4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt_mask.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/signal.h>
37 #include <linux/compiler.h>
38 #include <linux/completion.h>
39 #include <linux/pid.h>
40 #include <linux/percpu.h>
41 #include <linux/topology.h>
42 #include <linux/proportions.h>
43 #include <linux/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/task_io_accounting.h>
54 #include <linux/latencytop.h>
55 #include <linux/cred.h>
56 #include <linux/llist.h>
57 #include <linux/uidgid.h>
58 #include <linux/gfp.h>
60 #include <asm/processor.h>
62 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
65 * Extended scheduling parameters data structure.
67 * This is needed because the original struct sched_param can not be
68 * altered without introducing ABI issues with legacy applications
69 * (e.g., in sched_getparam()).
71 * However, the possibility of specifying more than just a priority for
72 * the tasks may be useful for a wide variety of application fields, e.g.,
73 * multimedia, streaming, automation and control, and many others.
75 * This variant (sched_attr) is meant at describing a so-called
76 * sporadic time-constrained task. In such model a task is specified by:
77 * - the activation period or minimum instance inter-arrival time;
78 * - the maximum (or average, depending on the actual scheduling
79 * discipline) computation time of all instances, a.k.a. runtime;
80 * - the deadline (relative to the actual activation time) of each
82 * Very briefly, a periodic (sporadic) task asks for the execution of
83 * some specific computation --which is typically called an instance--
84 * (at most) every period. Moreover, each instance typically lasts no more
85 * than the runtime and must be completed by time instant t equal to
86 * the instance activation time + the deadline.
88 * This is reflected by the actual fields of the sched_attr structure:
90 * @size size of the structure, for fwd/bwd compat.
92 * @sched_policy task's scheduling policy
93 * @sched_flags for customizing the scheduler behaviour
94 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
95 * @sched_priority task's static priority (SCHED_FIFO/RR)
96 * @sched_deadline representative of the task's deadline
97 * @sched_runtime representative of the task's runtime
98 * @sched_period representative of the task's period
100 * Given this task model, there are a multiplicity of scheduling algorithms
101 * and policies, that can be used to ensure all the tasks will make their
102 * timing constraints.
104 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
105 * only user of this new interface. More information about the algorithm
106 * available in the scheduling class file or in Documentation/.
114 /* SCHED_NORMAL, SCHED_BATCH */
117 /* SCHED_FIFO, SCHED_RR */
127 struct futex_pi_state;
128 struct robust_list_head;
131 struct perf_event_context;
135 #define VMACACHE_BITS 2
136 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
137 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
140 * These are the constant used to fake the fixed-point load-average
141 * counting. Some notes:
142 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
143 * a load-average precision of 10 bits integer + 11 bits fractional
144 * - if you want to count load-averages more often, you need more
145 * precision, or rounding will get you. With 2-second counting freq,
146 * the EXP_n values would be 1981, 2034 and 2043 if still using only
149 extern unsigned long avenrun[]; /* Load averages */
150 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
152 #define FSHIFT 11 /* nr of bits of precision */
153 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
154 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
155 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
156 #define EXP_5 2014 /* 1/exp(5sec/5min) */
157 #define EXP_15 2037 /* 1/exp(5sec/15min) */
159 #define CALC_LOAD(load,exp,n) \
161 load += n*(FIXED_1-exp); \
164 extern unsigned long total_forks;
165 extern int nr_threads;
166 DECLARE_PER_CPU(unsigned long, process_counts);
167 extern int nr_processes(void);
168 extern unsigned long nr_running(void);
169 extern unsigned long nr_iowait(void);
170 extern unsigned long nr_iowait_cpu(int cpu);
171 extern unsigned long this_cpu_load(void);
174 extern void calc_global_load(unsigned long ticks);
175 extern void update_cpu_load_nohz(void);
177 extern unsigned long get_parent_ip(unsigned long addr);
179 extern void dump_cpu_task(int cpu);
184 #ifdef CONFIG_SCHED_DEBUG
185 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
186 extern void proc_sched_set_task(struct task_struct *p);
188 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
192 * Task state bitmask. NOTE! These bits are also
193 * encoded in fs/proc/array.c: get_task_state().
195 * We have two separate sets of flags: task->state
196 * is about runnability, while task->exit_state are
197 * about the task exiting. Confusing, but this way
198 * modifying one set can't modify the other one by
201 #define TASK_RUNNING 0
202 #define TASK_INTERRUPTIBLE 1
203 #define TASK_UNINTERRUPTIBLE 2
204 #define __TASK_STOPPED 4
205 #define __TASK_TRACED 8
206 /* in tsk->exit_state */
208 #define EXIT_ZOMBIE 32
209 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
210 /* in tsk->state again */
212 #define TASK_WAKEKILL 128
213 #define TASK_WAKING 256
214 #define TASK_PARKED 512
215 #define TASK_STATE_MAX 1024
217 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
219 extern char ___assert_task_state[1 - 2*!!(
220 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
222 /* Convenience macros for the sake of set_task_state */
223 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
224 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
225 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
227 /* Convenience macros for the sake of wake_up */
228 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
229 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
231 /* get_task_state() */
232 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
233 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
234 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
236 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
237 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
238 #define task_is_stopped_or_traced(task) \
239 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
240 #define task_contributes_to_load(task) \
241 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
242 (task->flags & PF_FROZEN) == 0)
244 #define __set_task_state(tsk, state_value) \
245 do { (tsk)->state = (state_value); } while (0)
246 #define set_task_state(tsk, state_value) \
247 set_mb((tsk)->state, (state_value))
250 * set_current_state() includes a barrier so that the write of current->state
251 * is correctly serialised wrt the caller's subsequent test of whether to
254 * set_current_state(TASK_UNINTERRUPTIBLE);
255 * if (do_i_need_to_sleep())
258 * If the caller does not need such serialisation then use __set_current_state()
260 #define __set_current_state(state_value) \
261 do { current->state = (state_value); } while (0)
262 #define set_current_state(state_value) \
263 set_mb(current->state, (state_value))
265 /* Task command name length */
266 #define TASK_COMM_LEN 16
268 #include <linux/spinlock.h>
271 * This serializes "schedule()" and also protects
272 * the run-queue from deletions/modifications (but
273 * _adding_ to the beginning of the run-queue has
276 extern rwlock_t tasklist_lock;
277 extern spinlock_t mmlist_lock;
281 #ifdef CONFIG_PROVE_RCU
282 extern int lockdep_tasklist_lock_is_held(void);
283 #endif /* #ifdef CONFIG_PROVE_RCU */
285 extern void sched_init(void);
286 extern void sched_init_smp(void);
287 extern asmlinkage void schedule_tail(struct task_struct *prev);
288 extern void init_idle(struct task_struct *idle, int cpu);
289 extern void init_idle_bootup_task(struct task_struct *idle);
291 extern int runqueue_is_locked(int cpu);
293 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
294 extern void nohz_balance_enter_idle(int cpu);
295 extern void set_cpu_sd_state_idle(void);
296 extern int get_nohz_timer_target(int pinned);
298 static inline void nohz_balance_enter_idle(int cpu) { }
299 static inline void set_cpu_sd_state_idle(void) { }
300 static inline int get_nohz_timer_target(int pinned)
302 return smp_processor_id();
307 * Only dump TASK_* tasks. (0 for all tasks)
309 extern void show_state_filter(unsigned long state_filter);
311 static inline void show_state(void)
313 show_state_filter(0);
316 extern void show_regs(struct pt_regs *);
319 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
320 * task), SP is the stack pointer of the first frame that should be shown in the back
321 * trace (or NULL if the entire call-chain of the task should be shown).
