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;
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 smp_store_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 smp_store_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 smp_store_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 smp_store_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 cpumask_var_t cpu_isolated_map;
334 extern int runqueue_is_locked(int cpu);
336 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
337 extern void nohz_balance_enter_idle(int cpu);
338 extern void set_cpu_sd_state_idle(void);
339 extern int get_nohz_timer_target(int pinned);
341 static inline void nohz_balance_enter_idle(int cpu) { }
342 static inline void set_cpu_sd_state_idle(void) { }
343 static inline int get_nohz_timer_target(int pinned)
345 return smp_processor_id();
350 * Only dump TASK_* tasks. (0 for all tasks)
352 extern void show_state_filter(unsigned long state_filter);
354 static inline void show_state(void)
356 show_state_filter(0);
359 extern void show_regs(struct pt_regs *);
362 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
363 * task), SP is the stack pointer of the first frame that should be shown in the back
364 * trace (or NULL if the entire call-chain of the task should be shown).
366 extern void show_stack(struct task_struct *task, unsigned long *sp);
368 extern void cpu_init (void);
369 extern void trap_init(void);
370 extern void update_process_times(int user);
371 extern void scheduler_tick(void);
373 extern void sched_show_task(struct task_struct *p);
375 #ifdef CONFIG_LOCKUP_DETECTOR
376 extern void touch_softlockup_watchdog(void);
377 extern void touch_softlockup_watchdog_sync(void);
378 extern void touch_all_softlockup_watchdogs(void);
379 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
381 size_t *lenp, loff_t *ppos);
382 extern unsigned int softlockup_panic;
383 void lockup_detector_init(void);
385 static inline void touch_softlockup_watchdog(void)
388 static inline void touch_softlockup_watchdog_sync(void)
391 static inline void touch_all_softlockup_watchdogs(void)
394 static inline void lockup_detector_init(void)
399 #ifdef CONFIG_DETECT_HUNG_TASK
400 void reset_hung_task_detector(void);
402 static inline void reset_hung_task_detector(void)
407 /* Attach to any functions which should be ignored in wchan output. */
408 #define __sched __attribute__((__section__(".sched.text")))
410 /* Linker adds these: start and end of __sched functions */
411 extern char __sched_text_start[], __sched_text_end[];
413 /* Is this address in the __sched functions? */
414 extern int in_sched_functions(unsigned long addr);
416 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
417 extern signed long schedule_timeout(signed long timeout);
418 extern signed long schedule_timeout_interruptible(signed long timeout);
419 extern signed long schedule_timeout_killable(signed long timeout);
420 extern signed long schedule_timeout_uninterruptible(signed long timeout);
421 asmlinkage void schedule(void);
422 extern void schedule_preempt_disabled(void);
424 extern long io_schedule_timeout(long timeout);
426 static inline void io_schedule(void)
428 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
432 struct user_namespace;
435 extern void arch_pick_mmap_layout(struct mm_struct *mm);
437 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
438 unsigned long, unsigned long);
440 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
441 unsigned long len, unsigned long pgoff,
442 unsigned long flags);
444 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
447 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
448 #define SUID_DUMP_USER 1 /* Dump as user of process */
449 #define SUID_DUMP_ROOT 2 /* Dump as root */
453 /* for SUID_DUMP_* above */
454 #define MMF_DUMPABLE_BITS 2
455 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
457 extern void set_dumpable(struct mm_struct *mm, int value);
459 * This returns the actual value of the suid_dumpable flag. For things
460 * that are using this for checking for privilege transitions, it must
461 * test against SUID_DUMP_USER rather than treating it as a boolean
464 static inline int __get_dumpable(unsigned long mm_flags)
466 return mm_flags & MMF_DUMPABLE_MASK;
469 static inline int get_dumpable(struct mm_struct *mm)
471 return __get_dumpable(mm->flags);
474 /* coredump filter bits */
475 #define MMF_DUMP_ANON_PRIVATE 2
476 #define MMF_DUMP_ANON_SHARED 3
477 #define MMF_DUMP_MAPPED_PRIVATE 4
478 #define MMF_DUMP_MAPPED_SHARED 5
479 #define MMF_DUMP_ELF_HEADERS 6
480 #define MMF_DUMP_HUGETLB_PRIVATE 7
481 #define MMF_DUMP_HUGETLB_SHARED 8
483 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
484 #define MMF_DUMP_FILTER_BITS 7
485 #define MMF_DUMP_FILTER_MASK \
486 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
487 #define MMF_DUMP_FILTER_DEFAULT \
488 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
489 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
491 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
492 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
494 # define MMF_DUMP_MASK_DEFAULT_ELF 0
496 /* leave room for more dump flags */
497 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
498 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
499 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
501 #define MMF_HAS_UPROBES 19 /* has uprobes */
502 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
504 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
506 struct sighand_struct {
508 struct k_sigaction action[_NSIG];
510 wait_queue_head_t signalfd_wqh;
513 struct pacct_struct {
516 unsigned long ac_mem;
517 cputime_t ac_utime, ac_stime;
518 unsigned long ac_minflt, ac_majflt;
529 * struct cputime - snaphsot of system and user cputime
530 * @utime: time spent in user mode
531 * @stime: time spent in system mode
533 * Gathers a generic snapshot of user and system time.
541 * struct task_cputime - collected CPU time counts
542 * @utime: time spent in user mode, in &cputime_t units
543 * @stime: time spent in kernel mode, in &cputime_t units
544 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
546 * This is an extension of struct cputime that includes the total runtime
547 * spent by the task from the scheduler point of view.
549 * As a result, this structure groups together three kinds of CPU time
550 * that are tracked for threads and thread groups. Most things considering
551 * CPU time want to group these counts together and treat all three
552 * of them in parallel.
554 struct task_cputime {
557 unsigned long long sum_exec_runtime;
559 /* Alternate field names when used to cache expirations. */
560 #define prof_exp stime
561 #define virt_exp utime
562 #define sched_exp sum_exec_runtime
564 #define INIT_CPUTIME \
565 (struct task_cputime) { \
568 .sum_exec_runtime = 0, \
571 #ifdef CONFIG_PREEMPT_COUNT
572 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
574 #define PREEMPT_DISABLED PREEMPT_ENABLED
578 * Disable preemption until the scheduler is running.
579 * Reset by start_kernel()->sched_init()->init_idle().
581 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
582 * before the scheduler is active -- see should_resched().
584 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
587 * struct thread_group_cputimer - thread group interval timer counts
588 * @cputime: thread group interval timers.
589 * @running: non-zero when there are timers running and
590 * @cputime receives updates.
591 * @lock: lock for fields in this struct.
593 * This structure contains the version of task_cputime, above, that is
594 * used for thread group CPU timer calculations.
596 struct thread_group_cputimer {
597 struct task_cputime cputime;
602 #include <linux/rwsem.h>
606 * NOTE! "signal_struct" does not have its own
607 * locking, because a shared signal_struct always
608 * implies a shared sighand_struct, so locking
609 * sighand_struct is always a proper superset of
610 * the locking of signal_struct.
612 struct signal_struct {
616 struct list_head thread_head;
618 wait_queue_head_t wait_chldexit; /* for wait4() */
620 /* current thread group signal load-balancing target: */
621 struct task_struct *curr_target;
623 /* shared signal handling: */
624 struct sigpending shared_pending;
626 /* thread group exit support */
629 * - notify group_exit_task when ->count is equal to notify_count
630 * - everyone except group_exit_task is stopped during signal delivery
631 * of fatal signals, group_exit_task processes the signal.
634 struct task_struct *group_exit_task;
636 /* thread group stop support, overloads group_exit_code too */
637 int group_stop_count;
638 unsigned int flags; /* see SIGNAL_* flags below */
641 * PR_SET_CHILD_SUBREAPER marks a process, like a service
642 * manager, to re-parent orphan (double-forking) child processes
643 * to this process instead of 'init'. The service manager is
644 * able to receive SIGCHLD signals and is able to investigate
645 * the process until it calls wait(). All children of this
646 * process will inherit a flag if they should look for a
647 * child_subreaper process at exit.
649 unsigned int is_child_subreaper:1;
650 unsigned int has_child_subreaper:1;
652 /* POSIX.1b Interval Timers */
654 struct list_head posix_timers;
656 /* ITIMER_REAL timer for the process */
657 struct hrtimer real_timer;
658 struct pid *leader_pid;
659 ktime_t it_real_incr;
662 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
663 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
664 * values are defined to 0 and 1 respectively
666 struct cpu_itimer it[2];
669 * Thread group totals for process CPU timers.
