4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt_mask.h>
29 #include <asm/kmap_types.h>
32 #include <asm/ptrace.h>
33 #include <linux/cputime.h>
35 #include <linux/smp.h>
36 #include <linux/sem.h>
37 #include <linux/shm.h>
38 #include <linux/signal.h>
39 #include <linux/compiler.h>
40 #include <linux/completion.h>
41 #include <linux/pid.h>
42 #include <linux/percpu.h>
43 #include <linux/topology.h>
44 #include <linux/proportions.h>
45 #include <linux/seccomp.h>
46 #include <linux/rcupdate.h>
47 #include <linux/rculist.h>
48 #include <linux/rtmutex.h>
50 #include <linux/time.h>
51 #include <linux/param.h>
52 #include <linux/resource.h>
53 #include <linux/timer.h>
54 #include <linux/hrtimer.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>
63 #include <asm/processor.h>
65 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
68 * Extended scheduling parameters data structure.
70 * This is needed because the original struct sched_param can not be
71 * altered without introducing ABI issues with legacy applications
72 * (e.g., in sched_getparam()).
74 * However, the possibility of specifying more than just a priority for
75 * the tasks may be useful for a wide variety of application fields, e.g.,
76 * multimedia, streaming, automation and control, and many others.
78 * This variant (sched_attr) is meant at describing a so-called
79 * sporadic time-constrained task. In such model a task is specified by:
80 * - the activation period or minimum instance inter-arrival time;
81 * - the maximum (or average, depending on the actual scheduling
82 * discipline) computation time of all instances, a.k.a. runtime;
83 * - the deadline (relative to the actual activation time) of each
85 * Very briefly, a periodic (sporadic) task asks for the execution of
86 * some specific computation --which is typically called an instance--
87 * (at most) every period. Moreover, each instance typically lasts no more
88 * than the runtime and must be completed by time instant t equal to
89 * the instance activation time + the deadline.
91 * This is reflected by the actual fields of the sched_attr structure:
93 * @size size of the structure, for fwd/bwd compat.
95 * @sched_policy task's scheduling policy
96 * @sched_flags for customizing the scheduler behaviour
97 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
98 * @sched_priority task's static priority (SCHED_FIFO/RR)
99 * @sched_deadline representative of the task's deadline
100 * @sched_runtime representative of the task's runtime
101 * @sched_period representative of the task's period
103 * Given this task model, there are a multiplicity of scheduling algorithms
104 * and policies, that can be used to ensure all the tasks will make their
105 * timing constraints.
107 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
108 * only user of this new interface. More information about the algorithm
109 * available in the scheduling class file or in Documentation/.
117 /* SCHED_NORMAL, SCHED_BATCH */
120 /* SCHED_FIFO, SCHED_RR */
129 struct futex_pi_state;
130 struct robust_list_head;
133 struct perf_event_context;
137 #define VMACACHE_BITS 2
138 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
139 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
142 * These are the constant used to fake the fixed-point load-average
143 * counting. Some notes:
144 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
145 * a load-average precision of 10 bits integer + 11 bits fractional
146 * - if you want to count load-averages more often, you need more
147 * precision, or rounding will get you. With 2-second counting freq,
148 * the EXP_n values would be 1981, 2034 and 2043 if still using only
151 extern unsigned long avenrun[]; /* Load averages */
152 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
154 #define FSHIFT 11 /* nr of bits of precision */
155 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
156 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
157 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
158 #define EXP_5 2014 /* 1/exp(5sec/5min) */
159 #define EXP_15 2037 /* 1/exp(5sec/15min) */
161 #define CALC_LOAD(load,exp,n) \
163 load += n*(FIXED_1-exp); \
166 extern unsigned long total_forks;
167 extern int nr_threads;
168 DECLARE_PER_CPU(unsigned long, process_counts);
169 extern int nr_processes(void);
170 extern unsigned long nr_running(void);
171 extern bool single_task_running(void);
172 extern unsigned long nr_iowait(void);
173 extern unsigned long nr_iowait_cpu(int cpu);
174 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
176 extern void calc_global_load(unsigned long ticks);
177 extern void update_cpu_load_nohz(void);
179 extern unsigned long get_parent_ip(unsigned long addr);
181 extern void dump_cpu_task(int cpu);
186 #ifdef CONFIG_SCHED_DEBUG
187 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
188 extern void proc_sched_set_task(struct task_struct *p);
190 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
194 * Task state bitmask. NOTE! These bits are also
195 * encoded in fs/proc/array.c: get_task_state().
197 * We have two separate sets of flags: task->state
198 * is about runnability, while task->exit_state are
199 * about the task exiting. Confusing, but this way
200 * modifying one set can't modify the other one by
203 #define TASK_RUNNING 0
204 #define TASK_INTERRUPTIBLE 1
205 #define TASK_UNINTERRUPTIBLE 2
206 #define __TASK_STOPPED 4
207 #define __TASK_TRACED 8
208 /* in tsk->exit_state */
210 #define EXIT_ZOMBIE 32
211 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
212 /* in tsk->state again */
214 #define TASK_WAKEKILL 128
215 #define TASK_WAKING 256
216 #define TASK_PARKED 512
217 #define TASK_STATE_MAX 1024
219 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
221 extern char ___assert_task_state[1 - 2*!!(
222 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
224 /* Convenience macros for the sake of set_task_state */
225 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
226 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
227 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
229 /* Convenience macros for the sake of wake_up */
230 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
231 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
233 /* get_task_state() */
234 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
235 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
236 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
238 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
239 #define task_contributes_to_load(task) \
240 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
241 (task->flags & PF_FROZEN) == 0)
243 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
245 #define __set_task_state(tsk, state_value) \
247 (tsk)->task_state_change = _THIS_IP_; \
248 (tsk)->state = (state_value); \
250 #define set_task_state(tsk, state_value) \
252 (tsk)->task_state_change = _THIS_IP_; \
253 set_mb((tsk)->state, (state_value)); \
257 * set_current_state() includes a barrier so that the write of current->state
258 * is correctly serialised wrt the caller's subsequent test of whether to
261 * set_current_state(TASK_UNINTERRUPTIBLE);
262 * if (do_i_need_to_sleep())
265 * If the caller does not need such serialisation then use __set_current_state()
267 #define __set_current_state(state_value) \
269 current->task_state_change = _THIS_IP_; \
270 current->state = (state_value); \
272 #define set_current_state(state_value) \
274 current->task_state_change = _THIS_IP_; \
275 set_mb(current->state, (state_value)); \
280 #define __set_task_state(tsk, state_value) \
281 do { (tsk)->state = (state_value); } while (0)
282 #define set_task_state(tsk, state_value) \
283 set_mb((tsk)->state, (state_value))
286 * set_current_state() includes a barrier so that the write of current->state
287 * is correctly serialised wrt the caller's subsequent test of whether to
290 * set_current_state(TASK_UNINTERRUPTIBLE);
291 * if (do_i_need_to_sleep())
294 * If the caller does not need such serialisation then use __set_current_state()
296 #define __set_current_state(state_value) \
297 do { current->state = (state_value); } while (0)
298 #define set_current_state(state_value) \
299 set_mb(current->state, (state_value))
303 #define __set_current_state_no_track(state_value) \
304 do { current->state = (state_value); } while (0)
305 #define set_current_state_no_track(state_value) \
306 set_mb(current->state, (state_value))
308 /* Task command name length */
309 #define TASK_COMM_LEN 16
311 #include <linux/spinlock.h>
314 * This serializes "schedule()" and also protects
315 * the run-queue from deletions/modifications (but
316 * _adding_ to the beginning of the run-queue has
319 extern rwlock_t tasklist_lock;
320 extern spinlock_t mmlist_lock;
324 #ifdef CONFIG_PROVE_RCU
325 extern int lockdep_tasklist_lock_is_held(void);
326 #endif /* #ifdef CONFIG_PROVE_RCU */
328 extern void sched_init(void);
329 extern void sched_init_smp(void);
330 extern asmlinkage void schedule_tail(struct task_struct *prev);
331 extern void init_idle(struct task_struct *idle, int cpu);
332 extern void init_idle_bootup_task(struct task_struct *idle);
334 extern cpumask_var_t cpu_isolated_map;
336 extern int runqueue_is_locked(int cpu);
338 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
339 extern void nohz_balance_enter_idle(int cpu);
340 extern void set_cpu_sd_state_idle(void);
341 extern int get_nohz_timer_target(int pinned);
343 static inline void nohz_balance_enter_idle(int cpu) { }
344 static inline void set_cpu_sd_state_idle(void) { }
345 static inline int get_nohz_timer_target(int pinned)
347 return smp_processor_id();
352 * Only dump TASK_* tasks. (0 for all tasks)
354 extern void show_state_filter(unsigned long state_filter);
356 static inline void show_state(void)
358 show_state_filter(0);
361 extern void show_regs(struct pt_regs *);
364 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
365 * task), SP is the stack pointer of the first frame that should be shown in the back
366 * trace (or NULL if the entire call-chain of the task should be shown).