323 extern void show_stack(struct task_struct *task, unsigned long *sp);
325 void io_schedule(void);
326 long io_schedule_timeout(long timeout);
328 extern void cpu_init (void);
329 extern void trap_init(void);
330 extern void update_process_times(int user);
331 extern void scheduler_tick(void);
333 extern void sched_show_task(struct task_struct *p);
335 #ifdef CONFIG_LOCKUP_DETECTOR
336 extern void touch_softlockup_watchdog(void);
337 extern void touch_softlockup_watchdog_sync(void);
338 extern void touch_all_softlockup_watchdogs(void);
339 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
341 size_t *lenp, loff_t *ppos);
342 extern unsigned int softlockup_panic;
343 void lockup_detector_init(void);
345 static inline void touch_softlockup_watchdog(void)
348 static inline void touch_softlockup_watchdog_sync(void)
351 static inline void touch_all_softlockup_watchdogs(void)
354 static inline void lockup_detector_init(void)
359 #ifdef CONFIG_DETECT_HUNG_TASK
360 void reset_hung_task_detector(void);
362 static inline void reset_hung_task_detector(void)
367 /* Attach to any functions which should be ignored in wchan output. */
368 #define __sched __attribute__((__section__(".sched.text")))
370 /* Linker adds these: start and end of __sched functions */
371 extern char __sched_text_start[], __sched_text_end[];
373 /* Is this address in the __sched functions? */
374 extern int in_sched_functions(unsigned long addr);
376 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
377 extern signed long schedule_timeout(signed long timeout);
378 extern signed long schedule_timeout_interruptible(signed long timeout);
379 extern signed long schedule_timeout_killable(signed long timeout);
380 extern signed long schedule_timeout_uninterruptible(signed long timeout);
381 asmlinkage void schedule(void);
382 extern void schedule_preempt_disabled(void);
385 struct user_namespace;
388 extern void arch_pick_mmap_layout(struct mm_struct *mm);
390 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
391 unsigned long, unsigned long);
393 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
394 unsigned long len, unsigned long pgoff,
395 unsigned long flags);
397 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
400 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
401 #define SUID_DUMP_USER 1 /* Dump as user of process */
402 #define SUID_DUMP_ROOT 2 /* Dump as root */
406 /* for SUID_DUMP_* above */
407 #define MMF_DUMPABLE_BITS 2
408 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
410 extern void set_dumpable(struct mm_struct *mm, int value);
412 * This returns the actual value of the suid_dumpable flag. For things
413 * that are using this for checking for privilege transitions, it must
414 * test against SUID_DUMP_USER rather than treating it as a boolean
417 static inline int __get_dumpable(unsigned long mm_flags)
419 return mm_flags & MMF_DUMPABLE_MASK;
422 static inline int get_dumpable(struct mm_struct *mm)
424 return __get_dumpable(mm->flags);
427 /* coredump filter bits */
428 #define MMF_DUMP_ANON_PRIVATE 2
429 #define MMF_DUMP_ANON_SHARED 3
430 #define MMF_DUMP_MAPPED_PRIVATE 4
431 #define MMF_DUMP_MAPPED_SHARED 5
432 #define MMF_DUMP_ELF_HEADERS 6
433 #define MMF_DUMP_HUGETLB_PRIVATE 7
434 #define MMF_DUMP_HUGETLB_SHARED 8
436 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
437 #define MMF_DUMP_FILTER_BITS 7
438 #define MMF_DUMP_FILTER_MASK \
439 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
440 #define MMF_DUMP_FILTER_DEFAULT \
441 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
442 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
444 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
445 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
447 # define MMF_DUMP_MASK_DEFAULT_ELF 0
449 /* leave room for more dump flags */
450 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
451 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
452 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
454 #define MMF_HAS_UPROBES 19 /* has uprobes */
455 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
457 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
459 struct sighand_struct {
461 struct k_sigaction action[_NSIG];
463 wait_queue_head_t signalfd_wqh;
466 struct pacct_struct {
469 unsigned long ac_mem;
470 cputime_t ac_utime, ac_stime;
471 unsigned long ac_minflt, ac_majflt;
482 * struct cputime - snaphsot of system and user cputime
483 * @utime: time spent in user mode
484 * @stime: time spent in system mode
486 * Gathers a generic snapshot of user and system time.
494 * struct task_cputime - collected CPU time counts
495 * @utime: time spent in user mode, in &cputime_t units
496 * @stime: time spent in kernel mode, in &cputime_t units
497 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
499 * This is an extension of struct cputime that includes the total runtime
500 * spent by the task from the scheduler point of view.
502 * As a result, this structure groups together three kinds of CPU time
503 * that are tracked for threads and thread groups. Most things considering
504 * CPU time want to group these counts together and treat all three
505 * of them in parallel.
507 struct task_cputime {
510 unsigned long long sum_exec_runtime;
512 /* Alternate field names when used to cache expirations. */
513 #define prof_exp stime
514 #define virt_exp utime
515 #define sched_exp sum_exec_runtime
517 #define INIT_CPUTIME \
518 (struct task_cputime) { \
521 .sum_exec_runtime = 0, \
524 #ifdef CONFIG_PREEMPT_COUNT
525 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
527 #define PREEMPT_DISABLED PREEMPT_ENABLED
531 * Disable preemption until the scheduler is running.
532 * Reset by start_kernel()->sched_init()->init_idle().
534 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
535 * before the scheduler is active -- see should_resched().
537 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
540 * struct thread_group_cputimer - thread group interval timer counts
541 * @cputime: thread group interval timers.
542 * @running: non-zero when there are timers running and
543 * @cputime receives updates.
544 * @lock: lock for fields in this struct.
546 * This structure contains the version of task_cputime, above, that is
547 * used for thread group CPU timer calculations.
549 struct thread_group_cputimer {
550 struct task_cputime cputime;
555 #include <linux/rwsem.h>
559 * NOTE! "signal_struct" does not have its own
560 * locking, because a shared signal_struct always
561 * implies a shared sighand_struct, so locking
562 * sighand_struct is always a proper superset of
563 * the locking of signal_struct.
565 struct signal_struct {
569 struct list_head thread_head;
571 wait_queue_head_t wait_chldexit; /* for wait4() */
573 /* current thread group signal load-balancing target: */
574 struct task_struct *curr_target;
576 /* shared signal handling: */
577 struct sigpending shared_pending;
579 /* thread group exit support */
582 * - notify group_exit_task when ->count is equal to notify_count
583 * - everyone except group_exit_task is stopped during signal delivery
584 * of fatal signals, group_exit_task processes the signal.
587 struct task_struct *group_exit_task;
589 /* thread group stop support, overloads group_exit_code too */
590 int group_stop_count;
591 unsigned int flags; /* see SIGNAL_* flags below */
594 * PR_SET_CHILD_SUBREAPER marks a process, like a service
595 * manager, to re-parent orphan (double-forking) child processes
596 * to this process instead of 'init'. The service manager is
597 * able to receive SIGCHLD signals and is able to investigate
598 * the process until it calls wait(). All children of this
599 * process will inherit a flag if they should look for a
600 * child_subreaper process at exit.
602 unsigned int is_child_subreaper:1;
603 unsigned int has_child_subreaper:1;
605 /* POSIX.1b Interval Timers */
607 struct list_head posix_timers;
609 /* ITIMER_REAL timer for the process */
610 struct hrtimer real_timer;
611 struct pid *leader_pid;
612 ktime_t it_real_incr;
615 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
616 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
617 * values are defined to 0 and 1 respectively
619 struct cpu_itimer it[2];
622 * Thread group totals for process CPU timers.
623 * See thread_group_cputimer(), et al, for details.
625 struct thread_group_cputimer cputimer;
627 /* Earliest-expiration cache. */
628 struct task_cputime cputime_expires;
630 struct list_head cpu_timers[3];
632 struct pid *tty_old_pgrp;
634 /* boolean value for session group leader */
637 struct tty_struct *tty; /* NULL if no tty */
639 #ifdef CONFIG_SCHED_AUTOGROUP
640 struct autogroup *autogroup;
643 * Cumulative resource counters for dead threads in the group,
644 * and for reaped dead child processes forked by this group.
645 * Live threads maintain their own counters and add to these
646 * in __exit_signal, except for the group leader.
648 cputime_t utime, stime, cutime, cstime;
651 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
652 struct cputime prev_cputime;
654 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
655 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
656 unsigned long inblock, oublock, cinblock, coublock;
657 unsigned long maxrss, cmaxrss;
658 struct task_io_accounting ioac;
661 * Cumulative ns of schedule CPU time fo dead threads in the
662 * group, not including a zombie group leader, (This only differs
663 * from jiffies_to_ns(utime + stime) if sched_clock uses something
664 * other than jiffies.)
666 unsigned long long sum_sched_runtime;
669 * We don't bother to synchronize most readers of this at all,
670 * because there is no reader checking a limit that actually needs
671 * to get both rlim_cur and rlim_max atomically, and either one
672 * alone is a single word that can safely be read normally.
673 * getrlimit/setrlimit use task_lock(current->group_leader) to
674 * protect this instead of the siglock, because they really
675 * have no need to disable irqs.
677 struct rlimit rlim[RLIM_NLIMITS];
679 #ifdef CONFIG_BSD_PROCESS_ACCT
680 struct pacct_struct pacct; /* per-process accounting information */
682 #ifdef CONFIG_TASKSTATS
683 struct taskstats *stats;
687 unsigned audit_tty_log_passwd;
688 struct tty_audit_buf *tty_audit_buf;
690 #ifdef CONFIG_CGROUPS
692 * group_rwsem prevents new tasks from entering the threadgroup and
693 * member tasks from exiting,a more specifically, setting of
694 * PF_EXITING. fork and exit paths are protected with this rwsem
695 * using threadgroup_change_begin/end(). Users which require
696 * threadgroup to remain stable should use threadgroup_[un]lock()
697 * which also takes care of exec path. Currently, cgroup is the
700 struct rw_semaphore group_rwsem;
703 oom_flags_t oom_flags;
704 short oom_score_adj; /* OOM kill score adjustment */
705 short oom_score_adj_min; /* OOM kill score adjustment min value.
706 * Only settable by CAP_SYS_RESOURCE. */
708 struct mutex cred_guard_mutex; /* guard against foreign influences on
709 * credential calculations
710 * (notably. ptrace) */
714 * Bits in flags field of signal_struct.
716 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
717 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
718 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
719 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
721 * Pending notifications to parent.