670 * See thread_group_cputimer(), et al, for details.
672 struct thread_group_cputimer cputimer;
674 /* Earliest-expiration cache. */
675 struct task_cputime cputime_expires;
677 struct list_head cpu_timers[3];
679 struct pid *tty_old_pgrp;
681 /* boolean value for session group leader */
684 struct tty_struct *tty; /* NULL if no tty */
686 #ifdef CONFIG_SCHED_AUTOGROUP
687 struct autogroup *autogroup;
690 * Cumulative resource counters for dead threads in the group,
691 * and for reaped dead child processes forked by this group.
692 * Live threads maintain their own counters and add to these
693 * in __exit_signal, except for the group leader.
695 seqlock_t stats_lock;
696 cputime_t utime, stime, cutime, cstime;
699 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
700 struct cputime prev_cputime;
702 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
703 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
704 unsigned long inblock, oublock, cinblock, coublock;
705 unsigned long maxrss, cmaxrss;
706 struct task_io_accounting ioac;
709 * Cumulative ns of schedule CPU time fo dead threads in the
710 * group, not including a zombie group leader, (This only differs
711 * from jiffies_to_ns(utime + stime) if sched_clock uses something
712 * other than jiffies.)
714 unsigned long long sum_sched_runtime;
717 * We don't bother to synchronize most readers of this at all,
718 * because there is no reader checking a limit that actually needs
719 * to get both rlim_cur and rlim_max atomically, and either one
720 * alone is a single word that can safely be read normally.
721 * getrlimit/setrlimit use task_lock(current->group_leader) to
722 * protect this instead of the siglock, because they really
723 * have no need to disable irqs.
725 struct rlimit rlim[RLIM_NLIMITS];
727 #ifdef CONFIG_BSD_PROCESS_ACCT
728 struct pacct_struct pacct; /* per-process accounting information */
730 #ifdef CONFIG_TASKSTATS
731 struct taskstats *stats;
735 unsigned audit_tty_log_passwd;
736 struct tty_audit_buf *tty_audit_buf;
738 #ifdef CONFIG_CGROUPS
740 * group_rwsem prevents new tasks from entering the threadgroup and
741 * member tasks from exiting,a more specifically, setting of
742 * PF_EXITING. fork and exit paths are protected with this rwsem
743 * using threadgroup_change_begin/end(). Users which require
744 * threadgroup to remain stable should use threadgroup_[un]lock()
745 * which also takes care of exec path. Currently, cgroup is the
748 struct rw_semaphore group_rwsem;
751 oom_flags_t oom_flags;
752 short oom_score_adj; /* OOM kill score adjustment */
753 short oom_score_adj_min; /* OOM kill score adjustment min value.
754 * Only settable by CAP_SYS_RESOURCE. */
756 struct mutex cred_guard_mutex; /* guard against foreign influences on
757 * credential calculations
758 * (notably. ptrace) */
762 * Bits in flags field of signal_struct.
764 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
765 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
766 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
767 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
769 * Pending notifications to parent.
771 #define SIGNAL_CLD_STOPPED 0x00000010
772 #define SIGNAL_CLD_CONTINUED 0x00000020
773 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
775 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
777 /* If true, all threads except ->group_exit_task have pending SIGKILL */
778 static inline int signal_group_exit(const struct signal_struct *sig)
780 return (sig->flags & SIGNAL_GROUP_EXIT) ||
781 (sig->group_exit_task != NULL);
785 * Some day this will be a full-fledged user tracking system..
788 atomic_t __count; /* reference count */
789 atomic_t processes; /* How many processes does this user have? */
790 atomic_t sigpending; /* How many pending signals does this user have? */
791 #ifdef CONFIG_INOTIFY_USER
792 atomic_t inotify_watches; /* How many inotify watches does this user have? */
793 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
795 #ifdef CONFIG_FANOTIFY
796 atomic_t fanotify_listeners;
799 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
801 #ifdef CONFIG_POSIX_MQUEUE
802 /* protected by mq_lock */
803 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
805 unsigned long locked_shm; /* How many pages of mlocked shm ? */
808 struct key *uid_keyring; /* UID specific keyring */
809 struct key *session_keyring; /* UID's default session keyring */
812 /* Hash table maintenance information */
813 struct hlist_node uidhash_node;
816 #ifdef CONFIG_PERF_EVENTS
817 atomic_long_t locked_vm;
821 extern int uids_sysfs_init(void);
823 extern struct user_struct *find_user(kuid_t);
825 extern struct user_struct root_user;
826 #define INIT_USER (&root_user)
829 struct backing_dev_info;
830 struct reclaim_state;
832 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
834 /* cumulative counters */
835 unsigned long pcount; /* # of times run on this cpu */
836 unsigned long long run_delay; /* time spent waiting on a runqueue */
839 unsigned long long last_arrival,/* when we last ran on a cpu */
840 last_queued; /* when we were last queued to run */
842 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
844 #ifdef CONFIG_TASK_DELAY_ACCT
845 struct task_delay_info {
847 unsigned int flags; /* Private per-task flags */
849 /* For each stat XXX, add following, aligned appropriately
851 * struct timespec XXX_start, XXX_end;
855 * Atomicity of updates to XXX_delay, XXX_count protected by
856 * single lock above (split into XXX_lock if contention is an issue).
860 * XXX_count is incremented on every XXX operation, the delay
861 * associated with the operation is added to XXX_delay.
862 * XXX_delay contains the accumulated delay time in nanoseconds.
864 u64 blkio_start; /* Shared by blkio, swapin */
865 u64 blkio_delay; /* wait for sync block io completion */
866 u64 swapin_delay; /* wait for swapin block io completion */
867 u32 blkio_count; /* total count of the number of sync block */
868 /* io operations performed */
869 u32 swapin_count; /* total count of the number of swapin block */
870 /* io operations performed */
873 u64 freepages_delay; /* wait for memory reclaim */
874 u32 freepages_count; /* total count of memory reclaim */
876 #endif /* CONFIG_TASK_DELAY_ACCT */
878 static inline int sched_info_on(void)
880 #ifdef CONFIG_SCHEDSTATS
882 #elif defined(CONFIG_TASK_DELAY_ACCT)
883 extern int delayacct_on;
898 * Increase resolution of cpu_capacity calculations
900 #define SCHED_CAPACITY_SHIFT 10
901 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
904 * sched-domains (multiprocessor balancing) declarations:
907 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
908 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
909 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
910 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
911 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
912 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
913 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
914 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
915 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
916 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
917 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
918 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
919 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
920 #define SD_NUMA 0x4000 /* cross-node balancing */
922 #ifdef CONFIG_SCHED_SMT
923 static inline int cpu_smt_flags(void)
925 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
929 #ifdef CONFIG_SCHED_MC
930 static inline int cpu_core_flags(void)
932 return SD_SHARE_PKG_RESOURCES;
937 static inline int cpu_numa_flags(void)
943 struct sched_domain_attr {
944 int relax_domain_level;
947 #define SD_ATTR_INIT (struct sched_domain_attr) { \
948 .relax_domain_level = -1, \
951 extern int sched_domain_level_max;
955 struct sched_domain {
956 /* These fields must be setup */
957 struct sched_domain *parent; /* top domain must be null terminated */
958 struct sched_domain *child; /* bottom domain must be null terminated */
959 struct sched_group *groups; /* the balancing groups of the domain */
960 unsigned long min_interval; /* Minimum balance interval ms */
961 unsigned long max_interval; /* Maximum balance interval ms */
962 unsigned int busy_factor; /* less balancing by factor if busy */
963 unsigned int imbalance_pct; /* No balance until over watermark */
964 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
965 unsigned int busy_idx;
966 unsigned int idle_idx;
967 unsigned int newidle_idx;
968 unsigned int wake_idx;
969 unsigned int forkexec_idx;
970 unsigned int smt_gain;
972 int nohz_idle; /* NOHZ IDLE status */
973 int flags; /* See SD_* */
976 /* Runtime fields. */
977 unsigned long last_balance; /* init to jiffies. units in jiffies */
978 unsigned int balance_interval; /* initialise to 1. units in ms. */
979 unsigned int nr_balance_failed; /* initialise to 0 */
981 /* idle_balance() stats */
982 u64 max_newidle_lb_cost;
983 unsigned long next_decay_max_lb_cost;
985 #ifdef CONFIG_SCHEDSTATS
986 /* load_balance() stats */
987 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
988 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
989 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
990 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
991 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
992 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
993 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
994 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
996 /* Active load balancing */
997 unsigned int alb_count;
998 unsigned int alb_failed;
999 unsigned int alb_pushed;
1001 /* SD_BALANCE_EXEC stats */
1002 unsigned int sbe_count;
1003 unsigned int sbe_balanced;
1004 unsigned int sbe_pushed;
1006 /* SD_BALANCE_FORK stats */
1007 unsigned int sbf_count;
1008 unsigned int sbf_balanced;
1009 unsigned int sbf_pushed;
1011 /* try_to_wake_up() stats */
1012 unsigned int ttwu_wake_remote;
1013 unsigned int ttwu_move_affine;
1014 unsigned int ttwu_move_balance;
1016 #ifdef CONFIG_SCHED_DEBUG
1020 void *private; /* used during construction */
1021 struct rcu_head rcu; /* used during destruction */
1024 unsigned int span_weight;
1026 * Span of all CPUs in this domain.
