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 */
128 struct futex_pi_state;
129 struct robust_list_head;
132 struct perf_event_context;
136 #define VMACACHE_BITS 2
137 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
138 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
141 * These are the constant used to fake the fixed-point load-average
142 * counting. Some notes:
143 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
144 * a load-average precision of 10 bits integer + 11 bits fractional
145 * - if you want to count load-averages more often, you need more
146 * precision, or rounding will get you. With 2-second counting freq,
147 * the EXP_n values would be 1981, 2034 and 2043 if still using only
150 extern unsigned long avenrun[]; /* Load averages */
151 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
153 #define FSHIFT 11 /* nr of bits of precision */
154 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
155 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
156 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
157 #define EXP_5 2014 /* 1/exp(5sec/5min) */
158 #define EXP_15 2037 /* 1/exp(5sec/15min) */
160 #define CALC_LOAD(load,exp,n) \
162 load += n*(FIXED_1-exp); \
165 extern unsigned long total_forks;
166 extern int nr_threads;
167 DECLARE_PER_CPU(unsigned long, process_counts);
168 extern int nr_processes(void);
169 extern unsigned long nr_running(void);
170 extern bool single_task_running(void);
171 extern unsigned long nr_iowait(void);
172 extern unsigned long nr_iowait_cpu(int cpu);
173 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
175 extern void calc_global_load(unsigned long ticks);
176 extern void update_cpu_load_nohz(void);
178 /* Notifier for when a task gets migrated to a new CPU */
179 struct task_migration_notifier {
180 struct task_struct *task;
184 extern void register_task_migration_notifier(struct notifier_block *n);
186 extern unsigned long get_parent_ip(unsigned long addr);
188 extern void dump_cpu_task(int cpu);
193 #ifdef CONFIG_SCHED_DEBUG
194 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
195 extern void proc_sched_set_task(struct task_struct *p);
197 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
201 * Task state bitmask. NOTE! These bits are also
202 * encoded in fs/proc/array.c: get_task_state().
204 * We have two separate sets of flags: task->state
205 * is about runnability, while task->exit_state are
206 * about the task exiting. Confusing, but this way
207 * modifying one set can't modify the other one by
210 #define TASK_RUNNING 0
211 #define TASK_INTERRUPTIBLE 1
212 #define TASK_UNINTERRUPTIBLE 2
213 #define __TASK_STOPPED 4
214 #define __TASK_TRACED 8
215 /* in tsk->exit_state */
217 #define EXIT_ZOMBIE 32
218 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
219 /* in tsk->state again */
221 #define TASK_WAKEKILL 128
222 #define TASK_WAKING 256
223 #define TASK_PARKED 512
224 #define TASK_STATE_MAX 1024
226 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
228 extern char ___assert_task_state[1 - 2*!!(
229 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
231 /* Convenience macros for the sake of set_task_state */
232 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
233 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
234 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
236 /* Convenience macros for the sake of wake_up */
237 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
238 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
240 /* get_task_state() */
241 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
242 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
243 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
245 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
246 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
247 #define task_is_stopped_or_traced(task) \
248 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
249 #define task_contributes_to_load(task) \
250 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
251 (task->flags & PF_FROZEN) == 0)
253 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
255 #define __set_task_state(tsk, state_value) \
257 (tsk)->task_state_change = _THIS_IP_; \
258 (tsk)->state = (state_value); \
260 #define set_task_state(tsk, state_value) \
262 (tsk)->task_state_change = _THIS_IP_; \
263 set_mb((tsk)->state, (state_value)); \
267 * set_current_state() includes a barrier so that the write of current->state
268 * is correctly serialised wrt the caller's subsequent test of whether to
271 * set_current_state(TASK_UNINTERRUPTIBLE);
272 * if (do_i_need_to_sleep())
275 * If the caller does not need such serialisation then use __set_current_state()
277 #define __set_current_state(state_value) \
279 current->task_state_change = _THIS_IP_; \
280 current->state = (state_value); \
282 #define set_current_state(state_value) \
284 current->task_state_change = _THIS_IP_; \
285 set_mb(current->state, (state_value)); \
290 #define __set_task_state(tsk, state_value) \
291 do { (tsk)->state = (state_value); } while (0)
292 #define set_task_state(tsk, state_value) \
293 set_mb((tsk)->state, (state_value))
296 * set_current_state() includes a barrier so that the write of current->state
297 * is correctly serialised wrt the caller's subsequent test of whether to
300 * set_current_state(TASK_UNINTERRUPTIBLE);
301 * if (do_i_need_to_sleep())
304 * If the caller does not need such serialisation then use __set_current_state()
306 #define __set_current_state(state_value) \
307 do { current->state = (state_value); } while (0)
308 #define set_current_state(state_value) \
309 set_mb(current->state, (state_value))
313 /* Task command name length */
314 #define TASK_COMM_LEN 16
316 #include <linux/spinlock.h>
319 * This serializes "schedule()" and also protects
320 * the run-queue from deletions/modifications (but
321 * _adding_ to the beginning of the run-queue has
324 extern rwlock_t tasklist_lock;
325 extern spinlock_t mmlist_lock;
329 #ifdef CONFIG_PROVE_RCU
330 extern int lockdep_tasklist_lock_is_held(void);
331 #endif /* #ifdef CONFIG_PROVE_RCU */
333 extern void sched_init(void);
334 extern void sched_init_smp(void);
335 extern asmlinkage void schedule_tail(struct task_struct *prev);
336 extern void init_idle(struct task_struct *idle, int cpu);
337 extern void init_idle_bootup_task(struct task_struct *idle);
339 extern cpumask_var_t cpu_isolated_map;
341 extern int runqueue_is_locked(int cpu);
343 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
344 extern void nohz_balance_enter_idle(int cpu);
345 extern void set_cpu_sd_state_idle(void);
346 extern int get_nohz_timer_target(int pinned);
348 static inline void nohz_balance_enter_idle(int cpu) { }
349 static inline void set_cpu_sd_state_idle(void) { }
350 static inline int get_nohz_timer_target(int pinned)
352 return smp_processor_id();
357 * Only dump TASK_* tasks. (0 for all tasks)
359 extern void show_state_filter(unsigned long state_filter);
361 static inline void show_state(void)
363 show_state_filter(0);
366 extern void show_regs(struct pt_regs *);
369 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
370 * task), SP is the stack pointer of the first frame that should be shown in the back
371 * trace (or NULL if the entire call-chain of the task should be shown).
373 extern void show_stack(struct task_struct *task, unsigned long *sp);
375 extern void cpu_init (void);
376 extern void trap_init(void);
377 extern void update_process_times(int user);
378 extern void scheduler_tick(void);
380 extern void sched_show_task(struct task_struct *p);
382 #ifdef CONFIG_LOCKUP_DETECTOR
383 extern void touch_softlockup_watchdog(void);
384 extern void touch_softlockup_watchdog_sync(void);
385 extern void touch_all_softlockup_watchdogs(void);
386 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
388 size_t *lenp, loff_t *ppos);
389 extern unsigned int softlockup_panic;
390 void lockup_detector_init(void);
392 static inline void touch_softlockup_watchdog(void)
395 static inline void touch_softlockup_watchdog_sync(void)
398 static inline void touch_all_softlockup_watchdogs(void)
401 static inline void lockup_detector_init(void)
406 #ifdef CONFIG_DETECT_HUNG_TASK
407 void reset_hung_task_detector(void);
409 static inline void reset_hung_task_detector(void)
414 /* Attach to any functions which should be ignored in wchan output. */
415 #define __sched __attribute__((__section__(".sched.text")))
417 /* Linker adds these: start and end of __sched functions */
418 extern char __sched_text_start[], __sched_text_end[];
420 /* Is this address in the __sched functions? */
421 extern int in_sched_functions(unsigned long addr);
423 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
424 extern signed long schedule_timeout(signed long timeout);
425 extern signed long schedule_timeout_interruptible(signed long timeout);
426 extern signed long schedule_timeout_killable(signed long timeout);
427 extern signed long schedule_timeout_uninterruptible(signed long timeout);
428 asmlinkage void schedule(void);
429 extern void schedule_preempt_disabled(void);
431 extern long io_schedule_timeout(long timeout);
433 static inline void io_schedule(void)
435 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
439 struct user_namespace;
442 extern void arch_pick_mmap_layout(struct mm_struct *mm);
444 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
445 unsigned long, unsigned long);
447 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
448 unsigned long len, unsigned long pgoff,
449 unsigned long flags);
451 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
454 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
455 #define SUID_DUMP_USER 1 /* Dump as user of process */
456 #define SUID_DUMP_ROOT 2 /* Dump as root */
460 /* for SUID_DUMP_* above */
461 #define MMF_DUMPABLE_BITS 2
462 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
464 extern void set_dumpable(struct mm_struct *mm, int value);
466 * This returns the actual value of the suid_dumpable flag. For things
467 * that are using this for checking for privilege transitions, it must
468 * test against SUID_DUMP_USER rather than treating it as a boolean
471 static inline int __get_dumpable(unsigned long mm_flags)
473 return mm_flags & MMF_DUMPABLE_MASK;
476 static inline int get_dumpable(struct mm_struct *mm)
478 return __get_dumpable(mm->flags);
481 /* coredump filter bits */
482 #define MMF_DUMP_ANON_PRIVATE 2
483 #define MMF_DUMP_ANON_SHARED 3
484 #define MMF_DUMP_MAPPED_PRIVATE 4
485 #define MMF_DUMP_MAPPED_SHARED 5
486 #define MMF_DUMP_ELF_HEADERS 6
487 #define MMF_DUMP_HUGETLB_PRIVATE 7
488 #define MMF_DUMP_HUGETLB_SHARED 8
490 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
491 #define MMF_DUMP_FILTER_BITS 7
492 #define MMF_DUMP_FILTER_MASK \
493 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
494 #define MMF_DUMP_FILTER_DEFAULT \
495 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
496 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
498 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
499 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
501 # define MMF_DUMP_MASK_DEFAULT_ELF 0
503 /* leave room for more dump flags */
504 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
505 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
506 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
508 #define MMF_HAS_UPROBES 19 /* has uprobes */
509 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
511 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
513 struct sighand_struct {
515 struct k_sigaction action[_NSIG];
517 wait_queue_head_t signalfd_wqh;
520 struct pacct_struct {
523 unsigned long ac_mem;
524 cputime_t ac_utime, ac_stime;
525 unsigned long ac_minflt, ac_majflt;
536 * struct cputime - snaphsot of system and user cputime
537 * @utime: time spent in user mode
538 * @stime: time spent in system mode
540 * Gathers a generic snapshot of user and system time.
