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.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/kcov.h>
55 #include <linux/task_io_accounting.h>
56 #include <linux/latencytop.h>
57 #include <linux/cred.h>
58 #include <linux/llist.h>
59 #include <linux/uidgid.h>
60 #include <linux/gfp.h>
61 #include <linux/magic.h>
62 #include <linux/cgroup-defs.h>
64 #include <asm/processor.h>
66 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
69 * Extended scheduling parameters data structure.
71 * This is needed because the original struct sched_param can not be
72 * altered without introducing ABI issues with legacy applications
73 * (e.g., in sched_getparam()).
75 * However, the possibility of specifying more than just a priority for
76 * the tasks may be useful for a wide variety of application fields, e.g.,
77 * multimedia, streaming, automation and control, and many others.
79 * This variant (sched_attr) is meant at describing a so-called
80 * sporadic time-constrained task. In such model a task is specified by:
81 * - the activation period or minimum instance inter-arrival time;
82 * - the maximum (or average, depending on the actual scheduling
83 * discipline) computation time of all instances, a.k.a. runtime;
84 * - the deadline (relative to the actual activation time) of each
86 * Very briefly, a periodic (sporadic) task asks for the execution of
87 * some specific computation --which is typically called an instance--
88 * (at most) every period. Moreover, each instance typically lasts no more
89 * than the runtime and must be completed by time instant t equal to
90 * the instance activation time + the deadline.
92 * This is reflected by the actual fields of the sched_attr structure:
94 * @size size of the structure, for fwd/bwd compat.
96 * @sched_policy task's scheduling policy
97 * @sched_flags for customizing the scheduler behaviour
98 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
99 * @sched_priority task's static priority (SCHED_FIFO/RR)
100 * @sched_deadline representative of the task's deadline
101 * @sched_runtime representative of the task's runtime
102 * @sched_period representative of the task's period
104 * Given this task model, there are a multiplicity of scheduling algorithms
105 * and policies, that can be used to ensure all the tasks will make their
106 * timing constraints.
108 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
109 * only user of this new interface. More information about the algorithm
110 * available in the scheduling class file or in Documentation/.
118 /* SCHED_NORMAL, SCHED_BATCH */
121 /* SCHED_FIFO, SCHED_RR */
130 struct futex_pi_state;
131 struct robust_list_head;
134 struct perf_event_context;
139 #define VMACACHE_BITS 2
140 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
141 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
144 * These are the constant used to fake the fixed-point load-average
145 * counting. Some notes:
146 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
147 * a load-average precision of 10 bits integer + 11 bits fractional
148 * - if you want to count load-averages more often, you need more
149 * precision, or rounding will get you. With 2-second counting freq,
150 * the EXP_n values would be 1981, 2034 and 2043 if still using only
153 extern unsigned long avenrun[]; /* Load averages */
154 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
156 #define FSHIFT 11 /* nr of bits of precision */
157 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
158 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
159 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
160 #define EXP_5 2014 /* 1/exp(5sec/5min) */
161 #define EXP_15 2037 /* 1/exp(5sec/15min) */
163 #define CALC_LOAD(load,exp,n) \
165 load += n*(FIXED_1-exp); \
168 extern unsigned long total_forks;
169 extern int nr_threads;
170 DECLARE_PER_CPU(unsigned long, process_counts);
171 extern int nr_processes(void);
172 extern unsigned long nr_running(void);
173 extern bool single_task_running(void);
174 extern unsigned long nr_iowait(void);
175 extern unsigned long nr_iowait_cpu(int cpu);
176 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
178 extern void calc_global_load(unsigned long ticks);
180 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
181 extern void cpu_load_update_nohz_start(void);
182 extern void cpu_load_update_nohz_stop(void);
184 static inline void cpu_load_update_nohz_start(void) { }
185 static inline void cpu_load_update_nohz_stop(void) { }
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);
199 * Task state bitmask. NOTE! These bits are also
200 * encoded in fs/proc/array.c: get_task_state().
202 * We have two separate sets of flags: task->state
203 * is about runnability, while task->exit_state are
204 * about the task exiting. Confusing, but this way
205 * modifying one set can't modify the other one by
208 #define TASK_RUNNING 0
209 #define TASK_INTERRUPTIBLE 1
210 #define TASK_UNINTERRUPTIBLE 2
211 #define __TASK_STOPPED 4
212 #define __TASK_TRACED 8
213 /* in tsk->exit_state */
215 #define EXIT_ZOMBIE 32
216 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
217 /* in tsk->state again */
219 #define TASK_WAKEKILL 128
220 #define TASK_WAKING 256
221 #define TASK_PARKED 512
222 #define TASK_NOLOAD 1024
223 #define TASK_STATE_MAX 2048
225 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
227 extern char ___assert_task_state[1 - 2*!!(
228 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
230 /* Convenience macros for the sake of set_task_state */
231 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
232 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
233 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
235 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
237 /* Convenience macros for the sake of wake_up */
238 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
239 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
241 /* get_task_state() */
242 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
243 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
244 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
246 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
247 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
248 #define task_is_stopped_or_traced(task) \
249 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
250 #define task_contributes_to_load(task) \
251 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
252 (task->flags & PF_FROZEN) == 0 && \
253 (task->state & TASK_NOLOAD) == 0)
255 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
257 #define __set_task_state(tsk, state_value) \
259 (tsk)->task_state_change = _THIS_IP_; \
260 (tsk)->state = (state_value); \
262 #define set_task_state(tsk, state_value) \
264 (tsk)->task_state_change = _THIS_IP_; \
265 smp_store_mb((tsk)->state, (state_value)); \
269 * set_current_state() includes a barrier so that the write of current->state
270 * is correctly serialised wrt the caller's subsequent test of whether to
273 * set_current_state(TASK_UNINTERRUPTIBLE);
274 * if (do_i_need_to_sleep())
277 * If the caller does not need such serialisation then use __set_current_state()
279 #define __set_current_state(state_value) \
281 current->task_state_change = _THIS_IP_; \
282 current->state = (state_value); \
284 #define set_current_state(state_value) \
286 current->task_state_change = _THIS_IP_; \
287 smp_store_mb(current->state, (state_value)); \
292 #define __set_task_state(tsk, state_value) \
293 do { (tsk)->state = (state_value); } while (0)
294 #define set_task_state(tsk, state_value) \
295 smp_store_mb((tsk)->state, (state_value))
298 * set_current_state() includes a barrier so that the write of current->state
299 * is correctly serialised wrt the caller's subsequent test of whether to
302 * set_current_state(TASK_UNINTERRUPTIBLE);
303 * if (do_i_need_to_sleep())
306 * If the caller does not need such serialisation then use __set_current_state()
308 #define __set_current_state(state_value) \
309 do { current->state = (state_value); } while (0)
310 #define set_current_state(state_value) \
311 smp_store_mb(current->state, (state_value))
315 /* Task command name length */
316 #define TASK_COMM_LEN 16
318 #include <linux/spinlock.h>
321 * This serializes "schedule()" and also protects
322 * the run-queue from deletions/modifications (but
323 * _adding_ to the beginning of the run-queue has
326 extern rwlock_t tasklist_lock;
327 extern spinlock_t mmlist_lock;
331 #ifdef CONFIG_PROVE_RCU
332 extern int lockdep_tasklist_lock_is_held(void);
333 #endif /* #ifdef CONFIG_PROVE_RCU */
335 extern void sched_init(void);
336 extern void sched_init_smp(void);
337 extern asmlinkage void schedule_tail(struct task_struct *prev);
338 extern void init_idle(struct task_struct *idle, int cpu);
339 extern void init_idle_bootup_task(struct task_struct *idle);
341 extern cpumask_var_t cpu_isolated_map;
343 extern int runqueue_is_locked(int cpu);
345 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
346 extern void nohz_balance_enter_idle(int cpu);
347 extern void set_cpu_sd_state_idle(void);
348 extern int get_nohz_timer_target(void);
350 static inline void nohz_balance_enter_idle(int cpu) { }
351 static inline void set_cpu_sd_state_idle(void) { }
355 * Only dump TASK_* tasks. (0 for all tasks)
357 extern void show_state_filter(unsigned long state_filter);
359 static inline void show_state(void)
361 show_state_filter(0);
364 extern void show_regs(struct pt_regs *);
367 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
368 * task), SP is the stack pointer of the first frame that should be shown in the back
369 * trace (or NULL if the entire call-chain of the task should be shown).