723 #define SIGNAL_CLD_STOPPED 0x00000010
724 #define SIGNAL_CLD_CONTINUED 0x00000020
725 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
727 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
729 /* If true, all threads except ->group_exit_task have pending SIGKILL */
730 static inline int signal_group_exit(const struct signal_struct *sig)
732 return (sig->flags & SIGNAL_GROUP_EXIT) ||
733 (sig->group_exit_task != NULL);
737 * Some day this will be a full-fledged user tracking system..
740 atomic_t __count; /* reference count */
741 atomic_t processes; /* How many processes does this user have? */
742 atomic_t sigpending; /* How many pending signals does this user have? */
743 #ifdef CONFIG_INOTIFY_USER
744 atomic_t inotify_watches; /* How many inotify watches does this user have? */
745 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
747 #ifdef CONFIG_FANOTIFY
748 atomic_t fanotify_listeners;
751 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
753 #ifdef CONFIG_POSIX_MQUEUE
754 /* protected by mq_lock */
755 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
757 unsigned long locked_shm; /* How many pages of mlocked shm ? */
760 struct key *uid_keyring; /* UID specific keyring */
761 struct key *session_keyring; /* UID's default session keyring */
764 /* Hash table maintenance information */
765 struct hlist_node uidhash_node;
768 #ifdef CONFIG_PERF_EVENTS
769 atomic_long_t locked_vm;
773 extern int uids_sysfs_init(void);
775 extern struct user_struct *find_user(kuid_t);
777 extern struct user_struct root_user;
778 #define INIT_USER (&root_user)
781 struct backing_dev_info;
782 struct reclaim_state;
784 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
786 /* cumulative counters */
787 unsigned long pcount; /* # of times run on this cpu */
788 unsigned long long run_delay; /* time spent waiting on a runqueue */
791 unsigned long long last_arrival,/* when we last ran on a cpu */
792 last_queued; /* when we were last queued to run */
794 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
796 #ifdef CONFIG_TASK_DELAY_ACCT
797 struct task_delay_info {
799 unsigned int flags; /* Private per-task flags */
801 /* For each stat XXX, add following, aligned appropriately
803 * struct timespec XXX_start, XXX_end;
807 * Atomicity of updates to XXX_delay, XXX_count protected by
808 * single lock above (split into XXX_lock if contention is an issue).
812 * XXX_count is incremented on every XXX operation, the delay
813 * associated with the operation is added to XXX_delay.
814 * XXX_delay contains the accumulated delay time in nanoseconds.
816 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
817 u64 blkio_delay; /* wait for sync block io completion */
818 u64 swapin_delay; /* wait for swapin block io completion */
819 u32 blkio_count; /* total count of the number of sync block */
820 /* io operations performed */
821 u32 swapin_count; /* total count of the number of swapin block */
822 /* io operations performed */
824 struct timespec freepages_start, freepages_end;
825 u64 freepages_delay; /* wait for memory reclaim */
826 u32 freepages_count; /* total count of memory reclaim */
828 #endif /* CONFIG_TASK_DELAY_ACCT */
830 static inline int sched_info_on(void)
832 #ifdef CONFIG_SCHEDSTATS
834 #elif defined(CONFIG_TASK_DELAY_ACCT)
835 extern int delayacct_on;
850 * Increase resolution of cpu_power calculations
852 #define SCHED_POWER_SHIFT 10
853 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
856 * sched-domains (multiprocessor balancing) declarations:
859 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
860 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
861 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
862 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
863 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
864 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
865 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
866 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
867 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
868 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
869 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
870 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
871 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
872 #define SD_NUMA 0x4000 /* cross-node balancing */
874 #ifdef CONFIG_SCHED_SMT
875 static inline const int cpu_smt_flags(void)
877 return SD_SHARE_CPUPOWER | SD_SHARE_PKG_RESOURCES;
881 #ifdef CONFIG_SCHED_MC
882 static inline const int cpu_core_flags(void)
884 return SD_SHARE_PKG_RESOURCES;
889 static inline const int cpu_numa_flags(void)
895 struct sched_domain_attr {
896 int relax_domain_level;
899 #define SD_ATTR_INIT (struct sched_domain_attr) { \
900 .relax_domain_level = -1, \
903 extern int sched_domain_level_max;
907 struct sched_domain {
908 /* These fields must be setup */
909 struct sched_domain *parent; /* top domain must be null terminated */
910 struct sched_domain *child; /* bottom domain must be null terminated */
911 struct sched_group *groups; /* the balancing groups of the domain */
912 unsigned long min_interval; /* Minimum balance interval ms */
913 unsigned long max_interval; /* Maximum balance interval ms */
914 unsigned int busy_factor; /* less balancing by factor if busy */
915 unsigned int imbalance_pct; /* No balance until over watermark */
916 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
917 unsigned int busy_idx;
918 unsigned int idle_idx;
919 unsigned int newidle_idx;
920 unsigned int wake_idx;
921 unsigned int forkexec_idx;
922 unsigned int smt_gain;
924 int nohz_idle; /* NOHZ IDLE status */
925 int flags; /* See SD_* */
928 /* Runtime fields. */
929 unsigned long last_balance; /* init to jiffies. units in jiffies */
930 unsigned int balance_interval; /* initialise to 1. units in ms. */
931 unsigned int nr_balance_failed; /* initialise to 0 */
933 /* idle_balance() stats */
934 u64 max_newidle_lb_cost;
935 unsigned long next_decay_max_lb_cost;
937 #ifdef CONFIG_SCHEDSTATS
938 /* load_balance() stats */
939 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
940 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
948 /* Active load balancing */
949 unsigned int alb_count;
950 unsigned int alb_failed;
951 unsigned int alb_pushed;
953 /* SD_BALANCE_EXEC stats */
954 unsigned int sbe_count;
955 unsigned int sbe_balanced;
956 unsigned int sbe_pushed;
958 /* SD_BALANCE_FORK stats */
959 unsigned int sbf_count;
960 unsigned int sbf_balanced;
961 unsigned int sbf_pushed;
963 /* try_to_wake_up() stats */
964 unsigned int ttwu_wake_remote;
965 unsigned int ttwu_move_affine;
966 unsigned int ttwu_move_balance;
968 #ifdef CONFIG_SCHED_DEBUG
972 void *private; /* used during construction */
973 struct rcu_head rcu; /* used during destruction */
976 unsigned int span_weight;
978 * Span of all CPUs in this domain.
980 * NOTE: this field is variable length. (Allocated dynamically
981 * by attaching extra space to the end of the structure,
982 * depending on how many CPUs the kernel has booted up with)
984 unsigned long span[0];
987 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
989 return to_cpumask(sd->span);
992 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
993 struct sched_domain_attr *dattr_new);
995 /* Allocate an array of sched domains, for partition_sched_domains(). */
996 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
997 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
999 bool cpus_share_cache(int this_cpu, int that_cpu);
1001 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1002 typedef const int (*sched_domain_flags_f)(void);
1004 #define SDTL_OVERLAP 0x01
1007 struct sched_domain **__percpu sd;
1008 struct sched_group **__percpu sg;
1009 struct sched_group_power **__percpu sgp;
1012 struct sched_domain_topology_level {
1013 sched_domain_mask_f mask;
1014 sched_domain_flags_f sd_flags;
1017 struct sd_data data;
1018 #ifdef CONFIG_SCHED_DEBUG
1023 extern struct sched_domain_topology_level *sched_domain_topology;
1025 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1027 #ifdef CONFIG_SCHED_DEBUG
1028 # define SD_INIT_NAME(type) .name = #type
1030 # define SD_INIT_NAME(type)
1033 #else /* CONFIG_SMP */
1035 struct sched_domain_attr;
1038 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1039 struct sched_domain_attr *dattr_new)
1043 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1048 #endif /* !CONFIG_SMP */
1051 struct io_context; /* See blkdev.h */
1054 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1055 extern void prefetch_stack(struct task_struct *t);
1057 static inline void prefetch_stack(struct task_struct *t) { }
1060 struct audit_context; /* See audit.c */
1062 struct pipe_inode_info;
1063 struct uts_namespace;
1065 struct load_weight {
1066 unsigned long weight;
1072 * These sums represent an infinite geometric series and so are bound
1073 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1074 * choices of y < 1-2^(-32)*1024.