548 * struct task_cputime - collected CPU time counts
549 * @utime: time spent in user mode, in &cputime_t units
550 * @stime: time spent in kernel mode, in &cputime_t units
551 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
553 * This is an extension of struct cputime that includes the total runtime
554 * spent by the task from the scheduler point of view.
556 * As a result, this structure groups together three kinds of CPU time
557 * that are tracked for threads and thread groups. Most things considering
558 * CPU time want to group these counts together and treat all three
559 * of them in parallel.
561 struct task_cputime {
564 unsigned long long sum_exec_runtime;
566 /* Alternate field names when used to cache expirations. */
567 #define prof_exp stime
568 #define virt_exp utime
569 #define sched_exp sum_exec_runtime
571 #define INIT_CPUTIME \
572 (struct task_cputime) { \
575 .sum_exec_runtime = 0, \
578 #ifdef CONFIG_PREEMPT_COUNT
579 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
581 #define PREEMPT_DISABLED PREEMPT_ENABLED
585 * Disable preemption until the scheduler is running.
586 * Reset by start_kernel()->sched_init()->init_idle().
588 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
589 * before the scheduler is active -- see should_resched().
591 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
594 * struct thread_group_cputimer - thread group interval timer counts
595 * @cputime: thread group interval timers.
596 * @running: non-zero when there are timers running and
597 * @cputime receives updates.
598 * @lock: lock for fields in this struct.
600 * This structure contains the version of task_cputime, above, that is
601 * used for thread group CPU timer calculations.
603 struct thread_group_cputimer {
604 struct task_cputime cputime;
609 #include <linux/rwsem.h>
613 * NOTE! "signal_struct" does not have its own
614 * locking, because a shared signal_struct always
615 * implies a shared sighand_struct, so locking
616 * sighand_struct is always a proper superset of
617 * the locking of signal_struct.
619 struct signal_struct {
623 struct list_head thread_head;
625 wait_queue_head_t wait_chldexit; /* for wait4() */
627 /* current thread group signal load-balancing target: */
628 struct task_struct *curr_target;
630 /* shared signal handling: */
631 struct sigpending shared_pending;
633 /* thread group exit support */
636 * - notify group_exit_task when ->count is equal to notify_count
637 * - everyone except group_exit_task is stopped during signal delivery
638 * of fatal signals, group_exit_task processes the signal.
641 struct task_struct *group_exit_task;
643 /* thread group stop support, overloads group_exit_code too */
644 int group_stop_count;
645 unsigned int flags; /* see SIGNAL_* flags below */
648 * PR_SET_CHILD_SUBREAPER marks a process, like a service
649 * manager, to re-parent orphan (double-forking) child processes
650 * to this process instead of 'init'. The service manager is
651 * able to receive SIGCHLD signals and is able to investigate
652 * the process until it calls wait(). All children of this
653 * process will inherit a flag if they should look for a
654 * child_subreaper process at exit.
656 unsigned int is_child_subreaper:1;
657 unsigned int has_child_subreaper:1;
659 /* POSIX.1b Interval Timers */
661 struct list_head posix_timers;
663 /* ITIMER_REAL timer for the process */
664 struct hrtimer real_timer;
665 struct pid *leader_pid;
666 ktime_t it_real_incr;
669 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
670 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
671 * values are defined to 0 and 1 respectively
673 struct cpu_itimer it[2];
676 * Thread group totals for process CPU timers.
677 * See thread_group_cputimer(), et al, for details.
679 struct thread_group_cputimer cputimer;
681 /* Earliest-expiration cache. */
682 struct task_cputime cputime_expires;
684 struct list_head cpu_timers[3];
686 struct pid *tty_old_pgrp;
688 /* boolean value for session group leader */
691 struct tty_struct *tty; /* NULL if no tty */
693 #ifdef CONFIG_SCHED_AUTOGROUP
694 struct autogroup *autogroup;
697 * Cumulative resource counters for dead threads in the group,
698 * and for reaped dead child processes forked by this group.
699 * Live threads maintain their own counters and add to these
700 * in __exit_signal, except for the group leader.
702 seqlock_t stats_lock;
703 cputime_t utime, stime, cutime, cstime;
706 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
707 struct cputime prev_cputime;
709 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
710 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
711 unsigned long inblock, oublock, cinblock, coublock;
712 unsigned long maxrss, cmaxrss;
713 struct task_io_accounting ioac;
716 * Cumulative ns of schedule CPU time fo dead threads in the
717 * group, not including a zombie group leader, (This only differs
718 * from jiffies_to_ns(utime + stime) if sched_clock uses something
719 * other than jiffies.)
721 unsigned long long sum_sched_runtime;
724 * We don't bother to synchronize most readers of this at all,
725 * because there is no reader checking a limit that actually needs
726 * to get both rlim_cur and rlim_max atomically, and either one
727 * alone is a single word that can safely be read normally.
728 * getrlimit/setrlimit use task_lock(current->group_leader) to
729 * protect this instead of the siglock, because they really
730 * have no need to disable irqs.
732 struct rlimit rlim[RLIM_NLIMITS];
734 #ifdef CONFIG_BSD_PROCESS_ACCT
735 struct pacct_struct pacct; /* per-process accounting information */
737 #ifdef CONFIG_TASKSTATS
738 struct taskstats *stats;
742 unsigned audit_tty_log_passwd;
743 struct tty_audit_buf *tty_audit_buf;
745 #ifdef CONFIG_CGROUPS
747 * group_rwsem prevents new tasks from entering the threadgroup and
748 * member tasks from exiting,a more specifically, setting of
749 * PF_EXITING. fork and exit paths are protected with this rwsem
750 * using threadgroup_change_begin/end(). Users which require
751 * threadgroup to remain stable should use threadgroup_[un]lock()
752 * which also takes care of exec path. Currently, cgroup is the
755 struct rw_semaphore group_rwsem;
758 oom_flags_t oom_flags;
759 short oom_score_adj; /* OOM kill score adjustment */
760 short oom_score_adj_min; /* OOM kill score adjustment min value.
761 * Only settable by CAP_SYS_RESOURCE. */
763 struct mutex cred_guard_mutex; /* guard against foreign influences on
764 * credential calculations
765 * (notably. ptrace) */
769 * Bits in flags field of signal_struct.
771 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
772 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
773 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
774 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
776 * Pending notifications to parent.
778 #define SIGNAL_CLD_STOPPED 0x00000010
779 #define SIGNAL_CLD_CONTINUED 0x00000020
780 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
782 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
784 /* If true, all threads except ->group_exit_task have pending SIGKILL */
785 static inline int signal_group_exit(const struct signal_struct *sig)
787 return (sig->flags & SIGNAL_GROUP_EXIT) ||
788 (sig->group_exit_task != NULL);
792 * Some day this will be a full-fledged user tracking system..
795 atomic_t __count; /* reference count */
796 atomic_t processes; /* How many processes does this user have? */
797 atomic_t sigpending; /* How many pending signals does this user have? */
798 #ifdef CONFIG_INOTIFY_USER
799 atomic_t inotify_watches; /* How many inotify watches does this user have? */
800 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
802 #ifdef CONFIG_FANOTIFY
803 atomic_t fanotify_listeners;
806 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
808 #ifdef CONFIG_POSIX_MQUEUE
809 /* protected by mq_lock */
810 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
812 unsigned long locked_shm; /* How many pages of mlocked shm ? */
815 struct key *uid_keyring; /* UID specific keyring */
816 struct key *session_keyring; /* UID's default session keyring */
819 /* Hash table maintenance information */
820 struct hlist_node uidhash_node;
823 #ifdef CONFIG_PERF_EVENTS
824 atomic_long_t locked_vm;
828 extern int uids_sysfs_init(void);
830 extern struct user_struct *find_user(kuid_t);
832 extern struct user_struct root_user;
833 #define INIT_USER (&root_user)
836 struct backing_dev_info;
837 struct reclaim_state;
839 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
841 /* cumulative counters */
842 unsigned long pcount; /* # of times run on this cpu */
843 unsigned long long run_delay; /* time spent waiting on a runqueue */
846 unsigned long long last_arrival,/* when we last ran on a cpu */
847 last_queued; /* when we were last queued to run */
849 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
851 #ifdef CONFIG_TASK_DELAY_ACCT
852 struct task_delay_info {
854 unsigned int flags; /* Private per-task flags */
856 /* For each stat XXX, add following, aligned appropriately
858 * struct timespec XXX_start, XXX_end;
862 * Atomicity of updates to XXX_delay, XXX_count protected by
863 * single lock above (split into XXX_lock if contention is an issue).