371 extern void show_stack(struct task_struct *task, unsigned long *sp);
373 extern void cpu_init (void);
374 extern void trap_init(void);
375 extern void update_process_times(int user);
376 extern void scheduler_tick(void);
378 extern void sched_show_task(struct task_struct *p);
380 #ifdef CONFIG_LOCKUP_DETECTOR
381 extern void touch_softlockup_watchdog_sched(void);
382 extern void touch_softlockup_watchdog(void);
383 extern void touch_softlockup_watchdog_sync(void);
384 extern void touch_all_softlockup_watchdogs(void);
385 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
387 size_t *lenp, loff_t *ppos);
388 extern unsigned int softlockup_panic;
389 extern unsigned int hardlockup_panic;
390 void lockup_detector_init(void);
392 static inline void touch_softlockup_watchdog_sched(void)
395 static inline void touch_softlockup_watchdog(void)
398 static inline void touch_softlockup_watchdog_sync(void)
401 static inline void touch_all_softlockup_watchdogs(void)
404 static inline void lockup_detector_init(void)
409 #ifdef CONFIG_DETECT_HUNG_TASK
410 void reset_hung_task_detector(void);
412 static inline void reset_hung_task_detector(void)
417 /* Attach to any functions which should be ignored in wchan output. */
418 #define __sched __attribute__((__section__(".sched.text")))
420 /* Linker adds these: start and end of __sched functions */
421 extern char __sched_text_start[], __sched_text_end[];
423 /* Is this address in the __sched functions? */
424 extern int in_sched_functions(unsigned long addr);
426 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
427 extern signed long schedule_timeout(signed long timeout);
428 extern signed long schedule_timeout_interruptible(signed long timeout);
429 extern signed long schedule_timeout_killable(signed long timeout);
430 extern signed long schedule_timeout_uninterruptible(signed long timeout);
431 extern signed long schedule_timeout_idle(signed long timeout);
432 asmlinkage void schedule(void);
433 extern void schedule_preempt_disabled(void);
435 extern long io_schedule_timeout(long timeout);
437 static inline void io_schedule(void)
439 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
443 struct user_namespace;
446 extern void arch_pick_mmap_layout(struct mm_struct *mm);
448 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
449 unsigned long, unsigned long);
451 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
452 unsigned long len, unsigned long pgoff,
453 unsigned long flags);
455 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
458 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
459 #define SUID_DUMP_USER 1 /* Dump as user of process */
460 #define SUID_DUMP_ROOT 2 /* Dump as root */
464 /* for SUID_DUMP_* above */
465 #define MMF_DUMPABLE_BITS 2
466 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
468 extern void set_dumpable(struct mm_struct *mm, int value);
470 * This returns the actual value of the suid_dumpable flag. For things
471 * that are using this for checking for privilege transitions, it must
472 * test against SUID_DUMP_USER rather than treating it as a boolean
475 static inline int __get_dumpable(unsigned long mm_flags)
477 return mm_flags & MMF_DUMPABLE_MASK;
480 static inline int get_dumpable(struct mm_struct *mm)
482 return __get_dumpable(mm->flags);
485 /* coredump filter bits */
486 #define MMF_DUMP_ANON_PRIVATE 2
487 #define MMF_DUMP_ANON_SHARED 3
488 #define MMF_DUMP_MAPPED_PRIVATE 4
489 #define MMF_DUMP_MAPPED_SHARED 5
490 #define MMF_DUMP_ELF_HEADERS 6
491 #define MMF_DUMP_HUGETLB_PRIVATE 7
492 #define MMF_DUMP_HUGETLB_SHARED 8
493 #define MMF_DUMP_DAX_PRIVATE 9
494 #define MMF_DUMP_DAX_SHARED 10
496 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
497 #define MMF_DUMP_FILTER_BITS 9
498 #define MMF_DUMP_FILTER_MASK \
499 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
500 #define MMF_DUMP_FILTER_DEFAULT \
501 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
502 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
504 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
505 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
507 # define MMF_DUMP_MASK_DEFAULT_ELF 0
509 /* leave room for more dump flags */
510 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
511 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
512 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
514 #define MMF_HAS_UPROBES 19 /* has uprobes */
515 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
517 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
519 struct sighand_struct {
521 struct k_sigaction action[_NSIG];
523 wait_queue_head_t signalfd_wqh;
526 struct pacct_struct {
529 unsigned long ac_mem;
530 cputime_t ac_utime, ac_stime;
531 unsigned long ac_minflt, ac_majflt;
542 * struct prev_cputime - snaphsot of system and user cputime
543 * @utime: time spent in user mode
544 * @stime: time spent in system mode
545 * @lock: protects the above two fields
547 * Stores previous user/system time values such that we can guarantee
550 struct prev_cputime {
551 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
558 static inline void prev_cputime_init(struct prev_cputime *prev)
560 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
561 prev->utime = prev->stime = 0;
562 raw_spin_lock_init(&prev->lock);
567 * struct task_cputime - collected CPU time counts
568 * @utime: time spent in user mode, in &cputime_t units
569 * @stime: time spent in kernel mode, in &cputime_t units
570 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
572 * This structure groups together three kinds of CPU time that are tracked for
573 * threads and thread groups. Most things considering CPU time want to group
574 * these counts together and treat all three of them in parallel.
576 struct task_cputime {
579 unsigned long long sum_exec_runtime;
582 /* Alternate field names when used to cache expirations. */
583 #define virt_exp utime
584 #define prof_exp stime
585 #define sched_exp sum_exec_runtime
587 #define INIT_CPUTIME \
588 (struct task_cputime) { \
591 .sum_exec_runtime = 0, \
595 * This is the atomic variant of task_cputime, which can be used for
596 * storing and updating task_cputime statistics without locking.
598 struct task_cputime_atomic {
601 atomic64_t sum_exec_runtime;
604 #define INIT_CPUTIME_ATOMIC \
605 (struct task_cputime_atomic) { \
606 .utime = ATOMIC64_INIT(0), \
607 .stime = ATOMIC64_INIT(0), \
608 .sum_exec_runtime = ATOMIC64_INIT(0), \
611 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
614 * Disable preemption until the scheduler is running -- use an unconditional
615 * value so that it also works on !PREEMPT_COUNT kernels.
617 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
619 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
622 * Initial preempt_count value; reflects the preempt_count schedule invariant
623 * which states that during context switches:
625 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
627 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
628 * Note: See finish_task_switch().
630 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
633 * struct thread_group_cputimer - thread group interval timer counts
634 * @cputime_atomic: atomic thread group interval timers.
635 * @running: true when there are timers running and
636 * @cputime_atomic receives updates.
637 * @checking_timer: true when a thread in the group is in the
638 * process of checking for thread group timers.
640 * This structure contains the version of task_cputime, above, that is
641 * used for thread group CPU timer calculations.
643 struct thread_group_cputimer {
644 struct task_cputime_atomic cputime_atomic;
649 #include <linux/rwsem.h>
653 * NOTE! "signal_struct" does not have its own
654 * locking, because a shared signal_struct always
655 * implies a shared sighand_struct, so locking
656 * sighand_struct is always a proper superset of
657 * the locking of signal_struct.
659 struct signal_struct {
663 struct list_head thread_head;
665 wait_queue_head_t wait_chldexit; /* for wait4() */
667 /* current thread group signal load-balancing target: */
668 struct task_struct *curr_target;
670 /* shared signal handling: */
671 struct sigpending shared_pending;
673 /* thread group exit support */
676 * - notify group_exit_task when ->count is equal to notify_count
677 * - everyone except group_exit_task is stopped during signal delivery
678 * of fatal signals, group_exit_task processes the signal.
681 struct task_struct *group_exit_task;
683 /* thread group stop support, overloads group_exit_code too */
684 int group_stop_count;
685 unsigned int flags; /* see SIGNAL_* flags below */
688 * PR_SET_CHILD_SUBREAPER marks a process, like a service
689 * manager, to re-parent orphan (double-forking) child processes
690 * to this process instead of 'init'. The service manager is
691 * able to receive SIGCHLD signals and is able to investigate
692 * the process until it calls wait(). All children of this
693 * process will inherit a flag if they should look for a
694 * child_subreaper process at exit.
696 unsigned int is_child_subreaper:1;
697 unsigned int has_child_subreaper:1;
699 /* POSIX.1b Interval Timers */
701 struct list_head posix_timers;
703 /* ITIMER_REAL timer for the process */
704 struct hrtimer real_timer;
705 struct pid *leader_pid;
706 ktime_t it_real_incr;
709 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
710 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
711 * values are defined to 0 and 1 respectively
713 struct cpu_itimer it[2];
716 * Thread group totals for process CPU timers.
717 * See thread_group_cputimer(), et al, for details.
719 struct thread_group_cputimer cputimer;
721 /* Earliest-expiration cache. */
722 struct task_cputime cputime_expires;
724 #ifdef CONFIG_NO_HZ_FULL
725 atomic_t tick_dep_mask;
728 struct list_head cpu_timers[3];
730 struct pid *tty_old_pgrp;
732 /* boolean value for session group leader */
735 struct tty_struct *tty; /* NULL if no tty */
737 #ifdef CONFIG_SCHED_AUTOGROUP
738 struct autogroup *autogroup;
741 * Cumulative resource counters for dead threads in the group,
742 * and for reaped dead child processes forked by this group.
743 * Live threads maintain their own counters and add to these
744 * in __exit_signal, except for the group leader.
746 seqlock_t stats_lock;
747 cputime_t utime, stime, cutime, cstime;
750 struct prev_cputime prev_cputime;
751 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
752 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
753 unsigned long inblock, oublock, cinblock, coublock;
754 unsigned long maxrss, cmaxrss;
755 struct task_io_accounting ioac;
758 * Cumulative ns of schedule CPU time fo dead threads in the
759 * group, not including a zombie group leader, (This only differs
760 * from jiffies_to_ns(utime + stime) if sched_clock uses something
761 * other than jiffies.)
763 unsigned long long sum_sched_runtime;
766 * We don't bother to synchronize most readers of this at all,
767 * because there is no reader checking a limit that actually needs
768 * to get both rlim_cur and rlim_max atomically, and either one
769 * alone is a single word that can safely be read normally.
770 * getrlimit/setrlimit use task_lock(current->group_leader) to
771 * protect this instead of the siglock, because they really
772 * have no need to disable irqs.
774 struct rlimit rlim[RLIM_NLIMITS];
776 #ifdef CONFIG_BSD_PROCESS_ACCT
777 struct pacct_struct pacct; /* per-process accounting information */
779 #ifdef CONFIG_TASKSTATS
780 struct taskstats *stats;
784 struct tty_audit_buf *tty_audit_buf;
787 oom_flags_t oom_flags;
788 short oom_score_adj; /* OOM kill score adjustment */
789 short oom_score_adj_min; /* OOM kill score adjustment min value.
790 * Only settable by CAP_SYS_RESOURCE. */
792 struct mutex cred_guard_mutex; /* guard against foreign influences on
793 * credential calculations
794 * (notably. ptrace) */
798 * Bits in flags field of signal_struct.
800 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
801 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
802 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
803 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
805 * Pending notifications to parent.
807 #define SIGNAL_CLD_STOPPED 0x00000010
808 #define SIGNAL_CLD_CONTINUED 0x00000020
809 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
811 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
813 /* If true, all threads except ->group_exit_task have pending SIGKILL */
814 static inline int signal_group_exit(const struct signal_struct *sig)
816 return (sig->flags & SIGNAL_GROUP_EXIT) ||
817 (sig->group_exit_task != NULL);
821 * Some day this will be a full-fledged user tracking system..
