirq_disable_ip;
1427 unsigned int hardirq_enable_event;
1428 unsigned int hardirq_disable_event;
1429 int hardirqs_enabled;
1430 int hardirq_context;
1431 unsigned long softirq_disable_ip;
1432 unsigned long softirq_enable_ip;
1433 unsigned int softirq_disable_event;
1434 unsigned int softirq_enable_event;
1435 int softirqs_enabled;
1436 int softirq_context;
1438 #ifdef CONFIG_LOCKDEP
1439 # define MAX_LOCK_DEPTH 48UL
1442 unsigned int lockdep_recursion;
1443 struct held_lock held_locks[MAX_LOCK_DEPTH];
1444 gfp_t lockdep_reclaim_gfp;
1447 /* journalling filesystem info */
1450 /* stacked block device info */
1451 struct bio_list *bio_list;
1454 /* stack plugging */
1455 struct blk_plug *plug;
1459 struct reclaim_state *reclaim_state;
1461 struct backing_dev_info *backing_dev_info;
1463 struct io_context *io_context;
1465 unsigned long ptrace_message;
1466 siginfo_t *last_siginfo; /* For ptrace use. */
1467 struct task_io_accounting ioac;
1468 #if defined(CONFIG_TASK_XACCT)
1469 u64 acct_rss_mem1; /* accumulated rss usage */
1470 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1471 cputime_t acct_timexpd; /* stime + utime since last update */
1473 #ifdef CONFIG_CPUSETS
1474 nodemask_t mems_allowed; /* Protected by alloc_lock */
1475 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1476 int cpuset_mem_spread_rotor;
1477 int cpuset_slab_spread_rotor;
1479 #ifdef CONFIG_CGROUPS
1480 /* Control Group info protected by css_set_lock */
1481 struct css_set __rcu *cgroups;
1482 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1483 struct list_head cg_list;
1486 struct robust_list_head __user *robust_list;
1487 #ifdef CONFIG_COMPAT
1488 struct compat_robust_list_head __user *compat_robust_list;
1490 struct list_head pi_state_list;
1491 struct futex_pi_state *pi_state_cache;
1493 #ifdef CONFIG_PERF_EVENTS
1494 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1495 struct mutex perf_event_mutex;
1496 struct list_head perf_event_list;
1499 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1501 short pref_node_fork;
1503 #ifdef CONFIG_NUMA_BALANCING
1505 int numa_migrate_seq;
1506 unsigned int numa_scan_period;
1507 u64 node_stamp; /* migration stamp */
1508 struct callback_head numa_work;
1509 #endif /* CONFIG_NUMA_BALANCING */
1511 struct rcu_head rcu;
1514 * cache last used pipe for splice
1516 struct pipe_inode_info *splice_pipe;
1518 struct page_frag task_frag;
1520 #ifdef CONFIG_TASK_DELAY_ACCT
1521 struct task_delay_info *delays;
1523 #ifdef CONFIG_FAULT_INJECTION
1527 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1528 * balance_dirty_pages() for some dirty throttling pause
1531 int nr_dirtied_pause;
1532 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1534 #ifdef CONFIG_LATENCYTOP
1535 int latency_record_count;
1536 struct latency_record latency_record[LT_SAVECOUNT];
1539 * time slack values; these are used to round up poll() and
1540 * select() etc timeout values. These are in nanoseconds.
1542 unsigned long timer_slack_ns;
1543 unsigned long default_timer_slack_ns;
1545 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1546 /* Index of current stored address in ret_stack */
1548 /* Stack of return addresses for return function tracing */
1549 struct ftrace_ret_stack *ret_stack;
1550 /* time stamp for last schedule */
1551 unsigned long long ftrace_timestamp;
1553 * Number of functions that haven't been traced
1554 * because of depth overrun.