368 extern void show_stack(struct task_struct *task, unsigned long *sp);
370 extern void cpu_init (void);
371 extern void trap_init(void);
372 extern void update_process_times(int user);
373 extern void scheduler_tick(void);
375 extern void sched_show_task(struct task_struct *p);
377 #ifdef CONFIG_LOCKUP_DETECTOR
378 extern void touch_softlockup_watchdog(void);
379 extern void touch_softlockup_watchdog_sync(void);
380 extern void touch_all_softlockup_watchdogs(void);
381 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
383 size_t *lenp, loff_t *ppos);
384 extern unsigned int softlockup_panic;
385 void lockup_detector_init(void);
387 static inline void touch_softlockup_watchdog(void)
390 static inline void touch_softlockup_watchdog_sync(void)
393 static inline void touch_all_softlockup_watchdogs(void)
396 static inline void lockup_detector_init(void)
401 #ifdef CONFIG_DETECT_HUNG_TASK
402 void reset_hung_task_detector(void);
404 static inline void reset_hung_task_detector(void)
409 /* Attach to any functions which should be ignored in wchan output. */
410 #define __sched __attribute__((__section__(".sched.text")))
412 /* Linker adds these: start and end of __sched functions */
413 extern char __sched_text_start[], __sched_text_end[];
415 /* Is this address in the __sched functions? */
416 extern int in_sched_functions(unsigned long addr);
418 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
419 extern signed long schedule_timeout(signed long timeout);
420 extern signed long schedule_timeout_interruptible(signed long timeout);
421 extern signed long schedule_timeout_killable(signed long timeout);
422 extern signed long schedule_timeout_uninterruptible(signed long timeout);
423 asmlinkage void schedule(void);
424 extern void schedule_preempt_disabled(void);
426 extern long io_schedule_timeout(long timeout);
428 static inline void io_schedule(void)
430 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
434 struct user_namespace;
437 extern void arch_pick_mmap_layout(struct mm_struct *mm);
439 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
440 unsigned long, unsigned long);
442 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
443 unsigned long len, unsigned long pgoff,
444 unsigned long flags);
446 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
449 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
450 #define SUID_DUMP_USER 1 /* Dump as user of process */
451 #define SUID_DUMP_ROOT 2 /* Dump as root */
455 /* for SUID_DUMP_* above */
456 #define MMF_DUMPABLE_BITS 2
457 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
459 extern void set_dumpable(struct mm_struct *mm, int value);
461 * This returns the actual value of the suid_dumpable flag. For things
462 * that are using this for checking for privilege transitions, it must
463 * test against SUID_DUMP_USER rather than treating it as a boolean
466 static inline int __get_dumpable(unsigned long mm_flags)
468 return mm_flags & MMF_DUMPABLE_MASK;
471 static inline int get_dumpable(struct mm_struct *mm)
473 return __get_dumpable(mm->flags);
476 /* coredump filter bits */
477 #define MMF_DUMP_ANON_PRIVATE 2
478 #define MMF_DUMP_ANON_SHARED 3
479 #define MMF_DUMP_MAPPED_PRIVATE 4
480 #define MMF_DUMP_MAPPED_SHARED 5
481 #define MMF_DUMP_ELF_HEADERS 6
482 #define MMF_DUMP_HUGETLB_PRIVATE 7
483 #define MMF_DUMP_HUGETLB_SHARED 8
485 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
486 #define MMF_DUMP_FILTER_BITS 7
487 #define MMF_DUMP_FILTER_MASK \
488 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
489 #define MMF_DUMP_FILTER_DEFAULT \
490 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
491 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
493 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
494 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
496 # define MMF_DUMP_MASK_DEFAULT_ELF 0
498 /* leave room for more dump flags */
499 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
500 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
501 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
503 #define MMF_HAS_UPROBES 19 /* has uprobes */
504 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
506 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
508 struct sighand_struct {
510 struct k_sigaction action[_NSIG];
512 wait_queue_head_t signalfd_wqh;
515 struct pacct_struct {
518 unsigned long ac_mem;
519 cputime_t ac_utime, ac_stime;
520 unsigned long ac_minflt, ac_majflt;
531 * struct cputime - snaphsot of system and user cputime
532 * @utime: time spent in user mode
533 * @stime: time spent in system mode
535 * Gathers a generic snapshot of user and system time.
543 * struct task_cputime - collected CPU time counts
544 * @utime: time spent in user mode, in &cputime_t units
545 * @stime: time spent in kernel mode, in &cputime_t units
546 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
548 * This is an extension of struct cputime that includes the total runtime
549 * spent by the task from the scheduler point of view.
551 * As a result, this structure groups together three kinds of CPU time
552 * that are tracked for threads and thread groups. Most things considering
553 * CPU time want to group these counts together and treat all three
554 * of them in parallel.
556 struct task_cputime {
559 unsigned long long sum_exec_runtime;
561 /* Alternate field names when used to cache expirations. */
562 #define prof_exp stime
563 #define virt_exp utime
564 #define sched_exp sum_exec_runtime
566 #define INIT_CPUTIME \
567 (struct task_cputime) { \
570 .sum_exec_runtime = 0, \
573 #ifdef CONFIG_PREEMPT_COUNT
574 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
576 #define PREEMPT_DISABLED PREEMPT_ENABLED
580 * Disable preemption until the scheduler is running.
581 * Reset by start_kernel()->sched_init()->init_idle().
583 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
584 * before the scheduler is active -- see should_resched().
586 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
589 * struct thread_group_cputimer - thread group interval timer counts
590 * @cputime: thread group interval timers.
591 * @running: non-zero when there are timers running and
592 * @cputime receives updates.
593 * @lock: lock for fields in this struct.
595 * This structure contains the version of task_cputime, above, that is
596 * used for thread group CPU timer calculations.
598 struct thread_group_cputimer {
599 struct task_cputime cputime;
604 #include <linux/rwsem.h>
608 * NOTE! "signal_struct" does not have its own
609 * locking, because a shared signal_struct always
610 * implies a shared sighand_struct, so locking
611 * sighand_struct is always a proper superset of
612 * the locking of signal_struct.
614 struct signal_struct {
618 struct list_head thread_head;
620 wait_queue_head_t wait_chldexit; /* for wait4() */
622 /* current thread group signal load-balancing target: */
623 struct task_struct *curr_target;
625 /* shared signal handling: */
626 struct sigpending shared_pending;
628 /* thread group exit support */
631 * - notify group_exit_task when ->count is equal to notify_count
632 * - everyone except group_exit_task is stopped during signal delivery
633 * of fatal signals, group_exit_task processes the signal.
636 struct task_struct *group_exit_task;
638 /* thread group stop support, overloads group_exit_code too */
639 int group_stop_count;
640 unsigned int flags; /* see SIGNAL_* flags below */
643 * PR_SET_CHILD_SUBREAPER marks a process, like a service
644 * manager, to re-parent orphan (double-forking) child processes
645 * to this process instead of 'init'. The service manager is
646 * able to receive SIGCHLD signals and is able to investigate
647 * the process until it calls wait(). All children of this
648 * process will inherit a flag if they should look for a
649 * child_subreaper process at exit.
651 unsigned int is_child_subreaper:1;
652 unsigned int has_child_subreaper:1;
654 /* POSIX.1b Interval Timers */
656 struct list_head posix_timers;
658 /* ITIMER_REAL timer for the process */
659 struct hrtimer real_timer;
660 struct pid *leader_pid;
661 ktime_t it_real_incr;
664 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
665 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
666 * values are defined to 0 and 1 respectively
668 struct cpu_itimer it[2];
671 * Thread group totals for process CPU timers.
672 * See thread_group_cputimer(), et al, for details.
674 struct thread_group_cputimer cputimer;
676 /* Earliest-expiration cache. */
677 struct task_cputime cputime_expires;
679 struct list_head cpu_timers[3];
681 struct pid *tty_old_pgrp;
683 /* boolean value for session group leader */
686 struct tty_struct *tty; /* NULL if no tty */
688 #ifdef CONFIG_SCHED_AUTOGROUP
689 struct autogroup *autogroup;
692 * Cumulative resource counters for dead threads in the group,
693 * and for reaped dead child processes forked by this group.
694 * Live threads maintain their own counters and add to these
695 * in __exit_signal, except for the group leader.
697 seqlock_t stats_lock;
698 cputime_t utime, stime, cutime, cstime;
701 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
702 struct cputime prev_cputime;
704 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
705 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
706 unsigned long inblock, oublock, cinblock, coublock;
707 unsigned long maxrss, cmaxrss;
708 struct task_io_accounting ioac;
711 * Cumulative ns of schedule CPU time fo dead threads in the
712 * group, not including a zombie group leader, (This only differs
713 * from jiffies_to_ns(utime + stime) if sched_clock uses something
714 * other than jiffies.)
716 unsigned long long sum_sched_runtime;
719 * We don't bother to synchronize most readers of this at all,
720 * because there is no reader checking a limit that actually needs
721 * to get both rlim_cur and rlim_max atomically, and either one
722 * alone is a single word that can safely be read normally.
723 * getrlimit/setrlimit use task_lock(current->group_leader) to
724 * protect this instead of the siglock, because they really
725 * have no need to disable irqs.
727 struct rlimit rlim[RLIM_NLIMITS];
729 #ifdef CONFIG_BSD_PROCESS_ACCT
730 struct pacct_struct pacct; /* per-process accounting information */
732 #ifdef CONFIG_TASKSTATS
733 struct taskstats *stats;
737 unsigned audit_tty_log_passwd;
738 struct tty_audit_buf *tty_audit_buf;
740 #ifdef CONFIG_CGROUPS
742 * group_rwsem prevents new tasks from entering the threadgroup and
743 * member tasks from exiting,a more specifically, setting of
744 * PF_EXITING. fork and exit paths are protected with this rwsem
745 * using threadgroup_change_begin/end(). Users which require
746 * threadgroup to remain stable should use threadgroup_[un]lock()
747 * which also takes care of exec path. Currently, cgroup is the
750 struct rw_semaphore group_rwsem;
753 oom_flags_t oom_flags;
754 short oom_score_adj; /* OOM kill score adjustment */
755 short oom_score_adj_min; /* OOM kill score adjustment min value.
756 * Only settable by CAP_SYS_RESOURCE. */
758 struct mutex cred_guard_mutex; /* guard against foreign influences on
759 * credential calculations
760 * (notably. ptrace) */
764 * Bits in flags field of signal_struct.
766 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
767 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
768 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
769 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
771 * Pending notifications to parent.
773 #define SIGNAL_CLD_STOPPED 0x00000010
774 #define SIGNAL_CLD_CONTINUED 0x00000020
775 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
777 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
779 /* If true, all threads except ->group_exit_task have pending SIGKILL */
780 static inline int signal_group_exit(const struct signal_struct *sig)
782 return (sig->flags & SIGNAL_GROUP_EXIT) ||
783 (sig->group_exit_task != NULL);
787 * Some day this will be a full-fledged user tracking system..