1028 * NOTE: this field is variable length. (Allocated dynamically
1029 * by attaching extra space to the end of the structure,
1030 * depending on how many CPUs the kernel has booted up with)
1032 unsigned long span[0];
1035 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1037 return to_cpumask(sd->span);
1040 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1041 struct sched_domain_attr *dattr_new);
1043 /* Allocate an array of sched domains, for partition_sched_domains(). */
1044 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1045 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1047 bool cpus_share_cache(int this_cpu, int that_cpu);
1049 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1050 typedef int (*sched_domain_flags_f)(void);
1052 #define SDTL_OVERLAP 0x01
1055 struct sched_domain **__percpu sd;
1056 struct sched_group **__percpu sg;
1057 struct sched_group_capacity **__percpu sgc;
1060 struct sched_domain_topology_level {
1061 sched_domain_mask_f mask;
1062 sched_domain_flags_f sd_flags;
1065 struct sd_data data;
1066 #ifdef CONFIG_SCHED_DEBUG
1071 extern struct sched_domain_topology_level *sched_domain_topology;
1073 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1074 extern void wake_up_if_idle(int cpu);
1076 #ifdef CONFIG_SCHED_DEBUG
1077 # define SD_INIT_NAME(type) .name = #type
1079 # define SD_INIT_NAME(type)
1082 #else /* CONFIG_SMP */
1084 struct sched_domain_attr;
1087 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1088 struct sched_domain_attr *dattr_new)
1092 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1097 #endif /* !CONFIG_SMP */
1100 struct io_context; /* See blkdev.h */
1103 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1104 extern void prefetch_stack(struct task_struct *t);
1106 static inline void prefetch_stack(struct task_struct *t) { }
1109 struct audit_context; /* See audit.c */
1111 struct pipe_inode_info;
1112 struct uts_namespace;
1114 struct load_weight {
1115 unsigned long weight;
1120 u64 last_runnable_update;
1123 * utilization_avg_contrib describes the amount of time that a
1124 * sched_entity is running on a CPU. It is based on running_avg_sum
1125 * and is scaled in the range [0..SCHED_LOAD_SCALE].
1126 * load_avg_contrib described the amount of time that a sched_entity
1127 * is runnable on a rq. It is based on both runnable_avg_sum and the
1128 * weight of the task.
1130 unsigned long load_avg_contrib, utilization_avg_contrib;
1132 * These sums represent an infinite geometric series and so are bound
1133 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1134 * choices of y < 1-2^(-32)*1024.
1135 * running_avg_sum reflects the time that the sched_entity is
1136 * effectively running on the CPU.
1137 * runnable_avg_sum represents the amount of time a sched_entity is on
1138 * a runqueue which includes the running time that is monitored by
1141 u32 runnable_avg_sum, avg_period, running_avg_sum;
1144 #ifdef CONFIG_SCHEDSTATS
1145 struct sched_statistics {
1155 s64 sum_sleep_runtime;
1162 u64 nr_migrations_cold;
1163 u64 nr_failed_migrations_affine;
1164 u64 nr_failed_migrations_running;
1165 u64 nr_failed_migrations_hot;
1166 u64 nr_forced_migrations;
1169 u64 nr_wakeups_sync;
1170 u64 nr_wakeups_migrate;
1171 u64 nr_wakeups_local;
1172 u64 nr_wakeups_remote;
1173 u64 nr_wakeups_affine;
1174 u64 nr_wakeups_affine_attempts;
1175 u64 nr_wakeups_passive;
1176 u64 nr_wakeups_idle;
1180 struct sched_entity {
1181 struct load_weight load; /* for load-balancing */
1182 struct rb_node run_node;
1183 struct list_head group_node;
1187 u64 sum_exec_runtime;
1189 u64 prev_sum_exec_runtime;
1193 #ifdef CONFIG_SCHEDSTATS
1194 struct sched_statistics statistics;
1197 #ifdef CONFIG_FAIR_GROUP_SCHED
1199 struct sched_entity *parent;
1200 /* rq on which this entity is (to be) queued: */
1201 struct cfs_rq *cfs_rq;
1202 /* rq "owned" by this entity/group: */
1203 struct cfs_rq *my_q;
1207 /* Per-entity load-tracking */
1208 struct sched_avg avg;
1212 struct sched_rt_entity {
1213 struct list_head run_list;
1214 unsigned long timeout;
1215 unsigned long watchdog_stamp;
1216 unsigned int time_slice;
1218 struct sched_rt_entity *back;
1219 #ifdef CONFIG_RT_GROUP_SCHED
1220 struct sched_rt_entity *parent;
1221 /* rq on which this entity is (to be) queued: */
1222 struct rt_rq *rt_rq;
1223 /* rq "owned" by this entity/group: */
1228 struct sched_dl_entity {
1229 struct rb_node rb_node;
1232 * Original scheduling parameters. Copied here from sched_attr
1233 * during sched_setattr(), they will remain the same until
1234 * the next sched_setattr().
1236 u64 dl_runtime; /* maximum runtime for each instance */
1237 u64 dl_deadline; /* relative deadline of each instance */
1238 u64 dl_period; /* separation of two instances (period) */
1239 u64 dl_bw; /* dl_runtime / dl_deadline */
1242 * Actual scheduling parameters. Initialized with the values above,
1243 * they are continously updated during task execution. Note that
1244 * the remaining runtime could be < 0 in case we are in overrun.
1246 s64 runtime; /* remaining runtime for this instance */
1247 u64 deadline; /* absolute deadline for this instance */
1248 unsigned int flags; /* specifying the scheduler behaviour */
1253 * @dl_throttled tells if we exhausted the runtime. If so, the
1254 * task has to wait for a replenishment to be performed at the
1255 * next firing of dl_timer.
1257 * @dl_new tells if a new instance arrived. If so we must
1258 * start executing it with full runtime and reset its absolute
1261 * @dl_boosted tells if we are boosted due to DI. If so we are
1262 * outside bandwidth enforcement mechanism (but only until we
1263 * exit the critical section);
1265 * @dl_yielded tells if task gave up the cpu before consuming
1266 * all its available runtime during the last job.
1268 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1271 * Bandwidth enforcement timer. Each -deadline task has its
1272 * own bandwidth to be enforced, thus we need one timer per task.