867 * XXX_count is incremented on every XXX operation, the delay
868 * associated with the operation is added to XXX_delay.
869 * XXX_delay contains the accumulated delay time in nanoseconds.
871 u64 blkio_start; /* Shared by blkio, swapin */
872 u64 blkio_delay; /* wait for sync block io completion */
873 u64 swapin_delay; /* wait for swapin block io completion */
874 u32 blkio_count; /* total count of the number of sync block */
875 /* io operations performed */
876 u32 swapin_count; /* total count of the number of swapin block */
877 /* io operations performed */
880 u64 freepages_delay; /* wait for memory reclaim */
881 u32 freepages_count; /* total count of memory reclaim */
883 #endif /* CONFIG_TASK_DELAY_ACCT */
885 static inline int sched_info_on(void)
887 #ifdef CONFIG_SCHEDSTATS
889 #elif defined(CONFIG_TASK_DELAY_ACCT)
890 extern int delayacct_on;
905 * Increase resolution of cpu_capacity calculations
907 #define SCHED_CAPACITY_SHIFT 10
908 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
911 * sched-domains (multiprocessor balancing) declarations:
914 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
915 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
916 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
917 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
918 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
919 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
920 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
921 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
922 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
923 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
924 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
925 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
926 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
927 #define SD_NUMA 0x4000 /* cross-node balancing */
929 #ifdef CONFIG_SCHED_SMT
930 static inline int cpu_smt_flags(void)
932 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
936 #ifdef CONFIG_SCHED_MC
937 static inline int cpu_core_flags(void)
939 return SD_SHARE_PKG_RESOURCES;
944 static inline int cpu_numa_flags(void)
950 struct sched_domain_attr {
951 int relax_domain_level;
954 #define SD_ATTR_INIT (struct sched_domain_attr) { \
955 .relax_domain_level = -1, \
958 extern int sched_domain_level_max;
962 struct sched_domain {
963 /* These fields must be setup */
964 struct sched_domain *parent; /* top domain must be null terminated */
965 struct sched_domain *child; /* bottom domain must be null terminated */
966 struct sched_group *groups; /* the balancing groups of the domain */
967 unsigned long min_interval; /* Minimum balance interval ms */
968 unsigned long max_interval; /* Maximum balance interval ms */
969 unsigned int busy_factor; /* less balancing by factor if busy */
970 unsigned int imbalance_pct; /* No balance until over watermark */
971 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
972 unsigned int busy_idx;
973 unsigned int idle_idx;
974 unsigned int newidle_idx;
975 unsigned int wake_idx;
976 unsigned int forkexec_idx;
977 unsigned int smt_gain;
979 int nohz_idle; /* NOHZ IDLE status */
980 int flags; /* See SD_* */
983 /* Runtime fields. */
984 unsigned long last_balance; /* init to jiffies. units in jiffies */
985 unsigned int balance_interval; /* initialise to 1. units in ms. */
986 unsigned int nr_balance_failed; /* initialise to 0 */
988 /* idle_balance() stats */
989 u64 max_newidle_lb_cost;
990 unsigned long next_decay_max_lb_cost;
992 #ifdef CONFIG_SCHEDSTATS
993 /* load_balance() stats */
994 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
995 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
996 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
997 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
998 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
999 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1000 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1001 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1003 /* Active load balancing */
1004 unsigned int alb_count;
1005 unsigned int alb_failed;
1006 unsigned int alb_pushed;
1008 /* SD_BALANCE_EXEC stats */
1009 unsigned int sbe_count;
1010 unsigned int sbe_balanced;
1011 unsigned int sbe_pushed;
1013 /* SD_BALANCE_FORK stats */
1014 unsigned int sbf_count;
1015 unsigned int sbf_balanced;
1016 unsigned int sbf_pushed;
1018 /* try_to_wake_up() stats */
1019 unsigned int ttwu_wake_remote;
1020 unsigned int ttwu_move_affine;
1021 unsigned int ttwu_move_balance;
1023 #ifdef CONFIG_SCHED_DEBUG
1027 void *private; /* used during construction */
1028 struct rcu_head rcu; /* used during destruction */
1031 unsigned int span_weight;
1033 * Span of all CPUs in this domain.
1035 * NOTE: this field is variable length. (Allocated dynamically
1036 * by attaching extra space to the end of the structure,
1037 * depending on how many CPUs the kernel has booted up with)
1039 unsigned long span[0];
1042 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1044 return to_cpumask(sd->span);
1047 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1048 struct sched_domain_attr *dattr_new);
1050 /* Allocate an array of sched domains, for partition_sched_domains(). */
1051 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1052 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1054 bool cpus_share_cache(int this_cpu, int that_cpu);
1056 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1057 typedef int (*sched_domain_flags_f)(void);
1059 #define SDTL_OVERLAP 0x01
1062 struct sched_domain **__percpu sd;
1063 struct sched_group **__percpu sg;
1064 struct sched_group_capacity **__percpu sgc;
1067 struct sched_domain_topology_level {
1068 sched_domain_mask_f mask;
1069 sched_domain_flags_f sd_flags;
1072 struct sd_data data;
1073 #ifdef CONFIG_SCHED_DEBUG
1078 extern struct sched_domain_topology_level *sched_domain_topology;
1080 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1081 extern void wake_up_if_idle(int cpu);
1083 #ifdef CONFIG_SCHED_DEBUG
1084 # define SD_INIT_NAME(type) .name = #type
1086 # define SD_INIT_NAME(type)
1089 #else /* CONFIG_SMP */
1091 struct sched_domain_attr;
1094 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1095 struct sched_domain_attr *dattr_new)
1099 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1104 #endif /* !CONFIG_SMP */
1107 struct io_context; /* See blkdev.h */
1110 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1111 extern void prefetch_stack(struct task_struct *t);
1113 static inline void prefetch_stack(struct task_struct *t) { }
1116 struct audit_context; /* See audit.c */
1118 struct pipe_inode_info;
1119 struct uts_namespace;
1121 struct load_weight {
1122 unsigned long weight;
1127 u64 last_runnable_update;
1130 * utilization_avg_contrib describes the amount of time that a
1131 * sched_entity is running on a CPU. It is based on running_avg_sum
1132 * and is scaled in the range [0..SCHED_LOAD_SCALE].
1133 * load_avg_contrib described the amount of time that a sched_entity
1134 * is runnable on a rq. It is based on both runnable_avg_sum and the
1135 * weight of the task.
1137 unsigned long load_avg_contrib, utilization_avg_contrib;
1139 * These sums represent an infinite geometric series and so are bound
1140 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1141 * choices of y < 1-2^(-32)*1024.
1142 * running_avg_sum reflects the time that the sched_entity is
1143 * effectively running on the CPU.
1144 * runnable_avg_sum represents the amount of time a sched_entity is on
1145 * a runqueue which includes the running time that is monitored by
1148 u32 runnable_avg_sum, avg_period, running_avg_sum;
1151 #ifdef CONFIG_SCHEDSTATS
1152 struct sched_statistics {
1162 s64 sum_sleep_runtime;
1169 u64 nr_migrations_cold;
1170 u64 nr_failed_migrations_affine;
1171 u64 nr_failed_migrations_running;
1172 u64 nr_failed_migrations_hot;
1173 u64 nr_forced_migrations;
1176 u64 nr_wakeups_sync;
1177 u64 nr_wakeups_migrate;
1178 u64 nr_wakeups_local;
1179 u64 nr_wakeups_remote;
1180 u64 nr_wakeups_affine;
1181 u64 nr_wakeups_affine_attempts;
1182 u64 nr_wakeups_passive;
1183 u64 nr_wakeups_idle;
1187 struct sched_entity {
1188 struct load_weight load; /* for load-balancing */
1189 struct rb_node run_node;
1190 struct list_head group_node;
1194 u64 sum_exec_runtime;
1196 u64 prev_sum_exec_runtime;
1200 #ifdef CONFIG_SCHEDSTATS
1201 struct sched_statistics statistics;
1204 #ifdef CONFIG_FAIR_GROUP_SCHED
1206 struct sched_entity *parent;
1207 /* rq on which this entity is (to be) queued: */
1208 struct cfs_rq *cfs_rq;
1209 /* rq "owned" by this entity/group: */
1210 struct cfs_rq *my_q;
1214 /* Per-entity load-tracking */
1215 struct sched_avg avg;
1219 struct sched_rt_entity {
1220 struct list_head run_list;
1221 unsigned long timeout;
1222 unsigned long watchdog_stamp;
1223 unsigned int time_slice;
1225 struct sched_rt_entity *back;
1226 #ifdef CONFIG_RT_GROUP_SCHED
1227 struct sched_rt_entity *parent;
1228 /* rq on which this entity is (to be) queued: */
1229 struct rt_rq *rt_rq;
1230 /* rq "owned" by this entity/group: */
1235 struct sched_dl_entity {
1236 struct rb_node rb_node;
1239 * Original scheduling parameters. Copied here from sched_attr
1240 * during sched_setattr(), they will remain the same until
1241 * the next sched_setattr().