1076 u32 runnable_avg_sum, runnable_avg_period;
1077 u64 last_runnable_update;
1079 unsigned long load_avg_contrib;
1082 #ifdef CONFIG_SCHEDSTATS
1083 struct sched_statistics {
1093 s64 sum_sleep_runtime;
1100 u64 nr_migrations_cold;
1101 u64 nr_failed_migrations_affine;
1102 u64 nr_failed_migrations_running;
1103 u64 nr_failed_migrations_hot;
1104 u64 nr_forced_migrations;
1107 u64 nr_wakeups_sync;
1108 u64 nr_wakeups_migrate;
1109 u64 nr_wakeups_local;
1110 u64 nr_wakeups_remote;
1111 u64 nr_wakeups_affine;
1112 u64 nr_wakeups_affine_attempts;
1113 u64 nr_wakeups_passive;
1114 u64 nr_wakeups_idle;
1118 struct sched_entity {
1119 struct load_weight load; /* for load-balancing */
1120 struct rb_node run_node;
1121 struct list_head group_node;
1125 u64 sum_exec_runtime;
1127 u64 prev_sum_exec_runtime;
1131 #ifdef CONFIG_SCHEDSTATS
1132 struct sched_statistics statistics;
1135 #ifdef CONFIG_FAIR_GROUP_SCHED
1137 struct sched_entity *parent;
1138 /* rq on which this entity is (to be) queued: */
1139 struct cfs_rq *cfs_rq;
1140 /* rq "owned" by this entity/group: */
1141 struct cfs_rq *my_q;
1145 /* Per-entity load-tracking */
1146 struct sched_avg avg;
1150 struct sched_rt_entity {
1151 struct list_head run_list;
1152 unsigned long timeout;
1153 unsigned long watchdog_stamp;
1154 unsigned int time_slice;
1156 struct sched_rt_entity *back;
1157 #ifdef CONFIG_RT_GROUP_SCHED
1158 struct sched_rt_entity *parent;
1159 /* rq on which this entity is (to be) queued: */
1160 struct rt_rq *rt_rq;
1161 /* rq "owned" by this entity/group: */
1166 struct sched_dl_entity {
1167 struct rb_node rb_node;
1170 * Original scheduling parameters. Copied here from sched_attr
1171 * during sched_setattr(), they will remain the same until
1172 * the next sched_setattr().
1174 u64 dl_runtime; /* maximum runtime for each instance */
1175 u64 dl_deadline; /* relative deadline of each instance */
1176 u64 dl_period; /* separation of two instances (period) */
1177 u64 dl_bw; /* dl_runtime / dl_deadline */
1180 * Actual scheduling parameters. Initialized with the values above,
1181 * they are continously updated during task execution. Note that
1182 * the remaining runtime could be < 0 in case we are in overrun.
1184 s64 runtime; /* remaining runtime for this instance */
1185 u64 deadline; /* absolute deadline for this instance */
1186 unsigned int flags; /* specifying the scheduler behaviour */
1191 * @dl_throttled tells if we exhausted the runtime. If so, the
1192 * task has to wait for a replenishment to be performed at the
1193 * next firing of dl_timer.
1195 * @dl_new tells if a new instance arrived. If so we must
1196 * start executing it with full runtime and reset its absolute
1199 * @dl_boosted tells if we are boosted due to DI. If so we are
1200 * outside bandwidth enforcement mechanism (but only until we
1201 * exit the critical section);
1203 * @dl_yielded tells if task gave up the cpu before consuming
1204 * all its available runtime during the last job.
1206 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1209 * Bandwidth enforcement timer. Each -deadline task has its
1210 * own bandwidth to be enforced, thus we need one timer per task.
1212 struct hrtimer dl_timer;
1217 enum perf_event_task_context {
1218 perf_invalid_context = -1,
1219 perf_hw_context = 0,
1221 perf_nr_task_contexts,
1224 struct task_struct {
1225 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1228 unsigned int flags; /* per process flags, defined below */
1229 unsigned int ptrace;
1232 struct llist_node wake_entry;
1234 struct task_struct *last_wakee;
1235 unsigned long wakee_flips;
1236 unsigned long wakee_flip_decay_ts;
1242 int prio, static_prio, normal_prio;
1243 unsigned int rt_priority;
1244 const struct sched_class *sched_class;
1245 struct sched_entity se;
1246 struct sched_rt_entity rt;
1247 #ifdef CONFIG_CGROUP_SCHED
1248 struct task_group *sched_task_group;
1250 struct sched_dl_entity dl;
1252 #ifdef CONFIG_PREEMPT_NOTIFIERS
1253 /* list of struct preempt_notifier: */
1254 struct hlist_head preempt_notifiers;
1257 #ifdef CONFIG_BLK_DEV_IO_TRACE
1258 unsigned int btrace_seq;
1261 unsigned int policy;
1262 int nr_cpus_allowed;
1263 cpumask_t cpus_allowed;
1265 #ifdef CONFIG_PREEMPT_RCU
1266 int rcu_read_lock_nesting;
1267 char rcu_read_unlock_special;
1268 struct list_head rcu_node_entry;
1269 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1270 #ifdef CONFIG_TREE_PREEMPT_RCU
1271 struct rcu_node *rcu_blocked_node;
1272 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1273 #ifdef CONFIG_RCU_BOOST
1274 struct rt_mutex *rcu_boost_mutex;
1275 #endif /* #ifdef CONFIG_RCU_BOOST */
1277 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1278 struct sched_info sched_info;
1281 struct list_head tasks;
1283 struct plist_node pushable_tasks;
1284 struct rb_node pushable_dl_tasks;
1287 struct mm_struct *mm, *active_mm;
1288 #ifdef CONFIG_COMPAT_BRK
1289 unsigned brk_randomized:1;
1291 /* per-thread vma caching */
1292 u32 vmacache_seqnum;
1293 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1294 #if defined(SPLIT_RSS_COUNTING)
1295 struct task_rss_stat rss_stat;
1299 int exit_code, exit_signal;
1300 int pdeath_signal; /* The signal sent when the parent dies */
1301 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1303 /* Used for emulating ABI behavior of previous Linux versions */
1304 unsigned int personality;
1306 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1308 unsigned in_iowait:1;
1310 /* task may not gain privileges */
1311 unsigned no_new_privs:1;
1313 /* Revert to default priority/policy when forking */
1314 unsigned sched_reset_on_fork:1;
1315 unsigned sched_contributes_to_load:1;
1320 #ifdef CONFIG_CC_STACKPROTECTOR
1321 /* Canary value for the -fstack-protector gcc feature */
1322 unsigned long stack_canary;
1325 * pointers to (original) parent process, youngest child, younger sibling,
1326 * older sibling, respectively. (p->father can be replaced with
1327 * p->real_parent->pid)
1329 struct task_struct __rcu *real_parent; /* real parent process */
1330 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1332 * children/sibling forms the list of my natural children
1334 struct list_head children; /* list of my children */
1335 struct list_head sibling; /* linkage in my parent's children list */
1336 struct task_struct *group_leader; /* threadgroup leader */
1339 * ptraced is the list of tasks this task is using ptrace on.
1340 * This includes both natural children and PTRACE_ATTACH targets.
1341 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1343 struct list_head ptraced;
1344 struct list_head ptrace_entry;
1346 /* PID/PID hash table linkage. */
1347 struct pid_link pids[PIDTYPE_MAX];
1348 struct list_head thread_group;
1349 struct list_head thread_node;
1351 struct completion *vfork_done; /* for vfork() */
1352 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1353 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1355 cputime_t utime, stime, utimescaled, stimescaled;
1357 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1358 struct cputime prev_cputime;
1360 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1361 seqlock_t vtime_seqlock;
1362 unsigned long long vtime_snap;
1367 } vtime_snap_whence;
1369 unsigned long nvcsw, nivcsw; /* context switch counts */
1370 struct timespec start_time; /* monotonic time */
1371 struct timespec real_start_time; /* boot based time */
1372 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1373 unsigned long min_flt, maj_flt;
1375 struct task_cputime cputime_expires;
1376 struct list_head cpu_timers[3];
1378 /* process credentials */
1379 const struct cred __rcu *real_cred; /* objective and real subjective task
1380 * credentials (COW) */
1381 const struct cred __rcu *cred; /* effective (overridable) subjective task
1382 * credentials (COW) */
1383 char comm[TASK_COMM_LEN]; /* executable name excluding path
1384 - access with [gs]et_task_comm (which lock
1385 it with task_lock())
1386 - initialized normally by setup_new_exec */
1387 /* file system info */
1388 int link_count, total_link_count;
1389 #ifdef CONFIG_SYSVIPC
1391 struct sysv_sem sysvsem;
1393 #ifdef CONFIG_DETECT_HUNG_TASK
1394 /* hung task detection */
1395 unsigned long last_switch_count;
1397 /* CPU-specific state of this task */
1398 struct thread_struct thread;
1399 /* filesystem information */
1400 struct fs_struct *fs;
1401 /* open file information */
1402 struct files_struct *files;
1404 struct nsproxy *nsproxy;
1405 /* signal handlers */
1406 struct signal_struct *signal;
1407 struct sighand_struct *sighand;
1409 sigset_t blocked, real_blocked;
1410 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1411 struct sigpending pending;
1413 unsigned long sas_ss_sp;
1415 int (*notifier)(void *priv);
1416 void *notifier_data;
1417 sigset_t *notifier_mask;
1418 struct callback_head *task_works;
1420 struct audit_context *audit_context;
1421 #ifdef CONFIG_AUDITSYSCALL
1423 unsigned int sessionid;
1425 struct seccomp seccomp;
1427 /* Thread group tracking */
1430 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1432 spinlock_t alloc_lock;
1434 /* Protection of the PI data structures: */
1435 raw_spinlock_t pi_lock;
1437 #ifdef CONFIG_RT_MUTEXES
1438 /* PI waiters blocked on a rt_mutex held by this task */
1439 struct rb_root pi_waiters;
1440 struct rb_node *pi_waiters_leftmost;
1441 /* Deadlock detection and priority inheritance handling */
1442 struct rt_mutex_waiter *pi_blocked_on;
1443 /* Top pi_waiters task */
1444 struct task_struct *pi_top_task;
1447 #ifdef CONFIG_DEBUG_MUTEXES
1448 /* mutex deadlock detection */
