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/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60 #include <linux/magic.h>
62 #include <asm/processor.h>
64 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
67 * Extended scheduling parameters data structure.
69 * This is needed because the original struct sched_param can not be
70 * altered without introducing ABI issues with legacy applications
71 * (e.g., in sched_getparam()).
73 * However, the possibility of specifying more than just a priority for
74 * the tasks may be useful for a wide variety of application fields, e.g.,
75 * multimedia, streaming, automation and control, and many others.
77 * This variant (sched_attr) is meant at describing a so-called
78 * sporadic time-constrained task. In such model a task is specified by:
79 * - the activation period or minimum instance inter-arrival time;
80 * - the maximum (or average, depending on the actual scheduling
81 * discipline) computation time of all instances, a.k.a. runtime;
82 * - the deadline (relative to the actual activation time) of each
84 * Very briefly, a periodic (sporadic) task asks for the execution of
85 * some specific computation --which is typically called an instance--
86 * (at most) every period. Moreover, each instance typically lasts no more
87 * than the runtime and must be completed by time instant t equal to
88 * the instance activation time + the deadline.
90 * This is reflected by the actual fields of the sched_attr structure:
92 * @size size of the structure, for fwd/bwd compat.
94 * @sched_policy task's scheduling policy
95 * @sched_flags for customizing the scheduler behaviour
96 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
97 * @sched_priority task's static priority (SCHED_FIFO/RR)
98 * @sched_deadline representative of the task's deadline
99 * @sched_runtime representative of the task's runtime
100 * @sched_period representative of the task's period
102 * Given this task model, there are a multiplicity of scheduling algorithms
103 * and policies, that can be used to ensure all the tasks will make their
104 * timing constraints.
106 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
107 * only user of this new interface. More information about the algorithm
108 * available in the scheduling class file or in Documentation/.
116 /* SCHED_NORMAL, SCHED_BATCH */
119 /* SCHED_FIFO, SCHED_RR */
129 struct futex_pi_state;
130 struct robust_list_head;
133 struct perf_event_context;
137 #define VMACACHE_BITS 2
138 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
139 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
142 * These are the constant used to fake the fixed-point load-average
143 * counting. Some notes:
144 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
145 * a load-average precision of 10 bits integer + 11 bits fractional
146 * - if you want to count load-averages more often, you need more
147 * precision, or rounding will get you. With 2-second counting freq,
148 * the EXP_n values would be 1981, 2034 and 2043 if still using only
151 extern unsigned long avenrun[]; /* Load averages */
152 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
154 #define FSHIFT 11 /* nr of bits of precision */
155 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
156 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
157 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
158 #define EXP_5 2014 /* 1/exp(5sec/5min) */
159 #define EXP_15 2037 /* 1/exp(5sec/15min) */
161 #define CALC_LOAD(load,exp,n) \
163 load += n*(FIXED_1-exp); \
166 extern unsigned long total_forks;
167 extern int nr_threads;
168 DECLARE_PER_CPU(unsigned long, process_counts);
169 extern int nr_processes(void);
170 extern unsigned long nr_running(void);
171 extern bool single_task_running(void);
172 extern unsigned long nr_iowait(void);
173 extern unsigned long nr_iowait_cpu(int cpu);
174 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
176 extern void calc_global_load(unsigned long ticks);
177 extern void update_cpu_load_nohz(void);
179 extern unsigned long get_parent_ip(unsigned long addr);
181 extern void dump_cpu_task(int cpu);
186 #ifdef CONFIG_SCHED_DEBUG
187 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
188 extern void proc_sched_set_task(struct task_struct *p);
190 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
194 * Task state bitmask. NOTE! These bits are also
195 * encoded in fs/proc/array.c: get_task_state().
197 * We have two separate sets of flags: task->state
198 * is about runnability, while task->exit_state are
199 * about the task exiting. Confusing, but this way
200 * modifying one set can't modify the other one by
203 #define TASK_RUNNING 0
204 #define TASK_INTERRUPTIBLE 1
205 #define TASK_UNINTERRUPTIBLE 2
206 #define __TASK_STOPPED 4
207 #define __TASK_TRACED 8
208 /* in tsk->exit_state */
210 #define EXIT_ZOMBIE 32
211 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
212 /* in tsk->state again */
214 #define TASK_WAKEKILL 128
215 #define TASK_WAKING 256
216 #define TASK_PARKED 512
217 #define TASK_STATE_MAX 1024
219 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
221 extern char ___assert_task_state[1 - 2*!!(
222 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
224 /* Convenience macros for the sake of set_task_state */
225 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
226 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
227 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
229 /* Convenience macros for the sake of wake_up */
230 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
231 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
233 /* get_task_state() */
234 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
235 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
236 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
238 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
239 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
240 #define task_is_stopped_or_traced(task) \
241 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
242 #define task_contributes_to_load(task) \
243 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
244 (task->flags & PF_FROZEN) == 0)
246 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
248 #define __set_task_state(tsk, state_value) \
250 (tsk)->task_state_change = _THIS_IP_; \
251 (tsk)->state = (state_value); \
253 #define set_task_state(tsk, state_value) \
255 (tsk)->task_state_change = _THIS_IP_; \
256 set_mb((tsk)->state, (state_value)); \
260 * set_current_state() includes a barrier so that the write of current->state
261 * is correctly serialised wrt the caller's subsequent test of whether to
264 * set_current_state(TASK_UNINTERRUPTIBLE);
265 * if (do_i_need_to_sleep())
268 * If the caller does not need such serialisation then use __set_current_state()
270 #define __set_current_state(state_value) \
272 current->task_state_change = _THIS_IP_; \
273 current->state = (state_value); \
275 #define set_current_state(state_value) \
277 current->task_state_change = _THIS_IP_; \
278 set_mb(current->state, (state_value)); \
283 #define __set_task_state(tsk, state_value) \
284 do { (tsk)->state = (state_value); } while (0)
285 #define set_task_state(tsk, state_value) \
286 set_mb((tsk)->state, (state_value))
289 * set_current_state() includes a barrier so that the write of current->state
290 * is correctly serialised wrt the caller's subsequent test of whether to
293 * set_current_state(TASK_UNINTERRUPTIBLE);
294 * if (do_i_need_to_sleep())
297 * If the caller does not need such serialisation then use __set_current_state()
299 #define __set_current_state(state_value) \
300 do { current->state = (state_value); } while (0)
301 #define set_current_state(state_value) \
302 set_mb(current->state, (state_value))
306 /* Task command name length */
307 #define TASK_COMM_LEN 16
309 #include <linux/spinlock.h>
312 * This serializes "schedule()" and also protects
313 * the run-queue from deletions/modifications (but
314 * _adding_ to the beginning of the run-queue has
317 extern rwlock_t tasklist_lock;
318 extern spinlock_t mmlist_lock;
322 #ifdef CONFIG_PROVE_RCU
323 extern int lockdep_tasklist_lock_is_held(void);
324 #endif /* #ifdef CONFIG_PROVE_RCU */
326 extern void sched_init(void);
327 extern void sched_init_smp(void);
328 extern asmlinkage void schedule_tail(struct task_struct *prev);
329 extern void init_idle(struct task_struct *idle, int cpu);
330 extern void init_idle_bootup_task(struct task_struct *idle);
332 extern int runqueue_is_locked(int cpu);
334 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
335 extern void nohz_balance_enter_idle(int cpu);
336 extern void set_cpu_sd_state_idle(void);
337 extern int get_nohz_timer_target(int pinned);
339 static inline void nohz_balance_enter_idle(int cpu) { }
340 static inline void set_cpu_sd_state_idle(void) { }
341 static inline int get_nohz_timer_target(int pinned)
343 return smp_processor_id();
348 * Only dump TASK_* tasks. (0 for all tasks)
350 extern void show_state_filter(unsigned long state_filter);
352 static inline void show_state(void)
354 show_state_filter(0);
357 extern void show_regs(struct pt_regs *);
360 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
361 * task), SP is the stack pointer of the first frame that should be shown in the back
362 * trace (or NULL if the entire call-chain of the task should be shown).