1556 atomic_t trace_overrun;
1557 /* Pause for the tracing */
1558 atomic_t tracing_graph_pause;
1560 #ifdef CONFIG_TRACING
1561 /* state flags for use by tracers */
1562 unsigned long trace;
1563 /* bitmask and counter of trace recursion */
1564 unsigned long trace_recursion;
1565 #endif /* CONFIG_TRACING */
1566 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1567 struct memcg_batch_info {
1568 int do_batch; /* incremented when batch uncharge started */
1569 struct mem_cgroup *memcg; /* target memcg of uncharge */
1570 unsigned long nr_pages; /* uncharged usage */
1571 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1573 unsigned int memcg_kmem_skip_account;
1575 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1576 atomic_t ptrace_bp_refcnt;
1578 #ifdef CONFIG_UPROBES
1579 struct uprobe_task *utask;
1583 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1584 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1586 #ifdef CONFIG_NUMA_BALANCING
1587 extern void task_numa_fault(int node, int pages, bool migrated);
1588 extern void set_numabalancing_state(bool enabled);
1590 static inline void task_numa_fault(int node, int pages, bool migrated)
1593 static inline void set_numabalancing_state(bool enabled)
1598 static inline struct pid *task_pid(struct task_struct *task)
1600 return task->pids[PIDTYPE_PID].pid;
1603 static inline struct pid *task_tgid(struct task_struct *task)
1605 return task->group_leader->pids[PIDTYPE_PID].pid;
1609 * Without tasklist or rcu lock it is not safe to dereference
1610 * the result of task_pgrp/task_session even if task == current,
1611 * we can race with another thread doing sys_setsid/sys_setpgid.
1613 static inline struct pid *task_pgrp(struct task_struct *task)
1615 return task->group_leader->pids[PIDTYPE_PGID].pid;
1618 static inline struct pid *task_session(struct task_struct *task)
1620 return task->group_leader->pids[PIDTYPE_SID].pid;
1623 struct pid_namespace;
1626 * the helpers to get the task's different pids as they are seen
1627 * from various namespaces
1629 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1630 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1632 * task_xid_nr_ns() : id seen from the ns specified;
1634 * set_task_vxid() : assigns a virtual id to a task;
1636 * see also pid_nr() etc in include/linux/pid.h
1638 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1639 struct pid_namespace *ns);
1641 static inline pid_t task_pid_nr(struct task_struct *tsk)
1646 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1647 struct pid_namespace *ns)
1649 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1652 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1654 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1658 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1663 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1665 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1667 return pid_vnr(task_tgid(tsk));
1671 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1672 struct pid_namespace *ns)
1674 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1677 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1679 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1683 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1684 struct pid_namespace *ns)
1686 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1689 static inline pid_t task_session_vnr(struct task_struct *tsk)
1691 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1694 /* obsolete, do not use */
1695 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1697 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1701 * pid_alive - check that a task structure is not stale
1702 * @p: Task structure to be checked.
1704 * Test if a process is not yet dead (at most zombie state)
1705 * If pid_alive fails, then pointers within the task structure
1706 * can be stale and must not be dereferenced.
1708 static inline int pid_alive(struct task_struct *p)
1710 return p->pids[PIDTYPE_PID].pid != NULL;
1714 * is_global_init - check if a task structure is init
1715 * @tsk: Task structure to be checked.
1717 * Check if a task structure is the first user space task the kernel created.
1719 static inline int is_global_init(struct task_struct *tsk)
1721 return tsk->pid == 1;
1724 extern struct pid *cad_pid;
1726 extern void free_task(struct task_struct *tsk);
1727 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1729 extern void __put_task_struct(struct task_struct *t);
1731 static inline void put_task_struct(struct task_struct *t)
1733 if (atomic_dec_and_test(&t->usage))
1734 __put_task_struct(t);
1737 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1738 extern void task_cputime(struct task_struct *t,
1739 cputime_t *utime, cputime_t *stime);
1740 extern void task_cputime_scaled(struct task_struct *t,
1741 cputime_t *utimescaled, cputime_t *stimescaled);
1742 extern cputime_t task_gtime(struct task_struct *t);
1744 static inline void task_cputime(struct task_struct *t,
1745 cputime_t *utime, cputime_t *stime)
1753 static inline void task_cputime_scaled(struct task_struct *t,
1754 cputime_t *utimescaled,
1755 cputime_t *stimescaled)
1758 *utimescaled = t->utimescaled;
1760 *stimescaled = t->stimescaled;
1763 static inline cputime_t task_gtime(struct task_struct *t)
1768 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1769 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1774 #define PF_EXITING 0x00000004 /* getting shut down */
1775 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1776 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1777 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1778 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1779 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1780 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1781 #define PF_DUMPCORE 0x00000200 /* dumped core */
1782 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1783 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1784 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1785 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1786 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1787 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1788 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1789 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1790 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1791 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1792 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1793 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1794 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1795 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1796 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1797 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1798 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1799 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1800 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1801 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1802 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1805 * Only the _current_ task can read/write to tsk->flags, but other
1806 * tasks can access tsk->flags in readonly mode for example
1807 * with tsk_used_math (like during threaded core dumping).