1274 struct hrtimer dl_timer;
1286 enum perf_event_task_context {
1287 perf_invalid_context = -1,
1288 perf_hw_context = 0,
1290 perf_nr_task_contexts,
1293 struct task_struct {
1294 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1297 unsigned int flags; /* per process flags, defined below */
1298 unsigned int ptrace;
1301 struct llist_node wake_entry;
1303 struct task_struct *last_wakee;
1304 unsigned long wakee_flips;
1305 unsigned long wakee_flip_decay_ts;
1311 int prio, static_prio, normal_prio;
1312 unsigned int rt_priority;
1313 const struct sched_class *sched_class;
1314 struct sched_entity se;
1315 struct sched_rt_entity rt;
1316 #ifdef CONFIG_CGROUP_SCHED
1317 struct task_group *sched_task_group;
1319 struct sched_dl_entity dl;
1321 #ifdef CONFIG_PREEMPT_NOTIFIERS
1322 /* list of struct preempt_notifier: */
1323 struct hlist_head preempt_notifiers;
1326 #ifdef CONFIG_BLK_DEV_IO_TRACE
1327 unsigned int btrace_seq;
1330 unsigned int policy;
1331 int nr_cpus_allowed;
1332 cpumask_t cpus_allowed;
1334 #ifdef CONFIG_PREEMPT_RCU
1335 int rcu_read_lock_nesting;
1336 union rcu_special rcu_read_unlock_special;
1337 struct list_head rcu_node_entry;
1338 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1339 #ifdef CONFIG_PREEMPT_RCU
1340 struct rcu_node *rcu_blocked_node;
1341 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1342 #ifdef CONFIG_TASKS_RCU
1343 unsigned long rcu_tasks_nvcsw;
1344 bool rcu_tasks_holdout;
1345 struct list_head rcu_tasks_holdout_list;
1346 int rcu_tasks_idle_cpu;
1347 #endif /* #ifdef CONFIG_TASKS_RCU */
1349 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1350 struct sched_info sched_info;
1353 struct list_head tasks;
1355 struct plist_node pushable_tasks;
1356 struct rb_node pushable_dl_tasks;
1359 struct mm_struct *mm, *active_mm;
1360 /* per-thread vma caching */
1361 u32 vmacache_seqnum;
1362 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1363 #if defined(SPLIT_RSS_COUNTING)
1364 struct task_rss_stat rss_stat;
1368 int exit_code, exit_signal;
1369 int pdeath_signal; /* The signal sent when the parent dies */
1370 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1372 /* Used for emulating ABI behavior of previous Linux versions */
1373 unsigned int personality;
1375 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1377 unsigned in_iowait:1;
1379 /* Revert to default priority/policy when forking */
1380 unsigned sched_reset_on_fork:1;
1381 unsigned sched_contributes_to_load:1;
1382 unsigned sched_migrated:1;
1384 #ifdef CONFIG_MEMCG_KMEM
1385 unsigned memcg_kmem_skip_account:1;
1387 #ifdef CONFIG_COMPAT_BRK
1388 unsigned brk_randomized:1;
1391 unsigned long atomic_flags; /* Flags needing atomic access. */
1393 struct restart_block restart_block;
1398 #ifdef CONFIG_CC_STACKPROTECTOR
1399 /* Canary value for the -fstack-protector gcc feature */
1400 unsigned long stack_canary;
1403 * pointers to (original) parent process, youngest child, younger sibling,
1404 * older sibling, respectively. (p->father can be replaced with
1405 * p->real_parent->pid)
1407 struct task_struct __rcu *real_parent; /* real parent process */
1408 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1410 * children/sibling forms the list of my natural children
1412 struct list_head children; /* list of my children */
1413 struct list_head sibling; /* linkage in my parent's children list */
1414 struct task_struct *group_leader; /* threadgroup leader */
1417 * ptraced is the list of tasks this task is using ptrace on.
1418 * This includes both natural children and PTRACE_ATTACH targets.
1419 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1421 struct list_head ptraced;
1422 struct list_head ptrace_entry;
1424 /* PID/PID hash table linkage. */
1425 struct pid_link pids[PIDTYPE_MAX];
1426 struct list_head thread_group;
1427 struct list_head thread_node;
1429 struct completion *vfork_done; /* for vfork() */
1430 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1431 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1433 cputime_t utime, stime, utimescaled, stimescaled;
1435 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1436 struct cputime prev_cputime;
1438 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1439 seqlock_t vtime_seqlock;
1440 unsigned long long vtime_snap;
1445 } vtime_snap_whence;
1447 unsigned long nvcsw, nivcsw; /* context switch counts */
1448 u64 start_time; /* monotonic time in nsec */
1449 u64 real_start_time; /* boot based time in nsec */
1450 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1451 unsigned long min_flt, maj_flt;
1453 struct task_cputime cputime_expires;
1454 struct list_head cpu_timers[3];
1456 /* process credentials */
1457 const struct cred __rcu *real_cred; /* objective and real subjective task
1458 * credentials (COW) */
1459 const struct cred __rcu *cred; /* effective (overridable) subjective task
1460 * credentials (COW) */
1461 char comm[TASK_COMM_LEN]; /* executable name excluding path
1462 - access with [gs]et_task_comm (which lock
1463 it with task_lock())
1464 - initialized normally by setup_new_exec */
1465 /* file system info */
1466 struct nameidata *nameidata;
1467 #ifdef CONFIG_SYSVIPC
1469 struct sysv_sem sysvsem;
1470 struct sysv_shm sysvshm;
1472 #ifdef CONFIG_DETECT_HUNG_TASK
1473 /* hung task detection */
1474 unsigned long last_switch_count;
1476 /* CPU-specific state of this task */
1477 struct thread_struct thread;
1478 /* filesystem information */
1479 struct fs_struct *fs;
1480 /* open file information */
1481 struct files_struct *files;
1483 struct nsproxy *nsproxy;
1484 /* signal handlers */
1485 struct signal_struct *signal;
1486 struct sighand_struct *sighand;
1488 sigset_t blocked, real_blocked;
1489 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1490 struct sigpending pending;
1492 unsigned long sas_ss_sp;
1494 int (*notifier)(void *priv);
1495 void *notifier_data;
1496 sigset_t *notifier_mask;
1497 struct callback_head *task_works;
1499 struct audit_context *audit_context;
1500 #ifdef CONFIG_AUDITSYSCALL
1502 unsigned int sessionid;
1504 struct seccomp seccomp;
1506 /* Thread group tracking */
1509 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1511 spinlock_t alloc_lock;
1513 /* Protection of the PI data structures: */
1514 raw_spinlock_t pi_lock;
1516 #ifdef CONFIG_RT_MUTEXES
1517 /* PI waiters blocked on a rt_mutex held by this task */
1518 struct rb_root pi_waiters;
1519 struct rb_node *pi_waiters_leftmost;
1520 /* Deadlock detection and priority inheritance handling */
1521 struct rt_mutex_waiter *pi_blocked_on;
1524 #ifdef CONFIG_DEBUG_MUTEXES
1525 /* mutex deadlock detection */
1526 struct mutex_waiter *blocked_on;
1528 #ifdef CONFIG_TRACE_IRQFLAGS
1529 unsigned int irq_events;
1530 unsigned long hardirq_enable_ip;
1531 unsigned long hardirq_disable_ip;
1532 unsigned int hardirq_enable_event;
1533 unsigned int hardirq_disable_event;
1534 int hardirqs_enabled;
1535 int hardirq_context;
1536 unsigned long softirq_disable_ip;
1537 unsigned long softirq_enable_ip;
1538 unsigned int softirq_disable_event;
1539 unsigned int softirq_enable_event;
1540 int softirqs_enabled;
1541 int softirq_context;
1543 #ifdef CONFIG_LOCKDEP
1544 # define MAX_LOCK_DEPTH 48UL
1547 unsigned int lockdep_recursion;
1548 struct held_lock held_locks[MAX_LOCK_DEPTH];
1549 gfp_t lockdep_reclaim_gfp;
1552 /* journalling filesystem info */
1555 /* stacked block device info */
1556 struct bio_list *bio_list;
1559 /* stack plugging */
1560 struct blk_plug *plug;
1564 struct reclaim_state *reclaim_state;
1566 struct backing_dev_info *backing_dev_info;
1568 struct io_context *io_context;
1570 unsigned long ptrace_message;
1571 siginfo_t *last_siginfo; /* For ptrace use. */
1572 struct task_io_accounting ioac;
1573 #if defined(CONFIG_TASK_XACCT)
1574 u64 acct_rss_mem1; /* accumulated rss usage */
1575 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1576 cputime_t acct_timexpd; /* stime + utime since last update */
1578 #ifdef CONFIG_CPUSETS
1579 nodemask_t mems_allowed; /* Protected by alloc_lock */
1580 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1581 int cpuset_mem_spread_rotor;
1582 int cpuset_slab_spread_rotor;
1584 #ifdef CONFIG_CGROUPS
1585 /* Control Group info protected by css_set_lock */
1586 struct css_set __rcu *cgroups;
1587 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1588 struct list_head cg_list;
1591 struct robust_list_head __user *robust_list;
1592 #ifdef CONFIG_COMPAT
1593 struct compat_robust_list_head __user *compat_robust_list;
1595 struct list_head pi_state_list;
1596 struct futex_pi_state *pi_state_cache;
1598 #ifdef CONFIG_PERF_EVENTS
1599 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1600 struct mutex perf_event_mutex;
1601 struct list_head perf_event_list;
1603 #ifdef CONFIG_DEBUG_PREEMPT
1604 unsigned long preempt_disable_ip;
1607 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1609 short pref_node_fork;
1611 #ifdef CONFIG_NUMA_BALANCING
1613 unsigned int numa_scan_period;
1614 unsigned int numa_scan_period_max;
1615 int numa_preferred_nid;
1616 unsigned long numa_migrate_retry;
1617 u64 node_stamp; /* migration stamp */
1618 u64 last_task_numa_placement;
1619 u64 last_sum_exec_runtime;
1620 struct callback_head numa_work;
1622 struct list_head numa_entry;
1623 struct numa_group *numa_group;
1626 * numa_faults is an array split into four regions:
1627 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1628 * in this precise order.