364 extern void show_stack(struct task_struct *task, unsigned long *sp);
366 void io_schedule(void);
367 long io_schedule_timeout(long timeout);
369 extern void cpu_init (void);
370 extern void trap_init(void);
371 extern void update_process_times(int user);
372 extern void scheduler_tick(void);
374 extern void sched_show_task(struct task_struct *p);
376 #ifdef CONFIG_LOCKUP_DETECTOR
377 extern void touch_softlockup_watchdog(void);
378 extern void touch_softlockup_watchdog_sync(void);
379 extern void touch_all_softlockup_watchdogs(void);
380 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
382 size_t *lenp, loff_t *ppos);
383 extern unsigned int softlockup_panic;
384 void lockup_detector_init(void);
386 static inline void touch_softlockup_watchdog(void)
389 static inline void touch_softlockup_watchdog_sync(void)
392 static inline void touch_all_softlockup_watchdogs(void)
395 static inline void lockup_detector_init(void)
400 #ifdef CONFIG_DETECT_HUNG_TASK
401 void reset_hung_task_detector(void);
403 static inline void reset_hung_task_detector(void)
408 /* Attach to any functions which should be ignored in wchan output. */
409 #define __sched __attribute__((__section__(".sched.text")))
411 /* Linker adds these: start and end of __sched functions */
412 extern char __sched_text_start[], __sched_text_end[];
414 /* Is this address in the __sched functions? */
415 extern int in_sched_functions(unsigned long addr);
417 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
418 extern signed long schedule_timeout(signed long timeout);
419 extern signed long schedule_timeout_interruptible(signed long timeout);
420 extern signed long schedule_timeout_killable(signed long timeout);
421 extern signed long schedule_timeout_uninterruptible(signed long timeout);
422 asmlinkage void schedule(void);
423 extern void schedule_preempt_disabled(void);
426 struct user_namespace;
429 extern void arch_pick_mmap_layout(struct mm_struct *mm);
431 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
432 unsigned long, unsigned long);
434 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
435 unsigned long len, unsigned long pgoff,
436 unsigned long flags);
438 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
441 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
442 #define SUID_DUMP_USER 1 /* Dump as user of process */
443 #define SUID_DUMP_ROOT 2 /* Dump as root */
447 /* for SUID_DUMP_* above */
448 #define MMF_DUMPABLE_BITS 2
449 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
451 extern void set_dumpable(struct mm_struct *mm, int value);
453 * This returns the actual value of the suid_dumpable flag. For things
454 * that are using this for checking for privilege transitions, it must
455 * test against SUID_DUMP_USER rather than treating it as a boolean
458 static inline int __get_dumpable(unsigned long mm_flags)
460 return mm_flags & MMF_DUMPABLE_MASK;
463 static inline int get_dumpable(struct mm_struct *mm)
465 return __get_dumpable(mm->flags);
468 /* coredump filter bits */
469 #define MMF_DUMP_ANON_PRIVATE 2
470 #define MMF_DUMP_ANON_SHARED 3
471 #define MMF_DUMP_MAPPED_PRIVATE 4
472 #define MMF_DUMP_MAPPED_SHARED 5
473 #define MMF_DUMP_ELF_HEADERS 6
474 #define MMF_DUMP_HUGETLB_PRIVATE 7
475 #define MMF_DUMP_HUGETLB_SHARED 8
477 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
478 #define MMF_DUMP_FILTER_BITS 7
479 #define MMF_DUMP_FILTER_MASK \
480 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
481 #define MMF_DUMP_FILTER_DEFAULT \
482 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
483 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
485 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
486 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
488 # define MMF_DUMP_MASK_DEFAULT_ELF 0
490 /* leave room for more dump flags */
491 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
492 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
493 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
495 #define MMF_HAS_UPROBES 19 /* has uprobes */
496 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
498 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
500 struct sighand_struct {
502 struct k_sigaction action[_NSIG];
504 wait_queue_head_t signalfd_wqh;
507 struct pacct_struct {
510 unsigned long ac_mem;
511 cputime_t ac_utime, ac_stime;
512 unsigned long ac_minflt, ac_majflt;
523 * struct cputime - snaphsot of system and user cputime
524 * @utime: time spent in user mode
525 * @stime: time spent in system mode
527 * Gathers a generic snapshot of user and system time.
535 * struct task_cputime - collected CPU time counts
536 * @utime: time spent in user mode, in &cputime_t units
537 * @stime: time spent in kernel mode, in &cputime_t units
538 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
540 * This is an extension of struct cputime that includes the total runtime
541 * spent by the task from the scheduler point of view.
543 * As a result, this structure groups together three kinds of CPU time
544 * that are tracked for threads and thread groups. Most things considering
545 * CPU time want to group these counts together and treat all three
546 * of them in parallel.
548 struct task_cputime {
551 unsigned long long sum_exec_runtime;
553 /* Alternate field names when used to cache expirations. */
554 #define prof_exp stime
555 #define virt_exp utime
556 #define sched_exp sum_exec_runtime
558 #define INIT_CPUTIME \
559 (struct task_cputime) { \
562 .sum_exec_runtime = 0, \
565 #ifdef CONFIG_PREEMPT_COUNT
566 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
568 #define PREEMPT_DISABLED PREEMPT_ENABLED
572 * Disable preemption until the scheduler is running.
573 * Reset by start_kernel()->sched_init()->init_idle().
575 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
576 * before the scheduler is active -- see should_resched().
578 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
581 * struct thread_group_cputimer - thread group interval timer counts
582 * @cputime: thread group interval timers.
583 * @running: non-zero when there are timers running and
584 * @cputime receives updates.
585 * @lock: lock for fields in this struct.
587 * This structure contains the version of task_cputime, above, that is
588 * used for thread group CPU timer calculations.
590 struct thread_group_cputimer {
591 struct task_cputime cputime;
596 #include <linux/rwsem.h>
600 * NOTE! "signal_struct" does not have its own
601 * locking, because a shared signal_struct always
602 * implies a shared sighand_struct, so locking
603 * sighand_struct is always a proper superset of
604 * the locking of signal_struct.
606 struct signal_struct {
610 struct list_head thread_head;
612 wait_queue_head_t wait_chldexit; /* for wait4() */
614 /* current thread group signal load-balancing target: */
615 struct task_struct *curr_target;
617 /* shared signal handling: */
618 struct sigpending shared_pending;
620 /* thread group exit support */
623 * - notify group_exit_task when ->count is equal to notify_count
624 * - everyone except group_exit_task is stopped during signal delivery
625 * of fatal signals, group_exit_task processes the signal.
628 struct task_struct *group_exit_task;
630 /* thread group stop support, overloads group_exit_code too */
631 int group_stop_count;
632 unsigned int flags; /* see SIGNAL_* flags below */
635 * PR_SET_CHILD_SUBREAPER marks a process, like a service
636 * manager, to re-parent orphan (double-forking) child processes
637 * to this process instead of 'init'. The service manager is
638 * able to receive SIGCHLD signals and is able to investigate
639 * the process until it calls wait(). All children of this
640 * process will inherit a flag if they should look for a
641 * child_subreaper process at exit.
643 unsigned int is_child_subreaper:1;
644 unsigned int has_child_subreaper:1;
646 /* POSIX.1b Interval Timers */
648 struct list_head posix_timers;
650 /* ITIMER_REAL timer for the process */
651 struct hrtimer real_timer;
652 struct pid *leader_pid;
653 ktime_t it_real_incr;
656 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
657 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
658 * values are defined to 0 and 1 respectively
660 struct cpu_itimer it[2];
663 * Thread group totals for process CPU timers.
664 * See thread_group_cputimer(), et al, for details.
666 struct thread_group_cputimer cputimer;
668 /* Earliest-expiration cache. */
669 struct task_cputime cputime_expires;
671 struct list_head cpu_timers[3];
673 struct pid *tty_old_pgrp;
675 /* boolean value for session group leader */
678 struct tty_struct *tty; /* NULL if no tty */
680 #ifdef CONFIG_SCHED_AUTOGROUP
681 struct autogroup *autogroup;
684 * Cumulative resource counters for dead threads in the group,
685 * and for reaped dead child processes forked by this group.
686 * Live threads maintain their own counters and add to these
687 * in __exit_signal, except for the group leader.
689 seqlock_t stats_lock;
690 cputime_t utime, stime, cutime, cstime;
693 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
694 struct cputime prev_cputime;
696 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
697 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
698 unsigned long inblock, oublock, cinblock, coublock;
699 unsigned long maxrss, cmaxrss;
700 struct task_io_accounting ioac;
703 * Cumulative ns of schedule CPU time fo dead threads in the
704 * group, not including a zombie group leader, (This only differs
705 * from jiffies_to_ns(utime + stime) if sched_clock uses something
706 * other than jiffies.)
708 unsigned long long sum_sched_runtime;
711 * We don't bother to synchronize most readers of this at all,
712 * because there is no reader checking a limit that actually needs
713 * to get both rlim_cur and rlim_max atomically, and either one
714 * alone is a single word that can safely be read normally.
715 * getrlimit/setrlimit use task_lock(current->group_leader) to
716 * protect this instead of the siglock, because they really
717 * have no need to disable irqs.
719 struct rlimit rlim[RLIM_NLIMITS];
721 #ifdef CONFIG_BSD_PROCESS_ACCT
722 struct pacct_struct pacct; /* per-process accounting information */
724 #ifdef CONFIG_TASKSTATS
725 struct taskstats *stats;
729 unsigned audit_tty_log_passwd;
730 struct tty_audit_buf *tty_audit_buf;
732 #ifdef CONFIG_CGROUPS
734 * group_rwsem prevents new tasks from entering the threadgroup and
735 * member tasks from exiting,a more specifically, setting of
736 * PF_EXITING. fork and exit paths are protected with this rwsem
737 * using threadgroup_change_begin/end(). Users which require
738 * threadgroup to remain stable should use threadgroup_[un]lock()
739 * which also takes care of exec path. Currently, cgroup is the
742 struct rw_semaphore group_rwsem;
745 oom_flags_t oom_flags;
746 short oom_score_adj; /* OOM kill score adjustment */
747 short oom_score_adj_min; /* OOM kill score adjustment min value.
748 * Only settable by CAP_SYS_RESOURCE. */
750 struct mutex cred_guard_mutex; /* guard against foreign influences on
751 * credential calculations
752 * (notably. ptrace) */
756 * Bits in flags field of signal_struct.
758 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
759 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
760 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
761 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
763 * Pending notifications to parent.