1808 * There is however an exception to this rule during ptrace
1809 * or during fork: the ptracer task is allowed to write to the
1810 * child->flags of its traced child (same goes for fork, the parent
1811 * can write to the child->flags), because we're guaranteed the
1812 * child is not running and in turn not changing child->flags
1813 * at the same time the parent does it.
1815 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1816 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1817 #define clear_used_math() clear_stopped_child_used_math(current)
1818 #define set_used_math() set_stopped_child_used_math(current)
1819 #define conditional_stopped_child_used_math(condition, child) \
1820 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1821 #define conditional_used_math(condition) \
1822 conditional_stopped_child_used_math(condition, current)
1823 #define copy_to_stopped_child_used_math(child) \
1824 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1825 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1826 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1827 #define used_math() tsk_used_math(current)
1829 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1830 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1832 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1837 static inline unsigned int memalloc_noio_save(void)
1839 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1840 current->flags |= PF_MEMALLOC_NOIO;
1844 static inline void memalloc_noio_restore(unsigned int flags)
1846 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1850 * task->jobctl flags
1852 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1854 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1855 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1856 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1857 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1858 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1859 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1860 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1862 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1863 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1864 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1865 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1866 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1867 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1868 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1870 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1871 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1873 extern bool task_set_jobctl_pending(struct task_struct *task,
1875 extern void task_clear_jobctl_trapping(struct task_struct *task);
1876 extern void task_clear_jobctl_pending(struct task_struct *task,
1879 #ifdef CONFIG_PREEMPT_RCU
1881 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1882 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1884 static inline void rcu_copy_process(struct task_struct *p)
1886 p->rcu_read_lock_nesting = 0;
1887 p->rcu_read_unlock_special = 0;
1888 #ifdef CONFIG_TREE_PREEMPT_RCU
1889 p->rcu_blocked_node = NULL;
1890 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1891 #ifdef CONFIG_RCU_BOOST
1892 p->rcu_boost_mutex = NULL;
1893 #endif /* #ifdef CONFIG_RCU_BOOST */
1894 INIT_LIST_HEAD(&p->rcu_node_entry);
1899 static inline void rcu_copy_process(struct task_struct *p)
1905 static inline void tsk_restore_flags(struct task_struct *task,
1906 unsigned long orig_flags, unsigned long flags)
1908 task->flags &= ~flags;
1909 task->flags |= orig_flags & flags;
1913 extern void do_set_cpus_allowed(struct task_struct *p,
1914 const struct cpumask *new_mask);
1916 extern int set_cpus_allowed_ptr(struct task_struct *p,
1917 const struct cpumask *new_mask);
1919 static inline void do_set_cpus_allowed(struct task_struct *p,
1920 const struct cpumask *new_mask)
1923 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1924 const struct cpumask *new_mask)
1926 if (!cpumask_test_cpu(0, new_mask))
1933 void calc_load_enter_idle(void);
1934 void calc_load_exit_idle(void);
1936 static inline void calc_load_enter_idle(void) { }
1937 static inline void calc_load_exit_idle(void) { }
1938 #endif /* CONFIG_NO_HZ */
1940 #ifndef CONFIG_CPUMASK_OFFSTACK
1941 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1943 return set_cpus_allowed_ptr(p, &new_mask);
1948 * Do not use outside of architecture code which knows its limitations.