1449 struct mutex_waiter *blocked_on;
1451 #ifdef CONFIG_TRACE_IRQFLAGS
1452 unsigned int irq_events;
1453 unsigned long hardirq_enable_ip;
1454 unsigned long hardirq_disable_ip;
1455 unsigned int hardirq_enable_event;
1456 unsigned int hardirq_disable_event;
1457 int hardirqs_enabled;
1458 int hardirq_context;
1459 unsigned long softirq_disable_ip;
1460 unsigned long softirq_enable_ip;
1461 unsigned int softirq_disable_event;
1462 unsigned int softirq_enable_event;
1463 int softirqs_enabled;
1464 int softirq_context;
1466 #ifdef CONFIG_LOCKDEP
1467 # define MAX_LOCK_DEPTH 48UL
1470 unsigned int lockdep_recursion;
1471 struct held_lock held_locks[MAX_LOCK_DEPTH];
1472 gfp_t lockdep_reclaim_gfp;
1475 /* journalling filesystem info */
1478 /* stacked block device info */
1479 struct bio_list *bio_list;
1482 /* stack plugging */
1483 struct blk_plug *plug;
1487 struct reclaim_state *reclaim_state;
1489 struct backing_dev_info *backing_dev_info;
1491 struct io_context *io_context;
1493 unsigned long ptrace_message;
1494 siginfo_t *last_siginfo; /* For ptrace use. */
1495 struct task_io_accounting ioac;
1496 #if defined(CONFIG_TASK_XACCT)
1497 u64 acct_rss_mem1; /* accumulated rss usage */
1498 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1499 cputime_t acct_timexpd; /* stime + utime since last update */
1501 #ifdef CONFIG_CPUSETS
1502 nodemask_t mems_allowed; /* Protected by alloc_lock */
1503 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1504 int cpuset_mem_spread_rotor;
1505 int cpuset_slab_spread_rotor;
1507 #ifdef CONFIG_CGROUPS
1508 /* Control Group info protected by css_set_lock */
1509 struct css_set __rcu *cgroups;
1510 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1511 struct list_head cg_list;
1514 struct robust_list_head __user *robust_list;
1515 #ifdef CONFIG_COMPAT
1516 struct compat_robust_list_head __user *compat_robust_list;
1518 struct list_head pi_state_list;
1519 struct futex_pi_state *pi_state_cache;
1521 #ifdef CONFIG_PERF_EVENTS
1522 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1523 struct mutex perf_event_mutex;
1524 struct list_head perf_event_list;
1526 #ifdef CONFIG_DEBUG_PREEMPT
1527 unsigned long preempt_disable_ip;
1530 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1532 short pref_node_fork;
1534 #ifdef CONFIG_NUMA_BALANCING
1536 unsigned int numa_scan_period;
1537 unsigned int numa_scan_period_max;
1538 int numa_preferred_nid;
1539 unsigned long numa_migrate_retry;
1540 u64 node_stamp; /* migration stamp */
1541 u64 last_task_numa_placement;
1542 u64 last_sum_exec_runtime;
1543 struct callback_head numa_work;
1545 struct list_head numa_entry;
1546 struct numa_group *numa_group;
1549 * Exponential decaying average of faults on a per-node basis.
1550 * Scheduling placement decisions are made based on the these counts.
1551 * The values remain static for the duration of a PTE scan
1553 unsigned long *numa_faults_memory;
1554 unsigned long total_numa_faults;
1557 * numa_faults_buffer records faults per node during the current
1558 * scan window. When the scan completes, the counts in
1559 * numa_faults_memory decay and these values are copied.
1561 unsigned long *numa_faults_buffer_memory;
1564 * Track the nodes the process was running on when a NUMA hinting
1565 * fault was incurred.
1567 unsigned long *numa_faults_cpu;
1568 unsigned long *numa_faults_buffer_cpu;
1571 * numa_faults_locality tracks if faults recorded during the last
1572 * scan window were remote/local. The task scan period is adapted
1573 * based on the locality of the faults with different weights
1574 * depending on whether they were shared or private faults
1576 unsigned long numa_faults_locality[2];
1578 unsigned long numa_pages_migrated;
1579 #endif /* CONFIG_NUMA_BALANCING */
1581 struct rcu_head rcu;
1584 * cache last used pipe for splice
1586 struct pipe_inode_info *splice_pipe;
1588 struct page_frag task_frag;
1590 #ifdef CONFIG_TASK_DELAY_ACCT
1591 struct task_delay_info *delays;
1593 #ifdef CONFIG_FAULT_INJECTION
1597 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1598 * balance_dirty_pages() for some dirty throttling pause
1601 int nr_dirtied_pause;
1602 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1604 #ifdef CONFIG_LATENCYTOP
1605 int latency_record_count;
1606 struct latency_record latency_record[LT_SAVECOUNT];
1609 * time slack values; these are used to round up poll() and
1610 * select() etc timeout values. These are in nanoseconds.
1612 unsigned long timer_slack_ns;
1613 unsigned long default_timer_slack_ns;
1615 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1616 /* Index of current stored address in ret_stack */
1618 /* Stack of return addresses for return function tracing */
1619 struct ftrace_ret_stack *ret_stack;
1620 /* time stamp for last schedule */
1621 unsigned long long ftrace_timestamp;
1623 * Number of functions that haven't been traced
1624 * because of depth overrun.
1626 atomic_t trace_overrun;
1627 /* Pause for the tracing */
1628 atomic_t tracing_graph_pause;
1630 #ifdef CONFIG_TRACING
1631 /* state flags for use by tracers */
1632 unsigned long trace;
1633 /* bitmask and counter of trace recursion */
1634 unsigned long trace_recursion;
1635 #endif /* CONFIG_TRACING */
1636 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1637 struct memcg_batch_info {
1638 int do_batch; /* incremented when batch uncharge started */
1639 struct mem_cgroup *memcg; /* target memcg of uncharge */
1640 unsigned long nr_pages; /* uncharged usage */
1641 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1643 unsigned int memcg_kmem_skip_account;
1644 struct memcg_oom_info {
1645 struct mem_cgroup *memcg;
1648 unsigned int may_oom:1;
1651 #ifdef CONFIG_UPROBES
1652 struct uprobe_task *utask;
1654 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1655 unsigned int sequential_io;
1656 unsigned int sequential_io_avg;
1660 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1661 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1663 #define TNF_MIGRATED 0x01
1664 #define TNF_NO_GROUP 0x02
1665 #define TNF_SHARED 0x04
1666 #define TNF_FAULT_LOCAL 0x08
1668 #ifdef CONFIG_NUMA_BALANCING
1669 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1670 extern pid_t task_numa_group_id(struct task_struct *p);
1671 extern void set_numabalancing_state(bool enabled);
1672 extern void task_numa_free(struct task_struct *p);
1673 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1674 int src_nid, int dst_cpu);
1676 static inline void task_numa_fault(int last_node, int node, int pages,
1680 static inline pid_t task_numa_group_id(struct task_struct *p)
1684 static inline void set_numabalancing_state(bool enabled)
1687 static inline void task_numa_free(struct task_struct *p)
1690 static inline bool should_numa_migrate_memory(struct task_struct *p,
1691 struct page *page, int src_nid, int dst_cpu)
1697 static inline struct pid *task_pid(struct task_struct *task)
1699 return task->pids[PIDTYPE_PID].pid;
1702 static inline struct pid *task_tgid(struct task_struct *task)
1704 return task->group_leader->pids[PIDTYPE_PID].pid;
1708 * Without tasklist or rcu lock it is not safe to dereference
1709 * the result of task_pgrp/task_session even if task == current,
1710 * we can race with another thread doing sys_setsid/sys_setpgid.
1712 static inline struct pid *task_pgrp(struct task_struct *task)
1714 return task->group_leader->pids[PIDTYPE_PGID].pid;
1717 static inline struct pid *task_session(struct task_struct *task)
1719 return task->group_leader->pids[PIDTYPE_SID].pid;
1722 struct pid_namespace;
1725 * the helpers to get the task's different pids as they are seen
1726 * from various namespaces
1728 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1729 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1731 * task_xid_nr_ns() : id seen from the ns specified;
1733 * set_task_vxid() : assigns a virtual id to a task;
1735 * see also pid_nr() etc in include/linux/pid.h
1737 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1738 struct pid_namespace *ns);
1740 static inline pid_t task_pid_nr(struct task_struct *tsk)
1745 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1746 struct pid_namespace *ns)
1748 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1751 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1753 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1757 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1762 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1764 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1766 return pid_vnr(task_tgid(tsk));
1770 static inline int pid_alive(const struct task_struct *p);
1771 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1777 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1783 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1785 return task_ppid_nr_ns(tsk, &init_pid_ns);
1788 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1789 struct pid_namespace *ns)
1791 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1794 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1796 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1800 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1801 struct pid_namespace *ns)
1803 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1806 static inline pid_t task_session_vnr(struct task_struct *tsk)
1808 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1811 /* obsolete, do not use */
1812 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1814 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1818 * pid_alive - check that a task structure is not stale
1819 * @p: Task structure to be checked.