1243 u64 dl_runtime; /* maximum runtime for each instance */
1244 u64 dl_deadline; /* relative deadline of each instance */
1245 u64 dl_period; /* separation of two instances (period) */
1246 u64 dl_bw; /* dl_runtime / dl_deadline */
1249 * Actual scheduling parameters. Initialized with the values above,
1250 * they are continously updated during task execution. Note that
1251 * the remaining runtime could be < 0 in case we are in overrun.
1253 s64 runtime; /* remaining runtime for this instance */
1254 u64 deadline; /* absolute deadline for this instance */
1255 unsigned int flags; /* specifying the scheduler behaviour */
1260 * @dl_throttled tells if we exhausted the runtime. If so, the
1261 * task has to wait for a replenishment to be performed at the
1262 * next firing of dl_timer.
1264 * @dl_new tells if a new instance arrived. If so we must
1265 * start executing it with full runtime and reset its absolute
1268 * @dl_boosted tells if we are boosted due to DI. If so we are
1269 * outside bandwidth enforcement mechanism (but only until we
1270 * exit the critical section);
1272 * @dl_yielded tells if task gave up the cpu before consuming
1273 * all its available runtime during the last job.
1275 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1278 * Bandwidth enforcement timer. Each -deadline task has its
1279 * own bandwidth to be enforced, thus we need one timer per task.
1281 struct hrtimer dl_timer;
1293 enum perf_event_task_context {
1294 perf_invalid_context = -1,
1295 perf_hw_context = 0,
1297 perf_nr_task_contexts,
1300 struct task_struct {
1301 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1304 unsigned int flags; /* per process flags, defined below */
1305 unsigned int ptrace;
1308 struct llist_node wake_entry;
1310 struct task_struct *last_wakee;
1311 unsigned long wakee_flips;
1312 unsigned long wakee_flip_decay_ts;
1318 int prio, static_prio, normal_prio;
1319 unsigned int rt_priority;
1320 const struct sched_class *sched_class;
1321 struct sched_entity se;
1322 struct sched_rt_entity rt;
1323 #ifdef CONFIG_CGROUP_SCHED
1324 struct task_group *sched_task_group;
1326 struct sched_dl_entity dl;
1328 #ifdef CONFIG_PREEMPT_NOTIFIERS
1329 /* list of struct preempt_notifier: */
1330 struct hlist_head preempt_notifiers;
1333 #ifdef CONFIG_BLK_DEV_IO_TRACE
1334 unsigned int btrace_seq;
1337 unsigned int policy;
1338 int nr_cpus_allowed;
1339 cpumask_t cpus_allowed;
1341 #ifdef CONFIG_PREEMPT_RCU
1342 int rcu_read_lock_nesting;
1343 union rcu_special rcu_read_unlock_special;
1344 struct list_head rcu_node_entry;
1345 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1346 #ifdef CONFIG_PREEMPT_RCU
1347 struct rcu_node *rcu_blocked_node;
1348 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1349 #ifdef CONFIG_TASKS_RCU
1350 unsigned long rcu_tasks_nvcsw;
1351 bool rcu_tasks_holdout;
1352 struct list_head rcu_tasks_holdout_list;
1353 int rcu_tasks_idle_cpu;
1354 #endif /* #ifdef CONFIG_TASKS_RCU */
1356 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1357 struct sched_info sched_info;
1360 struct list_head tasks;
1362 struct plist_node pushable_tasks;
1363 struct rb_node pushable_dl_tasks;
1366 struct mm_struct *mm, *active_mm;
1367 #ifdef CONFIG_COMPAT_BRK
1368 unsigned brk_randomized:1;
1370 /* per-thread vma caching */
1371 u32 vmacache_seqnum;
1372 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1373 #if defined(SPLIT_RSS_COUNTING)
1374 struct task_rss_stat rss_stat;
1378 int exit_code, exit_signal;
1379 int pdeath_signal; /* The signal sent when the parent dies */
1380 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1382 /* Used for emulating ABI behavior of previous Linux versions */
1383 unsigned int personality;
1385 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1387 unsigned in_iowait:1;
1389 /* Revert to default priority/policy when forking */
1390 unsigned sched_reset_on_fork:1;
1391 unsigned sched_contributes_to_load:1;
1393 #ifdef CONFIG_MEMCG_KMEM
1394 unsigned memcg_kmem_skip_account:1;
1397 unsigned long atomic_flags; /* Flags needing atomic access. */
1399 struct restart_block restart_block;
1404 #ifdef CONFIG_CC_STACKPROTECTOR
1405 /* Canary value for the -fstack-protector gcc feature */
1406 unsigned long stack_canary;
1409 * pointers to (original) parent process, youngest child, younger sibling,
1410 * older sibling, respectively. (p->father can be replaced with
1411 * p->real_parent->pid)
1413 struct task_struct __rcu *real_parent; /* real parent process */
1414 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1416 * children/sibling forms the list of my natural children
1418 struct list_head children; /* list of my children */
1419 struct list_head sibling; /* linkage in my parent's children list */
1420 struct task_struct *group_leader; /* threadgroup leader */
1423 * ptraced is the list of tasks this task is using ptrace on.
1424 * This includes both natural children and PTRACE_ATTACH targets.
1425 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1427 struct list_head ptraced;
1428 struct list_head ptrace_entry;
1430 /* PID/PID hash table linkage. */
1431 struct pid_link pids[PIDTYPE_MAX];
1432 struct list_head thread_group;
1433 struct list_head thread_node;
1435 struct completion *vfork_done; /* for vfork() */
1436 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1437 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1439 cputime_t utime, stime, utimescaled, stimescaled;
1441 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1442 struct cputime prev_cputime;
1444 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1445 seqlock_t vtime_seqlock;
1446 unsigned long long vtime_snap;
1451 } vtime_snap_whence;
1453 unsigned long nvcsw, nivcsw; /* context switch counts */
1454 u64 start_time; /* monotonic time in nsec */
1455 u64 real_start_time; /* boot based time in nsec */
1456 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1457 unsigned long min_flt, maj_flt;
1459 struct task_cputime cputime_expires;
1460 struct list_head cpu_timers[3];
1462 /* process credentials */
1463 const struct cred __rcu *real_cred; /* objective and real subjective task
1464 * credentials (COW) */
1465 const struct cred __rcu *cred; /* effective (overridable) subjective task
1466 * credentials (COW) */
1467 char comm[TASK_COMM_LEN]; /* executable name excluding path
1468 - access with [gs]et_task_comm (which lock
1469 it with task_lock())
1470 - initialized normally by setup_new_exec */
1471 /* file system info */
1472 int link_count, total_link_count;
1473 #ifdef CONFIG_SYSVIPC
1475 struct sysv_sem sysvsem;
1476 struct sysv_shm sysvshm;
1478 #ifdef CONFIG_DETECT_HUNG_TASK
1479 /* hung task detection */
1480 unsigned long last_switch_count;
1482 /* CPU-specific state of this task */
1483 struct thread_struct thread;
1484 /* filesystem information */
1485 struct fs_struct *fs;
1486 /* open file information */
1487 struct files_struct *files;
1489 struct nsproxy *nsproxy;
1490 /* signal handlers */
1491 struct signal_struct *signal;
1492 struct sighand_struct *sighand;
1494 sigset_t blocked, real_blocked;
1495 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1496 struct sigpending pending;
1498 unsigned long sas_ss_sp;
1500 int (*notifier)(void *priv);
1501 void *notifier_data;
1502 sigset_t *notifier_mask;
1503 struct callback_head *task_works;
1505 struct audit_context *audit_context;
1506 #ifdef CONFIG_AUDITSYSCALL
1508 unsigned int sessionid;
1510 struct seccomp seccomp;
1512 /* Thread group tracking */
1515 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1517 spinlock_t alloc_lock;
1519 /* Protection of the PI data structures: */
1520 raw_spinlock_t pi_lock;
1522 #ifdef CONFIG_RT_MUTEXES
1523 /* PI waiters blocked on a rt_mutex held by this task */
1524 struct rb_root pi_waiters;
1525 struct rb_node *pi_waiters_leftmost;
1526 /* Deadlock detection and priority inheritance handling */
1527 struct rt_mutex_waiter *pi_blocked_on;
1530 #ifdef CONFIG_DEBUG_MUTEXES
1531 /* mutex deadlock detection */
1532 struct mutex_waiter *blocked_on;
1534 #ifdef CONFIG_TRACE_IRQFLAGS
1535 unsigned int irq_events;
1536 unsigned long hardirq_enable_ip;
1537 unsigned long hardirq_disable_ip;
1538 unsigned int hardirq_enable_event;
1539 unsigned int hardirq_disable_event;
1540 int hardirqs_enabled;
1541 int hardirq_context;
1542 unsigned long softirq_disable_ip;
1543 unsigned long softirq_enable_ip;
1544 unsigned int softirq_disable_event;
1545 unsigned int softirq_enable_event;
1546 int softirqs_enabled;
1547 int softirq_context;
1549 #ifdef CONFIG_LOCKDEP
1550 # define MAX_LOCK_DEPTH 48UL
1553 unsigned int lockdep_recursion;
1554 struct held_lock held_locks[MAX_LOCK_DEPTH];
1555 gfp_t lockdep_reclaim_gfp;
1558 /* journalling filesystem info */
1561 /* stacked block device info */
1562 struct bio_list *bio_list;
1565 /* stack plugging */
1566 struct blk_plug *plug;
1570 struct reclaim_state *reclaim_state;
1572 struct backing_dev_info *backing_dev_info;
1574 struct io_context *io_context;
1576 unsigned long ptrace_message;