1630 * faults_memory: Exponential decaying average of faults on a per-node
1631 * basis. Scheduling placement decisions are made based on these
1632 * counts. The values remain static for the duration of a PTE scan.
1633 * faults_cpu: Track the nodes the process was running on when a NUMA
1634 * hinting fault was incurred.
1635 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1636 * during the current scan window. When the scan completes, the counts
1637 * in faults_memory and faults_cpu decay and these values are copied.
1639 unsigned long *numa_faults;
1640 unsigned long total_numa_faults;
1643 * numa_faults_locality tracks if faults recorded during the last
1644 * scan window were remote/local or failed to migrate. The task scan
1645 * period is adapted based on the locality of the faults with different
1646 * weights depending on whether they were shared or private faults
1648 unsigned long numa_faults_locality[3];
1650 unsigned long numa_pages_migrated;
1651 #endif /* CONFIG_NUMA_BALANCING */
1653 struct rcu_head rcu;
1656 * cache last used pipe for splice
1658 struct pipe_inode_info *splice_pipe;
1660 struct page_frag task_frag;
1662 #ifdef CONFIG_TASK_DELAY_ACCT
1663 struct task_delay_info *delays;
1665 #ifdef CONFIG_FAULT_INJECTION
1669 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1670 * balance_dirty_pages() for some dirty throttling pause
1673 int nr_dirtied_pause;
1674 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1676 #ifdef CONFIG_LATENCYTOP
1677 int latency_record_count;
1678 struct latency_record latency_record[LT_SAVECOUNT];
1681 * time slack values; these are used to round up poll() and
1682 * select() etc timeout values. These are in nanoseconds.
1684 unsigned long timer_slack_ns;
1685 unsigned long default_timer_slack_ns;
1688 unsigned int kasan_depth;
1690 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1691 /* Index of current stored address in ret_stack */
1693 /* Stack of return addresses for return function tracing */
1694 struct ftrace_ret_stack *ret_stack;
1695 /* time stamp for last schedule */
1696 unsigned long long ftrace_timestamp;
1698 * Number of functions that haven't been traced
1699 * because of depth overrun.
1701 atomic_t trace_overrun;
1702 /* Pause for the tracing */
1703 atomic_t tracing_graph_pause;
1705 #ifdef CONFIG_TRACING
1706 /* state flags for use by tracers */
1707 unsigned long trace;
1708 /* bitmask and counter of trace recursion */
1709 unsigned long trace_recursion;
1710 #endif /* CONFIG_TRACING */
1712 struct memcg_oom_info {
1713 struct mem_cgroup *memcg;
1716 unsigned int may_oom:1;
1719 #ifdef CONFIG_UPROBES
1720 struct uprobe_task *utask;
1722 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1723 unsigned int sequential_io;
1724 unsigned int sequential_io_avg;
1726 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1727 unsigned long task_state_change;
1731 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1732 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1734 #define TNF_MIGRATED 0x01
1735 #define TNF_NO_GROUP 0x02
1736 #define TNF_SHARED 0x04
1737 #define TNF_FAULT_LOCAL 0x08
1738 #define TNF_MIGRATE_FAIL 0x10
1740 #ifdef CONFIG_NUMA_BALANCING
1741 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1742 extern pid_t task_numa_group_id(struct task_struct *p);
1743 extern void set_numabalancing_state(bool enabled);
1744 extern void task_numa_free(struct task_struct *p);
1745 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1746 int src_nid, int dst_cpu);
1748 static inline void task_numa_fault(int last_node, int node, int pages,
1752 static inline pid_t task_numa_group_id(struct task_struct *p)
1756 static inline void set_numabalancing_state(bool enabled)
1759 static inline void task_numa_free(struct task_struct *p)
1762 static inline bool should_numa_migrate_memory(struct task_struct *p,
1763 struct page *page, int src_nid, int dst_cpu)
1769 static inline struct pid *task_pid(struct task_struct *task)
1771 return task->pids[PIDTYPE_PID].pid;
1774 static inline struct pid *task_tgid(struct task_struct *task)
1776 return task->group_leader->pids[PIDTYPE_PID].pid;
1780 * Without tasklist or rcu lock it is not safe to dereference
1781 * the result of task_pgrp/task_session even if task == current,
1782 * we can race with another thread doing sys_setsid/sys_setpgid.
1784 static inline struct pid *task_pgrp(struct task_struct *task)
1786 return task->group_leader->pids[PIDTYPE_PGID].pid;
1789 static inline struct pid *task_session(struct task_struct *task)
1791 return task->group_leader->pids[PIDTYPE_SID].pid;
1794 struct pid_namespace;
1797 * the helpers to get the task's different pids as they are seen
1798 * from various namespaces
1800 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1801 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1803 * task_xid_nr_ns() : id seen from the ns specified;
1805 * set_task_vxid() : assigns a virtual id to a task;
1807 * see also pid_nr() etc in include/linux/pid.h
1809 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1810 struct pid_namespace *ns);
1812 static inline pid_t task_pid_nr(struct task_struct *tsk)
1817 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1818 struct pid_namespace *ns)
1820 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1823 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1825 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1829 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1834 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1836 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1838 return pid_vnr(task_tgid(tsk));
1842 static inline int pid_alive(const struct task_struct *p);
1843 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1849 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1855 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1857 return task_ppid_nr_ns(tsk, &init_pid_ns);
1860 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1861 struct pid_namespace *ns)
1863 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1866 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1868 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1872 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1873 struct pid_namespace *ns)
1875 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1878 static inline pid_t task_session_vnr(struct task_struct *tsk)
1880 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1883 /* obsolete, do not use */
1884 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1886 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1890 * pid_alive - check that a task structure is not stale
1891 * @p: Task structure to be checked.
1893 * Test if a process is not yet dead (at most zombie state)
1894 * If pid_alive fails, then pointers within the task structure
1895 * can be stale and must not be dereferenced.
1897 * Return: 1 if the process is alive. 0 otherwise.
1899 static inline int pid_alive(const struct task_struct *p)
1901 return p->pids[PIDTYPE_PID].pid != NULL;
1905 * is_global_init - check if a task structure is init
1906 * @tsk: Task structure to be checked.
1908 * Check if a task structure is the first user space task the kernel created.
1910 * Return: 1 if the task structure is init. 0 otherwise.
1912 static inline int is_global_init(struct task_struct *tsk)
1914 return tsk->pid == 1;
1917 extern struct pid *cad_pid;
1919 extern void free_task(struct task_struct *tsk);
1920 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1922 extern void __put_task_struct(struct task_struct *t);
1924 static inline void put_task_struct(struct task_struct *t)
1926 if (atomic_dec_and_test(&t->usage))
1927 __put_task_struct(t);
1930 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1931 extern void task_cputime(struct task_struct *t,
1932 cputime_t *utime, cputime_t *stime);
1933 extern void task_cputime_scaled(struct task_struct *t,
1934 cputime_t *utimescaled, cputime_t *stimescaled);
1935 extern cputime_t task_gtime(struct task_struct *t);
1937 static inline void task_cputime(struct task_struct *t,
1938 cputime_t *utime, cputime_t *stime)
1946 static inline void task_cputime_scaled(struct task_struct *t,
1947 cputime_t *utimescaled,
1948 cputime_t *stimescaled)
1951 *utimescaled = t->utimescaled;
1953 *stimescaled = t->stimescaled;
1956 static inline cputime_t task_gtime(struct task_struct *t)
1961 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1962 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1967 #define PF_EXITING 0x00000004 /* getting shut down */
1968 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1969 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1970 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1971 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1972 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1973 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1974 #define PF_DUMPCORE 0x00000200 /* dumped core */
1975 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1976 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1977 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1978 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1979 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1980 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1981 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1982 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1983 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1984 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1985 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1986 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1987 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1988 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1989 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1990 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1991 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1992 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1993 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1996 * Only the _current_ task can read/write to tsk->flags, but other
1997 * tasks can access tsk->flags in readonly mode for example
1998 * with tsk_used_math (like during threaded core dumping).
1999 * There is however an exception to this rule during ptrace
2000 * or during fork: the ptracer task is allowed to write to the
2001 * child->flags of its traced child (same goes for fork, the parent
2002 * can write to the child->flags), because we're guaranteed the
2003 * child is not running and in turn not changing child->flags
2004 * at the same time the parent does it.