1950 * sched_clock() has no promise of monotonicity or bounded drift between
1951 * CPUs, use (which you should not) requires disabling IRQs.
1953 * Please use one of the three interfaces below.
1955 extern unsigned long long notrace sched_clock(void);
1957 * See the comment in kernel/sched/clock.c
1959 extern u64 cpu_clock(int cpu);
1960 extern u64 local_clock(void);
1961 extern u64 sched_clock_cpu(int cpu);
1964 extern void sched_clock_init(void);
1966 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1967 static inline void sched_clock_tick(void)
1971 static inline void sched_clock_idle_sleep_event(void)
1975 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1980 * Architectures can set this to 1 if they have specified
1981 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1982 * but then during bootup it turns out that sched_clock()
1983 * is reliable after all:
1985 extern int sched_clock_stable;
1987 extern void sched_clock_tick(void);
1988 extern void sched_clock_idle_sleep_event(void);
1989 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1992 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1994 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1995 * The reason for this explicit opt-in is not to have perf penalty with
1996 * slow sched_clocks.
1998 extern void enable_sched_clock_irqtime(void);
1999 extern void disable_sched_clock_irqtime(void);
2001 static inline void enable_sched_clock_irqtime(void) {}
2002 static inline void disable_sched_clock_irqtime(void) {}
2005 extern unsigned long long
2006 task_sched_runtime(struct task_struct *task);
2008 /* sched_exec is called by processes performing an exec */
2010 extern void sched_exec(void);
2012 #define sched_exec() {}
2015 extern void sched_clock_idle_sleep_event(void);
2016 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2018 #ifdef CONFIG_HOTPLUG_CPU
2019 extern void idle_task_exit(void);
2021 static inline void idle_task_exit(void) {}
2024 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2025 extern void wake_up_idle_cpu(int cpu);
2027 static inline void wake_up_idle_cpu(int cpu) { }
2030 #ifdef CONFIG_SCHED_AUTOGROUP
2031 extern void sched_autogroup_create_attach(struct task_struct *p);
2032 extern void sched_autogroup_detach(struct task_struct *p);
2033 extern void sched_autogroup_fork(struct signal_struct *sig);
2034 extern void sched_autogroup_exit(struct signal_struct *sig);
2035 #ifdef CONFIG_PROC_FS
2036 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2037 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2040 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2041 static inline void sched_autogroup_detach(struct task_struct *p) { }
2042 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2043 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2046 extern bool yield_to(struct task_struct *p, bool preempt);
2047 extern void set_user_nice(struct task_struct *p, long nice);
2048 extern int task_prio(const struct task_struct *p);
2049 extern int task_nice(const struct task_struct *p);
2050 extern int can_nice(const struct task_struct *p, const int nice);
2051 extern int task_curr(const struct task_struct *p);
2052 extern int idle_cpu(int cpu);
2053 extern int sched_setscheduler(struct task_struct *, int,
2054 const struct sched_param *);
2055 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2056 const struct sched_param *);
2057 extern struct task_struct *idle_task(int cpu);
2059 * is_idle_task - is the specified task an idle task?
2060 * @p: the task in question.
2062 static inline bool is_idle_task(const struct task_struct *p)
2066 extern struct task_struct *curr_task(int cpu);
2067 extern void set_curr_task(int cpu, struct task_struct *p);
2072 * The default (Linux) execution domain.