790 atomic_t __count; /* reference count */
791 atomic_t processes; /* How many processes does this user have? */
792 atomic_t sigpending; /* How many pending signals does this user have? */
793 #ifdef CONFIG_INOTIFY_USER
794 atomic_t inotify_watches; /* How many inotify watches does this user have? */
795 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
797 #ifdef CONFIG_FANOTIFY
798 atomic_t fanotify_listeners;
801 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
803 #ifdef CONFIG_POSIX_MQUEUE
804 /* protected by mq_lock */
805 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
807 unsigned long locked_shm; /* How many pages of mlocked shm ? */
810 struct key *uid_keyring; /* UID specific keyring */
811 struct key *session_keyring; /* UID's default session keyring */
814 /* Hash table maintenance information */
815 struct hlist_node uidhash_node;
818 #ifdef CONFIG_PERF_EVENTS
819 atomic_long_t locked_vm;
823 extern int uids_sysfs_init(void);
825 extern struct user_struct *find_user(kuid_t);
827 extern struct user_struct root_user;
828 #define INIT_USER (&root_user)
831 struct backing_dev_info;
832 struct reclaim_state;
834 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
836 /* cumulative counters */
837 unsigned long pcount; /* # of times run on this cpu */
838 unsigned long long run_delay; /* time spent waiting on a runqueue */
841 unsigned long long last_arrival,/* when we last ran on a cpu */
842 last_queued; /* when we were last queued to run */
844 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
846 #ifdef CONFIG_TASK_DELAY_ACCT
847 struct task_delay_info {
849 unsigned int flags; /* Private per-task flags */
851 /* For each stat XXX, add following, aligned appropriately
853 * struct timespec XXX_start, XXX_end;
857 * Atomicity of updates to XXX_delay, XXX_count protected by
858 * single lock above (split into XXX_lock if contention is an issue).
862 * XXX_count is incremented on every XXX operation, the delay
863 * associated with the operation is added to XXX_delay.
864 * XXX_delay contains the accumulated delay time in nanoseconds.
866 u64 blkio_start; /* Shared by blkio, swapin */
867 u64 blkio_delay; /* wait for sync block io completion */
868 u64 swapin_delay; /* wait for swapin block io completion */
869 u32 blkio_count; /* total count of the number of sync block */
870 /* io operations performed */
871 u32 swapin_count; /* total count of the number of swapin block */
872 /* io operations performed */
875 u64 freepages_delay; /* wait for memory reclaim */
876 u32 freepages_count; /* total count of memory reclaim */
878 #endif /* CONFIG_TASK_DELAY_ACCT */
880 static inline int sched_info_on(void)
882 #ifdef CONFIG_SCHEDSTATS
884 #elif defined(CONFIG_TASK_DELAY_ACCT)
885 extern int delayacct_on;
900 * Increase resolution of cpu_capacity calculations
902 #define SCHED_CAPACITY_SHIFT 10
903 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
906 * Wake-queues are lists of tasks with a pending wakeup, whose
907 * callers have already marked the task as woken internally,
908 * and can thus carry on. A common use case is being able to
909 * do the wakeups once the corresponding user lock as been
912 * We hold reference to each task in the list across the wakeup,
913 * thus guaranteeing that the memory is still valid by the time
914 * the actual wakeups are performed in wake_up_q().
916 * One per task suffices, because there's never a need for a task to be
917 * in two wake queues simultaneously; it is forbidden to abandon a task
918 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
919 * already in a wake queue, the wakeup will happen soon and the second
920 * waker can just skip it.
922 * The WAKE_Q macro declares and initializes the list head.
923 * wake_up_q() does NOT reinitialize the list; it's expected to be
924 * called near the end of a function, where the fact that the queue is
925 * not used again will be easy to see by inspection.
927 * Note that this can cause spurious wakeups. schedule() callers
928 * must ensure the call is done inside a loop, confirming that the
929 * wakeup condition has in fact occurred.
932 struct wake_q_node *next;
936 struct wake_q_node *first;
937 struct wake_q_node **lastp;
940 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
942 #define WAKE_Q(name) \
943 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
945 extern void wake_q_add(struct wake_q_head *head,
946 struct task_struct *task);
947 extern void wake_up_q(struct wake_q_head *head);
950 * sched-domains (multiprocessor balancing) declarations:
953 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
954 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
955 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
956 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
957 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
958 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
959 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
960 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
961 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
962 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
963 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
964 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
965 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
966 #define SD_NUMA 0x4000 /* cross-node balancing */
968 #ifdef CONFIG_SCHED_SMT
969 static inline int cpu_smt_flags(void)
971 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
975 #ifdef CONFIG_SCHED_MC
976 static inline int cpu_core_flags(void)
978 return SD_SHARE_PKG_RESOURCES;
983 static inline int cpu_numa_flags(void)
989 struct sched_domain_attr {
990 int relax_domain_level;
993 #define SD_ATTR_INIT (struct sched_domain_attr) { \
994 .relax_domain_level = -1, \
997 extern int sched_domain_level_max;
1001 struct sched_domain {
1002 /* These fields must be setup */
1003 struct sched_domain *parent; /* top domain must be null terminated */
1004 struct sched_domain *child; /* bottom domain must be null terminated */
1005 struct sched_group *groups; /* the balancing groups of the domain */
1006 unsigned long min_interval; /* Minimum balance interval ms */
1007 unsigned long max_interval; /* Maximum balance interval ms */
1008 unsigned int busy_factor; /* less balancing by factor if busy */
1009 unsigned int imbalance_pct; /* No balance until over watermark */
1010 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1011 unsigned int busy_idx;
1012 unsigned int idle_idx;
1013 unsigned int newidle_idx;
1014 unsigned int wake_idx;
1015 unsigned int forkexec_idx;
1016 unsigned int smt_gain;
1018 int nohz_idle; /* NOHZ IDLE status */
1019 int flags; /* See SD_* */
1022 /* Runtime fields. */
1023 unsigned long last_balance; /* init to jiffies. units in jiffies */
1024 unsigned int balance_interval; /* initialise to 1. units in ms. */
1025 unsigned int nr_balance_failed; /* initialise to 0 */
1027 /* idle_balance() stats */
1028 u64 max_newidle_lb_cost;
1029 unsigned long next_decay_max_lb_cost;
1031 #ifdef CONFIG_SCHEDSTATS
1032 /* load_balance() stats */
1033 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1034 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1035 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1036 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1037 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1038 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1039 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1040 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1042 /* Active load balancing */
1043 unsigned int alb_count;
1044 unsigned int alb_failed;
1045 unsigned int alb_pushed;
1047 /* SD_BALANCE_EXEC stats */
1048 unsigned int sbe_count;
1049 unsigned int sbe_balanced;
1050 unsigned int sbe_pushed;
1052 /* SD_BALANCE_FORK stats */
1053 unsigned int sbf_count;
1054 unsigned int sbf_balanced;
1055 unsigned int sbf_pushed;
1057 /* try_to_wake_up() stats */
1058 unsigned int ttwu_wake_remote;
1059 unsigned int ttwu_move_affine;
1060 unsigned int ttwu_move_balance;
1062 #ifdef CONFIG_SCHED_DEBUG
1066 void *private; /* used during construction */
1067 struct rcu_head rcu; /* used during destruction */
1070 unsigned int span_weight;
1072 * Span of all CPUs in this domain.
1074 * NOTE: this field is variable length. (Allocated dynamically
1075 * by attaching extra space to the end of the structure,
1076 * depending on how many CPUs the kernel has booted up with)
1078 unsigned long span[0];
1081 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1083 return to_cpumask(sd->span);
1086 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1087 struct sched_domain_attr *dattr_new);
1089 /* Allocate an array of sched domains, for partition_sched_domains(). */
1090 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1091 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1093 bool cpus_share_cache(int this_cpu, int that_cpu);
1095 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1096 typedef int (*sched_domain_flags_f)(void);
1098 #define SDTL_OVERLAP 0x01
1101 struct sched_domain **__percpu sd;
1102 struct sched_group **__percpu sg;
1103 struct sched_group_capacity **__percpu sgc;
1106 struct sched_domain_topology_level {
1107 sched_domain_mask_f mask;
1108 sched_domain_flags_f sd_flags;
1111 struct sd_data data;
1112 #ifdef CONFIG_SCHED_DEBUG
1117 extern struct sched_domain_topology_level *sched_domain_topology;
1119 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1120 extern void wake_up_if_idle(int cpu);
1122 #ifdef CONFIG_SCHED_DEBUG
1123 # define SD_INIT_NAME(type) .name = #type
1125 # define SD_INIT_NAME(type)
1128 #else /* CONFIG_SMP */
1130 struct sched_domain_attr;
1133 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1134 struct sched_domain_attr *dattr_new)
1138 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1143 #endif /* !CONFIG_SMP */
1146 struct io_context; /* See blkdev.h */
1149 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1150 extern void prefetch_stack(struct task_struct *t);
1152 static inline void prefetch_stack(struct task_struct *t) { }
1155 struct audit_context; /* See audit.c */
1157 struct pipe_inode_info;
1158 struct uts_namespace;
1160 struct load_weight {
1161 unsigned long weight;
1166 u64 last_runnable_update;
1169 * utilization_avg_contrib describes the amount of time that a
1170 * sched_entity is running on a CPU. It is based on running_avg_sum
1171 * and is scaled in the range [0..SCHED_LOAD_SCALE].
1172 * load_avg_contrib described the amount of time that a sched_entity
1173 * is runnable on a rq. It is based on both runnable_avg_sum and the
1174 * weight of the task.
1176 unsigned long load_avg_contrib, utilization_avg_contrib;
1178 * These sums represent an infinite geometric series and so are bound
1179 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1180 * choices of y < 1-2^(-32)*1024.
1181 * running_avg_sum reflects the time that the sched_entity is
1182 * effectively running on the CPU.