1577 siginfo_t *last_siginfo; /* For ptrace use. */
1578 struct task_io_accounting ioac;
1579 #if defined(CONFIG_TASK_XACCT)
1580 u64 acct_rss_mem1; /* accumulated rss usage */
1581 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1582 cputime_t acct_timexpd; /* stime + utime since last update */
1584 #ifdef CONFIG_CPUSETS
1585 nodemask_t mems_allowed; /* Protected by alloc_lock */
1586 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1587 int cpuset_mem_spread_rotor;
1588 int cpuset_slab_spread_rotor;
1590 #ifdef CONFIG_CGROUPS
1591 /* Control Group info protected by css_set_lock */
1592 struct css_set __rcu *cgroups;
1593 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1594 struct list_head cg_list;
1597 struct robust_list_head __user *robust_list;
1598 #ifdef CONFIG_COMPAT
1599 struct compat_robust_list_head __user *compat_robust_list;
1601 struct list_head pi_state_list;
1602 struct futex_pi_state *pi_state_cache;
1604 #ifdef CONFIG_PERF_EVENTS
1605 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1606 struct mutex perf_event_mutex;
1607 struct list_head perf_event_list;
1609 #ifdef CONFIG_DEBUG_PREEMPT
1610 unsigned long preempt_disable_ip;
1613 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1615 short pref_node_fork;
1617 #ifdef CONFIG_NUMA_BALANCING
1619 unsigned int numa_scan_period;
1620 unsigned int numa_scan_period_max;
1621 int numa_preferred_nid;
1622 unsigned long numa_migrate_retry;
1623 u64 node_stamp; /* migration stamp */
1624 u64 last_task_numa_placement;
1625 u64 last_sum_exec_runtime;
1626 struct callback_head numa_work;
1628 struct list_head numa_entry;
1629 struct numa_group *numa_group;
1632 * numa_faults is an array split into four regions:
1633 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1634 * in this precise order.
1636 * faults_memory: Exponential decaying average of faults on a per-node
1637 * basis. Scheduling placement decisions are made based on these
1638 * counts. The values remain static for the duration of a PTE scan.
1639 * faults_cpu: Track the nodes the process was running on when a NUMA
1640 * hinting fault was incurred.
1641 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1642 * during the current scan window. When the scan completes, the counts
1643 * in faults_memory and faults_cpu decay and these values are copied.
1645 unsigned long *numa_faults;
1646 unsigned long total_numa_faults;
1649 * numa_faults_locality tracks if faults recorded during the last
1650 * scan window were remote/local or failed to migrate. The task scan
1651 * period is adapted based on the locality of the faults with different
1652 * weights depending on whether they were shared or private faults
1654 unsigned long numa_faults_locality[3];
1656 unsigned long numa_pages_migrated;
1657 #endif /* CONFIG_NUMA_BALANCING */
1659 struct rcu_head rcu;
1662 * cache last used pipe for splice
1664 struct pipe_inode_info *splice_pipe;
1666 struct page_frag task_frag;
1668 #ifdef CONFIG_TASK_DELAY_ACCT
1669 struct task_delay_info *delays;
1671 #ifdef CONFIG_FAULT_INJECTION
1675 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1676 * balance_dirty_pages() for some dirty throttling pause
1679 int nr_dirtied_pause;
1680 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1682 #ifdef CONFIG_LATENCYTOP
1683 int latency_record_count;
1684 struct latency_record latency_record[LT_SAVECOUNT];
1687 * time slack values; these are used to round up poll() and
1688 * select() etc timeout values. These are in nanoseconds.
1690 unsigned long timer_slack_ns;
1691 unsigned long default_timer_slack_ns;
1694 unsigned int kasan_depth;
1696 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1697 /* Index of current stored address in ret_stack */
1699 /* Stack of return addresses for return function tracing */
1700 struct ftrace_ret_stack *ret_stack;
1701 /* time stamp for last schedule */
1702 unsigned long long ftrace_timestamp;
1704 * Number of functions that haven't been traced
1705 * because of depth overrun.
1707 atomic_t trace_overrun;
1708 /* Pause for the tracing */
1709 atomic_t tracing_graph_pause;
1711 #ifdef CONFIG_TRACING
1712 /* state flags for use by tracers */
1713 unsigned long trace;
1714 /* bitmask and counter of trace recursion */
1715 unsigned long trace_recursion;
1716 #endif /* CONFIG_TRACING */
1718 struct memcg_oom_info {
1719 struct mem_cgroup *memcg;
1722 unsigned int may_oom:1;
1725 #ifdef CONFIG_UPROBES
1726 struct uprobe_task *utask;
1728 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1729 unsigned int sequential_io;
1730 unsigned int sequential_io_avg;
1732 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1733 unsigned long task_state_change;
1737 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1738 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1740 #define TNF_MIGRATED 0x01
1741 #define TNF_NO_GROUP 0x02
1742 #define TNF_SHARED 0x04
1743 #define TNF_FAULT_LOCAL 0x08
1744 #define TNF_MIGRATE_FAIL 0x10
1746 #ifdef CONFIG_NUMA_BALANCING
1747 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1748 extern pid_t task_numa_group_id(struct task_struct *p);
1749 extern void set_numabalancing_state(bool enabled);
1750 extern void task_numa_free(struct task_struct *p);
1751 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1752 int src_nid, int dst_cpu);
1754 static inline void task_numa_fault(int last_node, int node, int pages,
1758 static inline pid_t task_numa_group_id(struct task_struct *p)
1762 static inline void set_numabalancing_state(bool enabled)
1765 static inline void task_numa_free(struct task_struct *p)
1768 static inline bool should_numa_migrate_memory(struct task_struct *p,
1769 struct page *page, int src_nid, int dst_cpu)
1775 static inline struct pid *task_pid(struct task_struct *task)
1777 return task->pids[PIDTYPE_PID].pid;
1780 static inline struct pid *task_tgid(struct task_struct *task)
1782 return task->group_leader->pids[PIDTYPE_PID].pid;
1786 * Without tasklist or rcu lock it is not safe to dereference
1787 * the result of task_pgrp/task_session even if task == current,
1788 * we can race with another thread doing sys_setsid/sys_setpgid.
1790 static inline struct pid *task_pgrp(struct task_struct *task)
1792 return task->group_leader->pids[PIDTYPE_PGID].pid;
1795 static inline struct pid *task_session(struct task_struct *task)
1797 return task->group_leader->pids[PIDTYPE_SID].pid;
1800 struct pid_namespace;
1803 * the helpers to get the task's different pids as they are seen
1804 * from various namespaces
1806 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1807 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1809 * task_xid_nr_ns() : id seen from the ns specified;
1811 * set_task_vxid() : assigns a virtual id to a task;
1813 * see also pid_nr() etc in include/linux/pid.h
1815 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1816 struct pid_namespace *ns);
1818 static inline pid_t task_pid_nr(struct task_struct *tsk)
1823 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1824 struct pid_namespace *ns)
1826 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1829 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1831 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1835 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1840 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1842 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1844 return pid_vnr(task_tgid(tsk));
1848 static inline int pid_alive(const struct task_struct *p);
1849 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1855 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1861 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1863 return task_ppid_nr_ns(tsk, &init_pid_ns);
1866 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1867 struct pid_namespace *ns)
1869 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1872 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1874 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1878 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1879 struct pid_namespace *ns)
1881 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1884 static inline pid_t task_session_vnr(struct task_struct *tsk)
1886 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1889 /* obsolete, do not use */
1890 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1892 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1896 * pid_alive - check that a task structure is not stale
1897 * @p: Task structure to be checked.
1899 * Test if a process is not yet dead (at most zombie state)
1900 * If pid_alive fails, then pointers within the task structure
1901 * can be stale and must not be dereferenced.
1903 * Return: 1 if the process is alive. 0 otherwise.
1905 static inline int pid_alive(const struct task_struct *p)
1907 return p->pids[PIDTYPE_PID].pid != NULL;
1911 * is_global_init - check if a task structure is init
1912 * @tsk: Task structure to be checked.
1914 * Check if a task structure is the first user space task the kernel created.
1916 * Return: 1 if the task structure is init. 0 otherwise.