765 #define SIGNAL_CLD_STOPPED 0x00000010
766 #define SIGNAL_CLD_CONTINUED 0x00000020
767 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
769 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
771 /* If true, all threads except ->group_exit_task have pending SIGKILL */
772 static inline int signal_group_exit(const struct signal_struct *sig)
774 return (sig->flags & SIGNAL_GROUP_EXIT) ||
775 (sig->group_exit_task != NULL);
779 * Some day this will be a full-fledged user tracking system..
782 atomic_t __count; /* reference count */
783 atomic_t processes; /* How many processes does this user have? */
784 atomic_t sigpending; /* How many pending signals does this user have? */
785 #ifdef CONFIG_INOTIFY_USER
786 atomic_t inotify_watches; /* How many inotify watches does this user have? */
787 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
789 #ifdef CONFIG_FANOTIFY
790 atomic_t fanotify_listeners;
793 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
795 #ifdef CONFIG_POSIX_MQUEUE
796 /* protected by mq_lock */
797 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
799 unsigned long locked_shm; /* How many pages of mlocked shm ? */
802 struct key *uid_keyring; /* UID specific keyring */
803 struct key *session_keyring; /* UID's default session keyring */
806 /* Hash table maintenance information */
807 struct hlist_node uidhash_node;
810 #ifdef CONFIG_PERF_EVENTS
811 atomic_long_t locked_vm;
815 extern int uids_sysfs_init(void);
817 extern struct user_struct *find_user(kuid_t);
819 extern struct user_struct root_user;
820 #define INIT_USER (&root_user)
823 struct backing_dev_info;
824 struct reclaim_state;
826 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
828 /* cumulative counters */
829 unsigned long pcount; /* # of times run on this cpu */
830 unsigned long long run_delay; /* time spent waiting on a runqueue */
833 unsigned long long last_arrival,/* when we last ran on a cpu */
834 last_queued; /* when we were last queued to run */
836 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
838 #ifdef CONFIG_TASK_DELAY_ACCT
839 struct task_delay_info {
841 unsigned int flags; /* Private per-task flags */
843 /* For each stat XXX, add following, aligned appropriately
845 * struct timespec XXX_start, XXX_end;
849 * Atomicity of updates to XXX_delay, XXX_count protected by
850 * single lock above (split into XXX_lock if contention is an issue).
854 * XXX_count is incremented on every XXX operation, the delay
855 * associated with the operation is added to XXX_delay.
856 * XXX_delay contains the accumulated delay time in nanoseconds.
858 u64 blkio_start; /* Shared by blkio, swapin */
859 u64 blkio_delay; /* wait for sync block io completion */
860 u64 swapin_delay; /* wait for swapin block io completion */
861 u32 blkio_count; /* total count of the number of sync block */
862 /* io operations performed */
863 u32 swapin_count; /* total count of the number of swapin block */
864 /* io operations performed */
867 u64 freepages_delay; /* wait for memory reclaim */
868 u32 freepages_count; /* total count of memory reclaim */
870 #endif /* CONFIG_TASK_DELAY_ACCT */
872 static inline int sched_info_on(void)
874 #ifdef CONFIG_SCHEDSTATS
876 #elif defined(CONFIG_TASK_DELAY_ACCT)
877 extern int delayacct_on;
892 * Increase resolution of cpu_capacity calculations
894 #define SCHED_CAPACITY_SHIFT 10
895 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
898 * sched-domains (multiprocessor balancing) declarations:
901 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
902 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
903 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
904 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
905 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
906 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
907 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
908 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
909 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
910 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
911 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
912 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
913 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
914 #define SD_NUMA 0x4000 /* cross-node balancing */
916 #ifdef CONFIG_SCHED_SMT
917 static inline int cpu_smt_flags(void)
919 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
923 #ifdef CONFIG_SCHED_MC
924 static inline int cpu_core_flags(void)
926 return SD_SHARE_PKG_RESOURCES;
931 static inline int cpu_numa_flags(void)
937 struct sched_domain_attr {
938 int relax_domain_level;
941 #define SD_ATTR_INIT (struct sched_domain_attr) { \
942 .relax_domain_level = -1, \
945 extern int sched_domain_level_max;
949 struct sched_domain {
950 /* These fields must be setup */
951 struct sched_domain *parent; /* top domain must be null terminated */
952 struct sched_domain *child; /* bottom domain must be null terminated */
953 struct sched_group *groups; /* the balancing groups of the domain */
954 unsigned long min_interval; /* Minimum balance interval ms */
955 unsigned long max_interval; /* Maximum balance interval ms */
956 unsigned int busy_factor; /* less balancing by factor if busy */
957 unsigned int imbalance_pct; /* No balance until over watermark */
958 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
959 unsigned int busy_idx;
960 unsigned int idle_idx;
961 unsigned int newidle_idx;
962 unsigned int wake_idx;
963 unsigned int forkexec_idx;
964 unsigned int smt_gain;
966 int nohz_idle; /* NOHZ IDLE status */
967 int flags; /* See SD_* */
970 /* Runtime fields. */
971 unsigned long last_balance; /* init to jiffies. units in jiffies */
972 unsigned int balance_interval; /* initialise to 1. units in ms. */
973 unsigned int nr_balance_failed; /* initialise to 0 */
975 /* idle_balance() stats */
976 u64 max_newidle_lb_cost;
977 unsigned long next_decay_max_lb_cost;
979 #ifdef CONFIG_SCHEDSTATS
980 /* load_balance() stats */
981 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
982 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
983 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
984 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
985 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
986 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
987 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
988 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
990 /* Active load balancing */
991 unsigned int alb_count;
992 unsigned int alb_failed;
993 unsigned int alb_pushed;
995 /* SD_BALANCE_EXEC stats */
996 unsigned int sbe_count;
997 unsigned int sbe_balanced;
998 unsigned int sbe_pushed;
1000 /* SD_BALANCE_FORK stats */
1001 unsigned int sbf_count;
1002 unsigned int sbf_balanced;
1003 unsigned int sbf_pushed;
1005 /* try_to_wake_up() stats */
1006 unsigned int ttwu_wake_remote;
1007 unsigned int ttwu_move_affine;
1008 unsigned int ttwu_move_balance;
1010 #ifdef CONFIG_SCHED_DEBUG
1014 void *private; /* used during construction */
1015 struct rcu_head rcu; /* used during destruction */
1018 unsigned int span_weight;
1020 * Span of all CPUs in this domain.
1022 * NOTE: this field is variable length. (Allocated dynamically
1023 * by attaching extra space to the end of the structure,
1024 * depending on how many CPUs the kernel has booted up with)
1026 unsigned long span[0];
1029 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1031 return to_cpumask(sd->span);
1034 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1035 struct sched_domain_attr *dattr_new);
1037 /* Allocate an array of sched domains, for partition_sched_domains(). */
1038 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1039 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1041 bool cpus_share_cache(int this_cpu, int that_cpu);
1043 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1044 typedef int (*sched_domain_flags_f)(void);
1046 #define SDTL_OVERLAP 0x01
1049 struct sched_domain **__percpu sd;
1050 struct sched_group **__percpu sg;
1051 struct sched_group_capacity **__percpu sgc;
1054 struct sched_domain_topology_level {
1055 sched_domain_mask_f mask;
1056 sched_domain_flags_f sd_flags;
1059 struct sd_data data;
1060 #ifdef CONFIG_SCHED_DEBUG
1065 extern struct sched_domain_topology_level *sched_domain_topology;
1067 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1068 extern void wake_up_if_idle(int cpu);
1070 #ifdef CONFIG_SCHED_DEBUG
1071 # define SD_INIT_NAME(type) .name = #type
1073 # define SD_INIT_NAME(type)
1076 #else /* CONFIG_SMP */
1078 struct sched_domain_attr;
1081 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1082 struct sched_domain_attr *dattr_new)
1086 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1091 #endif /* !CONFIG_SMP */
1094 struct io_context; /* See blkdev.h */
1097 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1098 extern void prefetch_stack(struct task_struct *t);
1100 static inline void prefetch_stack(struct task_struct *t) { }
1103 struct audit_context; /* See audit.c */
1105 struct pipe_inode_info;
1106 struct uts_namespace;
1108 struct load_weight {
1109 unsigned long weight;
1115 * These sums represent an infinite geometric series and so are bound
1116 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1117 * choices of y < 1-2^(-32)*1024.