1183 * runnable_avg_sum represents the amount of time a sched_entity is on
1184 * a runqueue which includes the running time that is monitored by
1187 u32 runnable_avg_sum, avg_period, running_avg_sum;
1190 #ifdef CONFIG_SCHEDSTATS
1191 struct sched_statistics {
1201 s64 sum_sleep_runtime;
1208 u64 nr_migrations_cold;
1209 u64 nr_failed_migrations_affine;
1210 u64 nr_failed_migrations_running;
1211 u64 nr_failed_migrations_hot;
1212 u64 nr_forced_migrations;
1215 u64 nr_wakeups_sync;
1216 u64 nr_wakeups_migrate;
1217 u64 nr_wakeups_local;
1218 u64 nr_wakeups_remote;
1219 u64 nr_wakeups_affine;
1220 u64 nr_wakeups_affine_attempts;
1221 u64 nr_wakeups_passive;
1222 u64 nr_wakeups_idle;
1226 struct sched_entity {
1227 struct load_weight load; /* for load-balancing */
1228 struct rb_node run_node;
1229 struct list_head group_node;
1233 u64 sum_exec_runtime;
1235 u64 prev_sum_exec_runtime;
1239 #ifdef CONFIG_SCHEDSTATS
1240 struct sched_statistics statistics;
1243 #ifdef CONFIG_FAIR_GROUP_SCHED
1245 struct sched_entity *parent;
1246 /* rq on which this entity is (to be) queued: */
1247 struct cfs_rq *cfs_rq;
1248 /* rq "owned" by this entity/group: */
1249 struct cfs_rq *my_q;
1253 /* Per-entity load-tracking */
1254 struct sched_avg avg;
1258 struct sched_rt_entity {
1259 struct list_head run_list;
1260 unsigned long timeout;
1261 unsigned long watchdog_stamp;
1262 unsigned int time_slice;
1264 struct sched_rt_entity *back;
1265 #ifdef CONFIG_RT_GROUP_SCHED
1266 struct sched_rt_entity *parent;
1267 /* rq on which this entity is (to be) queued: */
1268 struct rt_rq *rt_rq;
1269 /* rq "owned" by this entity/group: */
1274 struct sched_dl_entity {
1275 struct rb_node rb_node;
1278 * Original scheduling parameters. Copied here from sched_attr
1279 * during sched_setattr(), they will remain the same until
1280 * the next sched_setattr().
1282 u64 dl_runtime; /* maximum runtime for each instance */
1283 u64 dl_deadline; /* relative deadline of each instance */
1284 u64 dl_period; /* separation of two instances (period) */
1285 u64 dl_bw; /* dl_runtime / dl_deadline */
1288 * Actual scheduling parameters. Initialized with the values above,
1289 * they are continously updated during task execution. Note that
1290 * the remaining runtime could be < 0 in case we are in overrun.
1292 s64 runtime; /* remaining runtime for this instance */
1293 u64 deadline; /* absolute deadline for this instance */
1294 unsigned int flags; /* specifying the scheduler behaviour */
1299 * @dl_throttled tells if we exhausted the runtime. If so, the
1300 * task has to wait for a replenishment to be performed at the
1301 * next firing of dl_timer.
1303 * @dl_new tells if a new instance arrived. If so we must
1304 * start executing it with full runtime and reset its absolute
1307 * @dl_boosted tells if we are boosted due to DI. If so we are
1308 * outside bandwidth enforcement mechanism (but only until we
1309 * exit the critical section);
1311 * @dl_yielded tells if task gave up the cpu before consuming
1312 * all its available runtime during the last job.
1314 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1317 * Bandwidth enforcement timer. Each -deadline task has its
1318 * own bandwidth to be enforced, thus we need one timer per task.
1320 struct hrtimer dl_timer;
1332 enum perf_event_task_context {
1333 perf_invalid_context = -1,
1334 perf_hw_context = 0,
1336 perf_nr_task_contexts,
1339 struct task_struct {
1340 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1341 volatile long saved_state; /* saved state for "spinlock sleepers" */
1344 unsigned int flags; /* per process flags, defined below */
1345 unsigned int ptrace;
1348 struct llist_node wake_entry;
1350 struct task_struct *last_wakee;
1351 unsigned long wakee_flips;
1352 unsigned long wakee_flip_decay_ts;
1358 int prio, static_prio, normal_prio;
1359 unsigned int rt_priority;
1360 const struct sched_class *sched_class;
1361 struct sched_entity se;
1362 struct sched_rt_entity rt;
1363 #ifdef CONFIG_CGROUP_SCHED
1364 struct task_group *sched_task_group;
1366 struct sched_dl_entity dl;
1368 #ifdef CONFIG_PREEMPT_NOTIFIERS
1369 /* list of struct preempt_notifier: */
1370 struct hlist_head preempt_notifiers;
1373 #ifdef CONFIG_BLK_DEV_IO_TRACE
1374 unsigned int btrace_seq;
1377 unsigned int policy;
1378 #ifdef CONFIG_PREEMPT_RT_FULL
1379 int migrate_disable;
1380 # ifdef CONFIG_SCHED_DEBUG
1381 int migrate_disable_atomic;
1384 int nr_cpus_allowed;
1385 cpumask_t cpus_allowed;
1387 #ifdef CONFIG_PREEMPT_RCU
1388 int rcu_read_lock_nesting;
1389 union rcu_special rcu_read_unlock_special;
1390 struct list_head rcu_node_entry;
1391 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1392 #ifdef CONFIG_PREEMPT_RCU
1393 struct rcu_node *rcu_blocked_node;
1394 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1395 #ifdef CONFIG_TASKS_RCU
1396 unsigned long rcu_tasks_nvcsw;
1397 bool rcu_tasks_holdout;
1398 struct list_head rcu_tasks_holdout_list;
1399 int rcu_tasks_idle_cpu;
1400 #endif /* #ifdef CONFIG_TASKS_RCU */
1402 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1403 struct sched_info sched_info;
1406 struct list_head tasks;
1408 struct plist_node pushable_tasks;
1409 struct rb_node pushable_dl_tasks;
1412 struct mm_struct *mm, *active_mm;
1413 #ifdef CONFIG_COMPAT_BRK
1414 unsigned brk_randomized:1;
1416 /* per-thread vma caching */
1417 u32 vmacache_seqnum;
1418 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1419 #if defined(SPLIT_RSS_COUNTING)
1420 struct task_rss_stat rss_stat;
1424 int exit_code, exit_signal;
1425 int pdeath_signal; /* The signal sent when the parent dies */
1426 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1428 /* Used for emulating ABI behavior of previous Linux versions */
1429 unsigned int personality;
1431 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1433 unsigned in_iowait:1;
1435 /* Revert to default priority/policy when forking */
1436 unsigned sched_reset_on_fork:1;
1437 unsigned sched_contributes_to_load:1;
1439 #ifdef CONFIG_MEMCG_KMEM
1440 unsigned memcg_kmem_skip_account:1;
1443 unsigned long atomic_flags; /* Flags needing atomic access. */
1445 struct restart_block restart_block;
1450 #ifdef CONFIG_CC_STACKPROTECTOR
1451 /* Canary value for the -fstack-protector gcc feature */
1452 unsigned long stack_canary;
1455 * pointers to (original) parent process, youngest child, younger sibling,
1456 * older sibling, respectively. (p->father can be replaced with
1457 * p->real_parent->pid)
1459 struct task_struct __rcu *real_parent; /* real parent process */
1460 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1462 * children/sibling forms the list of my natural children
1464 struct list_head children; /* list of my children */
1465 struct list_head sibling; /* linkage in my parent's children list */
1466 struct task_struct *group_leader; /* threadgroup leader */
1469 * ptraced is the list of tasks this task is using ptrace on.
1470 * This includes both natural children and PTRACE_ATTACH targets.