2074 extern struct exec_domain default_exec_domain;
2076 union thread_union {
2077 struct thread_info thread_info;
2078 unsigned long stack[THREAD_SIZE/sizeof(long)];
2081 #ifndef __HAVE_ARCH_KSTACK_END
2082 static inline int kstack_end(void *addr)
2084 /* Reliable end of stack detection:
2085 * Some APM bios versions misalign the stack
2087 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2091 extern union thread_union init_thread_union;
2092 extern struct task_struct init_task;
2094 extern struct mm_struct init_mm;
2096 extern struct pid_namespace init_pid_ns;
2099 * find a task by one of its numerical ids
2101 * find_task_by_pid_ns():
2102 * finds a task by its pid in the specified namespace
2103 * find_task_by_vpid():
2104 * finds a task by its virtual pid
2106 * see also find_vpid() etc in include/linux/pid.h
2109 extern struct task_struct *find_task_by_vpid(pid_t nr);
2110 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2111 struct pid_namespace *ns);
2113 extern void __set_special_pids(struct pid *pid);
2115 /* per-UID process charging. */
2116 extern struct user_struct * alloc_uid(kuid_t);
2117 static inline struct user_struct *get_uid(struct user_struct *u)
2119 atomic_inc(&u->__count);
2122 extern void free_uid(struct user_struct *);
2124 #include <asm/current.h>
2126 extern void xtime_update(unsigned long ticks);
2128 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2129 extern int wake_up_process(struct task_struct *tsk);
2130 extern void wake_up_new_task(struct task_struct *tsk);
2132 extern void kick_process(struct task_struct *tsk);
2134 static inline void kick_process(struct task_struct *tsk) { }
2136 extern void sched_fork(struct task_struct *p);
2137 extern void sched_dead(struct task_struct *p);
2139 extern void proc_caches_init(void);
2140 extern void flush_signals(struct task_struct *);
2141 extern void __flush_signals(struct task_struct *);
2142 extern void ignore_signals(struct task_struct *);
2143 extern void flush_signal_handlers(struct task_struct *, int force_default);
2144 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2146 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2148 unsigned long flags;
2151 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2152 ret = dequeue_signal(tsk, mask, info);
2153 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2158 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2160 extern void unblock_all_signals(void);
2161 extern void release_task(struct task_struct * p);
2162 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2163 extern int force_sigsegv(int, struct task_struct *);
2164 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2165 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2166 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2167 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2168 const struct cred *, u32);
2169 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2170 extern int kill_pid(struct pid *pid, int sig, int priv);
2171 extern int kill_proc_info(int, struct siginfo *, pid_t);
2172 extern __must_check bool do_notify_parent(struct task_struct *, int);
2173 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2174 extern void force_sig(int, struct task_struct *);
2175 extern int send_sig(int, struct task_struct *, int);
2176 extern int zap_other_threads(struct task_struct *p);
2177 extern struct sigqueue *sigqueue_alloc(void);
2178 extern void sigqueue_free(struct sigqueue *);
2179 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2180 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2182 static inline void restore_saved_sigmask(void)
2184 if (test_and_clear_restore_sigmask())
2185 __set_current_blocked(¤t->saved_sigmask);
2188 static inline sigset_t *sigmask_to_save(void)
2190 sigset_t *res = ¤t->blocked;
2191 if (unlikely(test_restore_sigmask()))
2192 res = ¤t->saved_sigmask;
2196 static inline int kill_cad_pid(int sig, int priv)
2198 return kill_pid(cad_pid, sig, priv);
2201 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2202 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2203 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2204 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2207 * True if we are on the alternate signal stack.