1821 * Test if a process is not yet dead (at most zombie state)
1822 * If pid_alive fails, then pointers within the task structure
1823 * can be stale and must not be dereferenced.
1825 * Return: 1 if the process is alive. 0 otherwise.
1827 static inline int pid_alive(const struct task_struct *p)
1829 return p->pids[PIDTYPE_PID].pid != NULL;
1833 * is_global_init - check if a task structure is init
1834 * @tsk: Task structure to be checked.
1836 * Check if a task structure is the first user space task the kernel created.
1838 * Return: 1 if the task structure is init. 0 otherwise.
1840 static inline int is_global_init(struct task_struct *tsk)
1842 return tsk->pid == 1;
1845 extern struct pid *cad_pid;
1847 extern void free_task(struct task_struct *tsk);
1848 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1850 extern void __put_task_struct(struct task_struct *t);
1852 static inline void put_task_struct(struct task_struct *t)
1854 if (atomic_dec_and_test(&t->usage))
1855 __put_task_struct(t);
1858 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1859 extern void task_cputime(struct task_struct *t,
1860 cputime_t *utime, cputime_t *stime);
1861 extern void task_cputime_scaled(struct task_struct *t,
1862 cputime_t *utimescaled, cputime_t *stimescaled);
1863 extern cputime_t task_gtime(struct task_struct *t);
1865 static inline void task_cputime(struct task_struct *t,
1866 cputime_t *utime, cputime_t *stime)
1874 static inline void task_cputime_scaled(struct task_struct *t,
1875 cputime_t *utimescaled,
1876 cputime_t *stimescaled)
1879 *utimescaled = t->utimescaled;
1881 *stimescaled = t->stimescaled;
1884 static inline cputime_t task_gtime(struct task_struct *t)
1889 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1890 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1895 #define PF_EXITING 0x00000004 /* getting shut down */
1896 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1897 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1898 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1899 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1900 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1901 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1902 #define PF_DUMPCORE 0x00000200 /* dumped core */
1903 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1904 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1905 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1906 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1907 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1908 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1909 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1910 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1911 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1912 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1913 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1914 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1915 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1916 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1917 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1918 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1919 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1920 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1921 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1922 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1923 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1926 * Only the _current_ task can read/write to tsk->flags, but other
1927 * tasks can access tsk->flags in readonly mode for example
1928 * with tsk_used_math (like during threaded core dumping).
1929 * There is however an exception to this rule during ptrace
1930 * or during fork: the ptracer task is allowed to write to the
1931 * child->flags of its traced child (same goes for fork, the parent
1932 * can write to the child->flags), because we're guaranteed the
1933 * child is not running and in turn not changing child->flags
1934 * at the same time the parent does it.
1936 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1937 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1938 #define clear_used_math() clear_stopped_child_used_math(current)
1939 #define set_used_math() set_stopped_child_used_math(current)
1940 #define conditional_stopped_child_used_math(condition, child) \
1941 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1942 #define conditional_used_math(condition) \
1943 conditional_stopped_child_used_math(condition, current)
1944 #define copy_to_stopped_child_used_math(child) \
1945 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1946 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1947 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1948 #define used_math() tsk_used_math(current)
1950 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1951 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1953 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1958 static inline unsigned int memalloc_noio_save(void)
1960 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1961 current->flags |= PF_MEMALLOC_NOIO;
1965 static inline void memalloc_noio_restore(unsigned int flags)
1967 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1971 * task->jobctl flags
1973 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1975 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1976 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1977 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1978 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1979 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1980 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1981 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1983 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1984 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1985 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1986 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1987 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1988 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1989 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1991 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1992 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1994 extern bool task_set_jobctl_pending(struct task_struct *task,
1996 extern void task_clear_jobctl_trapping(struct task_struct *task);
1997 extern void task_clear_jobctl_pending(struct task_struct *task,
2000 #ifdef CONFIG_PREEMPT_RCU
2002 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
2003 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
2005 static inline void rcu_copy_process(struct task_struct *p)
2007 p->rcu_read_lock_nesting = 0;
2008 p->rcu_read_unlock_special = 0;
2009 #ifdef CONFIG_TREE_PREEMPT_RCU
2010 p->rcu_blocked_node = NULL;
2011 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
2012 #ifdef CONFIG_RCU_BOOST
2013 p->rcu_boost_mutex = NULL;
2014 #endif /* #ifdef CONFIG_RCU_BOOST */
2015 INIT_LIST_HEAD(&p->rcu_node_entry);
2020 static inline void rcu_copy_process(struct task_struct *p)
2026 static inline void tsk_restore_flags(struct task_struct *task,
2027 unsigned long orig_flags, unsigned long flags)
2029 task->flags &= ~flags;
2030 task->flags |= orig_flags & flags;
2034 extern void do_set_cpus_allowed(struct task_struct *p,
2035 const struct cpumask *new_mask);
2037 extern int set_cpus_allowed_ptr(struct task_struct *p,
2038 const struct cpumask *new_mask);
2040 static inline void do_set_cpus_allowed(struct task_struct *p,
2041 const struct cpumask *new_mask)
2044 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2045 const struct cpumask *new_mask)
2047 if (!cpumask_test_cpu(0, new_mask))
2053 #ifdef CONFIG_NO_HZ_COMMON
2054 void calc_load_enter_idle(void);
2055 void calc_load_exit_idle(void);
2057 static inline void calc_load_enter_idle(void) { }
2058 static inline void calc_load_exit_idle(void) { }
2059 #endif /* CONFIG_NO_HZ_COMMON */
2061 #ifndef CONFIG_CPUMASK_OFFSTACK
2062 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2064 return set_cpus_allowed_ptr(p, &new_mask);
2069 * Do not use outside of architecture code which knows its limitations.
2071 * sched_clock() has no promise of monotonicity or bounded drift between
2072 * CPUs, use (which you should not) requires disabling IRQs.
2074 * Please use one of the three interfaces below.
2076 extern unsigned long long notrace sched_clock(void);
2078 * See the comment in kernel/sched/clock.c
2080 extern u64 cpu_clock(int cpu);
2081 extern u64 local_clock(void);
2082 extern u64 sched_clock_cpu(int cpu);
2085 extern void sched_clock_init(void);
2087 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2088 static inline void sched_clock_tick(void)
2092 static inline void sched_clock_idle_sleep_event(void)
2096 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2101 * Architectures can set this to 1 if they have specified
2102 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2103 * but then during bootup it turns out that sched_clock()
2104 * is reliable after all:
2106 extern int sched_clock_stable(void);
2107 extern void set_sched_clock_stable(void);
2108 extern void clear_sched_clock_stable(void);
2110 extern void sched_clock_tick(void);
2111 extern void sched_clock_idle_sleep_event(void);
2112 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2115 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2117 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2118 * The reason for this explicit opt-in is not to have perf penalty with
2119 * slow sched_clocks.
2121 extern void enable_sched_clock_irqtime(void);
2122 extern void disable_sched_clock_irqtime(void);
2124 static inline void enable_sched_clock_irqtime(void) {}
2125 static inline void disable_sched_clock_irqtime(void) {}
2128 extern unsigned long long
2129 task_sched_runtime(struct task_struct *task);
2131 /* sched_exec is called by processes performing an exec */
2133 extern void sched_exec(void);
2135 #define sched_exec() {}
2138 extern void sched_clock_idle_sleep_event(void);
2139 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2141 #ifdef CONFIG_HOTPLUG_CPU
2142 extern void idle_task_exit(void);
2144 static inline void idle_task_exit(void) {}
2147 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2148 extern void wake_up_nohz_cpu(int cpu);
2150 static inline void wake_up_nohz_cpu(int cpu) { }
2153 #ifdef CONFIG_NO_HZ_FULL
2154 extern bool sched_can_stop_tick(void);
2155 extern u64 scheduler_tick_max_deferment(void);
2157 static inline bool sched_can_stop_tick(void) { return false; }
2160 #ifdef CONFIG_SCHED_AUTOGROUP
2161 extern void sched_autogroup_create_attach(struct task_struct *p);
2162 extern void sched_autogroup_detach(struct task_struct *p);
2163 extern void sched_autogroup_fork(struct signal_struct *sig);
2164 extern void sched_autogroup_exit(struct signal_struct *sig);
2165 #ifdef CONFIG_PROC_FS
2166 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2167 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2170 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2171 static inline void sched_autogroup_detach(struct task_struct *p) { }
2172 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2173 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2176 extern bool yield_to(struct task_struct *p, bool preempt);
2177 extern void set_user_nice(struct task_struct *p, long nice);
2178 extern int task_prio(const struct task_struct *p);
2180 * task_nice - return the nice value of a given task.