1471 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1473 struct list_head ptraced;
1474 struct list_head ptrace_entry;
1476 /* PID/PID hash table linkage. */
1477 struct pid_link pids[PIDTYPE_MAX];
1478 struct list_head thread_group;
1479 struct list_head thread_node;
1481 struct completion *vfork_done; /* for vfork() */
1482 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1483 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1485 cputime_t utime, stime, utimescaled, stimescaled;
1487 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1488 struct cputime prev_cputime;
1490 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1491 raw_spinlock_t vtime_lock;
1492 seqcount_t vtime_seq;
1493 unsigned long long vtime_snap;
1498 } vtime_snap_whence;
1500 unsigned long nvcsw, nivcsw; /* context switch counts */
1501 u64 start_time; /* monotonic time in nsec */
1502 u64 real_start_time; /* boot based time in nsec */
1503 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1504 unsigned long min_flt, maj_flt;
1506 struct task_cputime cputime_expires;
1507 struct list_head cpu_timers[3];
1508 #ifdef CONFIG_PREEMPT_RT_BASE
1509 struct task_struct *posix_timer_list;
1512 /* process credentials */
1513 const struct cred __rcu *real_cred; /* objective and real subjective task
1514 * credentials (COW) */
1515 const struct cred __rcu *cred; /* effective (overridable) subjective task
1516 * credentials (COW) */
1517 char comm[TASK_COMM_LEN]; /* executable name excluding path
1518 - access with [gs]et_task_comm (which lock
1519 it with task_lock())
1520 - initialized normally by setup_new_exec */
1521 /* file system info */
1522 int link_count, total_link_count;
1523 #ifdef CONFIG_SYSVIPC
1525 struct sysv_sem sysvsem;
1526 struct sysv_shm sysvshm;
1528 #ifdef CONFIG_DETECT_HUNG_TASK
1529 /* hung task detection */
1530 unsigned long last_switch_count;
1532 /* CPU-specific state of this task */
1533 struct thread_struct thread;
1534 /* filesystem information */
1535 struct fs_struct *fs;
1536 /* open file information */
1537 struct files_struct *files;
1539 struct nsproxy *nsproxy;
1540 /* signal handlers */
1541 struct signal_struct *signal;
1542 struct sighand_struct *sighand;
1543 struct sigqueue *sigqueue_cache;
1545 sigset_t blocked, real_blocked;
1546 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1547 struct sigpending pending;
1548 #ifdef CONFIG_PREEMPT_RT_FULL
1549 /* TODO: move me into ->restart_block ? */
1550 struct siginfo forced_info;
1553 unsigned long sas_ss_sp;
1555 int (*notifier)(void *priv);
1556 void *notifier_data;
1557 sigset_t *notifier_mask;
1558 struct callback_head *task_works;
1560 struct audit_context *audit_context;
1561 #ifdef CONFIG_AUDITSYSCALL
1563 unsigned int sessionid;
1565 struct seccomp seccomp;
1567 /* Thread group tracking */
1570 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1572 spinlock_t alloc_lock;
1574 /* Protection of the PI data structures: */
1575 raw_spinlock_t pi_lock;
1577 struct wake_q_node wake_q;
1579 #ifdef CONFIG_RT_MUTEXES
1580 /* PI waiters blocked on a rt_mutex held by this task */
1581 struct rb_root pi_waiters;
1582 struct rb_node *pi_waiters_leftmost;
1583 /* Deadlock detection and priority inheritance handling */
1584 struct rt_mutex_waiter *pi_blocked_on;
1587 #ifdef CONFIG_DEBUG_MUTEXES
1588 /* mutex deadlock detection */
1589 struct mutex_waiter *blocked_on;
1591 #ifdef CONFIG_TRACE_IRQFLAGS
1592 unsigned int irq_events;
1593 unsigned long hardirq_enable_ip;
1594 unsigned long hardirq_disable_ip;
1595 unsigned int hardirq_enable_event;
1596 unsigned int hardirq_disable_event;
1597 int hardirqs_enabled;
1598 int hardirq_context;
1599 unsigned long softirq_disable_ip;
1600 unsigned long softirq_enable_ip;
1601 unsigned int softirq_disable_event;
1602 unsigned int softirq_enable_event;
1603 int softirqs_enabled;
1604 int softirq_context;
1606 #ifdef CONFIG_LOCKDEP
1607 # define MAX_LOCK_DEPTH 48UL
1610 unsigned int lockdep_recursion;
1611 struct held_lock held_locks[MAX_LOCK_DEPTH];
1612 gfp_t lockdep_reclaim_gfp;
1615 /* journalling filesystem info */
1618 /* stacked block device info */
1619 struct bio_list *bio_list;
1622 /* stack plugging */
1623 struct blk_plug *plug;
1627 struct reclaim_state *reclaim_state;
1629 struct backing_dev_info *backing_dev_info;
1631 struct io_context *io_context;
1633 unsigned long ptrace_message;
1634 siginfo_t *last_siginfo; /* For ptrace use. */
1635 struct task_io_accounting ioac;
1636 #if defined(CONFIG_TASK_XACCT)
1637 u64 acct_rss_mem1; /* accumulated rss usage */
1638 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1639 cputime_t acct_timexpd; /* stime + utime since last update */
1641 #ifdef CONFIG_CPUSETS
1642 nodemask_t mems_allowed; /* Protected by alloc_lock */
1643 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1644 int cpuset_mem_spread_rotor;
1645 int cpuset_slab_spread_rotor;
1647 #ifdef CONFIG_CGROUPS
1648 /* Control Group info protected by css_set_lock */
1649 struct css_set __rcu *cgroups;
1650 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1651 struct list_head cg_list;
1654 struct robust_list_head __user *robust_list;
1655 #ifdef CONFIG_COMPAT
1656 struct compat_robust_list_head __user *compat_robust_list;
1658 struct list_head pi_state_list;
1659 struct futex_pi_state *pi_state_cache;
1661 #ifdef CONFIG_PERF_EVENTS
1662 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1663 struct mutex perf_event_mutex;
1664 struct list_head perf_event_list;
1666 #ifdef CONFIG_DEBUG_PREEMPT
1667 unsigned long preempt_disable_ip;
1670 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1672 short pref_node_fork;
1674 #ifdef CONFIG_NUMA_BALANCING
1676 unsigned int numa_scan_period;
1677 unsigned int numa_scan_period_max;
1678 int numa_preferred_nid;
1679 unsigned long numa_migrate_retry;
1680 u64 node_stamp; /* migration stamp */
1681 u64 last_task_numa_placement;
1682 u64 last_sum_exec_runtime;
1683 struct callback_head numa_work;
1685 struct list_head numa_entry;
1686 struct numa_group *numa_group;
1689 * numa_faults is an array split into four regions:
1690 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1691 * in this precise order.
1693 * faults_memory: Exponential decaying average of faults on a per-node
1694 * basis. Scheduling placement decisions are made based on these
1695 * counts. The values remain static for the duration of a PTE scan.
1696 * faults_cpu: Track the nodes the process was running on when a NUMA
1697 * hinting fault was incurred.
1698 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1699 * during the current scan window. When the scan completes, the counts
1700 * in faults_memory and faults_cpu decay and these values are copied.
1702 unsigned long *numa_faults;
1703 unsigned long total_numa_faults;
1706 * numa_faults_locality tracks if faults recorded during the last
1707 * scan window were remote/local or failed to migrate. The task scan
1708 * period is adapted based on the locality of the faults with different
1709 * weights depending on whether they were shared or private faults
1711 unsigned long numa_faults_locality[3];
1713 unsigned long numa_pages_migrated;
1714 #endif /* CONFIG_NUMA_BALANCING */
1716 struct rcu_head rcu;
1719 * cache last used pipe for splice
1721 struct pipe_inode_info *splice_pipe;
1723 struct page_frag task_frag;
1725 #ifdef CONFIG_TASK_DELAY_ACCT
1726 struct task_delay_info *delays;
1728 #ifdef CONFIG_FAULT_INJECTION
1732 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1733 * balance_dirty_pages() for some dirty throttling pause
1736 int nr_dirtied_pause;
1737 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1739 #ifdef CONFIG_LATENCYTOP
1740 int latency_record_count;
1741 struct latency_record latency_record[LT_SAVECOUNT];
1744 * time slack values; these are used to round up poll() and
1745 * select() etc timeout values. These are in nanoseconds.
1747 unsigned long timer_slack_ns;
1748 unsigned long default_timer_slack_ns;
1751 unsigned int kasan_depth;
1753 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1754 /* Index of current stored address in ret_stack */
1756 /* Stack of return addresses for return function tracing */
1757 struct ftrace_ret_stack *ret_stack;
1758 /* time stamp for last schedule */
1759 unsigned long long ftrace_timestamp;
1761 * Number of functions that haven't been traced
1762 * because of depth overrun.
1764 atomic_t trace_overrun;
1765 /* Pause for the tracing */
1766 atomic_t tracing_graph_pause;
1768 #ifdef CONFIG_TRACING
1769 /* state flags for use by tracers */
1770 unsigned long trace;
1771 /* bitmask and counter of trace recursion */
1772 unsigned long trace_recursion;
1773 #ifdef CONFIG_WAKEUP_LATENCY_HIST
1774 u64 preempt_timestamp_hist;
1775 #ifdef CONFIG_MISSED_TIMER_OFFSETS_HIST
1779 #endif /* CONFIG_TRACING */
1781 struct memcg_oom_info {
1782 struct mem_cgroup *memcg;
1785 unsigned int may_oom:1;
1788 #ifdef CONFIG_UPROBES
1789 struct uprobe_task *utask;
1791 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1792 unsigned int sequential_io;
1793 unsigned int sequential_io_avg;
1795 #ifdef CONFIG_PREEMPT_RT_BASE
1796 struct rcu_head put_rcu;
1797 int softirq_nestcnt;
1798 unsigned int softirqs_raised;
1800 #ifdef CONFIG_PREEMPT_RT_FULL
1801 # if defined CONFIG_HIGHMEM || defined CONFIG_X86_32
1803 pte_t kmap_pte[KM_TYPE_NR];
1806 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1807 unsigned long task_state_change;
1809 int pagefault_disabled;
1812 #define TNF_MIGRATED 0x01
1813 #define TNF_NO_GROUP 0x02
1814 #define TNF_SHARED 0x04
1815 #define TNF_FAULT_LOCAL 0x08
1816 #define TNF_MIGRATE_FAIL 0x10
1818 #ifdef CONFIG_NUMA_BALANCING
1819 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1820 extern pid_t task_numa_group_id(struct task_struct *p);
1821 extern void set_numabalancing_state(bool enabled);
1822 extern void task_numa_free(struct task_struct *p);
1823 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1824 int src_nid, int dst_cpu);
1826 static inline void task_numa_fault(int last_node, int node, int pages,
1830 static inline pid_t task_numa_group_id(struct task_struct *p)
1834 static inline void set_numabalancing_state(bool enabled)
1837 static inline void task_numa_free(struct task_struct *p)
1840 static inline bool should_numa_migrate_memory(struct task_struct *p,
1841 struct page *page, int src_nid, int dst_cpu)
1847 static inline struct pid *task_pid(struct task_struct *task)
1849 return task->pids[PIDTYPE_PID].pid;
1852 static inline struct pid *task_tgid(struct task_struct *task)
1854 return task->group_leader->pids[PIDTYPE_PID].pid;
1858 * Without tasklist or rcu lock it is not safe to dereference
1859 * the result of task_pgrp/task_session even if task == current,
1860 * we can race with another thread doing sys_setsid/sys_setpgid.