1119 u32 runnable_avg_sum, runnable_avg_period;
1120 u64 last_runnable_update;
1122 unsigned long load_avg_contrib;
1125 #ifdef CONFIG_SCHEDSTATS
1126 struct sched_statistics {
1136 s64 sum_sleep_runtime;
1143 u64 nr_migrations_cold;
1144 u64 nr_failed_migrations_affine;
1145 u64 nr_failed_migrations_running;
1146 u64 nr_failed_migrations_hot;
1147 u64 nr_forced_migrations;
1150 u64 nr_wakeups_sync;
1151 u64 nr_wakeups_migrate;
1152 u64 nr_wakeups_local;
1153 u64 nr_wakeups_remote;
1154 u64 nr_wakeups_affine;
1155 u64 nr_wakeups_affine_attempts;
1156 u64 nr_wakeups_passive;
1157 u64 nr_wakeups_idle;
1161 struct sched_entity {
1162 struct load_weight load; /* for load-balancing */
1163 struct rb_node run_node;
1164 struct list_head group_node;
1168 u64 sum_exec_runtime;
1170 u64 prev_sum_exec_runtime;
1174 #ifdef CONFIG_SCHEDSTATS
1175 struct sched_statistics statistics;
1178 #ifdef CONFIG_FAIR_GROUP_SCHED
1180 struct sched_entity *parent;
1181 /* rq on which this entity is (to be) queued: */
1182 struct cfs_rq *cfs_rq;
1183 /* rq "owned" by this entity/group: */
1184 struct cfs_rq *my_q;
1188 /* Per-entity load-tracking */
1189 struct sched_avg avg;
1193 struct sched_rt_entity {
1194 struct list_head run_list;
1195 unsigned long timeout;
1196 unsigned long watchdog_stamp;
1197 unsigned int time_slice;
1199 struct sched_rt_entity *back;
1200 #ifdef CONFIG_RT_GROUP_SCHED
1201 struct sched_rt_entity *parent;
1202 /* rq on which this entity is (to be) queued: */
1203 struct rt_rq *rt_rq;
1204 /* rq "owned" by this entity/group: */
1209 struct sched_dl_entity {
1210 struct rb_node rb_node;
1213 * Original scheduling parameters. Copied here from sched_attr
1214 * during sched_setattr(), they will remain the same until
1215 * the next sched_setattr().
1217 u64 dl_runtime; /* maximum runtime for each instance */
1218 u64 dl_deadline; /* relative deadline of each instance */
1219 u64 dl_period; /* separation of two instances (period) */
1220 u64 dl_bw; /* dl_runtime / dl_deadline */
1223 * Actual scheduling parameters. Initialized with the values above,
1224 * they are continously updated during task execution. Note that
1225 * the remaining runtime could be < 0 in case we are in overrun.
1227 s64 runtime; /* remaining runtime for this instance */
1228 u64 deadline; /* absolute deadline for this instance */
1229 unsigned int flags; /* specifying the scheduler behaviour */
1234 * @dl_throttled tells if we exhausted the runtime. If so, the
1235 * task has to wait for a replenishment to be performed at the
1236 * next firing of dl_timer.
1238 * @dl_new tells if a new instance arrived. If so we must
1239 * start executing it with full runtime and reset its absolute
1242 * @dl_boosted tells if we are boosted due to DI. If so we are
1243 * outside bandwidth enforcement mechanism (but only until we
1244 * exit the critical section);
1246 * @dl_yielded tells if task gave up the cpu before consuming
1247 * all its available runtime during the last job.
1249 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1252 * Bandwidth enforcement timer. Each -deadline task has its
1253 * own bandwidth to be enforced, thus we need one timer per task.
1255 struct hrtimer dl_timer;
1267 enum perf_event_task_context {
1268 perf_invalid_context = -1,
1269 perf_hw_context = 0,
1271 perf_nr_task_contexts,
1274 struct task_struct {
1275 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1278 unsigned int flags; /* per process flags, defined below */
1279 unsigned int ptrace;
1282 struct llist_node wake_entry;
1284 struct task_struct *last_wakee;
1285 unsigned long wakee_flips;
1286 unsigned long wakee_flip_decay_ts;
1292 int prio, static_prio, normal_prio;
1293 unsigned int rt_priority;
1294 const struct sched_class *sched_class;
1295 struct sched_entity se;
1296 struct sched_rt_entity rt;
1297 #ifdef CONFIG_CGROUP_SCHED
1298 struct task_group *sched_task_group;
1300 struct sched_dl_entity dl;
1302 #ifdef CONFIG_PREEMPT_NOTIFIERS
1303 /* list of struct preempt_notifier: */
1304 struct hlist_head preempt_notifiers;
1307 #ifdef CONFIG_BLK_DEV_IO_TRACE
1308 unsigned int btrace_seq;
1311 unsigned int policy;
1312 int nr_cpus_allowed;
1313 cpumask_t cpus_allowed;
1315 #ifdef CONFIG_PREEMPT_RCU
1316 int rcu_read_lock_nesting;
1317 union rcu_special rcu_read_unlock_special;
1318 struct list_head rcu_node_entry;
1319 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1320 #ifdef CONFIG_PREEMPT_RCU
1321 struct rcu_node *rcu_blocked_node;
1322 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1323 #ifdef CONFIG_TASKS_RCU
1324 unsigned long rcu_tasks_nvcsw;
1325 bool rcu_tasks_holdout;
1326 struct list_head rcu_tasks_holdout_list;
1327 int rcu_tasks_idle_cpu;
1328 #endif /* #ifdef CONFIG_TASKS_RCU */
1330 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1331 struct sched_info sched_info;
1334 struct list_head tasks;
1336 struct plist_node pushable_tasks;
1337 struct rb_node pushable_dl_tasks;
1340 struct mm_struct *mm, *active_mm;
1341 #ifdef CONFIG_COMPAT_BRK
1342 unsigned brk_randomized:1;
1344 /* per-thread vma caching */
1345 u32 vmacache_seqnum;
1346 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1347 #if defined(SPLIT_RSS_COUNTING)
1348 struct task_rss_stat rss_stat;
1352 int exit_code, exit_signal;
1353 int pdeath_signal; /* The signal sent when the parent dies */
1354 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1356 /* Used for emulating ABI behavior of previous Linux versions */
1357 unsigned int personality;
1359 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1361 unsigned in_iowait:1;
1363 /* Revert to default priority/policy when forking */
1364 unsigned sched_reset_on_fork:1;
1365 unsigned sched_contributes_to_load:1;
1367 unsigned long atomic_flags; /* Flags needing atomic access. */
1372 #ifdef CONFIG_CC_STACKPROTECTOR
1373 /* Canary value for the -fstack-protector gcc feature */
1374 unsigned long stack_canary;
1377 * pointers to (original) parent process, youngest child, younger sibling,
1378 * older sibling, respectively. (p->father can be replaced with
1379 * p->real_parent->pid)
1381 struct task_struct __rcu *real_parent; /* real parent process */
1382 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1384 * children/sibling forms the list of my natural children
1386 struct list_head children; /* list of my children */
1387 struct list_head sibling; /* linkage in my parent's children list */
1388 struct task_struct *group_leader; /* threadgroup leader */
1391 * ptraced is the list of tasks this task is using ptrace on.
1392 * This includes both natural children and PTRACE_ATTACH targets.
1393 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1395 struct list_head ptraced;
1396 struct list_head ptrace_entry;
1398 /* PID/PID hash table linkage. */
1399 struct pid_link pids[PIDTYPE_MAX];
1400 struct list_head thread_group;
1401 struct list_head thread_node;
1403 struct completion *vfork_done; /* for vfork() */
1404 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1405 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1407 cputime_t utime, stime, utimescaled, stimescaled;
1409 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1410 struct cputime prev_cputime;
1412 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1413 seqlock_t vtime_seqlock;
1414 unsigned long long vtime_snap;
1419 } vtime_snap_whence;
1421 unsigned long nvcsw, nivcsw; /* context switch counts */
1422 u64 start_time; /* monotonic time in nsec */
1423 u64 real_start_time; /* boot based time in nsec */
1424 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1425 unsigned long min_flt, maj_flt;
1427 struct task_cputime cputime_expires;
1428 struct list_head cpu_timers[3];
1430 /* process credentials */
1431 const struct cred __rcu *real_cred; /* objective and real subjective task
1432 * credentials (COW) */
1433 const struct cred __rcu *cred; /* effective (overridable) subjective task
1434 * credentials (COW) */
1435 char comm[TASK_COMM_LEN]; /* executable name excluding path
1436 - access with [gs]et_task_comm (which lock
1437 it with task_lock())
1438 - initialized normally by setup_new_exec */
1439 /* file system info */
1440 int link_count, total_link_count;
1441 #ifdef CONFIG_SYSVIPC
1443 struct sysv_sem sysvsem;
1444 struct sysv_shm sysvshm;
1446 #ifdef CONFIG_DETECT_HUNG_TASK
1447 /* hung task detection */
1448 unsigned long last_switch_count;
1450 /* CPU-specific state of this task */
1451 struct thread_struct thread;
1452 /* filesystem information */
1453 struct fs_struct *fs;
1454 /* open file information */
1455 struct files_struct *files;
1457 struct nsproxy *nsproxy;
1458 /* signal handlers */
1459 struct signal_struct *signal;
1460 struct sighand_struct *sighand;
1462 sigset_t blocked, real_blocked;
1463 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1464 struct sigpending pending;
1466 unsigned long sas_ss_sp;
1468 int (*notifier)(void *priv);
1469 void *notifier_data;
1470 sigset_t *notifier_mask;
1471 struct callback_head *task_works;
1473 struct audit_context *audit_context;
1474 #ifdef CONFIG_AUDITSYSCALL
1476 unsigned int sessionid;
1478 struct seccomp seccomp;
1480 /* Thread group tracking */
1483 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1485 spinlock_t alloc_lock;
1487 /* Protection of the PI data structures: */
1488 raw_spinlock_t pi_lock;
1490 #ifdef CONFIG_RT_MUTEXES
1491 /* PI waiters blocked on a rt_mutex held by this task */
1492 struct rb_root pi_waiters;
1493 struct rb_node *pi_waiters_leftmost;
1494 /* Deadlock detection and priority inheritance handling */
1495 struct rt_mutex_waiter *pi_blocked_on;
1498 #ifdef CONFIG_DEBUG_MUTEXES
1499 /* mutex deadlock detection */
1500 struct mutex_waiter *blocked_on;