2209 static inline int on_sig_stack(unsigned long sp)
2211 #ifdef CONFIG_STACK_GROWSUP
2212 return sp >= current->sas_ss_sp &&
2213 sp - current->sas_ss_sp < current->sas_ss_size;
2215 return sp > current->sas_ss_sp &&
2216 sp - current->sas_ss_sp <= current->sas_ss_size;
2220 static inline int sas_ss_flags(unsigned long sp)
2222 return (current->sas_ss_size == 0 ? SS_DISABLE
2223 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2226 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2228 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2229 #ifdef CONFIG_STACK_GROWSUP
2230 return current->sas_ss_sp;
2232 return current->sas_ss_sp + current->sas_ss_size;
2238 * Routines for handling mm_structs
2240 extern struct mm_struct * mm_alloc(void);
2242 /* mmdrop drops the mm and the page tables */
2243 extern void __mmdrop(struct mm_struct *);
2244 static inline void mmdrop(struct mm_struct * mm)
2246 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2250 /* mmput gets rid of the mappings and all user-space */
2251 extern void mmput(struct mm_struct *);
2252 /* Grab a reference to a task's mm, if it is not already going away */
2253 extern struct mm_struct *get_task_mm(struct task_struct *task);
2255 * Grab a reference to a task's mm, if it is not already going away
2256 * and ptrace_may_access with the mode parameter passed to it
2259 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2260 /* Remove the current tasks stale references to the old mm_struct */
2261 extern void mm_release(struct task_struct *, struct mm_struct *);
2262 /* Allocate a new mm structure and copy contents from tsk->mm */
2263 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2265 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2266 struct task_struct *);
2267 extern void flush_thread(void);
2268 extern void exit_thread(void);
2270 extern void exit_files(struct task_struct *);
2271 extern void __cleanup_sighand(struct sighand_struct *);
2273 extern void exit_itimers(struct signal_struct *);
2274 extern void flush_itimer_signals(void);
2276 extern void do_group_exit(int);
2278 extern int allow_signal(int);
2279 extern int disallow_signal(int);
2281 extern int do_execve(const char *,
2282 const char __user * const __user *,
2283 const char __user * const __user *);
2284 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2285 struct task_struct *fork_idle(int);
2286 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2288 extern void set_task_comm(struct task_struct *tsk, char *from);
2289 extern char *get_task_comm(char *to, struct task_struct *tsk);
2292 void scheduler_ipi(void);
2293 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2295 static inline void scheduler_ipi(void) { }
2296 static inline unsigned long wait_task_inactive(struct task_struct *p,
2303 #define next_task(p) \
2304 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2306 #define for_each_process(p) \
2307 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2309 extern bool current_is_single_threaded(void);
2312 * Careful: do_each_thread/while_each_thread is a double loop so
2313 * 'break' will not work as expected - use goto instead.
2315 #define do_each_thread(g, t) \
2316 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2318 #define while_each_thread(g, t) \
2319 while ((t = next_thread(t)) != g)
2321 static inline int get_nr_threads(struct task_struct *tsk)
2323 return tsk->signal->nr_threads;
2326 static inline bool thread_group_leader(struct task_struct *p)
2328 return p->exit_signal >= 0;
2331 /* Do to the insanities of de_thread it is possible for a process
2332 * to have the pid of the thread group leader without actually being
2333 * the thread group leader. For iteration through the pids in proc
2334 * all we care about is that we have a task with the appropriate
2335 * pid, we don't actually care if we have the right task.
2337 static inline int has_group_leader_pid(struct task_struct *p)
2339 return p->pid == p->tgid;
2343 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2345 return p1->tgid == p2->tgid;
2348 static inline struct task_struct *next_thread(const struct task_struct *p)
2350 return list_entry_rcu(p->thread_group.next,
2351 struct task_struct, thread_group);
2354 static inline int thread_group_empty(struct task_struct *p)
2356 return list_empty(&p->thread_group);
2359 #define delay_group_leader(p) \
2360 (thread_group_leader(p) && !thread_group_empty(p))
2363 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2364 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2365 * pins the final release of task.io_context. Also protects ->cpuset and
2366 * ->cgroup.subsys[]. And ->vfork_done.
2368 * Nests both inside and outside of read_lock(&tasklist_lock).
2369 * It must not be nested with write_lock_irq(&tasklist_lock),
2370 * neither inside nor outside.
2372 static inline void task_lock(struct task_struct *p)
2374 spin_lock(&p->alloc_lock);
2377 static inline void task_unlock(struct task_struct *p)
2379 spin_unlock(&p->alloc_lock);
2382 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2383 unsigned long *flags);
2385 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2386 unsigned long *flags)
2388 struct sighand_struct *ret;
2390 ret = __lock_task_sighand(tsk, flags);
2391 (void)__cond_lock(&tsk->sighand->siglock, ret);
2395 static inline void unlock_task_sighand(struct task_struct *tsk,
2396 unsigned long *flags)
2398 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2401 #ifdef CONFIG_CGROUPS
2402 static inline void threadgroup_change_begin(struct task_struct *tsk)
2404 down_read(&tsk->signal->group_rwsem);
2406 static inline void threadgroup_change_end(struct task_struct *tsk)
2408 up_read(&tsk->signal->group_rwsem);
2412 * threadgroup_lock - lock threadgroup
2413 * @tsk: member task of the threadgroup to lock
2415 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2416 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2417 * perform exec. This is useful for cases where the threadgroup needs to
2418 * stay stable across blockable operations.