2006 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2007 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2008 #define clear_used_math() clear_stopped_child_used_math(current)
2009 #define set_used_math() set_stopped_child_used_math(current)
2010 #define conditional_stopped_child_used_math(condition, child) \
2011 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2012 #define conditional_used_math(condition) \
2013 conditional_stopped_child_used_math(condition, current)
2014 #define copy_to_stopped_child_used_math(child) \
2015 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2016 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2017 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2018 #define used_math() tsk_used_math(current)
2020 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2021 * __GFP_FS is also cleared as it implies __GFP_IO.
2023 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2025 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2026 flags &= ~(__GFP_IO | __GFP_FS);
2030 static inline unsigned int memalloc_noio_save(void)
2032 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2033 current->flags |= PF_MEMALLOC_NOIO;
2037 static inline void memalloc_noio_restore(unsigned int flags)
2039 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2042 /* Per-process atomic flags. */
2043 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2044 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2045 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2048 #define TASK_PFA_TEST(name, func) \
2049 static inline bool task_##func(struct task_struct *p) \
2050 { return test_bit(PFA_##name, &p->atomic_flags); }
2051 #define TASK_PFA_SET(name, func) \
2052 static inline void task_set_##func(struct task_struct *p) \
2053 { set_bit(PFA_##name, &p->atomic_flags); }
2054 #define TASK_PFA_CLEAR(name, func) \
2055 static inline void task_clear_##func(struct task_struct *p) \
2056 { clear_bit(PFA_##name, &p->atomic_flags); }
2058 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2059 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2061 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2062 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2063 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2065 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2066 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2067 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2070 * task->jobctl flags
2072 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2074 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2075 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2076 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2077 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2078 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2079 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2080 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2082 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
2083 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
2084 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
2085 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
2086 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
2087 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
2088 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
2090 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2091 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2093 extern bool task_set_jobctl_pending(struct task_struct *task,
2095 extern void task_clear_jobctl_trapping(struct task_struct *task);
2096 extern void task_clear_jobctl_pending(struct task_struct *task,
2099 static inline void rcu_copy_process(struct task_struct *p)
2101 #ifdef CONFIG_PREEMPT_RCU
2102 p->rcu_read_lock_nesting = 0;
2103 p->rcu_read_unlock_special.s = 0;
2104 p->rcu_blocked_node = NULL;
2105 INIT_LIST_HEAD(&p->rcu_node_entry);
2106 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2107 #ifdef CONFIG_TASKS_RCU
2108 p->rcu_tasks_holdout = false;
2109 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2110 p->rcu_tasks_idle_cpu = -1;
2111 #endif /* #ifdef CONFIG_TASKS_RCU */
2114 static inline void tsk_restore_flags(struct task_struct *task,
2115 unsigned long orig_flags, unsigned long flags)
2117 task->flags &= ~flags;
2118 task->flags |= orig_flags & flags;
2121 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2122 const struct cpumask *trial);
2123 extern int task_can_attach(struct task_struct *p,
2124 const struct cpumask *cs_cpus_allowed);
2126 extern void do_set_cpus_allowed(struct task_struct *p,
2127 const struct cpumask *new_mask);
2129 extern int set_cpus_allowed_ptr(struct task_struct *p,
2130 const struct cpumask *new_mask);
2132 static inline void do_set_cpus_allowed(struct task_struct *p,
2133 const struct cpumask *new_mask)
2136 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2137 const struct cpumask *new_mask)
2139 if (!cpumask_test_cpu(0, new_mask))
2145 #ifdef CONFIG_NO_HZ_COMMON
2146 void calc_load_enter_idle(void);
2147 void calc_load_exit_idle(void);
2149 static inline void calc_load_enter_idle(void) { }
2150 static inline void calc_load_exit_idle(void) { }
2151 #endif /* CONFIG_NO_HZ_COMMON */
2153 #ifndef CONFIG_CPUMASK_OFFSTACK
2154 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2156 return set_cpus_allowed_ptr(p, &new_mask);
2161 * Do not use outside of architecture code which knows its limitations.
2163 * sched_clock() has no promise of monotonicity or bounded drift between
2164 * CPUs, use (which you should not) requires disabling IRQs.
2166 * Please use one of the three interfaces below.
2168 extern unsigned long long notrace sched_clock(void);
2170 * See the comment in kernel/sched/clock.c
2172 extern u64 cpu_clock(int cpu);
2173 extern u64 local_clock(void);
2174 extern u64 running_clock(void);
2175 extern u64 sched_clock_cpu(int cpu);
2178 extern void sched_clock_init(void);
2180 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2181 static inline void sched_clock_tick(void)
2185 static inline void sched_clock_idle_sleep_event(void)
2189 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2194 * Architectures can set this to 1 if they have specified
2195 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2196 * but then during bootup it turns out that sched_clock()
2197 * is reliable after all:
2199 extern int sched_clock_stable(void);
2200 extern void set_sched_clock_stable(void);
2201 extern void clear_sched_clock_stable(void);
2203 extern void sched_clock_tick(void);
2204 extern void sched_clock_idle_sleep_event(void);
2205 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2208 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2210 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2211 * The reason for this explicit opt-in is not to have perf penalty with
2212 * slow sched_clocks.
2214 extern void enable_sched_clock_irqtime(void);
2215 extern void disable_sched_clock_irqtime(void);
2217 static inline void enable_sched_clock_irqtime(void) {}
2218 static inline void disable_sched_clock_irqtime(void) {}
2221 extern unsigned long long
2222 task_sched_runtime(struct task_struct *task);
2224 /* sched_exec is called by processes performing an exec */
2226 extern void sched_exec(void);
2228 #define sched_exec() {}
2231 extern void sched_clock_idle_sleep_event(void);
2232 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2234 #ifdef CONFIG_HOTPLUG_CPU
2235 extern void idle_task_exit(void);
2237 static inline void idle_task_exit(void) {}
2240 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2241 extern void wake_up_nohz_cpu(int cpu);
2243 static inline void wake_up_nohz_cpu(int cpu) { }
2246 #ifdef CONFIG_NO_HZ_FULL
2247 extern bool sched_can_stop_tick(void);
2248 extern u64 scheduler_tick_max_deferment(void);
2250 static inline bool sched_can_stop_tick(void) { return false; }
2253 #ifdef CONFIG_SCHED_AUTOGROUP
2254 extern void sched_autogroup_create_attach(struct task_struct *p);
2255 extern void sched_autogroup_detach(struct task_struct *p);
2256 extern void sched_autogroup_fork(struct signal_struct *sig);
2257 extern void sched_autogroup_exit(struct signal_struct *sig);
2258 #ifdef CONFIG_PROC_FS
2259 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2260 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2263 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2264 static inline void sched_autogroup_detach(struct task_struct *p) { }
2265 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2266 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2269 extern int yield_to(struct task_struct *p, bool preempt);
2270 extern void set_user_nice(struct task_struct *p, long nice);
2271 extern int task_prio(const struct task_struct *p);
2273 * task_nice - return the nice value of a given task.
2274 * @p: the task in question.
2276 * Return: The nice value [ -20 ... 0 ... 19 ].
2278 static inline int task_nice(const struct task_struct *p)
2280 return PRIO_TO_NICE((p)->static_prio);
2282 extern int can_nice(const struct task_struct *p, const int nice);
2283 extern int task_curr(const struct task_struct *p);
2284 extern int idle_cpu(int cpu);
2285 extern int sched_setscheduler(struct task_struct *, int,
2286 const struct sched_param *);
2287 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2288 const struct sched_param *);
2289 extern int sched_setattr(struct task_struct *,
2290 const struct sched_attr *);
2291 extern struct task_struct *idle_task(int cpu);
2293 * is_idle_task - is the specified task an idle task?
2294 * @p: the task in question.
2296 * Return: 1 if @p is an idle task. 0 otherwise.