824 atomic_t __count; /* reference count */
825 atomic_t processes; /* How many processes does this user have? */
826 atomic_t sigpending; /* How many pending signals does this user have? */
827 #ifdef CONFIG_INOTIFY_USER
828 atomic_t inotify_watches; /* How many inotify watches does this user have? */
829 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
831 #ifdef CONFIG_FANOTIFY
832 atomic_t fanotify_listeners;
835 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
837 #ifdef CONFIG_POSIX_MQUEUE
838 /* protected by mq_lock */
839 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
841 unsigned long locked_shm; /* How many pages of mlocked shm ? */
842 unsigned long unix_inflight; /* How many files in flight in unix sockets */
843 atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */
846 struct key *uid_keyring; /* UID specific keyring */
847 struct key *session_keyring; /* UID's default session keyring */
850 /* Hash table maintenance information */
851 struct hlist_node uidhash_node;
854 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
855 atomic_long_t locked_vm;
859 extern int uids_sysfs_init(void);
861 extern struct user_struct *find_user(kuid_t);
863 extern struct user_struct root_user;
864 #define INIT_USER (&root_user)
867 struct backing_dev_info;
868 struct reclaim_state;
870 #ifdef CONFIG_SCHED_INFO
872 /* cumulative counters */
873 unsigned long pcount; /* # of times run on this cpu */
874 unsigned long long run_delay; /* time spent waiting on a runqueue */
877 unsigned long long last_arrival,/* when we last ran on a cpu */
878 last_queued; /* when we were last queued to run */
880 #endif /* CONFIG_SCHED_INFO */
882 #ifdef CONFIG_TASK_DELAY_ACCT
883 struct task_delay_info {
885 unsigned int flags; /* Private per-task flags */
887 /* For each stat XXX, add following, aligned appropriately
889 * struct timespec XXX_start, XXX_end;
893 * Atomicity of updates to XXX_delay, XXX_count protected by
894 * single lock above (split into XXX_lock if contention is an issue).
898 * XXX_count is incremented on every XXX operation, the delay
899 * associated with the operation is added to XXX_delay.
900 * XXX_delay contains the accumulated delay time in nanoseconds.
902 u64 blkio_start; /* Shared by blkio, swapin */
903 u64 blkio_delay; /* wait for sync block io completion */
904 u64 swapin_delay; /* wait for swapin block io completion */
905 u32 blkio_count; /* total count of the number of sync block */
906 /* io operations performed */
907 u32 swapin_count; /* total count of the number of swapin block */
908 /* io operations performed */
911 u64 freepages_delay; /* wait for memory reclaim */
912 u32 freepages_count; /* total count of memory reclaim */
914 #endif /* CONFIG_TASK_DELAY_ACCT */
916 static inline int sched_info_on(void)
918 #ifdef CONFIG_SCHEDSTATS
920 #elif defined(CONFIG_TASK_DELAY_ACCT)
921 extern int delayacct_on;
928 #ifdef CONFIG_SCHEDSTATS
929 void force_schedstat_enabled(void);
940 * Increase resolution of cpu_capacity calculations
942 #define SCHED_CAPACITY_SHIFT 10
943 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
946 * Wake-queues are lists of tasks with a pending wakeup, whose
947 * callers have already marked the task as woken internally,
948 * and can thus carry on. A common use case is being able to
949 * do the wakeups once the corresponding user lock as been
952 * We hold reference to each task in the list across the wakeup,
953 * thus guaranteeing that the memory is still valid by the time
954 * the actual wakeups are performed in wake_up_q().
956 * One per task suffices, because there's never a need for a task to be
957 * in two wake queues simultaneously; it is forbidden to abandon a task
958 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
959 * already in a wake queue, the wakeup will happen soon and the second
960 * waker can just skip it.
962 * The WAKE_Q macro declares and initializes the list head.
963 * wake_up_q() does NOT reinitialize the list; it's expected to be
964 * called near the end of a function, where the fact that the queue is
965 * not used again will be easy to see by inspection.
967 * Note that this can cause spurious wakeups. schedule() callers
968 * must ensure the call is done inside a loop, confirming that the
969 * wakeup condition has in fact occurred.
972 struct wake_q_node *next;
976 struct wake_q_node *first;
977 struct wake_q_node **lastp;
980 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
982 #define WAKE_Q(name) \
983 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
985 extern void wake_q_add(struct wake_q_head *head,
986 struct task_struct *task);
987 extern void wake_up_q(struct wake_q_head *head);
990 * sched-domains (multiprocessor balancing) declarations:
993 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
994 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
995 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
996 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
997 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
998 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
999 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
1000 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
1001 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
1002 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1003 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1004 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1005 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1006 #define SD_NUMA 0x4000 /* cross-node balancing */
1008 #ifdef CONFIG_SCHED_SMT
1009 static inline int cpu_smt_flags(void)
1011 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1015 #ifdef CONFIG_SCHED_MC
1016 static inline int cpu_core_flags(void)
1018 return SD_SHARE_PKG_RESOURCES;
1023 static inline int cpu_numa_flags(void)
1029 struct sched_domain_attr {
1030 int relax_domain_level;
1033 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1034 .relax_domain_level = -1, \
1037 extern int sched_domain_level_max;
1041 struct sched_domain {
1042 /* These fields must be setup */
1043 struct sched_domain *parent; /* top domain must be null terminated */
1044 struct sched_domain *child; /* bottom domain must be null terminated */
1045 struct sched_group *groups; /* the balancing groups of the domain */
1046 unsigned long min_interval; /* Minimum balance interval ms */
1047 unsigned long max_interval; /* Maximum balance interval ms */
1048 unsigned int busy_factor; /* less balancing by factor if busy */
1049 unsigned int imbalance_pct; /* No balance until over watermark */
1050 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1051 unsigned int busy_idx;
1052 unsigned int idle_idx;
1053 unsigned int newidle_idx;
1054 unsigned int wake_idx;
1055 unsigned int forkexec_idx;
1056 unsigned int smt_gain;
1058 int nohz_idle; /* NOHZ IDLE status */
1059 int flags; /* See SD_* */
1062 /* Runtime fields. */
1063 unsigned long last_balance; /* init to jiffies. units in jiffies */
1064 unsigned int balance_interval; /* initialise to 1. units in ms. */
1065 unsigned int nr_balance_failed; /* initialise to 0 */
1067 /* idle_balance() stats */
1068 u64 max_newidle_lb_cost;
1069 unsigned long next_decay_max_lb_cost;
1071 #ifdef CONFIG_SCHEDSTATS
1072 /* load_balance() stats */
1073 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1074 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1075 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1076 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1077 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1078 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1079 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1080 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1082 /* Active load balancing */
1083 unsigned int alb_count;
1084 unsigned int alb_failed;
1085 unsigned int alb_pushed;
1087 /* SD_BALANCE_EXEC stats */
1088 unsigned int sbe_count;
1089 unsigned int sbe_balanced;
1090 unsigned int sbe_pushed;
1092 /* SD_BALANCE_FORK stats */
1093 unsigned int sbf_count;
1094 unsigned int sbf_balanced;
1095 unsigned int sbf_pushed;
1097 /* try_to_wake_up() stats */
1098 unsigned int ttwu_wake_remote;
1099 unsigned int ttwu_move_affine;
1100 unsigned int ttwu_move_balance;
1102 #ifdef CONFIG_SCHED_DEBUG
1106 void *private; /* used during construction */
1107 struct rcu_head rcu; /* used during destruction */
1110 unsigned int span_weight;
1112 * Span of all CPUs in this domain.
1114 * NOTE: this field is variable length. (Allocated dynamically
1115 * by attaching extra space to the end of the structure,
1116 * depending on how many CPUs the kernel has booted up with)
1118 unsigned long span[0];
1121 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1123 return to_cpumask(sd->span);
1126 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1127 struct sched_domain_attr *dattr_new);
1129 /* Allocate an array of sched domains, for partition_sched_domains(). */
1130 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1131 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1133 bool cpus_share_cache(int this_cpu, int that_cpu);
1135 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1136 typedef int (*sched_domain_flags_f)(void);
1138 #define SDTL_OVERLAP 0x01
1141 struct sched_domain **__percpu sd;
1142 struct sched_group **__percpu sg;
1143 struct sched_group_capacity **__percpu sgc;
1146 struct sched_domain_topology_level {
1147 sched_domain_mask_f mask;
1148 sched_domain_flags_f sd_flags;
1151 struct sd_data data;
1152 #ifdef CONFIG_SCHED_DEBUG
1157 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1158 extern void wake_up_if_idle(int cpu);
1160 #ifdef CONFIG_SCHED_DEBUG
1161 # define SD_INIT_NAME(type) .name = #type
1163 # define SD_INIT_NAME(type)
1166 #else /* CONFIG_SMP */
1168 struct sched_domain_attr;
1171 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1172 struct sched_domain_attr *dattr_new)
1176 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1181 #endif /* !CONFIG_SMP */
1184 struct io_context; /* See blkdev.h */
1187 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1188 extern void prefetch_stack(struct task_struct *t);
1190 static inline void prefetch_stack(struct task_struct *t) { }
1193 struct audit_context; /* See audit.c */
1195 struct pipe_inode_info;
1196 struct uts_namespace;
1198 struct load_weight {
1199 unsigned long weight;
1204 * The load_avg/util_avg accumulates an infinite geometric series.
1205 * 1) load_avg factors frequency scaling into the amount of time that a
1206 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1207 * aggregated such weights of all runnable and blocked sched_entities.
1208 * 2) util_avg factors frequency and cpu scaling into the amount of time
1209 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1210 * For cfs_rq, it is the aggregated such times of all runnable and
1211 * blocked sched_entities.
