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 #ifdef CONFIG_CPUQUIET_FRAMEWORK
104 extern u64 nr_running_integral(unsigned int cpu);
106 extern unsigned long nr_iowait_cpu(int cpu);
107 extern unsigned long this_cpu_load(void);
110 extern void calc_global_load(unsigned long ticks);
111 extern void update_cpu_load_nohz(void);
113 /* Notifier for when a task gets migrated to a new CPU */
114 struct task_migration_notifier {
115 struct task_struct *task;
119 extern void register_task_migration_notifier(struct notifier_block *n);
121 extern unsigned long get_parent_ip(unsigned long addr);
123 extern void dump_cpu_task(int cpu);
128 #ifdef CONFIG_SCHED_DEBUG
129 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
130 extern void proc_sched_set_task(struct task_struct *p);
132 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
136 * Task state bitmask. NOTE! These bits are also
137 * encoded in fs/proc/array.c: get_task_state().
139 * We have two separate sets of flags: task->state
140 * is about runnability, while task->exit_state are
141 * about the task exiting. Confusing, but this way
142 * modifying one set can't modify the other one by
145 #define TASK_RUNNING 0
146 #define TASK_INTERRUPTIBLE 1
147 #define TASK_UNINTERRUPTIBLE 2
148 #define __TASK_STOPPED 4
149 #define __TASK_TRACED 8
150 /* in tsk->exit_state */
151 #define EXIT_ZOMBIE 16
153 /* in tsk->state again */
155 #define TASK_WAKEKILL 128
156 #define TASK_WAKING 256
157 #define TASK_PARKED 512
158 #define TASK_STATE_MAX 1024
160 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
162 extern char ___assert_task_state[1 - 2*!!(
163 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
165 /* Convenience macros for the sake of set_task_state */
166 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
167 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
168 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
170 /* Convenience macros for the sake of wake_up */
171 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
172 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
174 /* get_task_state() */
175 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
176 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
179 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
180 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
181 #define task_is_dead(task) ((task)->exit_state != 0)
182 #define task_is_stopped_or_traced(task) \
183 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
184 #define task_contributes_to_load(task) \
185 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
186 (task->flags & PF_FROZEN) == 0)
188 #define __set_task_state(tsk, state_value) \
189 do { (tsk)->state = (state_value); } while (0)
190 #define set_task_state(tsk, state_value) \
191 set_mb((tsk)->state, (state_value))
194 * set_current_state() includes a barrier so that the write of current->state
195 * is correctly serialised wrt the caller's subsequent test of whether to
198 * set_current_state(TASK_UNINTERRUPTIBLE);
199 * if (do_i_need_to_sleep())
202 * If the caller does not need such serialisation then use __set_current_state()
204 #define __set_current_state(state_value) \
205 do { current->state = (state_value); } while (0)
206 #define set_current_state(state_value) \
207 set_mb(current->state, (state_value))
209 /* Task command name length */
210 #define TASK_COMM_LEN 16
212 #include <linux/spinlock.h>
215 * This serializes "schedule()" and also protects
216 * the run-queue from deletions/modifications (but
217 * _adding_ to the beginning of the run-queue has
220 extern rwlock_t tasklist_lock;
221 extern spinlock_t mmlist_lock;
225 #ifdef CONFIG_PROVE_RCU
226 extern int lockdep_tasklist_lock_is_held(void);
227 #endif /* #ifdef CONFIG_PROVE_RCU */
229 extern void sched_init(void);
230 extern void sched_init_smp(void);
231 extern asmlinkage void schedule_tail(struct task_struct *prev);
232 extern void init_idle(struct task_struct *idle, int cpu);
233 extern void init_idle_bootup_task(struct task_struct *idle);
235 extern int runqueue_is_locked(int cpu);
237 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
238 extern void nohz_balance_enter_idle(int cpu);
239 extern void set_cpu_sd_state_idle(void);
240 extern int get_nohz_timer_target(void);
242 static inline void nohz_balance_enter_idle(int cpu) { }
243 static inline void set_cpu_sd_state_idle(void) { }
247 * Only dump TASK_* tasks. (0 for all tasks)
249 extern void show_state_filter(unsigned long state_filter);
251 static inline void show_state(void)
253 show_state_filter(0);
256 extern void show_regs(struct pt_regs *);
259 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
260 * task), SP is the stack pointer of the first frame that should be shown in the back
261 * trace (or NULL if the entire call-chain of the task should be shown).
263 extern void show_stack(struct task_struct *task, unsigned long *sp);
265 void io_schedule(void);
266 long io_schedule_timeout(long timeout);
268 extern void cpu_init (void);
269 extern void trap_init(void);
270 extern void update_process_times(int user);
271 extern void scheduler_tick(void);
273 extern void sched_show_task(struct task_struct *p);
275 #ifdef CONFIG_LOCKUP_DETECTOR
276 extern void touch_softlockup_watchdog(void);
277 extern void touch_softlockup_watchdog_sync(void);
278 extern void touch_all_softlockup_watchdogs(void);
279 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
281 size_t *lenp, loff_t *ppos);
282 extern unsigned int softlockup_panic;
283 void lockup_detector_init(void);
285 static inline void touch_softlockup_watchdog(void)
288 static inline void touch_softlockup_watchdog_sync(void)
291 static inline void touch_all_softlockup_watchdogs(void)
294 static inline void lockup_detector_init(void)
299 /* Attach to any functions which should be ignored in wchan output. */
300 #define __sched __attribute__((__section__(".sched.text")))
302 /* Linker adds these: start and end of __sched functions */
303 extern char __sched_text_start[], __sched_text_end[];
305 /* Is this address in the __sched functions? */
306 extern int in_sched_functions(unsigned long addr);
308 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
309 extern signed long schedule_timeout(signed long timeout);
310 extern signed long schedule_timeout_interruptible(signed long timeout);
311 extern signed long schedule_timeout_killable(signed long timeout);
312 extern signed long schedule_timeout_uninterruptible(signed long timeout);
313 asmlinkage void schedule(void);
314 extern void schedule_preempt_disabled(void);
317 struct user_namespace;
320 extern void arch_pick_mmap_layout(struct mm_struct *mm);
322 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
323 unsigned long, unsigned long);
325 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
326 unsigned long len, unsigned long pgoff,
327 unsigned long flags);
328 extern void arch_unmap_area(struct mm_struct *, unsigned long);
329 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
331 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
335 extern void set_dumpable(struct mm_struct *mm, int value);
336 extern int get_dumpable(struct mm_struct *mm);
338 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
339 #define SUID_DUMP_USER 1 /* Dump as user of process */
340 #define SUID_DUMP_ROOT 2 /* Dump as root */
344 #define MMF_DUMPABLE 0 /* core dump is permitted */
345 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
347 #define MMF_DUMPABLE_BITS 2
348 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
350 /* coredump filter bits */
351 #define MMF_DUMP_ANON_PRIVATE 2
352 #define MMF_DUMP_ANON_SHARED 3
353 #define MMF_DUMP_MAPPED_PRIVATE 4
354 #define MMF_DUMP_MAPPED_SHARED 5
355 #define MMF_DUMP_ELF_HEADERS 6
356 #define MMF_DUMP_HUGETLB_PRIVATE 7
357 #define MMF_DUMP_HUGETLB_SHARED 8
359 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
360 #define MMF_DUMP_FILTER_BITS 7
361 #define MMF_DUMP_FILTER_MASK \
362 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
363 #define MMF_DUMP_FILTER_DEFAULT \
364 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
365 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
367 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
368 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
370 # define MMF_DUMP_MASK_DEFAULT_ELF 0
372 /* leave room for more dump flags */
373 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
374 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
375 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
377 #define MMF_HAS_UPROBES 19 /* has uprobes */
378 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
380 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
382 struct sighand_struct {
384 struct k_sigaction action[_NSIG];
386 wait_queue_head_t signalfd_wqh;
389 struct pacct_struct {
392 unsigned long ac_mem;
393 cputime_t ac_utime, ac_stime;
394 unsigned long ac_minflt, ac_majflt;
405 * struct cputime - snaphsot of system and user cputime
406 * @utime: time spent in user mode
407 * @stime: time spent in system mode
409 * Gathers a generic snapshot of user and system time.