1502 #ifdef CONFIG_TRACE_IRQFLAGS
1503 unsigned int irq_events;
1504 unsigned long hardirq_enable_ip;
1505 unsigned long hardirq_disable_ip;
1506 unsigned int hardirq_enable_event;
1507 unsigned int hardirq_disable_event;
1508 int hardirqs_enabled;
1509 int hardirq_context;
1510 unsigned long softirq_disable_ip;
1511 unsigned long softirq_enable_ip;
1512 unsigned int softirq_disable_event;
1513 unsigned int softirq_enable_event;
1514 int softirqs_enabled;
1515 int softirq_context;
1517 #ifdef CONFIG_LOCKDEP
1518 # define MAX_LOCK_DEPTH 48UL
1521 unsigned int lockdep_recursion;
1522 struct held_lock held_locks[MAX_LOCK_DEPTH];
1523 gfp_t lockdep_reclaim_gfp;
1526 /* journalling filesystem info */
1529 /* stacked block device info */
1530 struct bio_list *bio_list;
1533 /* stack plugging */
1534 struct blk_plug *plug;
1538 struct reclaim_state *reclaim_state;
1540 struct backing_dev_info *backing_dev_info;
1542 struct io_context *io_context;
1544 unsigned long ptrace_message;
1545 siginfo_t *last_siginfo; /* For ptrace use. */
1546 struct task_io_accounting ioac;
1547 #if defined(CONFIG_TASK_XACCT)
1548 u64 acct_rss_mem1; /* accumulated rss usage */
1549 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1550 cputime_t acct_timexpd; /* stime + utime since last update */
1552 #ifdef CONFIG_CPUSETS
1553 nodemask_t mems_allowed; /* Protected by alloc_lock */
1554 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1555 int cpuset_mem_spread_rotor;
1556 int cpuset_slab_spread_rotor;
1558 #ifdef CONFIG_CGROUPS
1559 /* Control Group info protected by css_set_lock */
1560 struct css_set __rcu *cgroups;
1561 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1562 struct list_head cg_list;
1565 struct robust_list_head __user *robust_list;
1566 #ifdef CONFIG_COMPAT
1567 struct compat_robust_list_head __user *compat_robust_list;
1569 struct list_head pi_state_list;
1570 struct futex_pi_state *pi_state_cache;
1572 #ifdef CONFIG_PERF_EVENTS
1573 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1574 struct mutex perf_event_mutex;
1575 struct list_head perf_event_list;
1577 #ifdef CONFIG_DEBUG_PREEMPT
1578 unsigned long preempt_disable_ip;
1581 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1583 short pref_node_fork;
1585 #ifdef CONFIG_NUMA_BALANCING
1587 unsigned int numa_scan_period;
1588 unsigned int numa_scan_period_max;
1589 int numa_preferred_nid;
1590 unsigned long numa_migrate_retry;
1591 u64 node_stamp; /* migration stamp */
1592 u64 last_task_numa_placement;
1593 u64 last_sum_exec_runtime;
1594 struct callback_head numa_work;
1596 struct list_head numa_entry;
1597 struct numa_group *numa_group;
1600 * numa_faults is an array split into four regions:
1601 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1602 * in this precise order.
1604 * faults_memory: Exponential decaying average of faults on a per-node
1605 * basis. Scheduling placement decisions are made based on these
1606 * counts. The values remain static for the duration of a PTE scan.
1607 * faults_cpu: Track the nodes the process was running on when a NUMA
1608 * hinting fault was incurred.
1609 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1610 * during the current scan window. When the scan completes, the counts
1611 * in faults_memory and faults_cpu decay and these values are copied.
1613 unsigned long *numa_faults;
1614 unsigned long total_numa_faults;
1617 * numa_faults_locality tracks if faults recorded during the last
1618 * scan window were remote/local. The task scan period is adapted
1619 * based on the locality of the faults with different weights
1620 * depending on whether they were shared or private faults
1622 unsigned long numa_faults_locality[2];
1624 unsigned long numa_pages_migrated;
1625 #endif /* CONFIG_NUMA_BALANCING */
1627 struct rcu_head rcu;
1630 * cache last used pipe for splice
1632 struct pipe_inode_info *splice_pipe;
1634 struct page_frag task_frag;
1636 #ifdef CONFIG_TASK_DELAY_ACCT
1637 struct task_delay_info *delays;
1639 #ifdef CONFIG_FAULT_INJECTION
1643 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1644 * balance_dirty_pages() for some dirty throttling pause
1647 int nr_dirtied_pause;
1648 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1650 #ifdef CONFIG_LATENCYTOP
1651 int latency_record_count;
1652 struct latency_record latency_record[LT_SAVECOUNT];
1655 * time slack values; these are used to round up poll() and
1656 * select() etc timeout values. These are in nanoseconds.
1658 unsigned long timer_slack_ns;
1659 unsigned long default_timer_slack_ns;
1661 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1662 /* Index of current stored address in ret_stack */
1664 /* Stack of return addresses for return function tracing */
1665 struct ftrace_ret_stack *ret_stack;
1666 /* time stamp for last schedule */
1667 unsigned long long ftrace_timestamp;
1669 * Number of functions that haven't been traced
1670 * because of depth overrun.
1672 atomic_t trace_overrun;
1673 /* Pause for the tracing */
1674 atomic_t tracing_graph_pause;
1676 #ifdef CONFIG_TRACING
1677 /* state flags for use by tracers */
1678 unsigned long trace;
1679 /* bitmask and counter of trace recursion */
1680 unsigned long trace_recursion;
1681 #endif /* CONFIG_TRACING */
1682 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1683 unsigned int memcg_kmem_skip_account;
1684 struct memcg_oom_info {
1685 struct mem_cgroup *memcg;
1688 unsigned int may_oom:1;
1691 #ifdef CONFIG_UPROBES
1692 struct uprobe_task *utask;
1694 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1695 unsigned int sequential_io;
1696 unsigned int sequential_io_avg;
1698 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1699 unsigned long task_state_change;
1703 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1704 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1706 #define TNF_MIGRATED 0x01
1707 #define TNF_NO_GROUP 0x02
1708 #define TNF_SHARED 0x04
1709 #define TNF_FAULT_LOCAL 0x08
1711 #ifdef CONFIG_NUMA_BALANCING
1712 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1713 extern pid_t task_numa_group_id(struct task_struct *p);
1714 extern void set_numabalancing_state(bool enabled);
1715 extern void task_numa_free(struct task_struct *p);
1716 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1717 int src_nid, int dst_cpu);
1719 static inline void task_numa_fault(int last_node, int node, int pages,
1723 static inline pid_t task_numa_group_id(struct task_struct *p)
1727 static inline void set_numabalancing_state(bool enabled)
1730 static inline void task_numa_free(struct task_struct *p)
1733 static inline bool should_numa_migrate_memory(struct task_struct *p,
1734 struct page *page, int src_nid, int dst_cpu)
1740 static inline struct pid *task_pid(struct task_struct *task)
1742 return task->pids[PIDTYPE_PID].pid;
1745 static inline struct pid *task_tgid(struct task_struct *task)
1747 return task->group_leader->pids[PIDTYPE_PID].pid;
1751 * Without tasklist or rcu lock it is not safe to dereference
1752 * the result of task_pgrp/task_session even if task == current,
1753 * we can race with another thread doing sys_setsid/sys_setpgid.
1755 static inline struct pid *task_pgrp(struct task_struct *task)
1757 return task->group_leader->pids[PIDTYPE_PGID].pid;
1760 static inline struct pid *task_session(struct task_struct *task)
1762 return task->group_leader->pids[PIDTYPE_SID].pid;
1765 struct pid_namespace;
1768 * the helpers to get the task's different pids as they are seen
1769 * from various namespaces
1771 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1772 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1774 * task_xid_nr_ns() : id seen from the ns specified;
1776 * set_task_vxid() : assigns a virtual id to a task;
1778 * see also pid_nr() etc in include/linux/pid.h
1780 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1781 struct pid_namespace *ns);
1783 static inline pid_t task_pid_nr(struct task_struct *tsk)
1788 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1789 struct pid_namespace *ns)
1791 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1794 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1796 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1800 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1805 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1807 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1809 return pid_vnr(task_tgid(tsk));
1813 static inline int pid_alive(const struct task_struct *p);
1814 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1820 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1826 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1828 return task_ppid_nr_ns(tsk, &init_pid_ns);
1831 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1832 struct pid_namespace *ns)
1834 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1837 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1839 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1843 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1844 struct pid_namespace *ns)
1846 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1849 static inline pid_t task_session_vnr(struct task_struct *tsk)
1851 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1854 /* obsolete, do not use */
1855 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1857 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1861 * pid_alive - check that a task structure is not stale
1862 * @p: Task structure to be checked.
1864 * Test if a process is not yet dead (at most zombie state)
1865 * If pid_alive fails, then pointers within the task structure
1866 * can be stale and must not be dereferenced.
1868 * Return: 1 if the process is alive. 0 otherwise.