1212 * The 64 bit load_sum can:
1213 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1214 * the highest weight (=88761) always runnable, we should not overflow
1215 * 2) for entity, support any load.weight always runnable
1218 u64 last_update_time, load_sum;
1219 u32 util_sum, period_contrib;
1220 unsigned long load_avg, util_avg;
1223 #ifdef CONFIG_SCHEDSTATS
1224 struct sched_statistics {
1234 s64 sum_sleep_runtime;
1241 u64 nr_migrations_cold;
1242 u64 nr_failed_migrations_affine;
1243 u64 nr_failed_migrations_running;
1244 u64 nr_failed_migrations_hot;
1245 u64 nr_forced_migrations;
1248 u64 nr_wakeups_sync;
1249 u64 nr_wakeups_migrate;
1250 u64 nr_wakeups_local;
1251 u64 nr_wakeups_remote;
1252 u64 nr_wakeups_affine;
1253 u64 nr_wakeups_affine_attempts;
1254 u64 nr_wakeups_passive;
1255 u64 nr_wakeups_idle;
1259 struct sched_entity {
1260 struct load_weight load; /* for load-balancing */
1261 struct rb_node run_node;
1262 struct list_head group_node;
1266 u64 sum_exec_runtime;
1268 u64 prev_sum_exec_runtime;
1272 #ifdef CONFIG_SCHEDSTATS
1273 struct sched_statistics statistics;
1276 #ifdef CONFIG_FAIR_GROUP_SCHED
1278 struct sched_entity *parent;
1279 /* rq on which this entity is (to be) queued: */
1280 struct cfs_rq *cfs_rq;
1281 /* rq "owned" by this entity/group: */
1282 struct cfs_rq *my_q;
1287 * Per entity load average tracking.
1289 * Put into separate cache line so it does not
1290 * collide with read-mostly values above.
1292 struct sched_avg avg ____cacheline_aligned_in_smp;
1296 struct sched_rt_entity {
1297 struct list_head run_list;
1298 unsigned long timeout;
1299 unsigned long watchdog_stamp;
1300 unsigned int time_slice;
1301 unsigned short on_rq;
1302 unsigned short on_list;
1304 struct sched_rt_entity *back;
1305 #ifdef CONFIG_RT_GROUP_SCHED
1306 struct sched_rt_entity *parent;
1307 /* rq on which this entity is (to be) queued: */
1308 struct rt_rq *rt_rq;
1309 /* rq "owned" by this entity/group: */
1314 struct sched_dl_entity {
1315 struct rb_node rb_node;
1318 * Original scheduling parameters. Copied here from sched_attr
1319 * during sched_setattr(), they will remain the same until
1320 * the next sched_setattr().
1322 u64 dl_runtime; /* maximum runtime for each instance */
1323 u64 dl_deadline; /* relative deadline of each instance */
1324 u64 dl_period; /* separation of two instances (period) */
1325 u64 dl_bw; /* dl_runtime / dl_deadline */
1328 * Actual scheduling parameters. Initialized with the values above,
1329 * they are continously updated during task execution. Note that
1330 * the remaining runtime could be < 0 in case we are in overrun.
1332 s64 runtime; /* remaining runtime for this instance */
1333 u64 deadline; /* absolute deadline for this instance */
1334 unsigned int flags; /* specifying the scheduler behaviour */
1339 * @dl_throttled tells if we exhausted the runtime. If so, the
1340 * task has to wait for a replenishment to be performed at the
1341 * next firing of dl_timer.
1343 * @dl_boosted tells if we are boosted due to DI. If so we are
1344 * outside bandwidth enforcement mechanism (but only until we
1345 * exit the critical section);
1347 * @dl_yielded tells if task gave up the cpu before consuming
1348 * all its available runtime during the last job.
1350 int dl_throttled, dl_boosted, dl_yielded;
1353 * Bandwidth enforcement timer. Each -deadline task has its
1354 * own bandwidth to be enforced, thus we need one timer per task.
1356 struct hrtimer dl_timer;
1364 u8 pad; /* Otherwise the compiler can store garbage here. */
1366 u32 s; /* Set of bits. */
1370 enum perf_event_task_context {
1371 perf_invalid_context = -1,
1372 perf_hw_context = 0,
1374 perf_nr_task_contexts,
1377 /* Track pages that require TLB flushes */
1378 struct tlbflush_unmap_batch {
1380 * Each bit set is a CPU that potentially has a TLB entry for one of
1381 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1383 struct cpumask cpumask;
1385 /* True if any bit in cpumask is set */
1386 bool flush_required;
1389 * If true then the PTE was dirty when unmapped. The entry must be
1390 * flushed before IO is initiated or a stale TLB entry potentially
1391 * allows an update without redirtying the page.
1396 struct task_struct {
1397 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1400 unsigned int flags; /* per process flags, defined below */
1401 unsigned int ptrace;
1404 struct llist_node wake_entry;
1406 unsigned int wakee_flips;
1407 unsigned long wakee_flip_decay_ts;
1408 struct task_struct *last_wakee;
1414 int prio, static_prio, normal_prio;
1415 unsigned int rt_priority;
1416 const struct sched_class *sched_class;
1417 struct sched_entity se;
1418 struct sched_rt_entity rt;
1419 #ifdef CONFIG_CGROUP_SCHED
1420 struct task_group *sched_task_group;
1422 struct sched_dl_entity dl;
1424 #ifdef CONFIG_PREEMPT_NOTIFIERS
1425 /* list of struct preempt_notifier: */
1426 struct hlist_head preempt_notifiers;
1429 #ifdef CONFIG_BLK_DEV_IO_TRACE
1430 unsigned int btrace_seq;
1433 unsigned int policy;
1434 int nr_cpus_allowed;
1435 cpumask_t cpus_allowed;
1437 #ifdef CONFIG_PREEMPT_RCU
1438 int rcu_read_lock_nesting;
1439 union rcu_special rcu_read_unlock_special;
1440 struct list_head rcu_node_entry;
1441 struct rcu_node *rcu_blocked_node;
1442 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1443 #ifdef CONFIG_TASKS_RCU
1444 unsigned long rcu_tasks_nvcsw;
1445 bool rcu_tasks_holdout;
1446 struct list_head rcu_tasks_holdout_list;
1447 int rcu_tasks_idle_cpu;
1448 #endif /* #ifdef CONFIG_TASKS_RCU */
1450 #ifdef CONFIG_SCHED_INFO
1451 struct sched_info sched_info;
1454 struct list_head tasks;
1456 struct plist_node pushable_tasks;
1457 struct rb_node pushable_dl_tasks;
1460 struct mm_struct *mm, *active_mm;
1461 /* per-thread vma caching */
1462 u32 vmacache_seqnum;
1463 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1464 #if defined(SPLIT_RSS_COUNTING)
1465 struct task_rss_stat rss_stat;
1469 int exit_code, exit_signal;
1470 int pdeath_signal; /* The signal sent when the parent dies */
1471 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1473 /* Used for emulating ABI behavior of previous Linux versions */
1474 unsigned int personality;
1476 /* scheduler bits, serialized by scheduler locks */
1477 unsigned sched_reset_on_fork:1;
1478 unsigned sched_contributes_to_load:1;
1479 unsigned sched_migrated:1;
1480 unsigned :0; /* force alignment to the next boundary */
1482 /* unserialized, strictly 'current' */
1483 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1484 unsigned in_iowait:1;
1486 unsigned memcg_may_oom:1;
1488 unsigned memcg_kmem_skip_account:1;
1491 #ifdef CONFIG_COMPAT_BRK
1492 unsigned brk_randomized:1;
1495 unsigned long atomic_flags; /* Flags needing atomic access. */
1497 struct restart_block restart_block;
1502 #ifdef CONFIG_CC_STACKPROTECTOR
1503 /* Canary value for the -fstack-protector gcc feature */
1504 unsigned long stack_canary;
1507 * pointers to (original) parent process, youngest child, younger sibling,
1508 * older sibling, respectively. (p->father can be replaced with
1509 * p->real_parent->pid)
1511 struct task_struct __rcu *real_parent; /* real parent process */
1512 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1514 * children/sibling forms the list of my natural children
1516 struct list_head children; /* list of my children */
1517 struct list_head sibling; /* linkage in my parent's children list */
1518 struct task_struct *group_leader; /* threadgroup leader */
1521 * ptraced is the list of tasks this task is using ptrace on.
1522 * This includes both natural children and PTRACE_ATTACH targets.