417 * struct task_cputime - collected CPU time counts
418 * @utime: time spent in user mode, in &cputime_t units
419 * @stime: time spent in kernel mode, in &cputime_t units
420 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
422 * This is an extension of struct cputime that includes the total runtime
423 * spent by the task from the scheduler point of view.
425 * As a result, this structure groups together three kinds of CPU time
426 * that are tracked for threads and thread groups. Most things considering
427 * CPU time want to group these counts together and treat all three
428 * of them in parallel.
430 struct task_cputime {
433 unsigned long long sum_exec_runtime;
435 /* Alternate field names when used to cache expirations. */
436 #define prof_exp stime
437 #define virt_exp utime
438 #define sched_exp sum_exec_runtime
440 #define INIT_CPUTIME \
441 (struct task_cputime) { \
444 .sum_exec_runtime = 0, \
448 * Disable preemption until the scheduler is running.
449 * Reset by start_kernel()->sched_init()->init_idle().
451 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
452 * before the scheduler is active -- see should_resched().
454 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
457 * struct thread_group_cputimer - thread group interval timer counts
458 * @cputime: thread group interval timers.
459 * @running: non-zero when there are timers running and
460 * @cputime receives updates.
461 * @lock: lock for fields in this struct.
463 * This structure contains the version of task_cputime, above, that is
464 * used for thread group CPU timer calculations.
466 struct thread_group_cputimer {
467 struct task_cputime cputime;
472 #include <linux/rwsem.h>
476 * NOTE! "signal_struct" does not have its own
477 * locking, because a shared signal_struct always
478 * implies a shared sighand_struct, so locking
479 * sighand_struct is always a proper superset of
480 * the locking of signal_struct.
482 struct signal_struct {
486 struct list_head thread_head;
488 wait_queue_head_t wait_chldexit; /* for wait4() */
490 /* current thread group signal load-balancing target: */
491 struct task_struct *curr_target;
493 /* shared signal handling: */
494 struct sigpending shared_pending;
496 /* thread group exit support */
499 * - notify group_exit_task when ->count is equal to notify_count
500 * - everyone except group_exit_task is stopped during signal delivery
501 * of fatal signals, group_exit_task processes the signal.
504 struct task_struct *group_exit_task;
506 /* thread group stop support, overloads group_exit_code too */
507 int group_stop_count;
508 unsigned int flags; /* see SIGNAL_* flags below */
511 * PR_SET_CHILD_SUBREAPER marks a process, like a service
512 * manager, to re-parent orphan (double-forking) child processes
513 * to this process instead of 'init'. The service manager is
514 * able to receive SIGCHLD signals and is able to investigate
515 * the process until it calls wait(). All children of this
516 * process will inherit a flag if they should look for a
517 * child_subreaper process at exit.
519 unsigned int is_child_subreaper:1;
520 unsigned int has_child_subreaper:1;
522 /* POSIX.1b Interval Timers */
524 struct list_head posix_timers;
526 /* ITIMER_REAL timer for the process */
527 struct hrtimer real_timer;
528 struct pid *leader_pid;
529 ktime_t it_real_incr;
532 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
533 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
534 * values are defined to 0 and 1 respectively
536 struct cpu_itimer it[2];
539 * Thread group totals for process CPU timers.
540 * See thread_group_cputimer(), et al, for details.
542 struct thread_group_cputimer cputimer;
544 /* Earliest-expiration cache. */
545 struct task_cputime cputime_expires;
547 struct list_head cpu_timers[3];
549 struct pid *tty_old_pgrp;
551 /* boolean value for session group leader */
554 struct tty_struct *tty; /* NULL if no tty */
556 #ifdef CONFIG_SCHED_AUTOGROUP
557 struct autogroup *autogroup;
560 * Cumulative resource counters for dead threads in the group,
561 * and for reaped dead child processes forked by this group.
562 * Live threads maintain their own counters and add to these
563 * in __exit_signal, except for the group leader.
565 cputime_t utime, stime, cutime, cstime;
568 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
569 struct cputime prev_cputime;
571 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
572 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
573 unsigned long inblock, oublock, cinblock, coublock;
574 unsigned long maxrss, cmaxrss;
575 struct task_io_accounting ioac;
578 * Cumulative ns of schedule CPU time fo dead threads in the
579 * group, not including a zombie group leader, (This only differs
580 * from jiffies_to_ns(utime + stime) if sched_clock uses something
581 * other than jiffies.)
583 unsigned long long sum_sched_runtime;
586 * We don't bother to synchronize most readers of this at all,
587 * because there is no reader checking a limit that actually needs
588 * to get both rlim_cur and rlim_max atomically, and either one
589 * alone is a single word that can safely be read normally.
590 * getrlimit/setrlimit use task_lock(current->group_leader) to
591 * protect this instead of the siglock, because they really
592 * have no need to disable irqs.
594 struct rlimit rlim[RLIM_NLIMITS];
596 #ifdef CONFIG_BSD_PROCESS_ACCT
597 struct pacct_struct pacct; /* per-process accounting information */
599 #ifdef CONFIG_TASKSTATS
600 struct taskstats *stats;
604 unsigned audit_tty_log_passwd;
605 struct tty_audit_buf *tty_audit_buf;
607 #ifdef CONFIG_CGROUPS
609 * group_rwsem prevents new tasks from entering the threadgroup and
610 * member tasks from exiting,a more specifically, setting of
611 * PF_EXITING. fork and exit paths are protected with this rwsem
612 * using threadgroup_change_begin/end(). Users which require
613 * threadgroup to remain stable should use threadgroup_[un]lock()
614 * which also takes care of exec path. Currently, cgroup is the
617 struct rw_semaphore group_rwsem;
620 oom_flags_t oom_flags;
621 short oom_score_adj; /* OOM kill score adjustment */
622 short oom_score_adj_min; /* OOM kill score adjustment min value.
623 * Only settable by CAP_SYS_RESOURCE. */
625 struct mutex cred_guard_mutex; /* guard against foreign influences on
626 * credential calculations
627 * (notably. ptrace) */
631 * Bits in flags field of signal_struct.
633 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
634 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
635 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
636 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
638 * Pending notifications to parent.
640 #define SIGNAL_CLD_STOPPED 0x00000010
641 #define SIGNAL_CLD_CONTINUED 0x00000020
642 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
644 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
646 /* If true, all threads except ->group_exit_task have pending SIGKILL */
647 static inline int signal_group_exit(const struct signal_struct *sig)
649 return (sig->flags & SIGNAL_GROUP_EXIT) ||
650 (sig->group_exit_task != NULL);
654 * Some day this will be a full-fledged user tracking system..
657 atomic_t __count; /* reference count */
658 atomic_t processes; /* How many processes does this user have? */
659 atomic_t files; /* How many open files does this user have? */
660 atomic_t sigpending; /* How many pending signals does this user have? */
661 #ifdef CONFIG_INOTIFY_USER
662 atomic_t inotify_watches; /* How many inotify watches does this user have? */
663 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
665 #ifdef CONFIG_FANOTIFY
666 atomic_t fanotify_listeners;
669 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
671 #ifdef CONFIG_POSIX_MQUEUE
672 /* protected by mq_lock */
673 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
675 unsigned long locked_shm; /* How many pages of mlocked shm ? */
678 struct key *uid_keyring; /* UID specific keyring */
679 struct key *session_keyring; /* UID's default session keyring */
682 /* Hash table maintenance information */
683 struct hlist_node uidhash_node;
686 #ifdef CONFIG_PERF_EVENTS
687 atomic_long_t locked_vm;
691 extern int uids_sysfs_init(void);
693 extern struct user_struct *find_user(kuid_t);
695 extern struct user_struct root_user;
696 #define INIT_USER (&root_user)
699 struct backing_dev_info;
700 struct reclaim_state;
702 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
704 /* cumulative counters */
705 unsigned long pcount; /* # of times run on this cpu */
706 unsigned long long run_delay; /* time spent waiting on a runqueue */
709 unsigned long long last_arrival,/* when we last ran on a cpu */
710 last_queued; /* when we were last queued to run */
712 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
714 #ifdef CONFIG_TASK_DELAY_ACCT
715 struct task_delay_info {
717 unsigned int flags; /* Private per-task flags */
719 /* For each stat XXX, add following, aligned appropriately
721 * struct timespec XXX_start, XXX_end;
725 * Atomicity of updates to XXX_delay, XXX_count protected by
726 * single lock above (split into XXX_lock if contention is an issue).