2298 static inline bool is_idle_task(const struct task_struct *p)
2302 extern struct task_struct *curr_task(int cpu);
2303 extern void set_curr_task(int cpu, struct task_struct *p);
2307 union thread_union {
2308 struct thread_info thread_info;
2309 unsigned long stack[THREAD_SIZE/sizeof(long)];
2312 #ifndef __HAVE_ARCH_KSTACK_END
2313 static inline int kstack_end(void *addr)
2315 /* Reliable end of stack detection:
2316 * Some APM bios versions misalign the stack
2318 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2322 extern union thread_union init_thread_union;
2323 extern struct task_struct init_task;
2325 extern struct mm_struct init_mm;
2327 extern struct pid_namespace init_pid_ns;
2330 * find a task by one of its numerical ids
2332 * find_task_by_pid_ns():
2333 * finds a task by its pid in the specified namespace
2334 * find_task_by_vpid():
2335 * finds a task by its virtual pid
2337 * see also find_vpid() etc in include/linux/pid.h
2340 extern struct task_struct *find_task_by_vpid(pid_t nr);
2341 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2342 struct pid_namespace *ns);
2344 /* per-UID process charging. */
2345 extern struct user_struct * alloc_uid(kuid_t);
2346 static inline struct user_struct *get_uid(struct user_struct *u)
2348 atomic_inc(&u->__count);
2351 extern void free_uid(struct user_struct *);
2353 #include <asm/current.h>
2355 extern void xtime_update(unsigned long ticks);
2357 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2358 extern int wake_up_process(struct task_struct *tsk);
2359 extern void wake_up_new_task(struct task_struct *tsk);
2361 extern void kick_process(struct task_struct *tsk);
2363 static inline void kick_process(struct task_struct *tsk) { }
2365 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2366 extern void sched_dead(struct task_struct *p);
2368 extern void proc_caches_init(void);
2369 extern void flush_signals(struct task_struct *);
2370 extern void __flush_signals(struct task_struct *);
2371 extern void ignore_signals(struct task_struct *);
2372 extern void flush_signal_handlers(struct task_struct *, int force_default);
2373 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2375 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2377 unsigned long flags;
2380 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2381 ret = dequeue_signal(tsk, mask, info);
2382 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2387 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2389 extern void unblock_all_signals(void);
2390 extern void release_task(struct task_struct * p);
2391 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2392 extern int force_sigsegv(int, struct task_struct *);
2393 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2394 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2395 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2396 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2397 const struct cred *, u32);
2398 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2399 extern int kill_pid(struct pid *pid, int sig, int priv);
2400 extern int kill_proc_info(int, struct siginfo *, pid_t);
2401 extern __must_check bool do_notify_parent(struct task_struct *, int);
2402 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2403 extern void force_sig(int, struct task_struct *);
2404 extern int send_sig(int, struct task_struct *, int);
2405 extern int zap_other_threads(struct task_struct *p);
2406 extern struct sigqueue *sigqueue_alloc(void);
2407 extern void sigqueue_free(struct sigqueue *);
2408 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2409 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2411 static inline void restore_saved_sigmask(void)
2413 if (test_and_clear_restore_sigmask())
2414 __set_current_blocked(¤t->saved_sigmask);
2417 static inline sigset_t *sigmask_to_save(void)
2419 sigset_t *res = ¤t->blocked;
2420 if (unlikely(test_restore_sigmask()))
2421 res = ¤t->saved_sigmask;
2425 static inline int kill_cad_pid(int sig, int priv)
2427 return kill_pid(cad_pid, sig, priv);
2430 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2431 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2432 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2433 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2436 * True if we are on the alternate signal stack.
2438 static inline int on_sig_stack(unsigned long sp)
2440 #ifdef CONFIG_STACK_GROWSUP
2441 return sp >= current->sas_ss_sp &&
2442 sp - current->sas_ss_sp < current->sas_ss_size;
2444 return sp > current->sas_ss_sp &&
2445 sp - current->sas_ss_sp <= current->sas_ss_size;
2449 static inline int sas_ss_flags(unsigned long sp)
2451 if (!current->sas_ss_size)
2454 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2457 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2459 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2460 #ifdef CONFIG_STACK_GROWSUP
2461 return current->sas_ss_sp;
2463 return current->sas_ss_sp + current->sas_ss_size;
2469 * Routines for handling mm_structs
2471 extern struct mm_struct * mm_alloc(void);
2473 /* mmdrop drops the mm and the page tables */
2474 extern void __mmdrop(struct mm_struct *);
2475 static inline void mmdrop(struct mm_struct * mm)
2477 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2481 /* mmput gets rid of the mappings and all user-space */
2482 extern void mmput(struct mm_struct *);
2483 /* Grab a reference to a task's mm, if it is not already going away */
2484 extern struct mm_struct *get_task_mm(struct task_struct *task);
2486 * Grab a reference to a task's mm, if it is not already going away
2487 * and ptrace_may_access with the mode parameter passed to it
2490 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2491 /* Remove the current tasks stale references to the old mm_struct */
2492 extern void mm_release(struct task_struct *, struct mm_struct *);
2494 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2495 struct task_struct *);
2496 extern void flush_thread(void);
2497 extern void exit_thread(void);
2499 extern void exit_files(struct task_struct *);
2500 extern void __cleanup_sighand(struct sighand_struct *);
2502 extern void exit_itimers(struct signal_struct *);
2503 extern void flush_itimer_signals(void);
2505 extern void do_group_exit(int);
2507 extern int do_execve(struct filename *,
2508 const char __user * const __user *,
2509 const char __user * const __user *);
2510 extern int do_execveat(int, struct filename *,
2511 const char __user * const __user *,
2512 const char __user * const __user *,
2514 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2515 struct task_struct *fork_idle(int);
2516 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2518 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2519 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2521 __set_task_comm(tsk, from, false);
2523 extern char *get_task_comm(char *to, struct task_struct *tsk);
2526 void scheduler_ipi(void);
2527 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2529 static inline void scheduler_ipi(void) { }
2530 static inline unsigned long wait_task_inactive(struct task_struct *p,
2537 #define next_task(p) \
2538 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2540 #define for_each_process(p) \
2541 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2543 extern bool current_is_single_threaded(void);
2546 * Careful: do_each_thread/while_each_thread is a double loop so
2547 * 'break' will not work as expected - use goto instead.
2549 #define do_each_thread(g, t) \
2550 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2552 #define while_each_thread(g, t) \
2553 while ((t = next_thread(t)) != g)
2555 #define __for_each_thread(signal, t) \
2556 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2558 #define for_each_thread(p, t) \
2559 __for_each_thread((p)->signal, t)
2561 /* Careful: this is a double loop, 'break' won't work as expected. */
2562 #define for_each_process_thread(p, t) \
2563 for_each_process(p) for_each_thread(p, t)
2565 static inline int get_nr_threads(struct task_struct *tsk)
2567 return tsk->signal->nr_threads;
2570 static inline bool thread_group_leader(struct task_struct *p)
2572 return p->exit_signal >= 0;
2575 /* Do to the insanities of de_thread it is possible for a process
2576 * to have the pid of the thread group leader without actually being
2577 * the thread group leader. For iteration through the pids in proc
2578 * all we care about is that we have a task with the appropriate
2579 * pid, we don't actually care if we have the right task.
2581 static inline bool has_group_leader_pid(struct task_struct *p)
2583 return task_pid(p) == p->signal->leader_pid;
2587 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2589 return p1->signal == p2->signal;
2592 static inline struct task_struct *next_thread(const struct task_struct *p)
2594 return list_entry_rcu(p->thread_group.next,
2595 struct task_struct, thread_group);
2598 static inline int thread_group_empty(struct task_struct *p)
2600 return list_empty(&p->thread_group);
2603 #define delay_group_leader(p) \
2604 (thread_group_leader(p) && !thread_group_empty(p))
2607 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2608 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2609 * pins the final release of task.io_context. Also protects ->cpuset and
2610 * ->cgroup.subsys[]. And ->vfork_done.
2612 * Nests both inside and outside of read_lock(&tasklist_lock).
2613 * It must not be nested with write_lock_irq(&tasklist_lock),
2614 * neither inside nor outside.
2616 static inline void task_lock(struct task_struct *p)
2618 spin_lock(&p->alloc_lock);
2621 static inline void task_unlock(struct task_struct *p)
2623 spin_unlock(&p->alloc_lock);
2626 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2627 unsigned long *flags);
2629 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2630 unsigned long *flags)
2632 struct sighand_struct *ret;
2634 ret = __lock_task_sighand(tsk, flags);
2635 (void)__cond_lock(&tsk->sighand->siglock, ret);
2639 static inline void unlock_task_sighand(struct task_struct *tsk,
2640 unsigned long *flags)
2642 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2645 #ifdef CONFIG_CGROUPS
2646 static inline void threadgroup_change_begin(struct task_struct *tsk)
2648 down_read(&tsk->signal->group_rwsem);
2650 static inline void threadgroup_change_end(struct task_struct *tsk)
2652 up_read(&tsk->signal->group_rwsem);
2656 * threadgroup_lock - lock threadgroup
2657 * @tsk: member task of the threadgroup to lock
2659 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2660 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2661 * change ->group_leader/pid. This is useful for cases where the threadgroup
2662 * needs to stay stable across blockable operations.