1523 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1525 struct list_head ptraced;
1526 struct list_head ptrace_entry;
1528 /* PID/PID hash table linkage. */
1529 struct pid_link pids[PIDTYPE_MAX];
1530 struct list_head thread_group;
1531 struct list_head thread_node;
1533 struct completion *vfork_done; /* for vfork() */
1534 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1535 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1537 cputime_t utime, stime, utimescaled, stimescaled;
1539 struct prev_cputime prev_cputime;
1540 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1541 seqcount_t vtime_seqcount;
1542 unsigned long long vtime_snap;
1544 /* Task is sleeping or running in a CPU with VTIME inactive */
1546 /* Task runs in userspace in a CPU with VTIME active */
1548 /* Task runs in kernelspace in a CPU with VTIME active */
1550 } vtime_snap_whence;
1553 #ifdef CONFIG_NO_HZ_FULL
1554 atomic_t tick_dep_mask;
1556 unsigned long nvcsw, nivcsw; /* context switch counts */
1557 u64 start_time; /* monotonic time in nsec */
1558 u64 real_start_time; /* boot based time in nsec */
1559 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1560 unsigned long min_flt, maj_flt;
1562 struct task_cputime cputime_expires;
1563 struct list_head cpu_timers[3];
1565 /* process credentials */
1566 const struct cred __rcu *real_cred; /* objective and real subjective task
1567 * credentials (COW) */
1568 const struct cred __rcu *cred; /* effective (overridable) subjective task
1569 * credentials (COW) */
1570 char comm[TASK_COMM_LEN]; /* executable name excluding path
1571 - access with [gs]et_task_comm (which lock
1572 it with task_lock())
1573 - initialized normally by setup_new_exec */
1574 /* file system info */
1575 struct nameidata *nameidata;
1576 #ifdef CONFIG_SYSVIPC
1578 struct sysv_sem sysvsem;
1579 struct sysv_shm sysvshm;
1581 #ifdef CONFIG_DETECT_HUNG_TASK
1582 /* hung task detection */
1583 unsigned long last_switch_count;
1585 /* filesystem information */
1586 struct fs_struct *fs;
1587 /* open file information */
1588 struct files_struct *files;
1590 struct nsproxy *nsproxy;
1591 /* signal handlers */
1592 struct signal_struct *signal;
1593 struct sighand_struct *sighand;
1595 sigset_t blocked, real_blocked;
1596 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1597 struct sigpending pending;
1599 unsigned long sas_ss_sp;
1602 struct callback_head *task_works;
1604 struct audit_context *audit_context;
1605 #ifdef CONFIG_AUDITSYSCALL
1607 unsigned int sessionid;
1609 struct seccomp seccomp;
1611 /* Thread group tracking */
1614 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1616 spinlock_t alloc_lock;
1618 /* Protection of the PI data structures: */
1619 raw_spinlock_t pi_lock;
1621 struct wake_q_node wake_q;
1623 #ifdef CONFIG_RT_MUTEXES
1624 /* PI waiters blocked on a rt_mutex held by this task */
1625 struct rb_root pi_waiters;
1626 struct rb_node *pi_waiters_leftmost;
1627 /* Deadlock detection and priority inheritance handling */
1628 struct rt_mutex_waiter *pi_blocked_on;
1631 #ifdef CONFIG_DEBUG_MUTEXES
1632 /* mutex deadlock detection */
1633 struct mutex_waiter *blocked_on;
1635 #ifdef CONFIG_TRACE_IRQFLAGS
1636 unsigned int irq_events;
1637 unsigned long hardirq_enable_ip;
1638 unsigned long hardirq_disable_ip;
1639 unsigned int hardirq_enable_event;
1640 unsigned int hardirq_disable_event;
1641 int hardirqs_enabled;
1642 int hardirq_context;
1643 unsigned long softirq_disable_ip;
1644 unsigned long softirq_enable_ip;
1645 unsigned int softirq_disable_event;
1646 unsigned int softirq_enable_event;
1647 int softirqs_enabled;
1648 int softirq_context;
1650 #ifdef CONFIG_LOCKDEP
1651 # define MAX_LOCK_DEPTH 48UL
1654 unsigned int lockdep_recursion;
1655 struct held_lock held_locks[MAX_LOCK_DEPTH];
1656 gfp_t lockdep_reclaim_gfp;
1659 unsigned int in_ubsan;
1662 /* journalling filesystem info */
1665 /* stacked block device info */
1666 struct bio_list *bio_list;
1669 /* stack plugging */
1670 struct blk_plug *plug;
1674 struct reclaim_state *reclaim_state;
1676 struct backing_dev_info *backing_dev_info;
1678 struct io_context *io_context;
1680 unsigned long ptrace_message;
1681 siginfo_t *last_siginfo; /* For ptrace use. */
1682 struct task_io_accounting ioac;
1683 #if defined(CONFIG_TASK_XACCT)
1684 u64 acct_rss_mem1; /* accumulated rss usage */
1685 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1686 cputime_t acct_timexpd; /* stime + utime since last update */
1688 #ifdef CONFIG_CPUSETS
1689 nodemask_t mems_allowed; /* Protected by alloc_lock */
1690 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1691 int cpuset_mem_spread_rotor;
1692 int cpuset_slab_spread_rotor;
1694 #ifdef CONFIG_CGROUPS
1695 /* Control Group info protected by css_set_lock */
1696 struct css_set __rcu *cgroups;
1697 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1698 struct list_head cg_list;
1701 struct robust_list_head __user *robust_list;
1702 #ifdef CONFIG_COMPAT
1703 struct compat_robust_list_head __user *compat_robust_list;
1705 struct list_head pi_state_list;
1706 struct futex_pi_state *pi_state_cache;
1708 #ifdef CONFIG_PERF_EVENTS
1709 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1710 struct mutex perf_event_mutex;
1711 struct list_head perf_event_list;
1713 #ifdef CONFIG_DEBUG_PREEMPT
1714 unsigned long preempt_disable_ip;
1717 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1719 short pref_node_fork;
1721 #ifdef CONFIG_NUMA_BALANCING
1723 unsigned int numa_scan_period;
1724 unsigned int numa_scan_period_max;
1725 int numa_preferred_nid;
1726 unsigned long numa_migrate_retry;
1727 u64 node_stamp; /* migration stamp */
1728 u64 last_task_numa_placement;
1729 u64 last_sum_exec_runtime;
1730 struct callback_head numa_work;
1732 struct list_head numa_entry;
1733 struct numa_group *numa_group;
1736 * numa_faults is an array split into four regions:
1737 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1738 * in this precise order.
1740 * faults_memory: Exponential decaying average of faults on a per-node
1741 * basis. Scheduling placement decisions are made based on these
1742 * counts. The values remain static for the duration of a PTE scan.
1743 * faults_cpu: Track the nodes the process was running on when a NUMA
1744 * hinting fault was incurred.
1745 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1746 * during the current scan window. When the scan completes, the counts
1747 * in faults_memory and faults_cpu decay and these values are copied.
1749 unsigned long *numa_faults;
1750 unsigned long total_numa_faults;
1753 * numa_faults_locality tracks if faults recorded during the last
1754 * scan window were remote/local or failed to migrate. The task scan
1755 * period is adapted based on the locality of the faults with different
1756 * weights depending on whether they were shared or private faults
1758 unsigned long numa_faults_locality[3];
1760 unsigned long numa_pages_migrated;
1761 #endif /* CONFIG_NUMA_BALANCING */
1763 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1764 struct tlbflush_unmap_batch tlb_ubc;
1767 struct rcu_head rcu;
1770 * cache last used pipe for splice
1772 struct pipe_inode_info *splice_pipe;
1774 struct page_frag task_frag;
1776 #ifdef CONFIG_TASK_DELAY_ACCT
1777 struct task_delay_info *delays;
1779 #ifdef CONFIG_FAULT_INJECTION
1783 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1784 * balance_dirty_pages() for some dirty throttling pause
1787 int nr_dirtied_pause;
1788 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1790 #ifdef CONFIG_LATENCYTOP
1791 int latency_record_count;
1792 struct latency_record latency_record[LT_SAVECOUNT];
1795 * time slack values; these are used to round up poll() and
1796 * select() etc timeout values. These are in nanoseconds.
1799 u64 default_timer_slack_ns;
1802 unsigned int kasan_depth;
1804 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1805 /* Index of current stored address in ret_stack */
1807 /* Stack of return addresses for return function tracing */
1808 struct ftrace_ret_stack *ret_stack;
1809 /* time stamp for last schedule */
1810 unsigned long long ftrace_timestamp;
1812 * Number of functions that haven't been traced
1813 * because of depth overrun.
1815 atomic_t trace_overrun;
1816 /* Pause for the tracing */
1817 atomic_t tracing_graph_pause;
1819 #ifdef CONFIG_TRACING
1820 /* state flags for use by tracers */
1821 unsigned long trace;
1822 /* bitmask and counter of trace recursion */
1823 unsigned long trace_recursion;
1824 #endif /* CONFIG_TRACING */
1826 /* Coverage collection mode enabled for this task (0 if disabled). */
1827 enum kcov_mode kcov_mode;
1828 /* Size of the kcov_area. */
1830 /* Buffer for coverage collection. */
1832 /* kcov desciptor wired with this task or NULL. */
1836 struct mem_cgroup *memcg_in_oom;
1837 gfp_t memcg_oom_gfp_mask;
1838 int memcg_oom_order;
1840 /* number of pages to reclaim on returning to userland */
1841 unsigned int memcg_nr_pages_over_high;
1843 #ifdef CONFIG_UPROBES
1844 struct uprobe_task *utask;
1846 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1847 unsigned int sequential_io;
1848 unsigned int sequential_io_avg;
1850 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1851 unsigned long task_state_change;
1853 int pagefault_disabled;
1855 struct task_struct *oom_reaper_list;
1857 /* CPU-specific state of this task */
1858 struct thread_struct thread;
1860 * WARNING: on x86, 'thread_struct' contains a variable-sized
1861 * structure. It *MUST* be at the end of 'task_struct'.
1863 * Do not put anything below here!
1867 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1868 extern int arch_task_struct_size __read_mostly;
1870 # define arch_task_struct_size (sizeof(struct task_struct))
1873 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1874 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1876 #define TNF_MIGRATED 0x01
1877 #define TNF_NO_GROUP 0x02
1878 #define TNF_SHARED 0x04
1879 #define TNF_FAULT_LOCAL 0x08
1880 #define TNF_MIGRATE_FAIL 0x10
1882 #ifdef CONFIG_NUMA_BALANCING
1883 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1884 extern pid_t task_numa_group_id(struct task_struct *p);
1885 extern void set_numabalancing_state(bool enabled);
1886 extern void task_numa_free(struct task_struct *p);
1887 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1888 int src_nid, int dst_cpu);
1890 static inline void task_numa_fault(int last_node, int node, int pages,
1894 static inline pid_t task_numa_group_id(struct task_struct *p)
1898 static inline void set_numabalancing_state(bool enabled)
1901 static inline void task_numa_free(struct task_struct *p)
1904 static inline bool should_numa_migrate_memory(struct task_struct *p,
1905 struct page *page, int src_nid, int dst_cpu)
1911 static inline struct pid *task_pid(struct task_struct *task)
1913 return task->pids[PIDTYPE_PID].pid;
1916 static inline struct pid *task_tgid(struct task_struct *task)
1918 return task->group_leader->pids[PIDTYPE_PID].pid;
1922 * Without tasklist or rcu lock it is not safe to dereference
1923 * the result of task_pgrp/task_session even if task == current,
1924 * we can race with another thread doing sys_setsid/sys_setpgid.