730 * XXX_count is incremented on every XXX operation, the delay
731 * associated with the operation is added to XXX_delay.
732 * XXX_delay contains the accumulated delay time in nanoseconds.
734 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
735 u64 blkio_delay; /* wait for sync block io completion */
736 u64 swapin_delay; /* wait for swapin block io completion */
737 u32 blkio_count; /* total count of the number of sync block */
738 /* io operations performed */
739 u32 swapin_count; /* total count of the number of swapin block */
740 /* io operations performed */
742 struct timespec freepages_start, freepages_end;
743 u64 freepages_delay; /* wait for memory reclaim */
744 u32 freepages_count; /* total count of memory reclaim */
746 #endif /* CONFIG_TASK_DELAY_ACCT */
748 static inline int sched_info_on(void)
750 #ifdef CONFIG_SCHEDSTATS
752 #elif defined(CONFIG_TASK_DELAY_ACCT)
753 extern int delayacct_on;
768 * Increase resolution of cpu_power calculations
770 #define SCHED_POWER_SHIFT 10
771 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
774 * sched-domains (multiprocessor balancing) declarations:
777 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
778 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
779 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
780 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
781 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
782 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
783 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
784 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
785 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
786 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
787 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
788 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
790 extern int __weak arch_sd_sibiling_asym_packing(void);
792 struct sched_domain_attr {
793 int relax_domain_level;
796 #define SD_ATTR_INIT (struct sched_domain_attr) { \
797 .relax_domain_level = -1, \
800 extern int sched_domain_level_max;
804 struct sched_domain {
805 /* These fields must be setup */
806 struct sched_domain *parent; /* top domain must be null terminated */
807 struct sched_domain *child; /* bottom domain must be null terminated */
808 struct sched_group *groups; /* the balancing groups of the domain */
809 unsigned long min_interval; /* Minimum balance interval ms */
810 unsigned long max_interval; /* Maximum balance interval ms */
811 unsigned int busy_factor; /* less balancing by factor if busy */
812 unsigned int imbalance_pct; /* No balance until over watermark */
813 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
814 unsigned int busy_idx;
815 unsigned int idle_idx;
816 unsigned int newidle_idx;
817 unsigned int wake_idx;
818 unsigned int forkexec_idx;
819 unsigned int smt_gain;
821 int nohz_idle; /* NOHZ IDLE status */
822 int flags; /* See SD_* */
825 /* Runtime fields. */
826 unsigned long last_balance; /* init to jiffies. units in jiffies */
827 unsigned int balance_interval; /* initialise to 1. units in ms. */
828 unsigned int nr_balance_failed; /* initialise to 0 */
832 #ifdef CONFIG_SCHEDSTATS
833 /* load_balance() stats */
834 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
835 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
836 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
837 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
838 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
839 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
840 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
841 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
843 /* Active load balancing */
844 unsigned int alb_count;
845 unsigned int alb_failed;
846 unsigned int alb_pushed;
848 /* SD_BALANCE_EXEC stats */
849 unsigned int sbe_count;
850 unsigned int sbe_balanced;
851 unsigned int sbe_pushed;
853 /* SD_BALANCE_FORK stats */
854 unsigned int sbf_count;
855 unsigned int sbf_balanced;
856 unsigned int sbf_pushed;
858 /* try_to_wake_up() stats */
859 unsigned int ttwu_wake_remote;
860 unsigned int ttwu_move_affine;
861 unsigned int ttwu_move_balance;
863 #ifdef CONFIG_SCHED_DEBUG
867 void *private; /* used during construction */
868 struct rcu_head rcu; /* used during destruction */
871 unsigned int span_weight;
873 * Span of all CPUs in this domain.
875 * NOTE: this field is variable length. (Allocated dynamically
876 * by attaching extra space to the end of the structure,
877 * depending on how many CPUs the kernel has booted up with)
879 unsigned long span[0];
882 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
884 return to_cpumask(sd->span);
887 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
888 struct sched_domain_attr *dattr_new);
890 /* Allocate an array of sched domains, for partition_sched_domains(). */
891 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
892 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
894 bool cpus_share_cache(int this_cpu, int that_cpu);
896 #ifdef CONFIG_SCHED_HMP
899 struct cpumask possible_cpus;
900 struct list_head hmp_domains;
902 #endif /* CONFIG_SCHED_HMP */
903 #else /* CONFIG_SMP */
905 struct sched_domain_attr;
908 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
909 struct sched_domain_attr *dattr_new)
913 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
918 #endif /* !CONFIG_SMP */
921 struct io_context; /* See blkdev.h */
924 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
925 extern void prefetch_stack(struct task_struct *t);
927 static inline void prefetch_stack(struct task_struct *t) { }
930 struct audit_context; /* See audit.c */
932 struct pipe_inode_info;
933 struct uts_namespace;
936 unsigned long weight, inv_weight;
941 * These sums represent an infinite geometric series and so are bound
942 * above by 1024/(1-y). Thus we only need a u32 to store them for for all
943 * choices of y < 1-2^(-32)*1024.
945 u32 runnable_avg_sum, runnable_avg_period;
946 u64 last_runnable_update;
948 unsigned long load_avg_contrib;
949 unsigned long load_avg_ratio;
950 #ifdef CONFIG_SCHED_HMP
951 u64 hmp_last_up_migration;
952 u64 hmp_last_down_migration;
957 #ifdef CONFIG_SCHED_HMP
959 * We want to avoid boosting any processes forked from init (PID 1)
960 * and kthreadd (assumed to be PID 2).
962 #define hmp_task_should_forkboost(task) ((task->parent && task->parent->pid > 2))
965 #ifdef CONFIG_SCHEDSTATS
966 struct sched_statistics {
976 s64 sum_sleep_runtime;
983 u64 nr_migrations_cold;
984 u64 nr_failed_migrations_affine;
985 u64 nr_failed_migrations_running;
986 u64 nr_failed_migrations_hot;
987 u64 nr_forced_migrations;
991 u64 nr_wakeups_migrate;
992 u64 nr_wakeups_local;
993 u64 nr_wakeups_remote;
994 u64 nr_wakeups_affine;
995 u64 nr_wakeups_affine_attempts;
996 u64 nr_wakeups_passive;
1001 struct sched_entity {
1002 struct load_weight load; /* for load-balancing */
1003 struct rb_node run_node;
1004 struct list_head group_node;
1008 u64 sum_exec_runtime;
1010 u64 prev_sum_exec_runtime;
1014 #ifdef CONFIG_SCHEDSTATS
1015 struct sched_statistics statistics;
1018 #ifdef CONFIG_FAIR_GROUP_SCHED
1019 struct sched_entity *parent;
1020 /* rq on which this entity is (to be) queued: */
1021 struct cfs_rq *cfs_rq;
1022 /* rq "owned" by this entity/group: */
1023 struct cfs_rq *my_q;
1027 * Load-tracking only depends on SMP, FAIR_GROUP_SCHED dependency below may be
1028 * removed when useful for applications beyond shares distribution (e.g.