1870 static inline int pid_alive(const struct task_struct *p)
1872 return p->pids[PIDTYPE_PID].pid != NULL;
1876 * is_global_init - check if a task structure is init
1877 * @tsk: Task structure to be checked.
1879 * Check if a task structure is the first user space task the kernel created.
1881 * Return: 1 if the task structure is init. 0 otherwise.
1883 static inline int is_global_init(struct task_struct *tsk)
1885 return tsk->pid == 1;
1888 extern struct pid *cad_pid;
1890 extern void free_task(struct task_struct *tsk);
1891 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1893 extern void __put_task_struct(struct task_struct *t);
1895 static inline void put_task_struct(struct task_struct *t)
1897 if (atomic_dec_and_test(&t->usage))
1898 __put_task_struct(t);
1901 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1902 extern void task_cputime(struct task_struct *t,
1903 cputime_t *utime, cputime_t *stime);
1904 extern void task_cputime_scaled(struct task_struct *t,
1905 cputime_t *utimescaled, cputime_t *stimescaled);
1906 extern cputime_t task_gtime(struct task_struct *t);
1908 static inline void task_cputime(struct task_struct *t,
1909 cputime_t *utime, cputime_t *stime)
1917 static inline void task_cputime_scaled(struct task_struct *t,
1918 cputime_t *utimescaled,
1919 cputime_t *stimescaled)
1922 *utimescaled = t->utimescaled;
1924 *stimescaled = t->stimescaled;
1927 static inline cputime_t task_gtime(struct task_struct *t)
1932 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1933 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1938 #define PF_EXITING 0x00000004 /* getting shut down */
1939 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1940 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1941 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1942 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1943 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1944 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1945 #define PF_DUMPCORE 0x00000200 /* dumped core */
1946 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1947 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1948 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1949 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1950 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1951 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1952 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1953 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1954 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1955 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1956 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1957 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1958 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1959 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1960 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1961 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1962 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1963 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1964 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1967 * Only the _current_ task can read/write to tsk->flags, but other
1968 * tasks can access tsk->flags in readonly mode for example
1969 * with tsk_used_math (like during threaded core dumping).
1970 * There is however an exception to this rule during ptrace
1971 * or during fork: the ptracer task is allowed to write to the
1972 * child->flags of its traced child (same goes for fork, the parent
1973 * can write to the child->flags), because we're guaranteed the
1974 * child is not running and in turn not changing child->flags
1975 * at the same time the parent does it.
1977 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1978 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1979 #define clear_used_math() clear_stopped_child_used_math(current)
1980 #define set_used_math() set_stopped_child_used_math(current)
1981 #define conditional_stopped_child_used_math(condition, child) \
1982 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1983 #define conditional_used_math(condition) \
1984 conditional_stopped_child_used_math(condition, current)
1985 #define copy_to_stopped_child_used_math(child) \
1986 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1987 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1988 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1989 #define used_math() tsk_used_math(current)
1991 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
1992 * __GFP_FS is also cleared as it implies __GFP_IO.
1994 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1996 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1997 flags &= ~(__GFP_IO | __GFP_FS);
2001 static inline unsigned int memalloc_noio_save(void)
2003 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2004 current->flags |= PF_MEMALLOC_NOIO;
2008 static inline void memalloc_noio_restore(unsigned int flags)
2010 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2013 /* Per-process atomic flags. */
2014 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2015 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2016 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2019 #define TASK_PFA_TEST(name, func) \
2020 static inline bool task_##func(struct task_struct *p) \
2021 { return test_bit(PFA_##name, &p->atomic_flags); }
2022 #define TASK_PFA_SET(name, func) \
2023 static inline void task_set_##func(struct task_struct *p) \
2024 { set_bit(PFA_##name, &p->atomic_flags); }
2025 #define TASK_PFA_CLEAR(name, func) \
2026 static inline void task_clear_##func(struct task_struct *p) \
2027 { clear_bit(PFA_##name, &p->atomic_flags); }
2029 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2030 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2032 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2033 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2034 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2036 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2037 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2038 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2041 * task->jobctl flags
2043 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2045 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2046 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2047 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2048 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2049 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2050 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2051 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2053 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
2054 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
2055 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
2056 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
2057 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
2058 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
2059 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
2061 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2062 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2064 extern bool task_set_jobctl_pending(struct task_struct *task,
2066 extern void task_clear_jobctl_trapping(struct task_struct *task);
2067 extern void task_clear_jobctl_pending(struct task_struct *task,
2070 static inline void rcu_copy_process(struct task_struct *p)
2072 #ifdef CONFIG_PREEMPT_RCU
2073 p->rcu_read_lock_nesting = 0;
2074 p->rcu_read_unlock_special.s = 0;
2075 p->rcu_blocked_node = NULL;
2076 INIT_LIST_HEAD(&p->rcu_node_entry);
2077 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2078 #ifdef CONFIG_TASKS_RCU
2079 p->rcu_tasks_holdout = false;
2080 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2081 p->rcu_tasks_idle_cpu = -1;
2082 #endif /* #ifdef CONFIG_TASKS_RCU */
2085 static inline void tsk_restore_flags(struct task_struct *task,
2086 unsigned long orig_flags, unsigned long flags)
2088 task->flags &= ~flags;
2089 task->flags |= orig_flags & flags;
2092 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2093 const struct cpumask *trial);
2094 extern int task_can_attach(struct task_struct *p,
2095 const struct cpumask *cs_cpus_allowed);
2097 extern void do_set_cpus_allowed(struct task_struct *p,
2098 const struct cpumask *new_mask);
2100 extern int set_cpus_allowed_ptr(struct task_struct *p,
2101 const struct cpumask *new_mask);
2103 static inline void do_set_cpus_allowed(struct task_struct *p,
2104 const struct cpumask *new_mask)
2107 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2108 const struct cpumask *new_mask)
2110 if (!cpumask_test_cpu(0, new_mask))
2116 #ifdef CONFIG_NO_HZ_COMMON
2117 void calc_load_enter_idle(void);
2118 void calc_load_exit_idle(void);
2120 static inline void calc_load_enter_idle(void) { }
2121 static inline void calc_load_exit_idle(void) { }
2122 #endif /* CONFIG_NO_HZ_COMMON */
2124 #ifndef CONFIG_CPUMASK_OFFSTACK
2125 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2127 return set_cpus_allowed_ptr(p, &new_mask);
2132 * Do not use outside of architecture code which knows its limitations.
2134 * sched_clock() has no promise of monotonicity or bounded drift between
2135 * CPUs, use (which you should not) requires disabling IRQs.
2137 * Please use one of the three interfaces below.
2139 extern unsigned long long notrace sched_clock(void);
2141 * See the comment in kernel/sched/clock.c
2143 extern u64 cpu_clock(int cpu);
2144 extern u64 local_clock(void);
2145 extern u64 sched_clock_cpu(int cpu);
2148 extern void sched_clock_init(void);
2150 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2151 static inline void sched_clock_tick(void)
2155 static inline void sched_clock_idle_sleep_event(void)
2159 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2164 * Architectures can set this to 1 if they have specified
2165 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2166 * but then during bootup it turns out that sched_clock()
2167 * is reliable after all:
2169 extern int sched_clock_stable(void);
2170 extern void set_sched_clock_stable(void);
2171 extern void clear_sched_clock_stable(void);
2173 extern void sched_clock_tick(void);
2174 extern void sched_clock_idle_sleep_event(void);
2175 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2178 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2180 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2181 * The reason for this explicit opt-in is not to have perf penalty with
2182 * slow sched_clocks.
2184 extern void enable_sched_clock_irqtime(void);
2185 extern void disable_sched_clock_irqtime(void);
2187 static inline void enable_sched_clock_irqtime(void) {}
2188 static inline void disable_sched_clock_irqtime(void) {}
2191 extern unsigned long long
2192 task_sched_runtime(struct task_struct *task);
2194 /* sched_exec is called by processes performing an exec */
2196 extern void sched_exec(void);
2198 #define sched_exec() {}
2201 extern void sched_clock_idle_sleep_event(void);
2202 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2204 #ifdef CONFIG_HOTPLUG_CPU
2205 extern void idle_task_exit(void);
2207 static inline void idle_task_exit(void) {}
2210 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2211 extern void wake_up_nohz_cpu(int cpu);
2213 static inline void wake_up_nohz_cpu(int cpu) { }
2216 #ifdef CONFIG_NO_HZ_FULL
2217 extern bool sched_can_stop_tick(void);
2218 extern u64 scheduler_tick_max_deferment(void);
2220 static inline bool sched_can_stop_tick(void) { return false; }
2223 #ifdef CONFIG_SCHED_AUTOGROUP
2224 extern void sched_autogroup_create_attach(struct task_struct *p);
2225 extern void sched_autogroup_detach(struct task_struct *p);
2226 extern void sched_autogroup_fork(struct signal_struct *sig);
2227 extern void sched_autogroup_exit(struct signal_struct *sig);
2228 #ifdef CONFIG_PROC_FS
2229 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2230 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2233 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2234 static inline void sched_autogroup_detach(struct task_struct *p) { }
2235 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2236 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2239 extern int yield_to(struct task_struct *p, bool preempt);
2240 extern void set_user_nice(struct task_struct *p, long nice);
2241 extern int task_prio(const struct task_struct *p);
2243 * task_nice - return the nice value of a given task.
2244 * @p: the task in question.