1862 static inline struct pid *task_pgrp(struct task_struct *task)
1864 return task->group_leader->pids[PIDTYPE_PGID].pid;
1867 static inline struct pid *task_session(struct task_struct *task)
1869 return task->group_leader->pids[PIDTYPE_SID].pid;
1872 struct pid_namespace;
1875 * the helpers to get the task's different pids as they are seen
1876 * from various namespaces
1878 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1879 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1881 * task_xid_nr_ns() : id seen from the ns specified;
1883 * set_task_vxid() : assigns a virtual id to a task;
1885 * see also pid_nr() etc in include/linux/pid.h
1887 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1888 struct pid_namespace *ns);
1890 static inline pid_t task_pid_nr(struct task_struct *tsk)
1895 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1896 struct pid_namespace *ns)
1898 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1901 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1903 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1907 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1912 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1914 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1916 return pid_vnr(task_tgid(tsk));
1920 static inline int pid_alive(const struct task_struct *p);
1921 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1927 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1933 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1935 return task_ppid_nr_ns(tsk, &init_pid_ns);
1938 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1939 struct pid_namespace *ns)
1941 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1944 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1946 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1950 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1951 struct pid_namespace *ns)
1953 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1956 static inline pid_t task_session_vnr(struct task_struct *tsk)
1958 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1961 /* obsolete, do not use */
1962 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1964 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1968 * pid_alive - check that a task structure is not stale
1969 * @p: Task structure to be checked.
1971 * Test if a process is not yet dead (at most zombie state)
1972 * If pid_alive fails, then pointers within the task structure
1973 * can be stale and must not be dereferenced.
1975 * Return: 1 if the process is alive. 0 otherwise.
1977 static inline int pid_alive(const struct task_struct *p)
1979 return p->pids[PIDTYPE_PID].pid != NULL;
1983 * is_global_init - check if a task structure is init
1984 * @tsk: Task structure to be checked.
1986 * Check if a task structure is the first user space task the kernel created.
1988 * Return: 1 if the task structure is init. 0 otherwise.
1990 static inline int is_global_init(struct task_struct *tsk)
1992 return tsk->pid == 1;
1995 extern struct pid *cad_pid;
1997 extern void free_task(struct task_struct *tsk);
1998 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2000 #ifdef CONFIG_PREEMPT_RT_BASE
2001 extern void __put_task_struct_cb(struct rcu_head *rhp);
2003 static inline void put_task_struct(struct task_struct *t)
2005 if (atomic_dec_and_test(&t->usage))
2006 call_rcu(&t->put_rcu, __put_task_struct_cb);
2009 extern void __put_task_struct(struct task_struct *t);
2011 static inline void put_task_struct(struct task_struct *t)
2013 if (atomic_dec_and_test(&t->usage))
2014 __put_task_struct(t);
2018 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2019 extern void task_cputime(struct task_struct *t,
2020 cputime_t *utime, cputime_t *stime);
2021 extern void task_cputime_scaled(struct task_struct *t,
2022 cputime_t *utimescaled, cputime_t *stimescaled);
2023 extern cputime_t task_gtime(struct task_struct *t);
2025 static inline void task_cputime(struct task_struct *t,
2026 cputime_t *utime, cputime_t *stime)
2034 static inline void task_cputime_scaled(struct task_struct *t,
2035 cputime_t *utimescaled,
2036 cputime_t *stimescaled)
2039 *utimescaled = t->utimescaled;
2041 *stimescaled = t->stimescaled;
2044 static inline cputime_t task_gtime(struct task_struct *t)
2049 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2050 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2055 #define PF_IN_SOFTIRQ 0x00000001 /* Task is serving softirq */
2056 #define PF_EXITING 0x00000004 /* getting shut down */
2057 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2058 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2059 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2060 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2061 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2062 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2063 #define PF_DUMPCORE 0x00000200 /* dumped core */
2064 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2065 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2066 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2067 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2068 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2069 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2070 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2071 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2072 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2073 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2074 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2075 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2076 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2077 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2078 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2079 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2080 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2081 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2082 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2085 * Only the _current_ task can read/write to tsk->flags, but other
2086 * tasks can access tsk->flags in readonly mode for example
2087 * with tsk_used_math (like during threaded core dumping).
2088 * There is however an exception to this rule during ptrace
2089 * or during fork: the ptracer task is allowed to write to the
2090 * child->flags of its traced child (same goes for fork, the parent
2091 * can write to the child->flags), because we're guaranteed the
2092 * child is not running and in turn not changing child->flags
2093 * at the same time the parent does it.
2095 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2096 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2097 #define clear_used_math() clear_stopped_child_used_math(current)
2098 #define set_used_math() set_stopped_child_used_math(current)
2099 #define conditional_stopped_child_used_math(condition, child) \
2100 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2101 #define conditional_used_math(condition) \
2102 conditional_stopped_child_used_math(condition, current)
2103 #define copy_to_stopped_child_used_math(child) \
2104 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2105 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2106 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2107 #define used_math() tsk_used_math(current)
2109 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2110 * __GFP_FS is also cleared as it implies __GFP_IO.
2112 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2114 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2115 flags &= ~(__GFP_IO | __GFP_FS);
2119 static inline unsigned int memalloc_noio_save(void)
2121 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2122 current->flags |= PF_MEMALLOC_NOIO;
2126 static inline void memalloc_noio_restore(unsigned int flags)
2128 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2131 /* Per-process atomic flags. */
2132 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2133 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2134 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2137 #define TASK_PFA_TEST(name, func) \
2138 static inline bool task_##func(struct task_struct *p) \
2139 { return test_bit(PFA_##name, &p->atomic_flags); }
2140 #define TASK_PFA_SET(name, func) \
2141 static inline void task_set_##func(struct task_struct *p) \
2142 { set_bit(PFA_##name, &p->atomic_flags); }
2143 #define TASK_PFA_CLEAR(name, func) \
2144 static inline void task_clear_##func(struct task_struct *p) \
2145 { clear_bit(PFA_##name, &p->atomic_flags); }
2147 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2148 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2150 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2151 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2152 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2154 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2155 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2156 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2159 * task->jobctl flags
2161 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2163 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2164 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2165 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2166 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2167 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2168 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2169 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2171 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
2172 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
2173 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
2174 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
2175 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
2176 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
2177 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
2179 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2180 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2182 extern bool task_set_jobctl_pending(struct task_struct *task,
2184 extern void task_clear_jobctl_trapping(struct task_struct *task);
2185 extern void task_clear_jobctl_pending(struct task_struct *task,
2188 static inline void rcu_copy_process(struct task_struct *p)
2190 #ifdef CONFIG_PREEMPT_RCU
2191 p->rcu_read_lock_nesting = 0;
2192 p->rcu_read_unlock_special.s = 0;
2193 p->rcu_blocked_node = NULL;
2194 INIT_LIST_HEAD(&p->rcu_node_entry);
2195 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2196 #ifdef CONFIG_TASKS_RCU
2197 p->rcu_tasks_holdout = false;
2198 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2199 p->rcu_tasks_idle_cpu = -1;
2200 #endif /* #ifdef CONFIG_TASKS_RCU */
2203 static inline void tsk_restore_flags(struct task_struct *task,
2204 unsigned long orig_flags, unsigned long flags)
2206 task->flags &= ~flags;
2207 task->flags |= orig_flags & flags;
2210 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2211 const struct cpumask *trial);
2212 extern int task_can_attach(struct task_struct *p,
2213 const struct cpumask *cs_cpus_allowed);
2215 extern void do_set_cpus_allowed(struct task_struct *p,
2216 const struct cpumask *new_mask);
2218 extern int set_cpus_allowed_ptr(struct task_struct *p,
2219 const struct cpumask *new_mask);
2220 int migrate_me(void);
2221 void tell_sched_cpu_down_begin(int cpu);
2222 void tell_sched_cpu_down_done(int cpu);
2225 static inline void do_set_cpus_allowed(struct task_struct *p,
2226 const struct cpumask *new_mask)
2229 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2230 const struct cpumask *new_mask)
2232 if (!cpumask_test_cpu(0, new_mask))
2236 static inline int migrate_me(void) { return 0; }
2237 static inline void tell_sched_cpu_down_begin(int cpu) { }
2238 static inline void tell_sched_cpu_down_done(int cpu) { }
2241 #ifdef CONFIG_NO_HZ_COMMON
2242 void calc_load_enter_idle(void);
2243 void calc_load_exit_idle(void);
2245 static inline void calc_load_enter_idle(void) { }
2246 static inline void calc_load_exit_idle(void) { }
2247 #endif /* CONFIG_NO_HZ_COMMON */
2249 #ifndef CONFIG_CPUMASK_OFFSTACK
2250 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2252 return set_cpus_allowed_ptr(p, &new_mask);
2257 * Do not use outside of architecture code which knows its limitations.
2259 * sched_clock() has no promise of monotonicity or bounded drift between
2260 * CPUs, use (which you should not) requires disabling IRQs.
2262 * Please use one of the three interfaces below.
2264 extern unsigned long long notrace sched_clock(void);
2266 * See the comment in kernel/sched/clock.c
2268 extern u64 cpu_clock(int cpu);
2269 extern u64 local_clock(void);
2270 extern u64 running_clock(void);
2271 extern u64 sched_clock_cpu(int cpu);
2274 extern void sched_clock_init(void);
2276 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2277 static inline void sched_clock_tick(void)
2281 static inline void sched_clock_idle_sleep_event(void)
2285 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2290 * Architectures can set this to 1 if they have specified
2291 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2292 * but then during bootup it turns out that sched_clock()
2293 * is reliable after all:
2295 extern int sched_clock_stable(void);
2296 extern void set_sched_clock_stable(void);
2297 extern void clear_sched_clock_stable(void);
2299 extern void sched_clock_tick(void);
2300 extern void sched_clock_idle_sleep_event(void);
2301 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2304 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2306 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2307 * The reason for this explicit opt-in is not to have perf penalty with
2308 * slow sched_clocks.