2420 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2421 * synchronization. While held, no new task will be added to threadgroup
2422 * and no existing live task will have its PF_EXITING set.
2424 * During exec, a task goes and puts its thread group through unusual
2425 * changes. After de-threading, exclusive access is assumed to resources
2426 * which are usually shared by tasks in the same group - e.g. sighand may
2427 * be replaced with a new one. Also, the exec'ing task takes over group
2428 * leader role including its pid. Exclude these changes while locked by
2429 * grabbing cred_guard_mutex which is used to synchronize exec path.
2431 static inline void threadgroup_lock(struct task_struct *tsk)
2434 * exec uses exit for de-threading nesting group_rwsem inside
2435 * cred_guard_mutex. Grab cred_guard_mutex first.
2437 mutex_lock(&tsk->signal->cred_guard_mutex);
2438 down_write(&tsk->signal->group_rwsem);
2442 * threadgroup_unlock - unlock threadgroup
2443 * @tsk: member task of the threadgroup to unlock
2445 * Reverse threadgroup_lock().
2447 static inline void threadgroup_unlock(struct task_struct *tsk)
2449 up_write(&tsk->signal->group_rwsem);
2450 mutex_unlock(&tsk->signal->cred_guard_mutex);
2453 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2454 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2455 static inline void threadgroup_lock(struct task_struct *tsk) {}
2456 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2459 #ifndef __HAVE_THREAD_FUNCTIONS
2461 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2462 #define task_stack_page(task) ((task)->stack)
2464 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2466 *task_thread_info(p) = *task_thread_info(org);
2467 task_thread_info(p)->task = p;
2470 static inline unsigned long *end_of_stack(struct task_struct *p)
2472 return (unsigned long *)(task_thread_info(p) + 1);
2477 static inline int object_is_on_stack(void *obj)
2479 void *stack = task_stack_page(current);
2481 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2484 extern void thread_info_cache_init(void);
2486 #ifdef CONFIG_DEBUG_STACK_USAGE
2487 static inline unsigned long stack_not_used(struct task_struct *p)
2489 unsigned long *n = end_of_stack(p);
2491 do { /* Skip over canary */
2495 return (unsigned long)n - (unsigned long)end_of_stack(p);
2499 /* set thread flags in other task's structures
2500 * - see asm/thread_info.h for TIF_xxxx flags available
2502 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2504 set_ti_thread_flag(task_thread_info(tsk), flag);
2507 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2509 clear_ti_thread_flag(task_thread_info(tsk), flag);
2512 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2514 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2517 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2519 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2522 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2524 return test_ti_thread_flag(task_thread_info(tsk), flag);
2527 static inline void set_tsk_need_resched(struct task_struct *tsk)
2529 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2532 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2534 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2537 static inline int test_tsk_need_resched(struct task_struct *tsk)
2539 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2542 static inline int restart_syscall(void)
2544 set_tsk_thread_flag(current, TIF_SIGPENDING);
2545 return -ERESTARTNOINTR;
2548 static inline int signal_pending(struct task_struct *p)
2550 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2553 static inline int __fatal_signal_pending(struct task_struct *p)
2555 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2558 static inline int fatal_signal_pending(struct task_struct *p)
2560 return signal_pending(p) && __fatal_signal_pending(p);
2563 static inline int signal_pending_state(long state, struct task_struct *p)
2565 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2567 if (!signal_pending(p))
2570 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2573 static inline int need_resched(void)
2575 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2579 * cond_resched() and cond_resched_lock(): latency reduction via
2580 * explicit rescheduling in places that are safe. The return
2581 * value indicates whether a reschedule was done in fact.
2582 * cond_resched_lock() will drop the spinlock before scheduling,
2583 * cond_resched_softirq() will enable bhs before scheduling.