1031 #if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)
1032 /* Per-entity load-tracking */
1033 struct sched_avg avg;
1037 struct sched_rt_entity {
1038 struct list_head run_list;
1039 unsigned long timeout;
1040 unsigned long watchdog_stamp;
1041 unsigned int time_slice;
1043 struct sched_rt_entity *back;
1044 #ifdef CONFIG_RT_GROUP_SCHED
1045 struct sched_rt_entity *parent;
1046 /* rq on which this entity is (to be) queued: */
1047 struct rt_rq *rt_rq;
1048 /* rq "owned" by this entity/group: */
1056 enum perf_event_task_context {
1057 perf_invalid_context = -1,
1058 perf_hw_context = 0,
1060 perf_nr_task_contexts,
1063 struct task_struct {
1064 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1067 unsigned int flags; /* per process flags, defined below */
1068 unsigned int ptrace;
1071 struct llist_node wake_entry;
1076 int prio, static_prio, normal_prio;
1077 unsigned int rt_priority;
1078 const struct sched_class *sched_class;
1079 struct sched_entity se;
1080 struct sched_rt_entity rt;
1081 #ifdef CONFIG_CGROUP_SCHED
1082 struct task_group *sched_task_group;
1085 #ifdef CONFIG_PREEMPT_NOTIFIERS
1086 /* list of struct preempt_notifier: */
1087 struct hlist_head preempt_notifiers;
1091 * fpu_counter contains the number of consecutive context switches
1092 * that the FPU is used. If this is over a threshold, the lazy fpu
1093 * saving becomes unlazy to save the trap. This is an unsigned char
1094 * so that after 256 times the counter wraps and the behavior turns
1095 * lazy again; this to deal with bursty apps that only use FPU for
1098 unsigned char fpu_counter;
1099 #ifdef CONFIG_BLK_DEV_IO_TRACE
1100 unsigned int btrace_seq;
1103 unsigned int policy;
1104 int nr_cpus_allowed;
1105 cpumask_t cpus_allowed;
1107 #ifdef CONFIG_PREEMPT_RCU
1108 int rcu_read_lock_nesting;
1109 char rcu_read_unlock_special;
1110 struct list_head rcu_node_entry;
1111 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1112 #ifdef CONFIG_TREE_PREEMPT_RCU
1113 struct rcu_node *rcu_blocked_node;
1114 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1115 #ifdef CONFIG_RCU_BOOST
1116 struct rt_mutex *rcu_boost_mutex;
1117 #endif /* #ifdef CONFIG_RCU_BOOST */
1119 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1120 struct sched_info sched_info;
1123 struct list_head tasks;
1125 struct plist_node pushable_tasks;
1128 struct mm_struct *mm, *active_mm;
1129 #ifdef CONFIG_COMPAT_BRK
1130 unsigned brk_randomized:1;
1132 #if defined(SPLIT_RSS_COUNTING)
1133 struct task_rss_stat rss_stat;
1137 int exit_code, exit_signal;
1138 int pdeath_signal; /* The signal sent when the parent dies */
1139 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1141 /* Used for emulating ABI behavior of previous Linux versions */
1142 unsigned int personality;
1144 unsigned did_exec:1;
1145 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1147 unsigned in_iowait:1;
1149 /* Revert to default priority/policy when forking */
1150 unsigned sched_reset_on_fork:1;
1151 unsigned sched_contributes_to_load:1;
1153 unsigned long atomic_flags; /* Flags needing atomic access. */
1158 #ifdef CONFIG_CC_STACKPROTECTOR
1159 /* Canary value for the -fstack-protector gcc feature */
1160 unsigned long stack_canary;
1163 * pointers to (original) parent process, youngest child, younger sibling,
1164 * older sibling, respectively. (p->father can be replaced with
1165 * p->real_parent->pid)
1167 struct task_struct __rcu *real_parent; /* real parent process */
1168 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1170 * children/sibling forms the list of my natural children
1172 struct list_head children; /* list of my children */
1173 struct list_head sibling; /* linkage in my parent's children list */
1174 struct task_struct *group_leader; /* threadgroup leader */
1177 * ptraced is the list of tasks this task is using ptrace on.
1178 * This includes both natural children and PTRACE_ATTACH targets.
1179 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1181 struct list_head ptraced;
1182 struct list_head ptrace_entry;
1184 /* PID/PID hash table linkage. */
1185 struct pid_link pids[PIDTYPE_MAX];
1186 struct list_head thread_group;
1187 struct list_head thread_node;
1189 struct completion *vfork_done; /* for vfork() */
1190 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1191 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1193 cputime_t utime, stime, utimescaled, stimescaled;
1195 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1196 struct cputime prev_cputime;
1198 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1199 seqlock_t vtime_seqlock;
1200 unsigned long long vtime_snap;
1205 } vtime_snap_whence;
1207 unsigned long nvcsw, nivcsw; /* context switch counts */
1208 struct timespec start_time; /* monotonic time */
1209 struct timespec real_start_time; /* boot based time */
1210 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1211 unsigned long min_flt, maj_flt;
1213 struct task_cputime cputime_expires;
1214 struct list_head cpu_timers[3];
1216 /* process credentials */
1217 const struct cred __rcu *real_cred; /* objective and real subjective task
1218 * credentials (COW) */
1219 const struct cred __rcu *cred; /* effective (overridable) subjective task
1220 * credentials (COW) */
1221 char comm[TASK_COMM_LEN]; /* executable name excluding path
1222 - access with [gs]et_task_comm (which lock
1223 it with task_lock())
1224 - initialized normally by setup_new_exec */
1225 /* file system info */
1226 int link_count, total_link_count;
1227 #ifdef CONFIG_SYSVIPC
1229 struct sysv_sem sysvsem;
1231 #ifdef CONFIG_DETECT_HUNG_TASK
1232 /* hung task detection */
1233 unsigned long last_switch_count;
1235 /* CPU-specific state of this task */
1236 struct thread_struct thread;
1237 /* filesystem information */
1238 struct fs_struct *fs;
1239 /* open file information */
1240 struct files_struct *files;
1242 struct nsproxy *nsproxy;
1243 /* signal handlers */
1244 struct signal_struct *signal;
1245 struct sighand_struct *sighand;
1247 sigset_t blocked, real_blocked;
1248 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1249 struct sigpending pending;
1251 unsigned long sas_ss_sp;
1253 int (*notifier)(void *priv);
1254 void *notifier_data;
1255 sigset_t *notifier_mask;
1256 struct callback_head *task_works;
1258 struct audit_context *audit_context;
1259 #ifdef CONFIG_AUDITSYSCALL
1261 unsigned int sessionid;
1263 struct seccomp seccomp;
1265 /* Thread group tracking */
1268 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1270 spinlock_t alloc_lock;
1272 /* Protection of the PI data structures: */
1273 raw_spinlock_t pi_lock;
1275 #ifdef CONFIG_RT_MUTEXES
1276 /* PI waiters blocked on a rt_mutex held by this task */
1277 struct plist_head pi_waiters;
1278 /* Deadlock detection and priority inheritance handling */
1279 struct rt_mutex_waiter *pi_blocked_on;
1282 #ifdef CONFIG_DEBUG_MUTEXES
1283 /* mutex deadlock detection */
1284 struct mutex_waiter *blocked_on;
1286 #ifdef CONFIG_TRACE_IRQFLAGS
1287 unsigned int irq_events;
1288 unsigned long hardirq_enable_ip;
1289 unsigned long hardirq_disable_ip;
1290 unsigned int hardirq_enable_event;
1291 unsigned int hardirq_disable_event;
1292 int hardirqs_enabled;
1293 int hardirq_context;
1294 unsigned long softirq_disable_ip;
1295 unsigned long softirq_enable_ip;
1296 unsigned int softirq_disable_event;
1297 unsigned int softirq_enable_event;
1298 int softirqs_enabled;
1299 int softirq_context;
1301 #ifdef CONFIG_LOCKDEP
1302 # define MAX_LOCK_DEPTH 48UL
1305 unsigned int lockdep_recursion;
1306 struct held_lock held_locks[MAX_LOCK_DEPTH];
1307 gfp_t lockdep_reclaim_gfp;
1310 /* journalling filesystem info */
1313 /* stacked block device info */
1314 struct bio_list *bio_list;
1317 /* stack plugging */
1318 struct blk_plug *plug;
1322 struct reclaim_state *reclaim_state;
1324 struct backing_dev_info *backing_dev_info;
1326 struct io_context *io_context;
1328 unsigned long ptrace_message;
1329 siginfo_t *last_siginfo; /* For ptrace use. */
1330 struct task_io_accounting ioac;
1331 #if defined(CONFIG_TASK_XACCT)
1332 u64 acct_rss_mem1; /* accumulated rss usage */
1333 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1334 cputime_t acct_timexpd; /* stime + utime since last update */
1336 #ifdef CONFIG_CPUSETS
1337 nodemask_t mems_allowed; /* Protected by alloc_lock */
1338 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1339 int cpuset_mem_spread_rotor;
1340 int cpuset_slab_spread_rotor;
1342 #ifdef CONFIG_CGROUPS
1343 /* Control Group info protected by css_set_lock */
1344 struct css_set __rcu *cgroups;
1345 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1346 struct list_head cg_list;
1349 struct robust_list_head __user *robust_list;
1350 #ifdef CONFIG_COMPAT
1351 struct compat_robust_list_head __user *compat_robust_list;
1353 struct list_head pi_state_list;
1354 struct futex_pi_state *pi_state_cache;
1356 #ifdef CONFIG_PERF_EVENTS
1357 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1358 struct mutex perf_event_mutex;
1359 struct list_head perf_event_list;
1362 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1364 short pref_node_fork;
1366 #ifdef CONFIG_NUMA_BALANCING
1368 int numa_migrate_seq;
1369 unsigned int numa_scan_period;
1370 u64 node_stamp; /* migration stamp */
1371 struct callback_head numa_work;
1372 #endif /* CONFIG_NUMA_BALANCING */
1374 struct rcu_head rcu;
1377 * cache last used pipe for splice
1379 struct pipe_inode_info *splice_pipe;
1381 struct page_frag task_frag;
1383 #ifdef CONFIG_TASK_DELAY_ACCT
1384 struct task_delay_info *delays;
1386 #ifdef CONFIG_FAULT_INJECTION
1390 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1391 * balance_dirty_pages() for some dirty throttling pause
1394 int nr_dirtied_pause;
1395 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1397 #ifdef CONFIG_LATENCYTOP
1398 int latency_record_count;
1399 struct latency_record latency_record[LT_SAVECOUNT];
1402 * time slack values; these are used to round up poll() and
1403 * select() etc timeout values. These are in nanoseconds.