2310 extern void enable_sched_clock_irqtime(void);
2311 extern void disable_sched_clock_irqtime(void);
2313 static inline void enable_sched_clock_irqtime(void) {}
2314 static inline void disable_sched_clock_irqtime(void) {}
2317 extern unsigned long long
2318 task_sched_runtime(struct task_struct *task);
2320 /* sched_exec is called by processes performing an exec */
2322 extern void sched_exec(void);
2324 #define sched_exec() {}
2327 extern void sched_clock_idle_sleep_event(void);
2328 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2330 #ifdef CONFIG_HOTPLUG_CPU
2331 extern void idle_task_exit(void);
2333 static inline void idle_task_exit(void) {}
2336 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2337 extern void wake_up_nohz_cpu(int cpu);
2339 static inline void wake_up_nohz_cpu(int cpu) { }
2342 #ifdef CONFIG_NO_HZ_FULL
2343 extern bool sched_can_stop_tick(void);
2344 extern u64 scheduler_tick_max_deferment(void);
2346 static inline bool sched_can_stop_tick(void) { return false; }
2349 #ifdef CONFIG_SCHED_AUTOGROUP
2350 extern void sched_autogroup_create_attach(struct task_struct *p);
2351 extern void sched_autogroup_detach(struct task_struct *p);
2352 extern void sched_autogroup_fork(struct signal_struct *sig);
2353 extern void sched_autogroup_exit(struct signal_struct *sig);
2354 #ifdef CONFIG_PROC_FS
2355 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2356 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2359 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2360 static inline void sched_autogroup_detach(struct task_struct *p) { }
2361 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2362 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2365 extern int yield_to(struct task_struct *p, bool preempt);
2366 extern void set_user_nice(struct task_struct *p, long nice);
2367 extern int task_prio(const struct task_struct *p);
2369 * task_nice - return the nice value of a given task.
2370 * @p: the task in question.
2372 * Return: The nice value [ -20 ... 0 ... 19 ].
2374 static inline int task_nice(const struct task_struct *p)
2376 return PRIO_TO_NICE((p)->static_prio);
2378 extern int can_nice(const struct task_struct *p, const int nice);
2379 extern int task_curr(const struct task_struct *p);
2380 extern int idle_cpu(int cpu);
2381 extern int sched_setscheduler(struct task_struct *, int,
2382 const struct sched_param *);
2383 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2384 const struct sched_param *);
2385 extern int sched_setattr(struct task_struct *,
2386 const struct sched_attr *);
2387 extern struct task_struct *idle_task(int cpu);
2389 * is_idle_task - is the specified task an idle task?
2390 * @p: the task in question.
2392 * Return: 1 if @p is an idle task. 0 otherwise.
2394 static inline bool is_idle_task(const struct task_struct *p)
2398 extern struct task_struct *curr_task(int cpu);
2399 extern void set_curr_task(int cpu, struct task_struct *p);
2403 union thread_union {
2404 struct thread_info thread_info;
2405 unsigned long stack[THREAD_SIZE/sizeof(long)];
2408 #ifndef __HAVE_ARCH_KSTACK_END
2409 static inline int kstack_end(void *addr)
2411 /* Reliable end of stack detection:
2412 * Some APM bios versions misalign the stack
2414 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2418 extern union thread_union init_thread_union;
2419 extern struct task_struct init_task;
2421 extern struct mm_struct init_mm;
2423 extern struct pid_namespace init_pid_ns;
2426 * find a task by one of its numerical ids
2428 * find_task_by_pid_ns():
2429 * finds a task by its pid in the specified namespace
2430 * find_task_by_vpid():
2431 * finds a task by its virtual pid
2433 * see also find_vpid() etc in include/linux/pid.h
2436 extern struct task_struct *find_task_by_vpid(pid_t nr);
2437 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2438 struct pid_namespace *ns);
2440 /* per-UID process charging. */
2441 extern struct user_struct * alloc_uid(kuid_t);
2442 static inline struct user_struct *get_uid(struct user_struct *u)
2444 atomic_inc(&u->__count);
2447 extern void free_uid(struct user_struct *);
2449 #include <asm/current.h>
2451 extern void xtime_update(unsigned long ticks);
2453 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2454 extern int wake_up_process(struct task_struct *tsk);
2455 extern int wake_up_lock_sleeper(struct task_struct * tsk);
2456 extern void wake_up_new_task(struct task_struct *tsk);
2458 extern void kick_process(struct task_struct *tsk);
2460 static inline void kick_process(struct task_struct *tsk) { }
2462 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2463 extern void sched_dead(struct task_struct *p);
2465 extern void proc_caches_init(void);
2466 extern void flush_signals(struct task_struct *);
2467 extern void __flush_signals(struct task_struct *);
2468 extern void ignore_signals(struct task_struct *);
2469 extern void flush_signal_handlers(struct task_struct *, int force_default);
2470 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2472 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2474 unsigned long flags;
2477 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2478 ret = dequeue_signal(tsk, mask, info);
2479 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2484 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2486 extern void unblock_all_signals(void);
2487 extern void release_task(struct task_struct * p);
2488 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2489 extern int force_sigsegv(int, struct task_struct *);
2490 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2491 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2492 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2493 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2494 const struct cred *, u32);
2495 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2496 extern int kill_pid(struct pid *pid, int sig, int priv);
2497 extern int kill_proc_info(int, struct siginfo *, pid_t);
2498 extern __must_check bool do_notify_parent(struct task_struct *, int);
2499 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2500 extern void force_sig(int, struct task_struct *);
2501 extern int send_sig(int, struct task_struct *, int);
2502 extern int zap_other_threads(struct task_struct *p);
2503 extern struct sigqueue *sigqueue_alloc(void);
2504 extern void sigqueue_free(struct sigqueue *);
2505 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2506 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2508 static inline void restore_saved_sigmask(void)
2510 if (test_and_clear_restore_sigmask())
2511 __set_current_blocked(¤t->saved_sigmask);
2514 static inline sigset_t *sigmask_to_save(void)
2516 sigset_t *res = ¤t->blocked;
2517 if (unlikely(test_restore_sigmask()))
2518 res = ¤t->saved_sigmask;
2522 static inline int kill_cad_pid(int sig, int priv)
2524 return kill_pid(cad_pid, sig, priv);
2527 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2528 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2529 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2530 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2533 * True if we are on the alternate signal stack.
2535 static inline int on_sig_stack(unsigned long sp)
2537 #ifdef CONFIG_STACK_GROWSUP
2538 return sp >= current->sas_ss_sp &&
2539 sp - current->sas_ss_sp < current->sas_ss_size;
2541 return sp > current->sas_ss_sp &&
2542 sp - current->sas_ss_sp <= current->sas_ss_size;
2546 static inline int sas_ss_flags(unsigned long sp)
2548 if (!current->sas_ss_size)
2551 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2554 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2556 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2557 #ifdef CONFIG_STACK_GROWSUP
2558 return current->sas_ss_sp;
2560 return current->sas_ss_sp + current->sas_ss_size;
2566 * Routines for handling mm_structs
2568 extern struct mm_struct * mm_alloc(void);
2570 /* mmdrop drops the mm and the page tables */
2571 extern void __mmdrop(struct mm_struct *);
2573 static inline void mmdrop(struct mm_struct * mm)
2575 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2579 #ifdef CONFIG_PREEMPT_RT_BASE
2580 extern void __mmdrop_delayed(struct rcu_head *rhp);
2581 static inline void mmdrop_delayed(struct mm_struct *mm)
2583 if (atomic_dec_and_test(&mm->mm_count))
2584 call_rcu(&mm->delayed_drop, __mmdrop_delayed);
2587 # define mmdrop_delayed(mm) mmdrop(mm)
2590 /* mmput gets rid of the mappings and all user-space */
2591 extern void mmput(struct mm_struct *);
2592 /* Grab a reference to a task's mm, if it is not already going away */
2593 extern struct mm_struct *get_task_mm(struct task_struct *task);
2595 * Grab a reference to a task's mm, if it is not already going away
2596 * and ptrace_may_access with the mode parameter passed to it
2599 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2600 /* Remove the current tasks stale references to the old mm_struct */
2601 extern void mm_release(struct task_struct *, struct mm_struct *);
2603 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2604 struct task_struct *);
2605 extern void flush_thread(void);
2606 extern void exit_thread(void);
2608 extern void exit_files(struct task_struct *);
2609 extern void __cleanup_sighand(struct sighand_struct *);
2611 extern void exit_itimers(struct signal_struct *);
2612 extern void flush_itimer_signals(void);
2614 extern void do_group_exit(int);
2616 extern int do_execve(struct filename *,
2617 const char __user * const __user *,
2618 const char __user * const __user *);
2619 extern int do_execveat(int, struct filename *,
2620 const char __user * const __user *,
2621 const char __user * const __user *,
2623 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2624 struct task_struct *fork_idle(int);
2625 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2627 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2628 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2630 __set_task_comm(tsk, from, false);
2632 extern char *get_task_comm(char *to, struct task_struct *tsk);
2635 void scheduler_ipi(void);
2636 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2638 static inline void scheduler_ipi(void) { }
2639 static inline unsigned long wait_task_inactive(struct task_struct *p,
2646 #define next_task(p) \
2647 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2649 #define for_each_process(p) \
2650 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2652 extern bool current_is_single_threaded(void);
2655 * Careful: do_each_thread/while_each_thread is a double loop so
2656 * 'break' will not work as expected - use goto instead.
2658 #define do_each_thread(g, t) \
2659 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2661 #define while_each_thread(g, t) \
2662 while ((t = next_thread(t)) != g)
2664 #define __for_each_thread(signal, t) \
2665 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2667 #define for_each_thread(p, t) \
2668 __for_each_thread((p)->signal, t)
2670 /* Careful: this is a double loop, 'break' won't work as expected. */
2671 #define for_each_process_thread(p, t) \
2672 for_each_process(p) for_each_thread(p, t)
2674 static inline int get_nr_threads(struct task_struct *tsk)
2676 return tsk->signal->nr_threads;
2679 static inline bool thread_group_leader(struct task_struct *p)
2681 return p->exit_signal >= 0;
2684 /* Do to the insanities of de_thread it is possible for a process
2685 * to have the pid of the thread group leader without actually being
2686 * the thread group leader. For iteration through the pids in proc
2687 * all we care about is that we have a task with the appropriate
2688 * pid, we don't actually care if we have the right task.