1918 static inline int is_global_init(struct task_struct *tsk)
1920 return tsk->pid == 1;
1923 extern struct pid *cad_pid;
1925 extern void free_task(struct task_struct *tsk);
1926 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1928 extern void __put_task_struct(struct task_struct *t);
1930 static inline void put_task_struct(struct task_struct *t)
1932 if (atomic_dec_and_test(&t->usage))
1933 __put_task_struct(t);
1936 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1937 extern void task_cputime(struct task_struct *t,
1938 cputime_t *utime, cputime_t *stime);
1939 extern void task_cputime_scaled(struct task_struct *t,
1940 cputime_t *utimescaled, cputime_t *stimescaled);
1941 extern cputime_t task_gtime(struct task_struct *t);
1943 static inline void task_cputime(struct task_struct *t,
1944 cputime_t *utime, cputime_t *stime)
1952 static inline void task_cputime_scaled(struct task_struct *t,
1953 cputime_t *utimescaled,
1954 cputime_t *stimescaled)
1957 *utimescaled = t->utimescaled;
1959 *stimescaled = t->stimescaled;
1962 static inline cputime_t task_gtime(struct task_struct *t)
1967 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1968 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1973 #define PF_EXITING 0x00000004 /* getting shut down */
1974 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1975 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1976 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1977 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1978 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1979 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1980 #define PF_DUMPCORE 0x00000200 /* dumped core */
1981 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1982 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1983 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1984 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1985 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1986 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1987 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1988 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1989 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1990 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1991 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1992 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1993 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1994 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1995 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1996 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1997 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1998 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1999 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2002 * Only the _current_ task can read/write to tsk->flags, but other
2003 * tasks can access tsk->flags in readonly mode for example
2004 * with tsk_used_math (like during threaded core dumping).
2005 * There is however an exception to this rule during ptrace
2006 * or during fork: the ptracer task is allowed to write to the
2007 * child->flags of its traced child (same goes for fork, the parent
2008 * can write to the child->flags), because we're guaranteed the
2009 * child is not running and in turn not changing child->flags
2010 * at the same time the parent does it.
2012 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2013 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2014 #define clear_used_math() clear_stopped_child_used_math(current)
2015 #define set_used_math() set_stopped_child_used_math(current)
2016 #define conditional_stopped_child_used_math(condition, child) \
2017 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2018 #define conditional_used_math(condition) \
2019 conditional_stopped_child_used_math(condition, current)
2020 #define copy_to_stopped_child_used_math(child) \
2021 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2022 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2023 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2024 #define used_math() tsk_used_math(current)
2026 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2027 * __GFP_FS is also cleared as it implies __GFP_IO.
2029 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2031 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2032 flags &= ~(__GFP_IO | __GFP_FS);
2036 static inline unsigned int memalloc_noio_save(void)
2038 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2039 current->flags |= PF_MEMALLOC_NOIO;
2043 static inline void memalloc_noio_restore(unsigned int flags)
2045 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2048 /* Per-process atomic flags. */
2049 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2050 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2051 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2054 #define TASK_PFA_TEST(name, func) \
2055 static inline bool task_##func(struct task_struct *p) \
2056 { return test_bit(PFA_##name, &p->atomic_flags); }
2057 #define TASK_PFA_SET(name, func) \
2058 static inline void task_set_##func(struct task_struct *p) \
2059 { set_bit(PFA_##name, &p->atomic_flags); }
2060 #define TASK_PFA_CLEAR(name, func) \
2061 static inline void task_clear_##func(struct task_struct *p) \
2062 { clear_bit(PFA_##name, &p->atomic_flags); }
2064 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2065 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2067 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2068 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2069 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2071 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2072 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2073 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2076 * task->jobctl flags
2078 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2080 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2081 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2082 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2083 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2084 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2085 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2086 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2088 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
2089 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
2090 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
2091 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
2092 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
2093 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
2094 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
2096 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2097 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2099 extern bool task_set_jobctl_pending(struct task_struct *task,
2101 extern void task_clear_jobctl_trapping(struct task_struct *task);
2102 extern void task_clear_jobctl_pending(struct task_struct *task,
2105 static inline void rcu_copy_process(struct task_struct *p)
2107 #ifdef CONFIG_PREEMPT_RCU
2108 p->rcu_read_lock_nesting = 0;
2109 p->rcu_read_unlock_special.s = 0;
2110 p->rcu_blocked_node = NULL;
2111 INIT_LIST_HEAD(&p->rcu_node_entry);
2112 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2113 #ifdef CONFIG_TASKS_RCU
2114 p->rcu_tasks_holdout = false;
2115 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2116 p->rcu_tasks_idle_cpu = -1;
2117 #endif /* #ifdef CONFIG_TASKS_RCU */
2120 static inline void tsk_restore_flags(struct task_struct *task,
2121 unsigned long orig_flags, unsigned long flags)
2123 task->flags &= ~flags;
2124 task->flags |= orig_flags & flags;
2127 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2128 const struct cpumask *trial);
2129 extern int task_can_attach(struct task_struct *p,
2130 const struct cpumask *cs_cpus_allowed);
2132 extern void do_set_cpus_allowed(struct task_struct *p,
2133 const struct cpumask *new_mask);
2135 extern int set_cpus_allowed_ptr(struct task_struct *p,
2136 const struct cpumask *new_mask);
2138 static inline void do_set_cpus_allowed(struct task_struct *p,
2139 const struct cpumask *new_mask)
2142 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2143 const struct cpumask *new_mask)
2145 if (!cpumask_test_cpu(0, new_mask))
2151 #ifdef CONFIG_NO_HZ_COMMON
2152 void calc_load_enter_idle(void);
2153 void calc_load_exit_idle(void);
2155 static inline void calc_load_enter_idle(void) { }
2156 static inline void calc_load_exit_idle(void) { }
2157 #endif /* CONFIG_NO_HZ_COMMON */
2159 #ifndef CONFIG_CPUMASK_OFFSTACK
2160 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2162 return set_cpus_allowed_ptr(p, &new_mask);
2167 * Do not use outside of architecture code which knows its limitations.
2169 * sched_clock() has no promise of monotonicity or bounded drift between
2170 * CPUs, use (which you should not) requires disabling IRQs.
2172 * Please use one of the three interfaces below.
2174 extern unsigned long long notrace sched_clock(void);
2176 * See the comment in kernel/sched/clock.c
2178 extern u64 cpu_clock(int cpu);
2179 extern u64 local_clock(void);
2180 extern u64 running_clock(void);
2181 extern u64 sched_clock_cpu(int cpu);
2184 extern void sched_clock_init(void);
2186 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2187 static inline void sched_clock_tick(void)
2191 static inline void sched_clock_idle_sleep_event(void)
2195 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2200 * Architectures can set this to 1 if they have specified
2201 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2202 * but then during bootup it turns out that sched_clock()
2203 * is reliable after all:
2205 extern int sched_clock_stable(void);
2206 extern void set_sched_clock_stable(void);
2207 extern void clear_sched_clock_stable(void);
2209 extern void sched_clock_tick(void);
2210 extern void sched_clock_idle_sleep_event(void);
2211 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2214 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2216 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2217 * The reason for this explicit opt-in is not to have perf penalty with
2218 * slow sched_clocks.
2220 extern void enable_sched_clock_irqtime(void);
2221 extern void disable_sched_clock_irqtime(void);
2223 static inline void enable_sched_clock_irqtime(void) {}
2224 static inline void disable_sched_clock_irqtime(void) {}
2227 extern unsigned long long
2228 task_sched_runtime(struct task_struct *task);
2230 /* sched_exec is called by processes performing an exec */
2232 extern void sched_exec(void);
2234 #define sched_exec() {}
2237 extern void sched_clock_idle_sleep_event(void);
2238 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2240 #ifdef CONFIG_HOTPLUG_CPU
2241 extern void idle_task_exit(void);
2243 static inline void idle_task_exit(void) {}
2246 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2247 extern void wake_up_nohz_cpu(int cpu);
2249 static inline void wake_up_nohz_cpu(int cpu) { }
2252 #ifdef CONFIG_NO_HZ_FULL
2253 extern bool sched_can_stop_tick(void);
2254 extern u64 scheduler_tick_max_deferment(void);
2256 static inline bool sched_can_stop_tick(void) { return false; }
2259 #ifdef CONFIG_SCHED_AUTOGROUP
2260 extern void sched_autogroup_create_attach(struct task_struct *p);
2261 extern void sched_autogroup_detach(struct task_struct *p);
2262 extern void sched_autogroup_fork(struct signal_struct *sig);
2263 extern void sched_autogroup_exit(struct signal_struct *sig);
2264 #ifdef CONFIG_PROC_FS
2265 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2266 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2269 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2270 static inline void sched_autogroup_detach(struct task_struct *p) { }
2271 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2272 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2275 extern int yield_to(struct task_struct *p, bool preempt);
2276 extern void set_user_nice(struct task_struct *p, long nice);
2277 extern int task_prio(const struct task_struct *p);
2279 * task_nice - return the nice value of a given task.
2280 * @p: the task in question.
2282 * Return: The nice value [ -20 ... 0 ... 19 ].