1926 static inline struct pid *task_pgrp(struct task_struct *task)
1928 return task->group_leader->pids[PIDTYPE_PGID].pid;
1931 static inline struct pid *task_session(struct task_struct *task)
1933 return task->group_leader->pids[PIDTYPE_SID].pid;
1936 struct pid_namespace;
1939 * the helpers to get the task's different pids as they are seen
1940 * from various namespaces
1942 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1943 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1945 * task_xid_nr_ns() : id seen from the ns specified;
1947 * set_task_vxid() : assigns a virtual id to a task;
1949 * see also pid_nr() etc in include/linux/pid.h
1951 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1952 struct pid_namespace *ns);
1954 static inline pid_t task_pid_nr(struct task_struct *tsk)
1959 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1960 struct pid_namespace *ns)
1962 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1965 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1967 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1971 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1976 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1978 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1980 return pid_vnr(task_tgid(tsk));
1984 static inline int pid_alive(const struct task_struct *p);
1985 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1991 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1997 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1999 return task_ppid_nr_ns(tsk, &init_pid_ns);
2002 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
2003 struct pid_namespace *ns)
2005 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
2008 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
2010 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
2014 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
2015 struct pid_namespace *ns)
2017 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
2020 static inline pid_t task_session_vnr(struct task_struct *tsk)
2022 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
2025 /* obsolete, do not use */
2026 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
2028 return task_pgrp_nr_ns(tsk, &init_pid_ns);
2032 * pid_alive - check that a task structure is not stale
2033 * @p: Task structure to be checked.
2035 * Test if a process is not yet dead (at most zombie state)
2036 * If pid_alive fails, then pointers within the task structure
2037 * can be stale and must not be dereferenced.
2039 * Return: 1 if the process is alive. 0 otherwise.
2041 static inline int pid_alive(const struct task_struct *p)
2043 return p->pids[PIDTYPE_PID].pid != NULL;
2047 * is_global_init - check if a task structure is init. Since init
2048 * is free to have sub-threads we need to check tgid.
2049 * @tsk: Task structure to be checked.
2051 * Check if a task structure is the first user space task the kernel created.
2053 * Return: 1 if the task structure is init. 0 otherwise.
2055 static inline int is_global_init(struct task_struct *tsk)
2057 return task_tgid_nr(tsk) == 1;
2060 extern struct pid *cad_pid;
2062 extern void free_task(struct task_struct *tsk);
2063 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2065 extern void __put_task_struct(struct task_struct *t);
2067 static inline void put_task_struct(struct task_struct *t)
2069 if (atomic_dec_and_test(&t->usage))
2070 __put_task_struct(t);
2073 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2074 extern void task_cputime(struct task_struct *t,
2075 cputime_t *utime, cputime_t *stime);
2076 extern void task_cputime_scaled(struct task_struct *t,
2077 cputime_t *utimescaled, cputime_t *stimescaled);
2078 extern cputime_t task_gtime(struct task_struct *t);
2080 static inline void task_cputime(struct task_struct *t,
2081 cputime_t *utime, cputime_t *stime)
2089 static inline void task_cputime_scaled(struct task_struct *t,
2090 cputime_t *utimescaled,
2091 cputime_t *stimescaled)
2094 *utimescaled = t->utimescaled;
2096 *stimescaled = t->stimescaled;
2099 static inline cputime_t task_gtime(struct task_struct *t)
2104 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2105 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2110 #define PF_EXITING 0x00000004 /* getting shut down */
2111 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2112 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2113 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2114 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2115 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2116 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2117 #define PF_DUMPCORE 0x00000200 /* dumped core */
2118 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2119 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2120 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2121 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2122 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2123 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2124 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2125 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2126 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2127 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2128 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2129 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2130 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2131 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2132 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2133 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2134 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2135 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2136 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2139 * Only the _current_ task can read/write to tsk->flags, but other
2140 * tasks can access tsk->flags in readonly mode for example
2141 * with tsk_used_math (like during threaded core dumping).
2142 * There is however an exception to this rule during ptrace
2143 * or during fork: the ptracer task is allowed to write to the
2144 * child->flags of its traced child (same goes for fork, the parent
2145 * can write to the child->flags), because we're guaranteed the
2146 * child is not running and in turn not changing child->flags
2147 * at the same time the parent does it.
2149 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2150 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2151 #define clear_used_math() clear_stopped_child_used_math(current)
2152 #define set_used_math() set_stopped_child_used_math(current)
2153 #define conditional_stopped_child_used_math(condition, child) \
2154 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2155 #define conditional_used_math(condition) \
2156 conditional_stopped_child_used_math(condition, current)
2157 #define copy_to_stopped_child_used_math(child) \
2158 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2159 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2160 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2161 #define used_math() tsk_used_math(current)
2163 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2164 * __GFP_FS is also cleared as it implies __GFP_IO.
2166 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2168 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2169 flags &= ~(__GFP_IO | __GFP_FS);
2173 static inline unsigned int memalloc_noio_save(void)
2175 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2176 current->flags |= PF_MEMALLOC_NOIO;
2180 static inline void memalloc_noio_restore(unsigned int flags)
2182 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2185 /* Per-process atomic flags. */
2186 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2187 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2188 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2191 #define TASK_PFA_TEST(name, func) \
2192 static inline bool task_##func(struct task_struct *p) \
2193 { return test_bit(PFA_##name, &p->atomic_flags); }
2194 #define TASK_PFA_SET(name, func) \
2195 static inline void task_set_##func(struct task_struct *p) \
2196 { set_bit(PFA_##name, &p->atomic_flags); }
2197 #define TASK_PFA_CLEAR(name, func) \
2198 static inline void task_clear_##func(struct task_struct *p) \
2199 { clear_bit(PFA_##name, &p->atomic_flags); }
2201 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2202 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2204 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2205 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2206 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2208 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2209 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2210 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2213 * task->jobctl flags
2215 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2217 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2218 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2219 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2220 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2221 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2222 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2223 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2225 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2226 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2227 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2228 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2229 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2230 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2231 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2233 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2234 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2236 extern bool task_set_jobctl_pending(struct task_struct *task,
2237 unsigned long mask);
2238 extern void task_clear_jobctl_trapping(struct task_struct *task);
2239 extern void task_clear_jobctl_pending(struct task_struct *task,
2240 unsigned long mask);
2242 static inline void rcu_copy_process(struct task_struct *p)
2244 #ifdef CONFIG_PREEMPT_RCU
2245 p->rcu_read_lock_nesting = 0;
2246 p->rcu_read_unlock_special.s = 0;
2247 p->rcu_blocked_node = NULL;
2248 INIT_LIST_HEAD(&p->rcu_node_entry);
2249 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2250 #ifdef CONFIG_TASKS_RCU
2251 p->rcu_tasks_holdout = false;
2252 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2253 p->rcu_tasks_idle_cpu = -1;
2254 #endif /* #ifdef CONFIG_TASKS_RCU */
2257 static inline void tsk_restore_flags(struct task_struct *task,
2258 unsigned long orig_flags, unsigned long flags)
2260 task->flags &= ~flags;
2261 task->flags |= orig_flags & flags;
2264 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2265 const struct cpumask *trial);
2266 extern int task_can_attach(struct task_struct *p,
2267 const struct cpumask *cs_cpus_allowed);
2269 extern void do_set_cpus_allowed(struct task_struct *p,
2270 const struct cpumask *new_mask);
2272 extern int set_cpus_allowed_ptr(struct task_struct *p,
2273 const struct cpumask *new_mask);
2275 static inline void do_set_cpus_allowed(struct task_struct *p,
2276 const struct cpumask *new_mask)
2279 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2280 const struct cpumask *new_mask)
2282 if (!cpumask_test_cpu(0, new_mask))
2288 #ifdef CONFIG_NO_HZ_COMMON
2289 void calc_load_enter_idle(void);
2290 void calc_load_exit_idle(void);
2292 static inline void calc_load_enter_idle(void) { }
2293 static inline void calc_load_exit_idle(void) { }
2294 #endif /* CONFIG_NO_HZ_COMMON */
2297 * Do not use outside of architecture code which knows its limitations.
2299 * sched_clock() has no promise of monotonicity or bounded drift between
2300 * CPUs, use (which you should not) requires disabling IRQs.
2302 * Please use one of the three interfaces below.
2304 extern unsigned long long notrace sched_clock(void);
2306 * See the comment in kernel/sched/clock.c
2308 extern u64 running_clock(void);
2309 extern u64 sched_clock_cpu(int cpu);
2312 extern void sched_clock_init(void);
2314 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2315 static inline void sched_clock_tick(void)
2319 static inline void sched_clock_idle_sleep_event(void)
2323 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2327 static inline u64 cpu_clock(int cpu)
2329 return sched_clock();
2332 static inline u64 local_clock(void)
2334 return sched_clock();
2338 * Architectures can set this to 1 if they have specified
2339 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2340 * but then during bootup it turns out that sched_clock()
2341 * is reliable after all:
2343 extern int sched_clock_stable(void);
2344 extern void set_sched_clock_stable(void);
2345 extern void clear_sched_clock_stable(void);
2347 extern void sched_clock_tick(void);
2348 extern void sched_clock_idle_sleep_event(void);
2349 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2352 * As outlined in clock.c, provides a fast, high resolution, nanosecond
2353 * time source that is monotonic per cpu argument and has bounded drift
2356 * ######################### BIG FAT WARNING ##########################
2357 * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can #
2358 * # go backwards !! #
2359 * ####################################################################
2361 static inline u64 cpu_clock(int cpu)
2363 return sched_clock_cpu(cpu);
2366 static inline u64 local_clock(void)
2368 return sched_clock_cpu(raw_smp_processor_id());
2372 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2374 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2375 * The reason for this explicit opt-in is not to have perf penalty with
2376 * slow sched_clocks.