2664 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2665 * synchronization. While held, no new task will be added to threadgroup
2666 * and no existing live task will have its PF_EXITING set.
2668 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2669 * sub-thread becomes a new leader.
2671 static inline void threadgroup_lock(struct task_struct *tsk)
2673 down_write(&tsk->signal->group_rwsem);
2677 * threadgroup_unlock - unlock threadgroup
2678 * @tsk: member task of the threadgroup to unlock
2680 * Reverse threadgroup_lock().
2682 static inline void threadgroup_unlock(struct task_struct *tsk)
2684 up_write(&tsk->signal->group_rwsem);
2687 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2688 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2689 static inline void threadgroup_lock(struct task_struct *tsk) {}
2690 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2693 #ifndef __HAVE_THREAD_FUNCTIONS
2695 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2696 #define task_stack_page(task) ((task)->stack)
2698 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2700 *task_thread_info(p) = *task_thread_info(org);
2701 task_thread_info(p)->task = p;
2705 * Return the address of the last usable long on the stack.
2707 * When the stack grows down, this is just above the thread
2708 * info struct. Going any lower will corrupt the threadinfo.
2710 * When the stack grows up, this is the highest address.
2711 * Beyond that position, we corrupt data on the next page.
2713 static inline unsigned long *end_of_stack(struct task_struct *p)
2715 #ifdef CONFIG_STACK_GROWSUP
2716 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2718 return (unsigned long *)(task_thread_info(p) + 1);
2723 #define task_stack_end_corrupted(task) \
2724 (*(end_of_stack(task)) != STACK_END_MAGIC)
2726 static inline int object_is_on_stack(void *obj)
2728 void *stack = task_stack_page(current);
2730 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2733 extern void thread_info_cache_init(void);
2735 #ifdef CONFIG_DEBUG_STACK_USAGE
2736 static inline unsigned long stack_not_used(struct task_struct *p)
2738 unsigned long *n = end_of_stack(p);
2740 do { /* Skip over canary */
2744 return (unsigned long)n - (unsigned long)end_of_stack(p);
2747 extern void set_task_stack_end_magic(struct task_struct *tsk);
2749 /* set thread flags in other task's structures
2750 * - see asm/thread_info.h for TIF_xxxx flags available
2752 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2754 set_ti_thread_flag(task_thread_info(tsk), flag);
2757 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2759 clear_ti_thread_flag(task_thread_info(tsk), flag);
2762 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2764 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2767 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2769 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2772 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2774 return test_ti_thread_flag(task_thread_info(tsk), flag);
2777 static inline void set_tsk_need_resched(struct task_struct *tsk)
2779 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2782 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2784 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2787 static inline int test_tsk_need_resched(struct task_struct *tsk)
2789 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2792 static inline int restart_syscall(void)
2794 set_tsk_thread_flag(current, TIF_SIGPENDING);
2795 return -ERESTARTNOINTR;
2798 static inline int signal_pending(struct task_struct *p)
2800 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2803 static inline int __fatal_signal_pending(struct task_struct *p)
2805 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2808 static inline int fatal_signal_pending(struct task_struct *p)
2810 return signal_pending(p) && __fatal_signal_pending(p);
2813 static inline int signal_pending_state(long state, struct task_struct *p)
2815 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2817 if (!signal_pending(p))
2820 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2824 * cond_resched() and cond_resched_lock(): latency reduction via
2825 * explicit rescheduling in places that are safe. The return
2826 * value indicates whether a reschedule was done in fact.
2827 * cond_resched_lock() will drop the spinlock before scheduling,
2828 * cond_resched_softirq() will enable bhs before scheduling.
2830 extern int _cond_resched(void);
2832 #define cond_resched() ({ \
2833 ___might_sleep(__FILE__, __LINE__, 0); \
2837 extern int __cond_resched_lock(spinlock_t *lock);
2839 #ifdef CONFIG_PREEMPT_COUNT
2840 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2842 #define PREEMPT_LOCK_OFFSET 0
2845 #define cond_resched_lock(lock) ({ \
2846 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2847 __cond_resched_lock(lock); \
2850 extern int __cond_resched_softirq(void);
2852 #define cond_resched_softirq() ({ \
2853 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2854 __cond_resched_softirq(); \
2857 static inline void cond_resched_rcu(void)
2859 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2867 * Does a critical section need to be broken due to another
2868 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2869 * but a general need for low latency)
2871 static inline int spin_needbreak(spinlock_t *lock)
2873 #ifdef CONFIG_PREEMPT
2874 return spin_is_contended(lock);
2881 * Idle thread specific functions to determine the need_resched
2884 #ifdef TIF_POLLING_NRFLAG
2885 static inline int tsk_is_polling(struct task_struct *p)
2887 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2890 static inline void __current_set_polling(void)
2892 set_thread_flag(TIF_POLLING_NRFLAG);
2895 static inline bool __must_check current_set_polling_and_test(void)
2897 __current_set_polling();
2900 * Polling state must be visible before we test NEED_RESCHED,
2901 * paired by resched_curr()
2903 smp_mb__after_atomic();
2905 return unlikely(tif_need_resched());
2908 static inline void __current_clr_polling(void)
2910 clear_thread_flag(TIF_POLLING_NRFLAG);
2913 static inline bool __must_check current_clr_polling_and_test(void)
2915 __current_clr_polling();
2918 * Polling state must be visible before we test NEED_RESCHED,
2919 * paired by resched_curr()
2921 smp_mb__after_atomic();
2923 return unlikely(tif_need_resched());
2927 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2928 static inline void __current_set_polling(void) { }
2929 static inline void __current_clr_polling(void) { }
2931 static inline bool __must_check current_set_polling_and_test(void)
2933 return unlikely(tif_need_resched());
2935 static inline bool __must_check current_clr_polling_and_test(void)
2937 return unlikely(tif_need_resched());
2941 static inline void current_clr_polling(void)
2943 __current_clr_polling();
2946 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2947 * Once the bit is cleared, we'll get IPIs with every new
2948 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2951 smp_mb(); /* paired with resched_curr() */
2953 preempt_fold_need_resched();
2956 static __always_inline bool need_resched(void)
2958 return unlikely(tif_need_resched());
2962 * Thread group CPU time accounting.
2964 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2965 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2967 static inline void thread_group_cputime_init(struct signal_struct *sig)
2969 raw_spin_lock_init(&sig->cputimer.lock);
2973 * Reevaluate whether the task has signals pending delivery.
2974 * Wake the task if so.
2975 * This is required every time the blocked sigset_t changes.
2976 * callers must hold sighand->siglock.
2978 extern void recalc_sigpending_and_wake(struct task_struct *t);
2979 extern void recalc_sigpending(void);
2981 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2983 static inline void signal_wake_up(struct task_struct *t, bool resume)
2985 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2987 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2989 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2993 * Wrappers for p->thread_info->cpu access. No-op on UP.
2997 static inline unsigned int task_cpu(const struct task_struct *p)
2999 return task_thread_info(p)->cpu;
3002 static inline int task_node(const struct task_struct *p)
3004 return cpu_to_node(task_cpu(p));
3007 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3011 static inline unsigned int task_cpu(const struct task_struct *p)
3016 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3020 #endif /* CONFIG_SMP */
3022 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3023 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3025 #ifdef CONFIG_CGROUP_SCHED
3026 extern struct task_group root_task_group;
3027 #endif /* CONFIG_CGROUP_SCHED */
3029 extern int task_can_switch_user(struct user_struct *up,
3030 struct task_struct *tsk);
3032 #ifdef CONFIG_TASK_XACCT
3033 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3035 tsk->ioac.rchar += amt;
3038 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3040 tsk->ioac.wchar += amt;
3043 static inline void inc_syscr(struct task_struct *tsk)
3048 static inline void inc_syscw(struct task_struct *tsk)
3053 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3057 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3061 static inline void inc_syscr(struct task_struct *tsk)
3065 static inline void inc_syscw(struct task_struct *tsk)
3070 #ifndef TASK_SIZE_OF
3071 #define TASK_SIZE_OF(tsk) TASK_SIZE
3075 extern void mm_update_next_owner(struct mm_struct *mm);
3077 static inline void mm_update_next_owner(struct mm_struct *mm)
3080 #endif /* CONFIG_MEMCG */
3082 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3085 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
3088 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3091 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
3094 static inline unsigned long rlimit(unsigned int limit)
3096 return task_rlimit(current, limit);
3099 static inline unsigned long rlimit_max(unsigned int limit)
3101 return task_rlimit_max(current, limit);