2246 * Return: The nice value [ -20 ... 0 ... 19 ].
2248 static inline int task_nice(const struct task_struct *p)
2250 return PRIO_TO_NICE((p)->static_prio);
2252 extern int can_nice(const struct task_struct *p, const int nice);
2253 extern int task_curr(const struct task_struct *p);
2254 extern int idle_cpu(int cpu);
2255 extern int sched_setscheduler(struct task_struct *, int,
2256 const struct sched_param *);
2257 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2258 const struct sched_param *);
2259 extern int sched_setattr(struct task_struct *,
2260 const struct sched_attr *);
2261 extern struct task_struct *idle_task(int cpu);
2263 * is_idle_task - is the specified task an idle task?
2264 * @p: the task in question.
2266 * Return: 1 if @p is an idle task. 0 otherwise.
2268 static inline bool is_idle_task(const struct task_struct *p)
2272 extern struct task_struct *curr_task(int cpu);
2273 extern void set_curr_task(int cpu, struct task_struct *p);
2278 * The default (Linux) execution domain.
2280 extern struct exec_domain default_exec_domain;
2282 union thread_union {
2283 struct thread_info thread_info;
2284 unsigned long stack[THREAD_SIZE/sizeof(long)];
2287 #ifndef __HAVE_ARCH_KSTACK_END
2288 static inline int kstack_end(void *addr)
2290 /* Reliable end of stack detection:
2291 * Some APM bios versions misalign the stack
2293 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2297 extern union thread_union init_thread_union;
2298 extern struct task_struct init_task;
2300 extern struct mm_struct init_mm;
2302 extern struct pid_namespace init_pid_ns;
2305 * find a task by one of its numerical ids
2307 * find_task_by_pid_ns():
2308 * finds a task by its pid in the specified namespace
2309 * find_task_by_vpid():
2310 * finds a task by its virtual pid
2312 * see also find_vpid() etc in include/linux/pid.h
2315 extern struct task_struct *find_task_by_vpid(pid_t nr);
2316 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2317 struct pid_namespace *ns);
2319 /* per-UID process charging. */
2320 extern struct user_struct * alloc_uid(kuid_t);
2321 static inline struct user_struct *get_uid(struct user_struct *u)
2323 atomic_inc(&u->__count);
2326 extern void free_uid(struct user_struct *);
2328 #include <asm/current.h>
2330 extern void xtime_update(unsigned long ticks);
2332 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2333 extern int wake_up_process(struct task_struct *tsk);
2334 extern void wake_up_new_task(struct task_struct *tsk);
2336 extern void kick_process(struct task_struct *tsk);
2338 static inline void kick_process(struct task_struct *tsk) { }
2340 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2341 extern void sched_dead(struct task_struct *p);
2343 extern void proc_caches_init(void);
2344 extern void flush_signals(struct task_struct *);
2345 extern void __flush_signals(struct task_struct *);
2346 extern void ignore_signals(struct task_struct *);
2347 extern void flush_signal_handlers(struct task_struct *, int force_default);
2348 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2350 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2352 unsigned long flags;
2355 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2356 ret = dequeue_signal(tsk, mask, info);
2357 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2362 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2364 extern void unblock_all_signals(void);
2365 extern void release_task(struct task_struct * p);
2366 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2367 extern int force_sigsegv(int, struct task_struct *);
2368 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2369 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2370 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2371 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2372 const struct cred *, u32);
2373 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2374 extern int kill_pid(struct pid *pid, int sig, int priv);
2375 extern int kill_proc_info(int, struct siginfo *, pid_t);
2376 extern __must_check bool do_notify_parent(struct task_struct *, int);
2377 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2378 extern void force_sig(int, struct task_struct *);
2379 extern int send_sig(int, struct task_struct *, int);
2380 extern int zap_other_threads(struct task_struct *p);
2381 extern struct sigqueue *sigqueue_alloc(void);
2382 extern void sigqueue_free(struct sigqueue *);
2383 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2384 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2386 static inline void restore_saved_sigmask(void)
2388 if (test_and_clear_restore_sigmask())
2389 __set_current_blocked(¤t->saved_sigmask);
2392 static inline sigset_t *sigmask_to_save(void)
2394 sigset_t *res = ¤t->blocked;
2395 if (unlikely(test_restore_sigmask()))
2396 res = ¤t->saved_sigmask;
2400 static inline int kill_cad_pid(int sig, int priv)
2402 return kill_pid(cad_pid, sig, priv);
2405 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2406 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2407 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2408 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2411 * True if we are on the alternate signal stack.
2413 static inline int on_sig_stack(unsigned long sp)
2415 #ifdef CONFIG_STACK_GROWSUP
2416 return sp >= current->sas_ss_sp &&
2417 sp - current->sas_ss_sp < current->sas_ss_size;
2419 return sp > current->sas_ss_sp &&
2420 sp - current->sas_ss_sp <= current->sas_ss_size;
2424 static inline int sas_ss_flags(unsigned long sp)
2426 if (!current->sas_ss_size)
2429 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2432 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2434 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2435 #ifdef CONFIG_STACK_GROWSUP
2436 return current->sas_ss_sp;
2438 return current->sas_ss_sp + current->sas_ss_size;
2444 * Routines for handling mm_structs
2446 extern struct mm_struct * mm_alloc(void);
2448 /* mmdrop drops the mm and the page tables */
2449 extern void __mmdrop(struct mm_struct *);
2450 static inline void mmdrop(struct mm_struct * mm)
2452 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2456 /* mmput gets rid of the mappings and all user-space */
2457 extern void mmput(struct mm_struct *);
2458 /* Grab a reference to a task's mm, if it is not already going away */
2459 extern struct mm_struct *get_task_mm(struct task_struct *task);
2461 * Grab a reference to a task's mm, if it is not already going away
2462 * and ptrace_may_access with the mode parameter passed to it
2465 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2466 /* Remove the current tasks stale references to the old mm_struct */
2467 extern void mm_release(struct task_struct *, struct mm_struct *);
2469 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2470 struct task_struct *);
2471 extern void flush_thread(void);
2472 extern void exit_thread(void);
2474 extern void exit_files(struct task_struct *);
2475 extern void __cleanup_sighand(struct sighand_struct *);
2477 extern void exit_itimers(struct signal_struct *);
2478 extern void flush_itimer_signals(void);
2480 extern void do_group_exit(int);
2482 extern int do_execve(struct filename *,
2483 const char __user * const __user *,
2484 const char __user * const __user *);
2485 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2486 struct task_struct *fork_idle(int);
2487 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2489 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2490 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2492 __set_task_comm(tsk, from, false);
2494 extern char *get_task_comm(char *to, struct task_struct *tsk);
2497 void scheduler_ipi(void);
2498 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2500 static inline void scheduler_ipi(void) { }
2501 static inline unsigned long wait_task_inactive(struct task_struct *p,
2508 #define next_task(p) \
2509 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2511 #define for_each_process(p) \
2512 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2514 extern bool current_is_single_threaded(void);
2517 * Careful: do_each_thread/while_each_thread is a double loop so
2518 * 'break' will not work as expected - use goto instead.
2520 #define do_each_thread(g, t) \
2521 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2523 #define while_each_thread(g, t) \
2524 while ((t = next_thread(t)) != g)
2526 #define __for_each_thread(signal, t) \
2527 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2529 #define for_each_thread(p, t) \
2530 __for_each_thread((p)->signal, t)
2532 /* Careful: this is a double loop, 'break' won't work as expected. */
2533 #define for_each_process_thread(p, t) \
2534 for_each_process(p) for_each_thread(p, t)
2536 static inline int get_nr_threads(struct task_struct *tsk)
2538 return tsk->signal->nr_threads;
2541 static inline bool thread_group_leader(struct task_struct *p)
2543 return p->exit_signal >= 0;
2546 /* Do to the insanities of de_thread it is possible for a process
2547 * to have the pid of the thread group leader without actually being
2548 * the thread group leader. For iteration through the pids in proc
2549 * all we care about is that we have a task with the appropriate
2550 * pid, we don't actually care if we have the right task.
2552 static inline bool has_group_leader_pid(struct task_struct *p)
2554 return task_pid(p) == p->signal->leader_pid;
2558 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2560 return p1->signal == p2->signal;
2563 static inline struct task_struct *next_thread(const struct task_struct *p)
2565 return list_entry_rcu(p->thread_group.next,
2566 struct task_struct, thread_group);
2569 static inline int thread_group_empty(struct task_struct *p)
2571 return list_empty(&p->thread_group);
2574 #define delay_group_leader(p) \
2575 (thread_group_leader(p) && !thread_group_empty(p))
2578 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2579 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2580 * pins the final release of task.io_context. Also protects ->cpuset and
2581 * ->cgroup.subsys[]. And ->vfork_done.
2583 * Nests both inside and outside of read_lock(&tasklist_lock).
2584 * It must not be nested with write_lock_irq(&tasklist_lock),
2585 * neither inside nor outside.