2378 extern void enable_sched_clock_irqtime(void);
2379 extern void disable_sched_clock_irqtime(void);
2381 static inline void enable_sched_clock_irqtime(void) {}
2382 static inline void disable_sched_clock_irqtime(void) {}
2385 extern unsigned long long
2386 task_sched_runtime(struct task_struct *task);
2388 /* sched_exec is called by processes performing an exec */
2390 extern void sched_exec(void);
2392 #define sched_exec() {}
2395 extern void sched_clock_idle_sleep_event(void);
2396 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2398 #ifdef CONFIG_HOTPLUG_CPU
2399 extern void idle_task_exit(void);
2401 static inline void idle_task_exit(void) {}
2404 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2405 extern void wake_up_nohz_cpu(int cpu);
2407 static inline void wake_up_nohz_cpu(int cpu) { }
2410 #ifdef CONFIG_NO_HZ_FULL
2411 extern u64 scheduler_tick_max_deferment(void);
2414 #ifdef CONFIG_SCHED_AUTOGROUP
2415 extern void sched_autogroup_create_attach(struct task_struct *p);
2416 extern void sched_autogroup_detach(struct task_struct *p);
2417 extern void sched_autogroup_fork(struct signal_struct *sig);
2418 extern void sched_autogroup_exit(struct signal_struct *sig);
2419 #ifdef CONFIG_PROC_FS
2420 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2421 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2424 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2425 static inline void sched_autogroup_detach(struct task_struct *p) { }
2426 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2427 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2430 extern int yield_to(struct task_struct *p, bool preempt);
2431 extern void set_user_nice(struct task_struct *p, long nice);
2432 extern int task_prio(const struct task_struct *p);
2434 * task_nice - return the nice value of a given task.
2435 * @p: the task in question.
2437 * Return: The nice value [ -20 ... 0 ... 19 ].
2439 static inline int task_nice(const struct task_struct *p)
2441 return PRIO_TO_NICE((p)->static_prio);
2443 extern int can_nice(const struct task_struct *p, const int nice);
2444 extern int task_curr(const struct task_struct *p);
2445 extern int idle_cpu(int cpu);
2446 extern int sched_setscheduler(struct task_struct *, int,
2447 const struct sched_param *);
2448 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2449 const struct sched_param *);
2450 extern int sched_setattr(struct task_struct *,
2451 const struct sched_attr *);
2452 extern struct task_struct *idle_task(int cpu);
2454 * is_idle_task - is the specified task an idle task?
2455 * @p: the task in question.
2457 * Return: 1 if @p is an idle task. 0 otherwise.
2459 static inline bool is_idle_task(const struct task_struct *p)
2463 extern struct task_struct *curr_task(int cpu);
2464 extern void set_curr_task(int cpu, struct task_struct *p);
2468 union thread_union {
2469 struct thread_info thread_info;
2470 unsigned long stack[THREAD_SIZE/sizeof(long)];
2473 #ifndef __HAVE_ARCH_KSTACK_END
2474 static inline int kstack_end(void *addr)
2476 /* Reliable end of stack detection:
2477 * Some APM bios versions misalign the stack
2479 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2483 extern union thread_union init_thread_union;
2484 extern struct task_struct init_task;
2486 extern struct mm_struct init_mm;
2488 extern struct pid_namespace init_pid_ns;
2491 * find a task by one of its numerical ids
2493 * find_task_by_pid_ns():
2494 * finds a task by its pid in the specified namespace
2495 * find_task_by_vpid():
2496 * finds a task by its virtual pid
2498 * see also find_vpid() etc in include/linux/pid.h
2501 extern struct task_struct *find_task_by_vpid(pid_t nr);
2502 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2503 struct pid_namespace *ns);
2505 /* per-UID process charging. */
2506 extern struct user_struct * alloc_uid(kuid_t);
2507 static inline struct user_struct *get_uid(struct user_struct *u)
2509 atomic_inc(&u->__count);
2512 extern void free_uid(struct user_struct *);
2514 #include <asm/current.h>
2516 extern void xtime_update(unsigned long ticks);
2518 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2519 extern int wake_up_process(struct task_struct *tsk);
2520 extern void wake_up_new_task(struct task_struct *tsk);
2522 extern void kick_process(struct task_struct *tsk);
2524 static inline void kick_process(struct task_struct *tsk) { }
2526 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2527 extern void sched_dead(struct task_struct *p);
2529 extern void proc_caches_init(void);
2530 extern void flush_signals(struct task_struct *);
2531 extern void ignore_signals(struct task_struct *);
2532 extern void flush_signal_handlers(struct task_struct *, int force_default);
2533 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2535 static inline int kernel_dequeue_signal(siginfo_t *info)
2537 struct task_struct *tsk = current;
2541 spin_lock_irq(&tsk->sighand->siglock);
2542 ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2543 spin_unlock_irq(&tsk->sighand->siglock);
2548 static inline void kernel_signal_stop(void)
2550 spin_lock_irq(¤t->sighand->siglock);
2551 if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2552 __set_current_state(TASK_STOPPED);
2553 spin_unlock_irq(¤t->sighand->siglock);
2558 extern void release_task(struct task_struct * p);
2559 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2560 extern int force_sigsegv(int, struct task_struct *);
2561 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2562 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2563 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2564 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2565 const struct cred *, u32);
2566 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2567 extern int kill_pid(struct pid *pid, int sig, int priv);
2568 extern int kill_proc_info(int, struct siginfo *, pid_t);
2569 extern __must_check bool do_notify_parent(struct task_struct *, int);
2570 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2571 extern void force_sig(int, struct task_struct *);
2572 extern int send_sig(int, struct task_struct *, int);
2573 extern int zap_other_threads(struct task_struct *p);
2574 extern struct sigqueue *sigqueue_alloc(void);
2575 extern void sigqueue_free(struct sigqueue *);
2576 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2577 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2579 static inline void restore_saved_sigmask(void)
2581 if (test_and_clear_restore_sigmask())
2582 __set_current_blocked(¤t->saved_sigmask);
2585 static inline sigset_t *sigmask_to_save(void)
2587 sigset_t *res = ¤t->blocked;
2588 if (unlikely(test_restore_sigmask()))
2589 res = ¤t->saved_sigmask;
2593 static inline int kill_cad_pid(int sig, int priv)
2595 return kill_pid(cad_pid, sig, priv);
2598 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2599 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2600 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2601 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2604 * True if we are on the alternate signal stack.
2606 static inline int on_sig_stack(unsigned long sp)
2608 #ifdef CONFIG_STACK_GROWSUP
2609 return sp >= current->sas_ss_sp &&
2610 sp - current->sas_ss_sp < current->sas_ss_size;
2612 return sp > current->sas_ss_sp &&
2613 sp - current->sas_ss_sp <= current->sas_ss_size;
2617 static inline int sas_ss_flags(unsigned long sp)
2619 if (!current->sas_ss_size)
2622 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2625 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2627 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2628 #ifdef CONFIG_STACK_GROWSUP
2629 return current->sas_ss_sp;
2631 return current->sas_ss_sp + current->sas_ss_size;
2637 * Routines for handling mm_structs
2639 extern struct mm_struct * mm_alloc(void);
2641 /* mmdrop drops the mm and the page tables */
2642 extern void __mmdrop(struct mm_struct *);
2643 static inline void mmdrop(struct mm_struct * mm)
2645 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2649 /* mmput gets rid of the mappings and all user-space */
2650 extern void mmput(struct mm_struct *);
2651 /* Grab a reference to a task's mm, if it is not already going away */
2652 extern struct mm_struct *get_task_mm(struct task_struct *task);
2654 * Grab a reference to a task's mm, if it is not already going away
2655 * and ptrace_may_access with the mode parameter passed to it
2658 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2659 /* Remove the current tasks stale references to the old mm_struct */
2660 extern void mm_release(struct task_struct *, struct mm_struct *);
2662 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2663 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2664 struct task_struct *, unsigned long);
2666 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2667 struct task_struct *);
2669 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2670 * via pt_regs, so ignore the tls argument passed via C. */
2671 static inline int copy_thread_tls(
2672 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2673 struct task_struct *p, unsigned long tls)
2675 return copy_thread(clone_flags, sp, arg, p);
2678 extern void flush_thread(void);
2679 extern void exit_thread(void);
2681 extern void exit_files(struct task_struct *);
2682 extern void __cleanup_sighand(struct sighand_struct *);
2684 extern void exit_itimers(struct signal_struct *);
2685 extern void flush_itimer_signals(void);
2687 extern void do_group_exit(int);
2689 extern int do_execve(struct filename *,
2690 const char __user * const __user *,
2691 const char __user * const __user *);
2692 extern int do_execveat(int, struct filename *,
2693 const char __user * const __user *,
2694 const char __user * const __user *,
2696 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2697 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2698 struct task_struct *fork_idle(int);
2699 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2701 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2702 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2704 __set_task_comm(tsk, from, false);
2706 extern char *get_task_comm(char *to, struct task_struct *tsk);
2709 void scheduler_ipi(void);
2710 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2712 static inline void scheduler_ipi(void) { }
2713 static inline unsigned long wait_task_inactive(struct task_struct *p,
2720 #define tasklist_empty() \
2721 list_empty(&init_task.tasks)
2723 #define next_task(p) \
2724 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2726 #define for_each_process(p) \
2727 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2729 extern bool current_is_single_threaded(void);
2732 * Careful: do_each_thread/while_each_thread is a double loop so
2733 * 'break' will not work as expected - use goto instead.
2735 #define do_each_thread(g, t) \
2736 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2738 #define while_each_thread(g, t) \
2739 while ((t = next_thread(t)) != g)
2741 #define __for_each_thread(signal, t) \
2742 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2744 #define for_each_thread(p, t) \
2745 __for_each_thread((p)->signal, t)
2747 /* Careful: this is a double loop, 'break' won't work as expected. */
2748 #define for_each_process_thread(p, t) \
2749 for_each_process(p) for_each_thread(p, t)
2751 static inline int get_nr_threads(struct task_struct *tsk)
2753 return tsk->signal->nr_threads;
2756 static inline bool thread_group_leader(struct task_struct *p)
2758 return p->exit_signal >= 0;
2761 /* Do to the insanities of de_thread it is possible for a process
2762 * to have the pid of the thread group leader without actually being
2763 * the thread group leader. For iteration through the pids in proc
2764 * all we care about is that we have a task with the appropriate
2765 * pid, we don't actually care if we have the right task.