2284 static inline int task_nice(const struct task_struct *p)
2286 return PRIO_TO_NICE((p)->static_prio);
2288 extern int can_nice(const struct task_struct *p, const int nice);
2289 extern int task_curr(const struct task_struct *p);
2290 extern int idle_cpu(int cpu);
2291 extern int sched_setscheduler(struct task_struct *, int,
2292 const struct sched_param *);
2293 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2294 const struct sched_param *);
2295 extern int sched_setattr(struct task_struct *,
2296 const struct sched_attr *);
2297 extern struct task_struct *idle_task(int cpu);
2299 * is_idle_task - is the specified task an idle task?
2300 * @p: the task in question.
2302 * Return: 1 if @p is an idle task. 0 otherwise.
2304 static inline bool is_idle_task(const struct task_struct *p)
2308 extern struct task_struct *curr_task(int cpu);
2309 extern void set_curr_task(int cpu, struct task_struct *p);
2313 union thread_union {
2314 struct thread_info thread_info;
2315 unsigned long stack[THREAD_SIZE/sizeof(long)];
2318 #ifndef __HAVE_ARCH_KSTACK_END
2319 static inline int kstack_end(void *addr)
2321 /* Reliable end of stack detection:
2322 * Some APM bios versions misalign the stack
2324 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2328 extern union thread_union init_thread_union;
2329 extern struct task_struct init_task;
2331 extern struct mm_struct init_mm;
2333 extern struct pid_namespace init_pid_ns;
2336 * find a task by one of its numerical ids
2338 * find_task_by_pid_ns():
2339 * finds a task by its pid in the specified namespace
2340 * find_task_by_vpid():
2341 * finds a task by its virtual pid
2343 * see also find_vpid() etc in include/linux/pid.h
2346 extern struct task_struct *find_task_by_vpid(pid_t nr);
2347 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2348 struct pid_namespace *ns);
2350 /* per-UID process charging. */
2351 extern struct user_struct * alloc_uid(kuid_t);
2352 static inline struct user_struct *get_uid(struct user_struct *u)
2354 atomic_inc(&u->__count);
2357 extern void free_uid(struct user_struct *);
2359 #include <asm/current.h>
2361 extern void xtime_update(unsigned long ticks);
2363 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2364 extern int wake_up_process(struct task_struct *tsk);
2365 extern void wake_up_new_task(struct task_struct *tsk);
2367 extern void kick_process(struct task_struct *tsk);
2369 static inline void kick_process(struct task_struct *tsk) { }
2371 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2372 extern void sched_dead(struct task_struct *p);
2374 extern void proc_caches_init(void);
2375 extern void flush_signals(struct task_struct *);
2376 extern void __flush_signals(struct task_struct *);
2377 extern void ignore_signals(struct task_struct *);
2378 extern void flush_signal_handlers(struct task_struct *, int force_default);
2379 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2381 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2383 unsigned long flags;
2386 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2387 ret = dequeue_signal(tsk, mask, info);
2388 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2393 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2395 extern void unblock_all_signals(void);
2396 extern void release_task(struct task_struct * p);
2397 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2398 extern int force_sigsegv(int, struct task_struct *);
2399 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2400 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2401 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2402 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2403 const struct cred *, u32);
2404 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2405 extern int kill_pid(struct pid *pid, int sig, int priv);
2406 extern int kill_proc_info(int, struct siginfo *, pid_t);
2407 extern __must_check bool do_notify_parent(struct task_struct *, int);
2408 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2409 extern void force_sig(int, struct task_struct *);
2410 extern int send_sig(int, struct task_struct *, int);
2411 extern int zap_other_threads(struct task_struct *p);
2412 extern struct sigqueue *sigqueue_alloc(void);
2413 extern void sigqueue_free(struct sigqueue *);
2414 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2415 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2417 static inline void restore_saved_sigmask(void)
2419 if (test_and_clear_restore_sigmask())
2420 __set_current_blocked(¤t->saved_sigmask);
2423 static inline sigset_t *sigmask_to_save(void)
2425 sigset_t *res = ¤t->blocked;
2426 if (unlikely(test_restore_sigmask()))
2427 res = ¤t->saved_sigmask;
2431 static inline int kill_cad_pid(int sig, int priv)
2433 return kill_pid(cad_pid, sig, priv);
2436 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2437 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2438 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2439 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2442 * True if we are on the alternate signal stack.
2444 static inline int on_sig_stack(unsigned long sp)
2446 #ifdef CONFIG_STACK_GROWSUP
2447 return sp >= current->sas_ss_sp &&
2448 sp - current->sas_ss_sp < current->sas_ss_size;
2450 return sp > current->sas_ss_sp &&
2451 sp - current->sas_ss_sp <= current->sas_ss_size;
2455 static inline int sas_ss_flags(unsigned long sp)
2457 if (!current->sas_ss_size)
2460 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2463 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2465 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2466 #ifdef CONFIG_STACK_GROWSUP
2467 return current->sas_ss_sp;
2469 return current->sas_ss_sp + current->sas_ss_size;
2475 * Routines for handling mm_structs
2477 extern struct mm_struct * mm_alloc(void);
2479 /* mmdrop drops the mm and the page tables */
2480 extern void __mmdrop(struct mm_struct *);
2481 static inline void mmdrop(struct mm_struct * mm)
2483 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2487 /* mmput gets rid of the mappings and all user-space */
2488 extern void mmput(struct mm_struct *);
2489 /* Grab a reference to a task's mm, if it is not already going away */
2490 extern struct mm_struct *get_task_mm(struct task_struct *task);
2492 * Grab a reference to a task's mm, if it is not already going away
2493 * and ptrace_may_access with the mode parameter passed to it
2496 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2497 /* Remove the current tasks stale references to the old mm_struct */
2498 extern void mm_release(struct task_struct *, struct mm_struct *);
2500 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2501 struct task_struct *);
2502 extern void flush_thread(void);
2503 extern void exit_thread(void);
2505 extern void exit_files(struct task_struct *);
2506 extern void __cleanup_sighand(struct sighand_struct *);
2508 extern void exit_itimers(struct signal_struct *);
2509 extern void flush_itimer_signals(void);
2511 extern void do_group_exit(int);
2513 extern int do_execve(struct filename *,
2514 const char __user * const __user *,
2515 const char __user * const __user *);
2516 extern int do_execveat(int, struct filename *,
2517 const char __user * const __user *,
2518 const char __user * const __user *,
2520 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2521 struct task_struct *fork_idle(int);
2522 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2524 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2525 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2527 __set_task_comm(tsk, from, false);
2529 extern char *get_task_comm(char *to, struct task_struct *tsk);
2532 void scheduler_ipi(void);
2533 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2535 static inline void scheduler_ipi(void) { }
2536 static inline unsigned long wait_task_inactive(struct task_struct *p,
2543 #define next_task(p) \
2544 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2546 #define for_each_process(p) \
2547 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2549 extern bool current_is_single_threaded(void);
2552 * Careful: do_each_thread/while_each_thread is a double loop so
2553 * 'break' will not work as expected - use goto instead.
2555 #define do_each_thread(g, t) \
2556 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2558 #define while_each_thread(g, t) \
2559 while ((t = next_thread(t)) != g)
2561 #define __for_each_thread(signal, t) \
2562 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2564 #define for_each_thread(p, t) \
2565 __for_each_thread((p)->signal, t)
2567 /* Careful: this is a double loop, 'break' won't work as expected. */
2568 #define for_each_process_thread(p, t) \
2569 for_each_process(p) for_each_thread(p, t)
2571 static inline int get_nr_threads(struct task_struct *tsk)
2573 return tsk->signal->nr_threads;
2576 static inline bool thread_group_leader(struct task_struct *p)
2578 return p->exit_signal >= 0;
2581 /* Do to the insanities of de_thread it is possible for a process
2582 * to have the pid of the thread group leader without actually being
2583 * the thread group leader. For iteration through the pids in proc
2584 * all we care about is that we have a task with the appropriate
2585 * pid, we don't actually care if we have the right task.
2587 static inline bool has_group_leader_pid(struct task_struct *p)
2589 return task_pid(p) == p->signal->leader_pid;
2593 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2595 return p1->signal == p2->signal;
2598 static inline struct task_struct *next_thread(const struct task_struct *p)
2600 return list_entry_rcu(p->thread_group.next,
2601 struct task_struct, thread_group);
2604 static inline int thread_group_empty(struct task_struct *p)
2606 return list_empty(&p->thread_group);
2609 #define delay_group_leader(p) \
2610 (thread_group_leader(p) && !thread_group_empty(p))
2613 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2614 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2615 * pins the final release of task.io_context. Also protects ->cpuset and
2616 * ->cgroup.subsys[]. And ->vfork_done.
2618 * Nests both inside and outside of read_lock(&tasklist_lock).
2619 * It must not be nested with write_lock_irq(&tasklist_lock),
2620 * neither inside nor outside.