2585 extern int _cond_resched(void);
2587 #define cond_resched() ({ \
2588 __might_sleep(__FILE__, __LINE__, 0); \
2592 extern int __cond_resched_lock(spinlock_t *lock);
2594 #ifdef CONFIG_PREEMPT_COUNT
2595 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2597 #define PREEMPT_LOCK_OFFSET 0
2600 #define cond_resched_lock(lock) ({ \
2601 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2602 __cond_resched_lock(lock); \
2605 extern int __cond_resched_softirq(void);
2607 #define cond_resched_softirq() ({ \
2608 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2609 __cond_resched_softirq(); \
2613 * Does a critical section need to be broken due to another
2614 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2615 * but a general need for low latency)
2617 static inline int spin_needbreak(spinlock_t *lock)
2619 #ifdef CONFIG_PREEMPT
2620 return spin_is_contended(lock);
2627 * Thread group CPU time accounting.
2629 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2630 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2632 static inline void thread_group_cputime_init(struct signal_struct *sig)
2634 raw_spin_lock_init(&sig->cputimer.lock);
2638 * Reevaluate whether the task has signals pending delivery.
2639 * Wake the task if so.
2640 * This is required every time the blocked sigset_t changes.
2641 * callers must hold sighand->siglock.
2643 extern void recalc_sigpending_and_wake(struct task_struct *t);
2644 extern void recalc_sigpending(void);
2646 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2648 static inline void signal_wake_up(struct task_struct *t, bool resume)
2650 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2652 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2654 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2658 * Wrappers for p->thread_info->cpu access. No-op on UP.
2662 static inline unsigned int task_cpu(const struct task_struct *p)
2664 return task_thread_info(p)->cpu;
2667 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2671 static inline unsigned int task_cpu(const struct task_struct *p)
2676 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2680 #endif /* CONFIG_SMP */
2682 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2683 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2685 #ifdef CONFIG_CGROUP_SCHED
2687 extern struct task_group root_task_group;
2689 extern struct task_group *sched_create_group(struct task_group *parent);
2690 extern void sched_online_group(struct task_group *tg,
2691 struct task_group *parent);
2692 extern void sched_destroy_group(struct task_group *tg);
2693 extern void sched_offline_group(struct task_group *tg);
2694 extern void sched_move_task(struct task_struct *tsk);
2695 #ifdef CONFIG_FAIR_GROUP_SCHED
2696 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2697 extern unsigned long sched_group_shares(struct task_group *tg);
2699 #ifdef CONFIG_RT_GROUP_SCHED
2700 extern int sched_group_set_rt_runtime(struct task_group *tg,
2701 long rt_runtime_us);
2702 extern long sched_group_rt_runtime(struct task_group *tg);
2703 extern int sched_group_set_rt_period(struct task_group *tg,
2705 extern long sched_group_rt_period(struct task_group *tg);
2706 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2708 #endif /* CONFIG_CGROUP_SCHED */
2710 extern int task_can_switch_user(struct user_struct *up,
2711 struct task_struct *tsk);
2713 #ifdef CONFIG_TASK_XACCT
2714 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2716 tsk->ioac.rchar += amt;
2719 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2721 tsk->ioac.wchar += amt;
2724 static inline void inc_syscr(struct task_struct *tsk)
2729 static inline void inc_syscw(struct task_struct *tsk)
2734 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2738 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2742 static inline void inc_syscr(struct task_struct *tsk)
2746 static inline void inc_syscw(struct task_struct *tsk)
2751 #ifndef TASK_SIZE_OF
2752 #define TASK_SIZE_OF(tsk) TASK_SIZE
2755 #ifdef CONFIG_MM_OWNER
2756 extern void mm_update_next_owner(struct mm_struct *mm);
2757 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2759 static inline void mm_update_next_owner(struct mm_struct *mm)
2763 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2766 #endif /* CONFIG_MM_OWNER */
2768 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2771 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2774 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2777 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2780 static inline unsigned long rlimit(unsigned int limit)
2782 return task_rlimit(current, limit);
2785 static inline unsigned long rlimit_max(unsigned int limit)
2787 return task_rlimit_max(current, limit);