2767 static inline bool has_group_leader_pid(struct task_struct *p)
2769 return task_pid(p) == p->signal->leader_pid;
2773 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2775 return p1->signal == p2->signal;
2778 static inline struct task_struct *next_thread(const struct task_struct *p)
2780 return list_entry_rcu(p->thread_group.next,
2781 struct task_struct, thread_group);
2784 static inline int thread_group_empty(struct task_struct *p)
2786 return list_empty(&p->thread_group);
2789 #define delay_group_leader(p) \
2790 (thread_group_leader(p) && !thread_group_empty(p))
2793 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2794 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2795 * pins the final release of task.io_context. Also protects ->cpuset and
2796 * ->cgroup.subsys[]. And ->vfork_done.
2798 * Nests both inside and outside of read_lock(&tasklist_lock).
2799 * It must not be nested with write_lock_irq(&tasklist_lock),
2800 * neither inside nor outside.
2802 static inline void task_lock(struct task_struct *p)
2804 spin_lock(&p->alloc_lock);
2807 static inline void task_unlock(struct task_struct *p)
2809 spin_unlock(&p->alloc_lock);
2812 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2813 unsigned long *flags);
2815 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2816 unsigned long *flags)
2818 struct sighand_struct *ret;
2820 ret = __lock_task_sighand(tsk, flags);
2821 (void)__cond_lock(&tsk->sighand->siglock, ret);
2825 static inline void unlock_task_sighand(struct task_struct *tsk,
2826 unsigned long *flags)
2828 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2832 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2833 * @tsk: task causing the changes
2835 * All operations which modify a threadgroup - a new thread joining the
2836 * group, death of a member thread (the assertion of PF_EXITING) and
2837 * exec(2) dethreading the process and replacing the leader - are wrapped
2838 * by threadgroup_change_{begin|end}(). This is to provide a place which
2839 * subsystems needing threadgroup stability can hook into for
2842 static inline void threadgroup_change_begin(struct task_struct *tsk)
2845 cgroup_threadgroup_change_begin(tsk);
2849 * threadgroup_change_end - mark the end of changes to a threadgroup
2850 * @tsk: task causing the changes
2852 * See threadgroup_change_begin().
2854 static inline void threadgroup_change_end(struct task_struct *tsk)
2856 cgroup_threadgroup_change_end(tsk);
2859 #ifndef __HAVE_THREAD_FUNCTIONS
2861 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2862 #define task_stack_page(task) ((task)->stack)
2864 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2866 *task_thread_info(p) = *task_thread_info(org);
2867 task_thread_info(p)->task = p;
2871 * Return the address of the last usable long on the stack.
2873 * When the stack grows down, this is just above the thread
2874 * info struct. Going any lower will corrupt the threadinfo.
2876 * When the stack grows up, this is the highest address.
2877 * Beyond that position, we corrupt data on the next page.
2879 static inline unsigned long *end_of_stack(struct task_struct *p)
2881 #ifdef CONFIG_STACK_GROWSUP
2882 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2884 return (unsigned long *)(task_thread_info(p) + 1);
2889 #define task_stack_end_corrupted(task) \
2890 (*(end_of_stack(task)) != STACK_END_MAGIC)
2892 static inline int object_is_on_stack(void *obj)
2894 void *stack = task_stack_page(current);
2896 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2899 extern void thread_info_cache_init(void);
2901 #ifdef CONFIG_DEBUG_STACK_USAGE
2902 static inline unsigned long stack_not_used(struct task_struct *p)
2904 unsigned long *n = end_of_stack(p);
2906 do { /* Skip over canary */
2907 # ifdef CONFIG_STACK_GROWSUP
2914 # ifdef CONFIG_STACK_GROWSUP
2915 return (unsigned long)end_of_stack(p) - (unsigned long)n;
2917 return (unsigned long)n - (unsigned long)end_of_stack(p);
2921 extern void set_task_stack_end_magic(struct task_struct *tsk);
2923 /* set thread flags in other task's structures
2924 * - see asm/thread_info.h for TIF_xxxx flags available
2926 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2928 set_ti_thread_flag(task_thread_info(tsk), flag);
2931 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2933 clear_ti_thread_flag(task_thread_info(tsk), flag);
2936 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2938 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2941 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2943 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2946 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2948 return test_ti_thread_flag(task_thread_info(tsk), flag);
2951 static inline void set_tsk_need_resched(struct task_struct *tsk)
2953 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2956 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2958 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2961 static inline int test_tsk_need_resched(struct task_struct *tsk)
2963 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2966 static inline int restart_syscall(void)
2968 set_tsk_thread_flag(current, TIF_SIGPENDING);
2969 return -ERESTARTNOINTR;
2972 static inline int signal_pending(struct task_struct *p)
2974 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2977 static inline int __fatal_signal_pending(struct task_struct *p)
2979 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2982 static inline int fatal_signal_pending(struct task_struct *p)
2984 return signal_pending(p) && __fatal_signal_pending(p);
2987 static inline int signal_pending_state(long state, struct task_struct *p)
2989 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2991 if (!signal_pending(p))
2994 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2998 * cond_resched() and cond_resched_lock(): latency reduction via
2999 * explicit rescheduling in places that are safe. The return
3000 * value indicates whether a reschedule was done in fact.
3001 * cond_resched_lock() will drop the spinlock before scheduling,
3002 * cond_resched_softirq() will enable bhs before scheduling.
3004 extern int _cond_resched(void);
3006 #define cond_resched() ({ \
3007 ___might_sleep(__FILE__, __LINE__, 0); \
3011 extern int __cond_resched_lock(spinlock_t *lock);
3013 #define cond_resched_lock(lock) ({ \
3014 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3015 __cond_resched_lock(lock); \
3018 extern int __cond_resched_softirq(void);
3020 #define cond_resched_softirq() ({ \
3021 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3022 __cond_resched_softirq(); \
3025 static inline void cond_resched_rcu(void)
3027 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3035 * Does a critical section need to be broken due to another
3036 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3037 * but a general need for low latency)
3039 static inline int spin_needbreak(spinlock_t *lock)
3041 #ifdef CONFIG_PREEMPT
3042 return spin_is_contended(lock);
3049 * Idle thread specific functions to determine the need_resched
3052 #ifdef TIF_POLLING_NRFLAG
3053 static inline int tsk_is_polling(struct task_struct *p)
3055 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3058 static inline void __current_set_polling(void)
3060 set_thread_flag(TIF_POLLING_NRFLAG);
3063 static inline bool __must_check current_set_polling_and_test(void)
3065 __current_set_polling();
3068 * Polling state must be visible before we test NEED_RESCHED,
3069 * paired by resched_curr()
3071 smp_mb__after_atomic();
3073 return unlikely(tif_need_resched());
3076 static inline void __current_clr_polling(void)
3078 clear_thread_flag(TIF_POLLING_NRFLAG);
3081 static inline bool __must_check current_clr_polling_and_test(void)
3083 __current_clr_polling();
3086 * Polling state must be visible before we test NEED_RESCHED,
3087 * paired by resched_curr()
3089 smp_mb__after_atomic();
3091 return unlikely(tif_need_resched());
3095 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3096 static inline void __current_set_polling(void) { }
3097 static inline void __current_clr_polling(void) { }
3099 static inline bool __must_check current_set_polling_and_test(void)
3101 return unlikely(tif_need_resched());
3103 static inline bool __must_check current_clr_polling_and_test(void)
3105 return unlikely(tif_need_resched());
3109 static inline void current_clr_polling(void)
3111 __current_clr_polling();
3114 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3115 * Once the bit is cleared, we'll get IPIs with every new
3116 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3119 smp_mb(); /* paired with resched_curr() */
3121 preempt_fold_need_resched();
3124 static __always_inline bool need_resched(void)
3126 return unlikely(tif_need_resched());
3130 * Thread group CPU time accounting.
3132 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3133 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3136 * Reevaluate whether the task has signals pending delivery.
3137 * Wake the task if so.
3138 * This is required every time the blocked sigset_t changes.
3139 * callers must hold sighand->siglock.
3141 extern void recalc_sigpending_and_wake(struct task_struct *t);
3142 extern void recalc_sigpending(void);
3144 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3146 static inline void signal_wake_up(struct task_struct *t, bool resume)
3148 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3150 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3152 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3156 * Wrappers for p->thread_info->cpu access. No-op on UP.
3160 static inline unsigned int task_cpu(const struct task_struct *p)
3162 return task_thread_info(p)->cpu;
3165 static inline int task_node(const struct task_struct *p)
3167 return cpu_to_node(task_cpu(p));
3170 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3174 static inline unsigned int task_cpu(const struct task_struct *p)
3179 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3183 #endif /* CONFIG_SMP */
3185 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3186 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3188 #ifdef CONFIG_CGROUP_SCHED
3189 extern struct task_group root_task_group;
3190 #endif /* CONFIG_CGROUP_SCHED */
3192 extern int task_can_switch_user(struct user_struct *up,
3193 struct task_struct *tsk);
3195 #ifdef CONFIG_TASK_XACCT
3196 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3198 tsk->ioac.rchar += amt;
3201 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3203 tsk->ioac.wchar += amt;
3206 static inline void inc_syscr(struct task_struct *tsk)
3211 static inline void inc_syscw(struct task_struct *tsk)
3216 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3220 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3224 static inline void inc_syscr(struct task_struct *tsk)
3228 static inline void inc_syscw(struct task_struct *tsk)
3233 #ifndef TASK_SIZE_OF
3234 #define TASK_SIZE_OF(tsk) TASK_SIZE
3238 extern void mm_update_next_owner(struct mm_struct *mm);
3240 static inline void mm_update_next_owner(struct mm_struct *mm)
3243 #endif /* CONFIG_MEMCG */
3245 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3248 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3251 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3254 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3257 static inline unsigned long rlimit(unsigned int limit)
3259 return task_rlimit(current, limit);
3262 static inline unsigned long rlimit_max(unsigned int limit)
3264 return task_rlimit_max(current, limit);
3267 #ifdef CONFIG_CPU_FREQ
3268 struct update_util_data {
3269 void (*func)(struct update_util_data *data,
3270 u64 time, unsigned long util, unsigned long max);
3273 void cpufreq_set_update_util_data(int cpu, struct update_util_data *data);
3274 #endif /* CONFIG_CPU_FREQ */