2181 * @p: the task in question.
2183 * Return: The nice value [ -20 ... 0 ... 19 ].
2185 static inline int task_nice(const struct task_struct *p)
2187 return PRIO_TO_NICE((p)->static_prio);
2189 extern int can_nice(const struct task_struct *p, const int nice);
2190 extern int task_curr(const struct task_struct *p);
2191 extern int idle_cpu(int cpu);
2192 extern int sched_setscheduler(struct task_struct *, int,
2193 const struct sched_param *);
2194 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2195 const struct sched_param *);
2196 extern int sched_setattr(struct task_struct *,
2197 const struct sched_attr *);
2198 extern struct task_struct *idle_task(int cpu);
2200 * is_idle_task - is the specified task an idle task?
2201 * @p: the task in question.
2203 * Return: 1 if @p is an idle task. 0 otherwise.
2205 static inline bool is_idle_task(const struct task_struct *p)
2209 extern struct task_struct *curr_task(int cpu);
2210 extern void set_curr_task(int cpu, struct task_struct *p);
2215 * The default (Linux) execution domain.
2217 extern struct exec_domain default_exec_domain;
2219 union thread_union {
2220 struct thread_info thread_info;
2221 unsigned long stack[THREAD_SIZE/sizeof(long)];
2224 #ifndef __HAVE_ARCH_KSTACK_END
2225 static inline int kstack_end(void *addr)
2227 /* Reliable end of stack detection:
2228 * Some APM bios versions misalign the stack
2230 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2234 extern union thread_union init_thread_union;
2235 extern struct task_struct init_task;
2237 extern struct mm_struct init_mm;
2239 extern struct pid_namespace init_pid_ns;
2242 * find a task by one of its numerical ids
2244 * find_task_by_pid_ns():
2245 * finds a task by its pid in the specified namespace
2246 * find_task_by_vpid():
2247 * finds a task by its virtual pid
2249 * see also find_vpid() etc in include/linux/pid.h
2252 extern struct task_struct *find_task_by_vpid(pid_t nr);
2253 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2254 struct pid_namespace *ns);
2256 /* per-UID process charging. */
2257 extern struct user_struct * alloc_uid(kuid_t);
2258 static inline struct user_struct *get_uid(struct user_struct *u)
2260 atomic_inc(&u->__count);
2263 extern void free_uid(struct user_struct *);
2265 #include <asm/current.h>
2267 extern void xtime_update(unsigned long ticks);
2269 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2270 extern int wake_up_process(struct task_struct *tsk);
2271 extern void wake_up_new_task(struct task_struct *tsk);
2273 extern void kick_process(struct task_struct *tsk);
2275 static inline void kick_process(struct task_struct *tsk) { }
2277 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2278 extern void sched_dead(struct task_struct *p);
2280 extern void proc_caches_init(void);
2281 extern void flush_signals(struct task_struct *);
2282 extern void __flush_signals(struct task_struct *);
2283 extern void ignore_signals(struct task_struct *);
2284 extern void flush_signal_handlers(struct task_struct *, int force_default);
2285 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2287 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2289 unsigned long flags;
2292 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2293 ret = dequeue_signal(tsk, mask, info);
2294 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2299 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2301 extern void unblock_all_signals(void);
2302 extern void release_task(struct task_struct * p);
2303 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2304 extern int force_sigsegv(int, struct task_struct *);
2305 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2306 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2307 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2308 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2309 const struct cred *, u32);
2310 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2311 extern int kill_pid(struct pid *pid, int sig, int priv);
2312 extern int kill_proc_info(int, struct siginfo *, pid_t);
2313 extern __must_check bool do_notify_parent(struct task_struct *, int);
2314 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2315 extern void force_sig(int, struct task_struct *);
2316 extern int send_sig(int, struct task_struct *, int);
2317 extern int zap_other_threads(struct task_struct *p);
2318 extern struct sigqueue *sigqueue_alloc(void);
2319 extern void sigqueue_free(struct sigqueue *);
2320 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2321 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2323 static inline void restore_saved_sigmask(void)
2325 if (test_and_clear_restore_sigmask())
2326 __set_current_blocked(¤t->saved_sigmask);
2329 static inline sigset_t *sigmask_to_save(void)
2331 sigset_t *res = ¤t->blocked;
2332 if (unlikely(test_restore_sigmask()))
2333 res = ¤t->saved_sigmask;
2337 static inline int kill_cad_pid(int sig, int priv)
2339 return kill_pid(cad_pid, sig, priv);
2342 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2343 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2344 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2345 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2348 * True if we are on the alternate signal stack.
2350 static inline int on_sig_stack(unsigned long sp)
2352 #ifdef CONFIG_STACK_GROWSUP
2353 return sp >= current->sas_ss_sp &&
2354 sp - current->sas_ss_sp < current->sas_ss_size;
2356 return sp > current->sas_ss_sp &&
2357 sp - current->sas_ss_sp <= current->sas_ss_size;
2361 static inline int sas_ss_flags(unsigned long sp)
2363 return (current->sas_ss_size == 0 ? SS_DISABLE
2364 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2367 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2369 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2370 #ifdef CONFIG_STACK_GROWSUP
2371 return current->sas_ss_sp;
2373 return current->sas_ss_sp + current->sas_ss_size;
2379 * Routines for handling mm_structs
2381 extern struct mm_struct * mm_alloc(void);
2383 /* mmdrop drops the mm and the page tables */
2384 extern void __mmdrop(struct mm_struct *);
2385 static inline void mmdrop(struct mm_struct * mm)
2387 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2391 /* mmput gets rid of the mappings and all user-space */
2392 extern void mmput(struct mm_struct *);
2393 /* Grab a reference to a task's mm, if it is not already going away */
2394 extern struct mm_struct *get_task_mm(struct task_struct *task);
2396 * Grab a reference to a task's mm, if it is not already going away
2397 * and ptrace_may_access with the mode parameter passed to it
2400 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2401 /* Remove the current tasks stale references to the old mm_struct */
2402 extern void mm_release(struct task_struct *, struct mm_struct *);
2404 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2405 struct task_struct *);
2406 extern void flush_thread(void);
2407 extern void exit_thread(void);
2409 extern void exit_files(struct task_struct *);
2410 extern void __cleanup_sighand(struct sighand_struct *);
2412 extern void exit_itimers(struct signal_struct *);
2413 extern void flush_itimer_signals(void);
2415 extern void do_group_exit(int);
2417 extern int do_execve(struct filename *,
2418 const char __user * const __user *,
2419 const char __user * const __user *);
2420 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2421 struct task_struct *fork_idle(int);
2422 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2424 extern void set_task_comm(struct task_struct *tsk, const char *from);
2425 extern char *get_task_comm(char *to, struct task_struct *tsk);
2428 void scheduler_ipi(void);
2429 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2431 static inline void scheduler_ipi(void) { }
2432 static inline unsigned long wait_task_inactive(struct task_struct *p,
2439 #define next_task(p) \
2440 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2442 #define for_each_process(p) \
2443 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2445 extern bool current_is_single_threaded(void);
2448 * Careful: do_each_thread/while_each_thread is a double loop so
2449 * 'break' will not work as expected - use goto instead.
2451 #define do_each_thread(g, t) \
2452 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2454 #define while_each_thread(g, t) \
2455 while ((t = next_thread(t)) != g)
2457 #define __for_each_thread(signal, t) \
2458 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2460 #define for_each_thread(p, t) \
2461 __for_each_thread((p)->signal, t)
2463 /* Careful: this is a double loop, 'break' won't work as expected. */
2464 #define for_each_process_thread(p, t) \
2465 for_each_process(p) for_each_thread(p, t)
2467 static inline int get_nr_threads(struct task_struct *tsk)
2469 return tsk->signal->nr_threads;
2472 static inline bool thread_group_leader(struct task_struct *p)
2474 return p->exit_signal >= 0;
2477 /* Do to the insanities of de_thread it is possible for a process
2478 * to have the pid of the thread group leader without actually being
2479 * the thread group leader. For iteration through the pids in proc
2480 * all we care about is that we have a task with the appropriate
2481 * pid, we don't actually care if we have the right task.
2483 static inline bool has_group_leader_pid(struct task_struct *p)
2485 return task_pid(p) == p->signal->leader_pid;
2489 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2491 return p1->signal == p2->signal;
2494 static inline struct task_struct *next_thread(const struct task_struct *p)
2496 return list_entry_rcu(p->thread_group.next,
2497 struct task_struct, thread_group);
2500 static inline int thread_group_empty(struct task_struct *p)
2502 return list_empty(&p->thread_group);
2505 #define delay_group_leader(p) \
2506 (thread_group_leader(p) && !thread_group_empty(p))
2509 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2510 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2511 * pins the final release of task.io_context. Also protects ->cpuset and
2512 * ->cgroup.subsys[]. And ->vfork_done.
2514 * Nests both inside and outside of read_lock(&tasklist_lock).
2515 * It must not be nested with write_lock_irq(&tasklist_lock),
2516 * neither inside nor outside.