1405 unsigned long timer_slack_ns;
1406 unsigned long default_timer_slack_ns;
1408 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1409 /* Index of current stored address in ret_stack */
1411 /* Stack of return addresses for return function tracing */
1412 struct ftrace_ret_stack *ret_stack;
1413 /* time stamp for last schedule */
1414 unsigned long long ftrace_timestamp;
1416 * Number of functions that haven't been traced
1417 * because of depth overrun.
1419 atomic_t trace_overrun;
1420 /* Pause for the tracing */
1421 atomic_t tracing_graph_pause;
1423 #ifdef CONFIG_TRACING
1424 /* state flags for use by tracers */
1425 unsigned long trace;
1426 /* bitmask and counter of trace recursion */
1427 unsigned long trace_recursion;
1428 #endif /* CONFIG_TRACING */
1429 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1430 struct memcg_batch_info {
1431 int do_batch; /* incremented when batch uncharge started */
1432 struct mem_cgroup *memcg; /* target memcg of uncharge */
1433 unsigned long nr_pages; /* uncharged usage */
1434 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1436 unsigned int memcg_kmem_skip_account;
1437 struct memcg_oom_info {
1438 struct mem_cgroup *memcg;
1441 unsigned int may_oom:1;
1444 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1445 atomic_t ptrace_bp_refcnt;
1447 #ifdef CONFIG_UPROBES
1448 struct uprobe_task *utask;
1450 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1451 unsigned int sequential_io;
1452 unsigned int sequential_io_avg;
1456 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1457 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1459 #ifdef CONFIG_NUMA_BALANCING
1460 extern void task_numa_fault(int node, int pages, bool migrated);
1461 extern void set_numabalancing_state(bool enabled);
1463 static inline void task_numa_fault(int node, int pages, bool migrated)
1466 static inline void set_numabalancing_state(bool enabled)
1471 static inline struct pid *task_pid(struct task_struct *task)
1473 return task->pids[PIDTYPE_PID].pid;
1476 static inline struct pid *task_tgid(struct task_struct *task)
1478 return task->group_leader->pids[PIDTYPE_PID].pid;
1482 * Without tasklist or rcu lock it is not safe to dereference
1483 * the result of task_pgrp/task_session even if task == current,
1484 * we can race with another thread doing sys_setsid/sys_setpgid.
1486 static inline struct pid *task_pgrp(struct task_struct *task)
1488 return task->group_leader->pids[PIDTYPE_PGID].pid;
1491 static inline struct pid *task_session(struct task_struct *task)
1493 return task->group_leader->pids[PIDTYPE_SID].pid;
1496 struct pid_namespace;
1499 * the helpers to get the task's different pids as they are seen
1500 * from various namespaces
1502 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1503 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1505 * task_xid_nr_ns() : id seen from the ns specified;
1507 * set_task_vxid() : assigns a virtual id to a task;
1509 * see also pid_nr() etc in include/linux/pid.h
1511 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1512 struct pid_namespace *ns);
1514 static inline pid_t task_pid_nr(struct task_struct *tsk)
1519 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1520 struct pid_namespace *ns)
1522 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1525 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1527 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1531 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1536 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1538 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1540 return pid_vnr(task_tgid(tsk));
1544 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1545 struct pid_namespace *ns)
1547 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1550 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1552 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1556 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1557 struct pid_namespace *ns)
1559 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1562 static inline pid_t task_session_vnr(struct task_struct *tsk)
1564 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1567 /* obsolete, do not use */
1568 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1570 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1574 * pid_alive - check that a task structure is not stale
1575 * @p: Task structure to be checked.
1577 * Test if a process is not yet dead (at most zombie state)
1578 * If pid_alive fails, then pointers within the task structure
1579 * can be stale and must not be dereferenced.
1581 static inline int pid_alive(struct task_struct *p)
1583 return p->pids[PIDTYPE_PID].pid != NULL;
1587 * is_global_init - check if a task structure is init
1588 * @tsk: Task structure to be checked.
1590 * Check if a task structure is the first user space task the kernel created.
1592 static inline int is_global_init(struct task_struct *tsk)
1594 return tsk->pid == 1;
1597 extern struct pid *cad_pid;
1599 extern void free_task(struct task_struct *tsk);
1600 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1602 extern void __put_task_struct(struct task_struct *t);
1604 static inline void put_task_struct(struct task_struct *t)
1606 if (atomic_dec_and_test(&t->usage))
1607 __put_task_struct(t);
1610 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1611 extern void task_cputime(struct task_struct *t,
1612 cputime_t *utime, cputime_t *stime);
1613 extern void task_cputime_scaled(struct task_struct *t,
1614 cputime_t *utimescaled, cputime_t *stimescaled);
1615 extern cputime_t task_gtime(struct task_struct *t);
1617 static inline void task_cputime(struct task_struct *t,
1618 cputime_t *utime, cputime_t *stime)
1626 static inline void task_cputime_scaled(struct task_struct *t,
1627 cputime_t *utimescaled,
1628 cputime_t *stimescaled)
1631 *utimescaled = t->utimescaled;
1633 *stimescaled = t->stimescaled;
1636 static inline cputime_t task_gtime(struct task_struct *t)
1641 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1642 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1644 extern int task_free_register(struct notifier_block *n);
1645 extern int task_free_unregister(struct notifier_block *n);
1650 #define PF_EXITING 0x00000004 /* getting shut down */
1651 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1652 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1653 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1654 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1655 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1656 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1657 #define PF_DUMPCORE 0x00000200 /* dumped core */
1658 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1659 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1660 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1661 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1662 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1663 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1664 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1665 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1666 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1667 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1668 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1669 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1670 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1671 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1672 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1673 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1674 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1675 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1676 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1677 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1678 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1681 * Only the _current_ task can read/write to tsk->flags, but other
1682 * tasks can access tsk->flags in readonly mode for example
1683 * with tsk_used_math (like during threaded core dumping).
1684 * There is however an exception to this rule during ptrace
1685 * or during fork: the ptracer task is allowed to write to the
1686 * child->flags of its traced child (same goes for fork, the parent
1687 * can write to the child->flags), because we're guaranteed the
1688 * child is not running and in turn not changing child->flags
1689 * at the same time the parent does it.
1691 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1692 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1693 #define clear_used_math() clear_stopped_child_used_math(current)
1694 #define set_used_math() set_stopped_child_used_math(current)
1695 #define conditional_stopped_child_used_math(condition, child) \
1696 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1697 #define conditional_used_math(condition) \
1698 conditional_stopped_child_used_math(condition, current)
1699 #define copy_to_stopped_child_used_math(child) \
1700 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1701 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1702 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1703 #define used_math() tsk_used_math(current)
1705 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
1706 * __GFP_FS is also cleared as it implies __GFP_IO.