2587 static inline void task_lock(struct task_struct *p)
2589 spin_lock(&p->alloc_lock);
2592 static inline void task_unlock(struct task_struct *p)
2594 spin_unlock(&p->alloc_lock);
2597 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2598 unsigned long *flags);
2600 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2601 unsigned long *flags)
2603 struct sighand_struct *ret;
2605 ret = __lock_task_sighand(tsk, flags);
2606 (void)__cond_lock(&tsk->sighand->siglock, ret);
2610 static inline void unlock_task_sighand(struct task_struct *tsk,
2611 unsigned long *flags)
2613 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2616 #ifdef CONFIG_CGROUPS
2617 static inline void threadgroup_change_begin(struct task_struct *tsk)
2619 down_read(&tsk->signal->group_rwsem);
2621 static inline void threadgroup_change_end(struct task_struct *tsk)
2623 up_read(&tsk->signal->group_rwsem);
2627 * threadgroup_lock - lock threadgroup
2628 * @tsk: member task of the threadgroup to lock
2630 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2631 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2632 * change ->group_leader/pid. This is useful for cases where the threadgroup
2633 * needs to stay stable across blockable operations.
2635 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2636 * synchronization. While held, no new task will be added to threadgroup
2637 * and no existing live task will have its PF_EXITING set.
2639 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2640 * sub-thread becomes a new leader.
2642 static inline void threadgroup_lock(struct task_struct *tsk)
2644 down_write(&tsk->signal->group_rwsem);
2648 * threadgroup_unlock - unlock threadgroup
2649 * @tsk: member task of the threadgroup to unlock
2651 * Reverse threadgroup_lock().
2653 static inline void threadgroup_unlock(struct task_struct *tsk)
2655 up_write(&tsk->signal->group_rwsem);
2658 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2659 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2660 static inline void threadgroup_lock(struct task_struct *tsk) {}
2661 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2664 #ifndef __HAVE_THREAD_FUNCTIONS
2666 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2667 #define task_stack_page(task) ((task)->stack)
2669 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2671 *task_thread_info(p) = *task_thread_info(org);
2672 task_thread_info(p)->task = p;
2676 * Return the address of the last usable long on the stack.
2678 * When the stack grows down, this is just above the thread
2679 * info struct. Going any lower will corrupt the threadinfo.
2681 * When the stack grows up, this is the highest address.
2682 * Beyond that position, we corrupt data on the next page.
2684 static inline unsigned long *end_of_stack(struct task_struct *p)
2686 #ifdef CONFIG_STACK_GROWSUP
2687 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2689 return (unsigned long *)(task_thread_info(p) + 1);
2694 #define task_stack_end_corrupted(task) \
2695 (*(end_of_stack(task)) != STACK_END_MAGIC)
2697 static inline int object_is_on_stack(void *obj)
2699 void *stack = task_stack_page(current);
2701 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2704 extern void thread_info_cache_init(void);
2706 #ifdef CONFIG_DEBUG_STACK_USAGE
2707 static inline unsigned long stack_not_used(struct task_struct *p)
2709 unsigned long *n = end_of_stack(p);
2711 do { /* Skip over canary */
2715 return (unsigned long)n - (unsigned long)end_of_stack(p);
2718 extern void set_task_stack_end_magic(struct task_struct *tsk);
2720 /* set thread flags in other task's structures
2721 * - see asm/thread_info.h for TIF_xxxx flags available
2723 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2725 set_ti_thread_flag(task_thread_info(tsk), flag);
2728 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2730 clear_ti_thread_flag(task_thread_info(tsk), flag);
2733 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2735 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2738 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2740 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2743 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2745 return test_ti_thread_flag(task_thread_info(tsk), flag);
2748 static inline void set_tsk_need_resched(struct task_struct *tsk)
2750 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2753 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2755 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2758 static inline int test_tsk_need_resched(struct task_struct *tsk)
2760 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2763 static inline int restart_syscall(void)
2765 set_tsk_thread_flag(current, TIF_SIGPENDING);
2766 return -ERESTARTNOINTR;
2769 static inline int signal_pending(struct task_struct *p)
2771 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2774 static inline int __fatal_signal_pending(struct task_struct *p)
2776 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2779 static inline int fatal_signal_pending(struct task_struct *p)
2781 return signal_pending(p) && __fatal_signal_pending(p);
2784 static inline int signal_pending_state(long state, struct task_struct *p)
2786 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2788 if (!signal_pending(p))
2791 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2795 * cond_resched() and cond_resched_lock(): latency reduction via
2796 * explicit rescheduling in places that are safe. The return
2797 * value indicates whether a reschedule was done in fact.
2798 * cond_resched_lock() will drop the spinlock before scheduling,
2799 * cond_resched_softirq() will enable bhs before scheduling.
2801 extern int _cond_resched(void);
2803 #define cond_resched() ({ \
2804 ___might_sleep(__FILE__, __LINE__, 0); \
2808 extern int __cond_resched_lock(spinlock_t *lock);
2810 #ifdef CONFIG_PREEMPT_COUNT
2811 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2813 #define PREEMPT_LOCK_OFFSET 0
2816 #define cond_resched_lock(lock) ({ \
2817 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2818 __cond_resched_lock(lock); \
2821 extern int __cond_resched_softirq(void);
2823 #define cond_resched_softirq() ({ \
2824 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2825 __cond_resched_softirq(); \
2828 static inline void cond_resched_rcu(void)
2830 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2838 * Does a critical section need to be broken due to another
2839 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2840 * but a general need for low latency)
2842 static inline int spin_needbreak(spinlock_t *lock)
2844 #ifdef CONFIG_PREEMPT
2845 return spin_is_contended(lock);
2852 * Idle thread specific functions to determine the need_resched
2855 #ifdef TIF_POLLING_NRFLAG
2856 static inline int tsk_is_polling(struct task_struct *p)
2858 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2861 static inline void __current_set_polling(void)
2863 set_thread_flag(TIF_POLLING_NRFLAG);
2866 static inline bool __must_check current_set_polling_and_test(void)
2868 __current_set_polling();
2871 * Polling state must be visible before we test NEED_RESCHED,
2872 * paired by resched_curr()
2874 smp_mb__after_atomic();
2876 return unlikely(tif_need_resched());
2879 static inline void __current_clr_polling(void)
2881 clear_thread_flag(TIF_POLLING_NRFLAG);
2884 static inline bool __must_check current_clr_polling_and_test(void)
2886 __current_clr_polling();
2889 * Polling state must be visible before we test NEED_RESCHED,
2890 * paired by resched_curr()
2892 smp_mb__after_atomic();
2894 return unlikely(tif_need_resched());
2898 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2899 static inline void __current_set_polling(void) { }
2900 static inline void __current_clr_polling(void) { }
2902 static inline bool __must_check current_set_polling_and_test(void)
2904 return unlikely(tif_need_resched());
2906 static inline bool __must_check current_clr_polling_and_test(void)
2908 return unlikely(tif_need_resched());
2912 static inline void current_clr_polling(void)
2914 __current_clr_polling();
2917 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2918 * Once the bit is cleared, we'll get IPIs with every new
2919 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2922 smp_mb(); /* paired with resched_curr() */
2924 preempt_fold_need_resched();
2927 static __always_inline bool need_resched(void)
2929 return unlikely(tif_need_resched());
2933 * Thread group CPU time accounting.
2935 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2936 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2938 static inline void thread_group_cputime_init(struct signal_struct *sig)
2940 raw_spin_lock_init(&sig->cputimer.lock);
2944 * Reevaluate whether the task has signals pending delivery.
2945 * Wake the task if so.
2946 * This is required every time the blocked sigset_t changes.
2947 * callers must hold sighand->siglock.
2949 extern void recalc_sigpending_and_wake(struct task_struct *t);
2950 extern void recalc_sigpending(void);
2952 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2954 static inline void signal_wake_up(struct task_struct *t, bool resume)
2956 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2958 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2960 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2964 * Wrappers for p->thread_info->cpu access. No-op on UP.
2968 static inline unsigned int task_cpu(const struct task_struct *p)
2970 return task_thread_info(p)->cpu;
2973 static inline int task_node(const struct task_struct *p)
2975 return cpu_to_node(task_cpu(p));
2978 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2982 static inline unsigned int task_cpu(const struct task_struct *p)
2987 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2991 #endif /* CONFIG_SMP */
2993 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2994 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2996 #ifdef CONFIG_CGROUP_SCHED
2997 extern struct task_group root_task_group;
2998 #endif /* CONFIG_CGROUP_SCHED */
3000 extern int task_can_switch_user(struct user_struct *up,
3001 struct task_struct *tsk);
3003 #ifdef CONFIG_TASK_XACCT
3004 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3006 tsk->ioac.rchar += amt;
3009 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3011 tsk->ioac.wchar += amt;
3014 static inline void inc_syscr(struct task_struct *tsk)
3019 static inline void inc_syscw(struct task_struct *tsk)
3024 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3028 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3032 static inline void inc_syscr(struct task_struct *tsk)
3036 static inline void inc_syscw(struct task_struct *tsk)
3041 #ifndef TASK_SIZE_OF
3042 #define TASK_SIZE_OF(tsk) TASK_SIZE
3046 extern void mm_update_next_owner(struct mm_struct *mm);
3048 static inline void mm_update_next_owner(struct mm_struct *mm)
3051 #endif /* CONFIG_MEMCG */
3053 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3056 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
3059 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3062 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
3065 static inline unsigned long rlimit(unsigned int limit)
3067 return task_rlimit(current, limit);
3070 static inline unsigned long rlimit_max(unsigned int limit)
3072 return task_rlimit_max(current, limit);