2518 static inline void task_lock(struct task_struct *p)
2520 spin_lock(&p->alloc_lock);
2523 static inline void task_unlock(struct task_struct *p)
2525 spin_unlock(&p->alloc_lock);
2528 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2529 unsigned long *flags);
2531 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2532 unsigned long *flags)
2534 struct sighand_struct *ret;
2536 ret = __lock_task_sighand(tsk, flags);
2537 (void)__cond_lock(&tsk->sighand->siglock, ret);
2541 static inline void unlock_task_sighand(struct task_struct *tsk,
2542 unsigned long *flags)
2544 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2547 #ifdef CONFIG_CGROUPS
2548 static inline void threadgroup_change_begin(struct task_struct *tsk)
2550 down_read(&tsk->signal->group_rwsem);
2552 static inline void threadgroup_change_end(struct task_struct *tsk)
2554 up_read(&tsk->signal->group_rwsem);
2558 * threadgroup_lock - lock threadgroup
2559 * @tsk: member task of the threadgroup to lock
2561 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2562 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2563 * change ->group_leader/pid. This is useful for cases where the threadgroup
2564 * needs to stay stable across blockable operations.
2566 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2567 * synchronization. While held, no new task will be added to threadgroup
2568 * and no existing live task will have its PF_EXITING set.
2570 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2571 * sub-thread becomes a new leader.
2573 static inline void threadgroup_lock(struct task_struct *tsk)
2575 down_write(&tsk->signal->group_rwsem);
2579 * threadgroup_unlock - unlock threadgroup
2580 * @tsk: member task of the threadgroup to unlock
2582 * Reverse threadgroup_lock().
2584 static inline void threadgroup_unlock(struct task_struct *tsk)
2586 up_write(&tsk->signal->group_rwsem);
2589 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2590 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2591 static inline void threadgroup_lock(struct task_struct *tsk) {}
2592 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2595 #ifndef __HAVE_THREAD_FUNCTIONS
2597 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2598 #define task_stack_page(task) ((task)->stack)
2600 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2602 *task_thread_info(p) = *task_thread_info(org);
2603 task_thread_info(p)->task = p;
2606 static inline unsigned long *end_of_stack(struct task_struct *p)
2608 return (unsigned long *)(task_thread_info(p) + 1);
2613 static inline int object_is_on_stack(void *obj)
2615 void *stack = task_stack_page(current);
2617 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2620 extern void thread_info_cache_init(void);
2622 #ifdef CONFIG_DEBUG_STACK_USAGE
2623 static inline unsigned long stack_not_used(struct task_struct *p)
2625 unsigned long *n = end_of_stack(p);
2627 do { /* Skip over canary */
2631 return (unsigned long)n - (unsigned long)end_of_stack(p);
2635 /* set thread flags in other task's structures
2636 * - see asm/thread_info.h for TIF_xxxx flags available
2638 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2640 set_ti_thread_flag(task_thread_info(tsk), flag);
2643 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2645 clear_ti_thread_flag(task_thread_info(tsk), flag);
2648 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2650 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2653 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2655 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2658 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2660 return test_ti_thread_flag(task_thread_info(tsk), flag);
2663 static inline void set_tsk_need_resched(struct task_struct *tsk)
2665 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2668 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2670 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2673 static inline int test_tsk_need_resched(struct task_struct *tsk)
2675 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2678 static inline int restart_syscall(void)
2680 set_tsk_thread_flag(current, TIF_SIGPENDING);
2681 return -ERESTARTNOINTR;
2684 static inline int signal_pending(struct task_struct *p)
2686 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2689 static inline int __fatal_signal_pending(struct task_struct *p)
2691 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2694 static inline int fatal_signal_pending(struct task_struct *p)
2696 return signal_pending(p) && __fatal_signal_pending(p);
2699 static inline int signal_pending_state(long state, struct task_struct *p)
2701 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2703 if (!signal_pending(p))
2706 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2710 * cond_resched() and cond_resched_lock(): latency reduction via
2711 * explicit rescheduling in places that are safe. The return
2712 * value indicates whether a reschedule was done in fact.
2713 * cond_resched_lock() will drop the spinlock before scheduling,
2714 * cond_resched_softirq() will enable bhs before scheduling.
2716 extern int _cond_resched(void);
2718 #define cond_resched() ({ \
2719 __might_sleep(__FILE__, __LINE__, 0); \
2723 extern int __cond_resched_lock(spinlock_t *lock);
2725 #ifdef CONFIG_PREEMPT_COUNT
2726 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2728 #define PREEMPT_LOCK_OFFSET 0
2731 #define cond_resched_lock(lock) ({ \
2732 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2733 __cond_resched_lock(lock); \
2736 extern int __cond_resched_softirq(void);
2738 #define cond_resched_softirq() ({ \
2739 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2740 __cond_resched_softirq(); \
2743 static inline void cond_resched_rcu(void)
2745 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2753 * Does a critical section need to be broken due to another
2754 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2755 * but a general need for low latency)
2757 static inline int spin_needbreak(spinlock_t *lock)
2759 #ifdef CONFIG_PREEMPT
2760 return spin_is_contended(lock);
2767 * Idle thread specific functions to determine the need_resched
2770 #ifdef TIF_POLLING_NRFLAG
2771 static inline int tsk_is_polling(struct task_struct *p)
2773 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2776 static inline void __current_set_polling(void)
2778 set_thread_flag(TIF_POLLING_NRFLAG);
2781 static inline bool __must_check current_set_polling_and_test(void)
2783 __current_set_polling();
2786 * Polling state must be visible before we test NEED_RESCHED,
2787 * paired by resched_task()
2789 smp_mb__after_atomic();
2791 return unlikely(tif_need_resched());
2794 static inline void __current_clr_polling(void)
2796 clear_thread_flag(TIF_POLLING_NRFLAG);
2799 static inline bool __must_check current_clr_polling_and_test(void)
2801 __current_clr_polling();
2804 * Polling state must be visible before we test NEED_RESCHED,
2805 * paired by resched_task()
2807 smp_mb__after_atomic();
2809 return unlikely(tif_need_resched());
2813 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2814 static inline void __current_set_polling(void) { }
2815 static inline void __current_clr_polling(void) { }
2817 static inline bool __must_check current_set_polling_and_test(void)
2819 return unlikely(tif_need_resched());
2821 static inline bool __must_check current_clr_polling_and_test(void)
2823 return unlikely(tif_need_resched());
2827 static inline void current_clr_polling(void)
2829 __current_clr_polling();
2832 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2833 * Once the bit is cleared, we'll get IPIs with every new
2834 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2837 smp_mb(); /* paired with resched_task() */
2839 preempt_fold_need_resched();
2842 static __always_inline bool need_resched(void)
2844 return unlikely(tif_need_resched());
2848 * Thread group CPU time accounting.
2850 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2851 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2853 static inline void thread_group_cputime_init(struct signal_struct *sig)
2855 raw_spin_lock_init(&sig->cputimer.lock);
2859 * Reevaluate whether the task has signals pending delivery.
2860 * Wake the task if so.
2861 * This is required every time the blocked sigset_t changes.
2862 * callers must hold sighand->siglock.
2864 extern void recalc_sigpending_and_wake(struct task_struct *t);
2865 extern void recalc_sigpending(void);
2867 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2869 static inline void signal_wake_up(struct task_struct *t, bool resume)
2871 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2873 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2875 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2879 * Wrappers for p->thread_info->cpu access. No-op on UP.
2883 static inline unsigned int task_cpu(const struct task_struct *p)
2885 return task_thread_info(p)->cpu;
2888 static inline int task_node(const struct task_struct *p)
2890 return cpu_to_node(task_cpu(p));
2893 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2897 static inline unsigned int task_cpu(const struct task_struct *p)
2902 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2906 #endif /* CONFIG_SMP */
2908 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2909 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2911 #ifdef CONFIG_CGROUP_SCHED
2912 extern struct task_group root_task_group;
2913 #endif /* CONFIG_CGROUP_SCHED */
2915 extern int task_can_switch_user(struct user_struct *up,
2916 struct task_struct *tsk);
2918 #ifdef CONFIG_TASK_XACCT
2919 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2921 tsk->ioac.rchar += amt;
2924 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2926 tsk->ioac.wchar += amt;
2929 static inline void inc_syscr(struct task_struct *tsk)
2934 static inline void inc_syscw(struct task_struct *tsk)
2939 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2943 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2947 static inline void inc_syscr(struct task_struct *tsk)
2951 static inline void inc_syscw(struct task_struct *tsk)
2956 #ifndef TASK_SIZE_OF
2957 #define TASK_SIZE_OF(tsk) TASK_SIZE
2961 extern void mm_update_next_owner(struct mm_struct *mm);
2962 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2964 static inline void mm_update_next_owner(struct mm_struct *mm)
2968 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2971 #endif /* CONFIG_MEMCG */
2973 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2976 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2979 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2982 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2985 static inline unsigned long rlimit(unsigned int limit)
2987 return task_rlimit(current, limit);
2990 static inline unsigned long rlimit_max(unsigned int limit)
2992 return task_rlimit_max(current, limit);