1708 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1710 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1711 flags &= ~(__GFP_IO | __GFP_FS);
1715 static inline unsigned int memalloc_noio_save(void)
1717 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1718 current->flags |= PF_MEMALLOC_NOIO;
1722 static inline void memalloc_noio_restore(unsigned int flags)
1724 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1727 /* Per-process atomic flags. */
1728 #define PFA_NO_NEW_PRIVS 0x00000001 /* May not gain new privileges. */
1730 static inline bool task_no_new_privs(struct task_struct *p)
1732 return test_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1735 static inline void task_set_no_new_privs(struct task_struct *p)
1737 set_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1741 * task->jobctl flags
1743 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1745 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1746 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1747 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1748 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1749 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1750 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1751 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1753 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1754 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1755 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1756 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1757 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1758 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1759 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1761 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1762 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1764 extern bool task_set_jobctl_pending(struct task_struct *task,
1766 extern void task_clear_jobctl_trapping(struct task_struct *task);
1767 extern void task_clear_jobctl_pending(struct task_struct *task,
1770 #ifdef CONFIG_PREEMPT_RCU
1772 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1773 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1775 static inline void rcu_copy_process(struct task_struct *p)
1777 p->rcu_read_lock_nesting = 0;
1778 p->rcu_read_unlock_special = 0;
1779 #ifdef CONFIG_TREE_PREEMPT_RCU
1780 p->rcu_blocked_node = NULL;
1781 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1782 #ifdef CONFIG_RCU_BOOST
1783 p->rcu_boost_mutex = NULL;
1784 #endif /* #ifdef CONFIG_RCU_BOOST */
1785 INIT_LIST_HEAD(&p->rcu_node_entry);
1790 static inline void rcu_copy_process(struct task_struct *p)
1796 static inline void tsk_restore_flags(struct task_struct *task,
1797 unsigned long orig_flags, unsigned long flags)
1799 task->flags &= ~flags;
1800 task->flags |= orig_flags & flags;
1804 extern void do_set_cpus_allowed(struct task_struct *p,
1805 const struct cpumask *new_mask);
1807 extern int set_cpus_allowed_ptr(struct task_struct *p,
1808 const struct cpumask *new_mask);
1810 static inline void do_set_cpus_allowed(struct task_struct *p,
1811 const struct cpumask *new_mask)
1814 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1815 const struct cpumask *new_mask)
1817 if (!cpumask_test_cpu(0, new_mask))
1823 #ifdef CONFIG_NO_HZ_COMMON
1824 void calc_load_enter_idle(void);
1825 void calc_load_exit_idle(void);
1827 static inline void calc_load_enter_idle(void) { }
1828 static inline void calc_load_exit_idle(void) { }
1829 #endif /* CONFIG_NO_HZ_COMMON */
1831 #ifndef CONFIG_CPUMASK_OFFSTACK
1832 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1834 return set_cpus_allowed_ptr(p, &new_mask);
1839 * Do not use outside of architecture code which knows its limitations.
1841 * sched_clock() has no promise of monotonicity or bounded drift between
1842 * CPUs, use (which you should not) requires disabling IRQs.
1844 * Please use one of the three interfaces below.
1846 extern unsigned long long notrace sched_clock(void);
1848 * See the comment in kernel/sched/clock.c
1850 extern u64 cpu_clock(int cpu);
1851 extern u64 local_clock(void);
1852 extern u64 sched_clock_cpu(int cpu);
1855 extern void sched_clock_init(void);
1857 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1858 static inline void sched_clock_tick(void)
1862 static inline void sched_clock_idle_sleep_event(void)
1866 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1871 * Architectures can set this to 1 if they have specified
1872 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1873 * but then during bootup it turns out that sched_clock()
1874 * is reliable after all:
1876 extern int sched_clock_stable;
1878 extern void sched_clock_tick(void);
1879 extern void sched_clock_idle_sleep_event(void);
1880 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1883 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1885 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1886 * The reason for this explicit opt-in is not to have perf penalty with
1887 * slow sched_clocks.
1889 extern void enable_sched_clock_irqtime(void);
1890 extern void disable_sched_clock_irqtime(void);
1892 static inline void enable_sched_clock_irqtime(void) {}
1893 static inline void disable_sched_clock_irqtime(void) {}
1896 extern unsigned long long
1897 task_sched_runtime(struct task_struct *task);
1899 /* sched_exec is called by processes performing an exec */
1901 extern void sched_exec(void);
1903 #define sched_exec() {}
1906 extern void sched_clock_idle_sleep_event(void);
1907 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1909 #ifdef CONFIG_HOTPLUG_CPU
1910 extern void idle_task_exit(void);
1912 static inline void idle_task_exit(void) {}
1915 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1916 extern void wake_up_nohz_cpu(int cpu);
1918 static inline void wake_up_nohz_cpu(int cpu) { }
1921 #ifdef CONFIG_NO_HZ_FULL
1922 extern bool sched_can_stop_tick(void);
1923 extern u64 scheduler_tick_max_deferment(void);
1925 static inline bool sched_can_stop_tick(void) { return false; }
1928 #ifdef CONFIG_SCHED_AUTOGROUP
1929 extern void sched_autogroup_create_attach(struct task_struct *p);
1930 extern void sched_autogroup_detach(struct task_struct *p);
1931 extern void sched_autogroup_fork(struct signal_struct *sig);
1932 extern void sched_autogroup_exit(struct signal_struct *sig);
1933 #ifdef CONFIG_PROC_FS
1934 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1935 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
1938 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1939 static inline void sched_autogroup_detach(struct task_struct *p) { }
1940 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1941 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1944 extern bool yield_to(struct task_struct *p, bool preempt);
1945 extern void set_user_nice(struct task_struct *p, long nice);
1946 extern int task_prio(const struct task_struct *p);
1947 extern int task_nice(const struct task_struct *p);
1948 extern int can_nice(const struct task_struct *p, const int nice);
1949 extern int task_curr(const struct task_struct *p);
1950 extern int idle_cpu(int cpu);
1951 extern int sched_setscheduler(struct task_struct *, int,
1952 const struct sched_param *);
1953 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1954 const struct sched_param *);
1955 extern struct task_struct *idle_task(int cpu);
1957 * is_idle_task - is the specified task an idle task?
1958 * @p: the task in question.
1960 static inline bool is_idle_task(const struct task_struct *p)
1964 extern struct task_struct *curr_task(int cpu);
1965 extern void set_curr_task(int cpu, struct task_struct *p);
1970 * The default (Linux) execution domain.