2622 static inline void task_lock(struct task_struct *p)
2624 spin_lock(&p->alloc_lock);
2627 static inline void task_unlock(struct task_struct *p)
2629 spin_unlock(&p->alloc_lock);
2632 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2633 unsigned long *flags);
2635 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2636 unsigned long *flags)
2638 struct sighand_struct *ret;
2640 ret = __lock_task_sighand(tsk, flags);
2641 (void)__cond_lock(&tsk->sighand->siglock, ret);
2645 static inline void unlock_task_sighand(struct task_struct *tsk,
2646 unsigned long *flags)
2648 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2651 #ifdef CONFIG_CGROUPS
2652 static inline void threadgroup_change_begin(struct task_struct *tsk)
2654 down_read(&tsk->signal->group_rwsem);
2656 static inline void threadgroup_change_end(struct task_struct *tsk)
2658 up_read(&tsk->signal->group_rwsem);
2662 * threadgroup_lock - lock threadgroup
2663 * @tsk: member task of the threadgroup to lock
2665 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2666 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2667 * change ->group_leader/pid. This is useful for cases where the threadgroup
2668 * needs to stay stable across blockable operations.
2670 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2671 * synchronization. While held, no new task will be added to threadgroup
2672 * and no existing live task will have its PF_EXITING set.
2674 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2675 * sub-thread becomes a new leader.
2677 static inline void threadgroup_lock(struct task_struct *tsk)
2679 down_write(&tsk->signal->group_rwsem);
2683 * threadgroup_unlock - unlock threadgroup
2684 * @tsk: member task of the threadgroup to unlock
2686 * Reverse threadgroup_lock().
2688 static inline void threadgroup_unlock(struct task_struct *tsk)
2690 up_write(&tsk->signal->group_rwsem);
2693 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2694 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2695 static inline void threadgroup_lock(struct task_struct *tsk) {}
2696 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2699 #ifndef __HAVE_THREAD_FUNCTIONS
2701 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2702 #define task_stack_page(task) ((task)->stack)
2704 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2706 *task_thread_info(p) = *task_thread_info(org);
2707 task_thread_info(p)->task = p;
2711 * Return the address of the last usable long on the stack.
2713 * When the stack grows down, this is just above the thread
2714 * info struct. Going any lower will corrupt the threadinfo.
2716 * When the stack grows up, this is the highest address.
2717 * Beyond that position, we corrupt data on the next page.
2719 static inline unsigned long *end_of_stack(struct task_struct *p)
2721 #ifdef CONFIG_STACK_GROWSUP
2722 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2724 return (unsigned long *)(task_thread_info(p) + 1);
2729 #define task_stack_end_corrupted(task) \
2730 (*(end_of_stack(task)) != STACK_END_MAGIC)
2732 static inline int object_is_on_stack(void *obj)
2734 void *stack = task_stack_page(current);
2736 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2739 extern void thread_info_cache_init(void);
2741 #ifdef CONFIG_DEBUG_STACK_USAGE
2742 static inline unsigned long stack_not_used(struct task_struct *p)
2744 unsigned long *n = end_of_stack(p);
2746 do { /* Skip over canary */
2750 return (unsigned long)n - (unsigned long)end_of_stack(p);
2753 extern void set_task_stack_end_magic(struct task_struct *tsk);
2755 /* set thread flags in other task's structures
2756 * - see asm/thread_info.h for TIF_xxxx flags available
2758 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2760 set_ti_thread_flag(task_thread_info(tsk), flag);
2763 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2765 clear_ti_thread_flag(task_thread_info(tsk), flag);
2768 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2770 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2773 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2775 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2778 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2780 return test_ti_thread_flag(task_thread_info(tsk), flag);
2783 static inline void set_tsk_need_resched(struct task_struct *tsk)
2785 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2788 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2790 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2793 static inline int test_tsk_need_resched(struct task_struct *tsk)
2795 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2798 static inline int restart_syscall(void)
2800 set_tsk_thread_flag(current, TIF_SIGPENDING);
2801 return -ERESTARTNOINTR;
2804 static inline int signal_pending(struct task_struct *p)
2806 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2809 static inline int __fatal_signal_pending(struct task_struct *p)
2811 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2814 static inline int fatal_signal_pending(struct task_struct *p)
2816 return signal_pending(p) && __fatal_signal_pending(p);
2819 static inline int signal_pending_state(long state, struct task_struct *p)
2821 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2823 if (!signal_pending(p))
2826 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2830 * cond_resched() and cond_resched_lock(): latency reduction via
2831 * explicit rescheduling in places that are safe. The return
2832 * value indicates whether a reschedule was done in fact.
2833 * cond_resched_lock() will drop the spinlock before scheduling,
2834 * cond_resched_softirq() will enable bhs before scheduling.
2836 extern int _cond_resched(void);
2838 #define cond_resched() ({ \
2839 ___might_sleep(__FILE__, __LINE__, 0); \
2843 extern int __cond_resched_lock(spinlock_t *lock);
2845 #ifdef CONFIG_PREEMPT_COUNT
2846 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2848 #define PREEMPT_LOCK_OFFSET 0
2851 #define cond_resched_lock(lock) ({ \
2852 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2853 __cond_resched_lock(lock); \
2856 extern int __cond_resched_softirq(void);
2858 #define cond_resched_softirq() ({ \
2859 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2860 __cond_resched_softirq(); \
2863 static inline void cond_resched_rcu(void)
2865 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2873 * Does a critical section need to be broken due to another
2874 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2875 * but a general need for low latency)
2877 static inline int spin_needbreak(spinlock_t *lock)
2879 #ifdef CONFIG_PREEMPT
2880 return spin_is_contended(lock);
2887 * Idle thread specific functions to determine the need_resched
2890 #ifdef TIF_POLLING_NRFLAG
2891 static inline int tsk_is_polling(struct task_struct *p)
2893 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2896 static inline void __current_set_polling(void)
2898 set_thread_flag(TIF_POLLING_NRFLAG);
2901 static inline bool __must_check current_set_polling_and_test(void)
2903 __current_set_polling();
2906 * Polling state must be visible before we test NEED_RESCHED,
2907 * paired by resched_curr()
2909 smp_mb__after_atomic();
2911 return unlikely(tif_need_resched());
2914 static inline void __current_clr_polling(void)
2916 clear_thread_flag(TIF_POLLING_NRFLAG);
2919 static inline bool __must_check current_clr_polling_and_test(void)
2921 __current_clr_polling();
2924 * Polling state must be visible before we test NEED_RESCHED,
2925 * paired by resched_curr()
2927 smp_mb__after_atomic();
2929 return unlikely(tif_need_resched());
2933 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2934 static inline void __current_set_polling(void) { }
2935 static inline void __current_clr_polling(void) { }
2937 static inline bool __must_check current_set_polling_and_test(void)
2939 return unlikely(tif_need_resched());
2941 static inline bool __must_check current_clr_polling_and_test(void)
2943 return unlikely(tif_need_resched());
2947 static inline void current_clr_polling(void)
2949 __current_clr_polling();
2952 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2953 * Once the bit is cleared, we'll get IPIs with every new
2954 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2957 smp_mb(); /* paired with resched_curr() */
2959 preempt_fold_need_resched();
2962 static __always_inline bool need_resched(void)
2964 return unlikely(tif_need_resched());
2968 * Thread group CPU time accounting.
2970 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2971 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2973 static inline void thread_group_cputime_init(struct signal_struct *sig)
2975 raw_spin_lock_init(&sig->cputimer.lock);
2979 * Reevaluate whether the task has signals pending delivery.
2980 * Wake the task if so.
2981 * This is required every time the blocked sigset_t changes.
2982 * callers must hold sighand->siglock.
2984 extern void recalc_sigpending_and_wake(struct task_struct *t);
2985 extern void recalc_sigpending(void);
2987 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2989 static inline void signal_wake_up(struct task_struct *t, bool resume)
2991 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2993 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2995 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2999 * Wrappers for p->thread_info->cpu access. No-op on UP.
3003 static inline unsigned int task_cpu(const struct task_struct *p)
3005 return task_thread_info(p)->cpu;
3008 static inline int task_node(const struct task_struct *p)
3010 return cpu_to_node(task_cpu(p));
3013 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3017 static inline unsigned int task_cpu(const struct task_struct *p)
3022 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3026 #endif /* CONFIG_SMP */
3028 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3029 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3031 #ifdef CONFIG_CGROUP_SCHED
3032 extern struct task_group root_task_group;
3033 #endif /* CONFIG_CGROUP_SCHED */
3035 extern int task_can_switch_user(struct user_struct *up,
3036 struct task_struct *tsk);
3038 #ifdef CONFIG_TASK_XACCT
3039 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3041 tsk->ioac.rchar += amt;
3044 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3046 tsk->ioac.wchar += amt;
3049 static inline void inc_syscr(struct task_struct *tsk)
3054 static inline void inc_syscw(struct task_struct *tsk)
3059 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3063 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3067 static inline void inc_syscr(struct task_struct *tsk)
3071 static inline void inc_syscw(struct task_struct *tsk)
3076 #ifndef TASK_SIZE_OF
3077 #define TASK_SIZE_OF(tsk) TASK_SIZE
3081 extern void mm_update_next_owner(struct mm_struct *mm);
3083 static inline void mm_update_next_owner(struct mm_struct *mm)
3086 #endif /* CONFIG_MEMCG */
3088 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3091 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
3094 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3097 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
3100 static inline unsigned long rlimit(unsigned int limit)
3102 return task_rlimit(current, limit);
3105 static inline unsigned long rlimit_max(unsigned int limit)
3107 return task_rlimit_max(current, limit);