1972 extern struct exec_domain default_exec_domain;
1974 union thread_union {
1975 struct thread_info thread_info;
1976 unsigned long stack[THREAD_SIZE/sizeof(long)];
1979 #ifndef __HAVE_ARCH_KSTACK_END
1980 static inline int kstack_end(void *addr)
1982 /* Reliable end of stack detection:
1983 * Some APM bios versions misalign the stack
1985 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1989 extern union thread_union init_thread_union;
1990 extern struct task_struct init_task;
1992 extern struct mm_struct init_mm;
1994 extern struct pid_namespace init_pid_ns;
1997 * find a task by one of its numerical ids
1999 * find_task_by_pid_ns():
2000 * finds a task by its pid in the specified namespace
2001 * find_task_by_vpid():
2002 * finds a task by its virtual pid
2004 * see also find_vpid() etc in include/linux/pid.h
2007 extern struct task_struct *find_task_by_vpid(pid_t nr);
2008 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2009 struct pid_namespace *ns);
2011 extern void __set_special_pids(struct pid *pid);
2013 /* per-UID process charging. */
2014 extern struct user_struct * alloc_uid(kuid_t);
2015 static inline struct user_struct *get_uid(struct user_struct *u)
2017 atomic_inc(&u->__count);
2020 extern void free_uid(struct user_struct *);
2022 #include <asm/current.h>
2024 extern void xtime_update(unsigned long ticks);
2026 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2027 extern int wake_up_process(struct task_struct *tsk);
2028 extern void wake_up_new_task(struct task_struct *tsk);
2030 extern void kick_process(struct task_struct *tsk);
2032 static inline void kick_process(struct task_struct *tsk) { }
2034 extern void sched_fork(struct task_struct *p);
2035 extern void sched_dead(struct task_struct *p);
2037 extern void proc_caches_init(void);
2038 extern void flush_signals(struct task_struct *);
2039 extern void __flush_signals(struct task_struct *);
2040 extern void ignore_signals(struct task_struct *);
2041 extern void flush_signal_handlers(struct task_struct *, int force_default);
2042 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2044 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2046 unsigned long flags;
2049 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2050 ret = dequeue_signal(tsk, mask, info);
2051 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2056 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2058 extern void unblock_all_signals(void);
2059 extern void release_task(struct task_struct * p);
2060 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2061 extern int force_sigsegv(int, struct task_struct *);
2062 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2063 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2064 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2065 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2066 const struct cred *, u32);
2067 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2068 extern int kill_pid(struct pid *pid, int sig, int priv);
2069 extern int kill_proc_info(int, struct siginfo *, pid_t);
2070 extern __must_check bool do_notify_parent(struct task_struct *, int);
2071 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2072 extern void force_sig(int, struct task_struct *);
2073 extern int send_sig(int, struct task_struct *, int);
2074 extern int zap_other_threads(struct task_struct *p);
2075 extern struct sigqueue *sigqueue_alloc(void);
2076 extern void sigqueue_free(struct sigqueue *);
2077 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2078 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2080 static inline void restore_saved_sigmask(void)
2082 if (test_and_clear_restore_sigmask())
2083 __set_current_blocked(¤t->saved_sigmask);
2086 static inline sigset_t *sigmask_to_save(void)
2088 sigset_t *res = ¤t->blocked;
2089 if (unlikely(test_restore_sigmask()))
2090 res = ¤t->saved_sigmask;
2094 static inline int kill_cad_pid(int sig, int priv)
2096 return kill_pid(cad_pid, sig, priv);
2099 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2100 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2101 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2102 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2105 * True if we are on the alternate signal stack.
2107 static inline int on_sig_stack(unsigned long sp)
2109 #ifdef CONFIG_STACK_GROWSUP
2110 return sp >= current->sas_ss_sp &&
2111 sp - current->sas_ss_sp < current->sas_ss_size;
2113 return sp > current->sas_ss_sp &&
2114 sp - current->sas_ss_sp <= current->sas_ss_size;
2118 static inline int sas_ss_flags(unsigned long sp)
2120 return (current->sas_ss_size == 0 ? SS_DISABLE
2121 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2124 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2126 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2127 #ifdef CONFIG_STACK_GROWSUP
2128 return current->sas_ss_sp;
2130 return current->sas_ss_sp + current->sas_ss_size;
2136 * Routines for handling mm_structs
2138 extern struct mm_struct * mm_alloc(void);
2140 /* mmdrop drops the mm and the page tables */
2141 extern void __mmdrop(struct mm_struct *);
2142 static inline void mmdrop(struct mm_struct * mm)
2144 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2148 /* mmput gets rid of the mappings and all user-space */
2149 extern void mmput(struct mm_struct *);
2150 /* Grab a reference to a task's mm, if it is not already going away */
2151 extern struct mm_struct *get_task_mm(struct task_struct *task);
2153 * Grab a reference to a task's mm, if it is not already going away
2154 * and ptrace_may_access with the mode parameter passed to it
2157 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2158 /* Remove the current tasks stale references to the old mm_struct */
2159 extern void mm_release(struct task_struct *, struct mm_struct *);
2160 /* Allocate a new mm structure and copy contents from tsk->mm */
2161 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2163 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2164 struct task_struct *);
2165 extern void flush_thread(void);
2166 extern void exit_thread(void);
2168 extern void exit_files(struct task_struct *);
2169 extern void __cleanup_sighand(struct sighand_struct *);
2171 extern void exit_itimers(struct signal_struct *);
2172 extern void flush_itimer_signals(void);
2174 extern void do_group_exit(int);
2176 extern int allow_signal(int);
2177 extern int disallow_signal(int);
2179 extern int do_execve(const char *,
2180 const char __user * const __user *,
2181 const char __user * const __user *);
2182 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2183 struct task_struct *fork_idle(int);
2184 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2186 extern void set_task_comm(struct task_struct *tsk, char *from);
2187 extern char *get_task_comm(char *to, struct task_struct *tsk);
2190 void scheduler_ipi(void);
2191 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2193 static inline void scheduler_ipi(void) { }
2194 static inline unsigned long wait_task_inactive(struct task_struct *p,
2201 #define next_task(p) \
2202 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2204 #define for_each_process(p) \
2205 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2207 extern bool current_is_single_threaded(void);
2210 * Careful: do_each_thread/while_each_thread is a double loop so
2211 * 'break' will not work as expected - use goto instead.
2213 #define do_each_thread(g, t) \
2214 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2216 #define while_each_thread(g, t) \
2217 while ((t = next_thread(t)) != g)
2219 #define __for_each_thread(signal, t) \
2220 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2222 #define for_each_thread(p, t) \
2223 __for_each_thread((p)->signal, t)
2225 /* Careful: this is a double loop, 'break' won't work as expected. */
2226 #define for_each_process_thread(p, t) \
2227 for_each_process(p) for_each_thread(p, t)
2229 static inline int get_nr_threads(struct task_struct *tsk)
2231 return tsk->signal->nr_threads;
2234 static inline bool thread_group_leader(struct task_struct *p)
2236 return p->exit_signal >= 0;
2239 /* Do to the insanities of de_thread it is possible for a process
2240 * to have the pid of the thread group leader without actually being
2241 * the thread group leader. For iteration through the pids in proc
2242 * all we care about is that we have a task with the appropriate
2243 * pid, we don't actually care if we have the right task.
2245 static inline int has_group_leader_pid(struct task_struct *p)
2247 return p->pid == p->tgid;
2251 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2253 return p1->tgid == p2->tgid;
2256 static inline struct task_struct *next_thread(const struct task_struct *p)
2258 return list_entry_rcu(p->thread_group.next,
2259 struct task_struct, thread_group);
2262 static inline int thread_group_empty(struct task_struct *p)
2264 return list_empty(&p->thread_group);
2267 #define delay_group_leader(p) \
2268 (thread_group_leader(p) && !thread_group_empty(p))
2271 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2272 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2273 * pins the final release of task.io_context. Also protects ->cpuset and
2274 * ->cgroup.subsys[]. And ->vfork_done.
2276 * Nests both inside and outside of read_lock(&tasklist_lock).
2277 * It must not be nested with write_lock_irq(&tasklist_lock),
2278 * neither inside nor outside.
2280 static inline void task_lock(struct task_struct *p)
2282 spin_lock(&p->alloc_lock);
2285 static inline void task_unlock(struct task_struct *p)
2287 spin_unlock(&p->alloc_lock);
2290 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2291 unsigned long *flags);
2293 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2294 unsigned long *flags)
2296 struct sighand_struct *ret;
2298 ret = __lock_task_sighand(tsk, flags);
2299 (void)__cond_lock(&tsk->sighand->siglock, ret);
2303 static inline void unlock_task_sighand(struct task_struct *tsk,
2304 unsigned long *flags)
2306 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2309 #ifdef CONFIG_CGROUPS
2310 static inline void threadgroup_change_begin(struct task_struct *tsk)
2312 down_read(&tsk->signal->group_rwsem);
2314 static inline void threadgroup_change_end(struct task_struct *tsk)
2316 up_read(&tsk->signal->group_rwsem);
2320 * threadgroup_lock - lock threadgroup
2321 * @tsk: member task of the threadgroup to lock
2323 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2324 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2325 * change ->group_leader/pid. This is useful for cases where the threadgroup
2326 * needs to stay stable across blockable operations.
2328 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2329 * synchronization. While held, no new task will be added to threadgroup
2330 * and no existing live task will have its PF_EXITING set.
2332 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2333 * sub-thread becomes a new leader.
2335 static inline void threadgroup_lock(struct task_struct *tsk)
2337 down_write(&tsk->signal->group_rwsem);
2341 * threadgroup_unlock - unlock threadgroup
2342 * @tsk: member task of the threadgroup to unlock
2344 * Reverse threadgroup_lock().
2346 static inline void threadgroup_unlock(struct task_struct *tsk)
2348 up_write(&tsk->signal->group_rwsem);
2351 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2352 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2353 static inline void threadgroup_lock(struct task_struct *tsk) {}
2354 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2357 #ifndef __HAVE_THREAD_FUNCTIONS
2359 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2360 #define task_stack_page(task) ((task)->stack)
2362 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2364 *task_thread_info(p) = *task_thread_info(org);
2365 task_thread_info(p)->task = p;
2368 static inline unsigned long *end_of_stack(struct task_struct *p)
2370 return (unsigned long *)(task_thread_info(p) + 1);
2375 static inline int object_is_on_stack(void *obj)
2377 void *stack = task_stack_page(current);
2379 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2382 extern void thread_info_cache_init(void);
2384 #ifdef CONFIG_DEBUG_STACK_USAGE
2385 static inline unsigned long stack_not_used(struct task_struct *p)
2387 unsigned long *n = end_of_stack(p);
2389 do { /* Skip over canary */
2393 return (unsigned long)n - (unsigned long)end_of_stack(p);
2397 /* set thread flags in other task's structures
2398 * - see asm/thread_info.h for TIF_xxxx flags available
2400 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2402 set_ti_thread_flag(task_thread_info(tsk), flag);
2405 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2407 clear_ti_thread_flag(task_thread_info(tsk), flag);
2410 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2412 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2415 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2417 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2420 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2422 return test_ti_thread_flag(task_thread_info(tsk), flag);
2425 static inline void set_tsk_need_resched(struct task_struct *tsk)
2427 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2430 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2432 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2435 static inline int test_tsk_need_resched(struct task_struct *tsk)
2437 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2440 static inline int restart_syscall(void)
2442 set_tsk_thread_flag(current, TIF_SIGPENDING);
2443 return -ERESTARTNOINTR;
2446 static inline int signal_pending(struct task_struct *p)
2448 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2451 static inline int __fatal_signal_pending(struct task_struct *p)
2453 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2456 static inline int fatal_signal_pending(struct task_struct *p)
2458 return signal_pending(p) && __fatal_signal_pending(p);
2461 static inline int signal_pending_state(long state, struct task_struct *p)
2463 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2465 if (!signal_pending(p))
2468 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2471 static inline int need_resched(void)
2473 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2477 * cond_resched() and cond_resched_lock(): latency reduction via
2478 * explicit rescheduling in places that are safe. The return
2479 * value indicates whether a reschedule was done in fact.
2480 * cond_resched_lock() will drop the spinlock before scheduling,
2481 * cond_resched_softirq() will enable bhs before scheduling.
2483 extern int _cond_resched(void);
2485 #define cond_resched() ({ \
2486 __might_sleep(__FILE__, __LINE__, 0); \
2490 extern int __cond_resched_lock(spinlock_t *lock);
2492 #ifdef CONFIG_PREEMPT_COUNT
2493 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2495 #define PREEMPT_LOCK_OFFSET 0
2498 #define cond_resched_lock(lock) ({ \
2499 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2500 __cond_resched_lock(lock); \
2503 extern int __cond_resched_softirq(void);
2505 #define cond_resched_softirq() ({ \
2506 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2507 __cond_resched_softirq(); \
2511 * Does a critical section need to be broken due to another
2512 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2513 * but a general need for low latency)
2515 static inline int spin_needbreak(spinlock_t *lock)
2517 #ifdef CONFIG_PREEMPT
2518 return spin_is_contended(lock);
2525 * Idle thread specific functions to determine the need_resched
2526 * polling state. We have two versions, one based on TS_POLLING in
2527 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2531 static inline int tsk_is_polling(struct task_struct *p)
2533 return task_thread_info(p)->status & TS_POLLING;
2535 static inline void __current_set_polling(void)
2537 current_thread_info()->status |= TS_POLLING;
2540 static inline bool __must_check current_set_polling_and_test(void)
2542 __current_set_polling();
2545 * Polling state must be visible before we test NEED_RESCHED,
2546 * paired by resched_task()
2550 return unlikely(tif_need_resched());
2553 static inline void __current_clr_polling(void)
2555 current_thread_info()->status &= ~TS_POLLING;
2558 static inline bool __must_check current_clr_polling_and_test(void)
2560 __current_clr_polling();
2563 * Polling state must be visible before we test NEED_RESCHED,
2564 * paired by resched_task()
2568 return unlikely(tif_need_resched());
2570 #elif defined(TIF_POLLING_NRFLAG)
2571 static inline int tsk_is_polling(struct task_struct *p)
2573 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2576 static inline void __current_set_polling(void)
2578 set_thread_flag(TIF_POLLING_NRFLAG);
2581 static inline bool __must_check current_set_polling_and_test(void)
2583 __current_set_polling();
2586 * Polling state must be visible before we test NEED_RESCHED,
2587 * paired by resched_task()
2589 * XXX: assumes set/clear bit are identical barrier wise.
2591 smp_mb__after_clear_bit();
2593 return unlikely(tif_need_resched());
2596 static inline void __current_clr_polling(void)
2598 clear_thread_flag(TIF_POLLING_NRFLAG);
2601 static inline bool __must_check current_clr_polling_and_test(void)
2603 __current_clr_polling();
2606 * Polling state must be visible before we test NEED_RESCHED,
2607 * paired by resched_task()
2609 smp_mb__after_clear_bit();
2611 return unlikely(tif_need_resched());
2615 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2616 static inline void __current_set_polling(void) { }
2617 static inline void __current_clr_polling(void) { }
2619 static inline bool __must_check current_set_polling_and_test(void)
2621 return unlikely(tif_need_resched());
2623 static inline bool __must_check current_clr_polling_and_test(void)
2625 return unlikely(tif_need_resched());
2630 * Thread group CPU time accounting.
2632 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2633 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2635 static inline void thread_group_cputime_init(struct signal_struct *sig)
2637 raw_spin_lock_init(&sig->cputimer.lock);
2641 * Reevaluate whether the task has signals pending delivery.
2642 * Wake the task if so.
2643 * This is required every time the blocked sigset_t changes.
2644 * callers must hold sighand->siglock.
2646 extern void recalc_sigpending_and_wake(struct task_struct *t);
2647 extern void recalc_sigpending(void);
2649 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2651 static inline void signal_wake_up(struct task_struct *t, bool resume)
2653 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2655 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2657 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2661 * Wrappers for p->thread_info->cpu access. No-op on UP.
2665 static inline unsigned int task_cpu(const struct task_struct *p)
2667 return task_thread_info(p)->cpu;
2670 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2674 static inline unsigned int task_cpu(const struct task_struct *p)
2679 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2683 #endif /* CONFIG_SMP */
2685 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2686 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2688 #ifdef CONFIG_CGROUP_SCHED
2689 extern struct task_group root_task_group;
2690 #endif /* CONFIG_CGROUP_SCHED */
2692 extern int task_can_switch_user(struct user_struct *up,
2693 struct task_struct *tsk);
2695 #ifdef CONFIG_TASK_XACCT
2696 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2698 tsk->ioac.rchar += amt;
2701 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2703 tsk->ioac.wchar += amt;
2706 static inline void inc_syscr(struct task_struct *tsk)
2711 static inline void inc_syscw(struct task_struct *tsk)
2716 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2720 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2724 static inline void inc_syscr(struct task_struct *tsk)
2728 static inline void inc_syscw(struct task_struct *tsk)
2733 #ifndef TASK_SIZE_OF
2734 #define TASK_SIZE_OF(tsk) TASK_SIZE
2737 #ifdef CONFIG_MM_OWNER
2738 extern void mm_update_next_owner(struct mm_struct *mm);
2739 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2741 static inline void mm_update_next_owner(struct mm_struct *mm)
2745 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2748 #endif /* CONFIG_MM_OWNER */
2750 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2753 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2756 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2759 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2762 static inline unsigned long rlimit(unsigned int limit)
2764 return task_rlimit(current, limit);
2767 static inline unsigned long rlimit_max(unsigned int limit)
2769 return task_rlimit_max(current, limit);