2690 static inline bool has_group_leader_pid(struct task_struct *p)
2692 return task_pid(p) == p->signal->leader_pid;
2696 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2698 return p1->signal == p2->signal;
2701 static inline struct task_struct *next_thread(const struct task_struct *p)
2703 return list_entry_rcu(p->thread_group.next,
2704 struct task_struct, thread_group);
2707 static inline int thread_group_empty(struct task_struct *p)
2709 return list_empty(&p->thread_group);
2712 #define delay_group_leader(p) \
2713 (thread_group_leader(p) && !thread_group_empty(p))
2716 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2717 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2718 * pins the final release of task.io_context. Also protects ->cpuset and
2719 * ->cgroup.subsys[]. And ->vfork_done.
2721 * Nests both inside and outside of read_lock(&tasklist_lock).
2722 * It must not be nested with write_lock_irq(&tasklist_lock),
2723 * neither inside nor outside.
2725 static inline void task_lock(struct task_struct *p)
2727 spin_lock(&p->alloc_lock);
2730 static inline void task_unlock(struct task_struct *p)
2732 spin_unlock(&p->alloc_lock);
2735 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2736 unsigned long *flags);
2738 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2739 unsigned long *flags)
2741 struct sighand_struct *ret;
2743 ret = __lock_task_sighand(tsk, flags);
2744 (void)__cond_lock(&tsk->sighand->siglock, ret);
2748 static inline void unlock_task_sighand(struct task_struct *tsk,
2749 unsigned long *flags)
2751 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2754 #ifdef CONFIG_CGROUPS
2755 static inline void threadgroup_change_begin(struct task_struct *tsk)
2757 down_read(&tsk->signal->group_rwsem);
2759 static inline void threadgroup_change_end(struct task_struct *tsk)
2761 up_read(&tsk->signal->group_rwsem);
2765 * threadgroup_lock - lock threadgroup
2766 * @tsk: member task of the threadgroup to lock
2768 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2769 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2770 * change ->group_leader/pid. This is useful for cases where the threadgroup
2771 * needs to stay stable across blockable operations.
2773 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2774 * synchronization. While held, no new task will be added to threadgroup
2775 * and no existing live task will have its PF_EXITING set.
2777 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2778 * sub-thread becomes a new leader.
2780 static inline void threadgroup_lock(struct task_struct *tsk)
2782 down_write(&tsk->signal->group_rwsem);
2786 * threadgroup_unlock - unlock threadgroup
2787 * @tsk: member task of the threadgroup to unlock
2789 * Reverse threadgroup_lock().
2791 static inline void threadgroup_unlock(struct task_struct *tsk)
2793 up_write(&tsk->signal->group_rwsem);
2796 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2797 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2798 static inline void threadgroup_lock(struct task_struct *tsk) {}
2799 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2802 #ifndef __HAVE_THREAD_FUNCTIONS
2804 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2805 #define task_stack_page(task) ((task)->stack)
2807 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2809 *task_thread_info(p) = *task_thread_info(org);
2810 task_thread_info(p)->task = p;
2814 * Return the address of the last usable long on the stack.
2816 * When the stack grows down, this is just above the thread
2817 * info struct. Going any lower will corrupt the threadinfo.
2819 * When the stack grows up, this is the highest address.
2820 * Beyond that position, we corrupt data on the next page.
2822 static inline unsigned long *end_of_stack(struct task_struct *p)
2824 #ifdef CONFIG_STACK_GROWSUP
2825 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2827 return (unsigned long *)(task_thread_info(p) + 1);
2832 #define task_stack_end_corrupted(task) \
2833 (*(end_of_stack(task)) != STACK_END_MAGIC)
2835 static inline int object_is_on_stack(void *obj)
2837 void *stack = task_stack_page(current);
2839 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2842 extern void thread_info_cache_init(void);
2844 #ifdef CONFIG_DEBUG_STACK_USAGE
2845 static inline unsigned long stack_not_used(struct task_struct *p)
2847 unsigned long *n = end_of_stack(p);
2849 do { /* Skip over canary */
2853 return (unsigned long)n - (unsigned long)end_of_stack(p);
2856 extern void set_task_stack_end_magic(struct task_struct *tsk);
2858 /* set thread flags in other task's structures
2859 * - see asm/thread_info.h for TIF_xxxx flags available
2861 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2863 set_ti_thread_flag(task_thread_info(tsk), flag);
2866 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2868 clear_ti_thread_flag(task_thread_info(tsk), flag);
2871 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2873 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2876 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2878 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2881 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2883 return test_ti_thread_flag(task_thread_info(tsk), flag);
2886 static inline void set_tsk_need_resched(struct task_struct *tsk)
2888 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2891 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2893 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2896 static inline int test_tsk_need_resched(struct task_struct *tsk)
2898 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2901 #ifdef CONFIG_PREEMPT_LAZY
2902 static inline void set_tsk_need_resched_lazy(struct task_struct *tsk)
2904 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY);
2907 static inline void clear_tsk_need_resched_lazy(struct task_struct *tsk)
2909 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY);
2912 static inline int test_tsk_need_resched_lazy(struct task_struct *tsk)
2914 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY));
2917 static inline int need_resched_lazy(void)
2919 return test_thread_flag(TIF_NEED_RESCHED_LAZY);
2922 static inline int need_resched_now(void)
2924 return test_thread_flag(TIF_NEED_RESCHED);
2928 static inline void clear_tsk_need_resched_lazy(struct task_struct *tsk) { }
2929 static inline int need_resched_lazy(void) { return 0; }
2931 static inline int need_resched_now(void)
2933 return test_thread_flag(TIF_NEED_RESCHED);
2938 static inline int restart_syscall(void)
2940 set_tsk_thread_flag(current, TIF_SIGPENDING);
2941 return -ERESTARTNOINTR;
2944 static inline int signal_pending(struct task_struct *p)
2946 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2949 static inline int __fatal_signal_pending(struct task_struct *p)
2951 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2954 static inline int fatal_signal_pending(struct task_struct *p)
2956 return signal_pending(p) && __fatal_signal_pending(p);
2959 static inline int signal_pending_state(long state, struct task_struct *p)
2961 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2963 if (!signal_pending(p))
2966 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2969 static inline bool __task_is_stopped_or_traced(struct task_struct *task)
2971 if (task->state & (__TASK_STOPPED | __TASK_TRACED))
2973 #ifdef CONFIG_PREEMPT_RT_FULL
2974 if (task->saved_state & (__TASK_STOPPED | __TASK_TRACED))
2980 static inline bool task_is_stopped_or_traced(struct task_struct *task)
2982 bool traced_stopped;
2984 #ifdef CONFIG_PREEMPT_RT_FULL
2985 unsigned long flags;
2987 raw_spin_lock_irqsave(&task->pi_lock, flags);
2988 traced_stopped = __task_is_stopped_or_traced(task);
2989 raw_spin_unlock_irqrestore(&task->pi_lock, flags);
2991 traced_stopped = __task_is_stopped_or_traced(task);
2993 return traced_stopped;
2996 static inline bool task_is_traced(struct task_struct *task)
2998 bool traced = false;
3000 if (task->state & __TASK_TRACED)
3002 #ifdef CONFIG_PREEMPT_RT_FULL
3003 /* in case the task is sleeping on tasklist_lock */
3004 raw_spin_lock_irq(&task->pi_lock);
3005 if (task->state & __TASK_TRACED)
3007 else if (task->saved_state & __TASK_TRACED)
3009 raw_spin_unlock_irq(&task->pi_lock);
3015 * cond_resched() and cond_resched_lock(): latency reduction via
3016 * explicit rescheduling in places that are safe. The return
3017 * value indicates whether a reschedule was done in fact.
3018 * cond_resched_lock() will drop the spinlock before scheduling,
3019 * cond_resched_softirq() will enable bhs before scheduling.
3021 extern int _cond_resched(void);
3023 #define cond_resched() ({ \
3024 ___might_sleep(__FILE__, __LINE__, 0); \
3028 extern int __cond_resched_lock(spinlock_t *lock);
3030 #if defined(CONFIG_PREEMPT_COUNT) && !defined(CONFIG_PREEMPT_RT_FULL)
3031 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
3033 #define PREEMPT_LOCK_OFFSET 0
3036 #define cond_resched_lock(lock) ({ \
3037 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3038 __cond_resched_lock(lock); \
3041 #ifndef CONFIG_PREEMPT_RT_FULL
3042 extern int __cond_resched_softirq(void);
3044 #define cond_resched_softirq() ({ \
3045 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3046 __cond_resched_softirq(); \
3049 # define cond_resched_softirq() cond_resched()
3052 static inline void cond_resched_rcu(void)
3054 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3062 * Does a critical section need to be broken due to another
3063 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3064 * but a general need for low latency)
3066 static inline int spin_needbreak(spinlock_t *lock)
3068 #ifdef CONFIG_PREEMPT
3069 return spin_is_contended(lock);
3076 * Idle thread specific functions to determine the need_resched
3079 #ifdef TIF_POLLING_NRFLAG
3080 static inline int tsk_is_polling(struct task_struct *p)
3082 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3085 static inline void __current_set_polling(void)
3087 set_thread_flag(TIF_POLLING_NRFLAG);
3090 static inline bool __must_check current_set_polling_and_test(void)
3092 __current_set_polling();
3095 * Polling state must be visible before we test NEED_RESCHED,
3096 * paired by resched_curr()
3098 smp_mb__after_atomic();
3100 return unlikely(tif_need_resched());
3103 static inline void __current_clr_polling(void)
3105 clear_thread_flag(TIF_POLLING_NRFLAG);
3108 static inline bool __must_check current_clr_polling_and_test(void)
3110 __current_clr_polling();
3113 * Polling state must be visible before we test NEED_RESCHED,
3114 * paired by resched_curr()
3116 smp_mb__after_atomic();
3118 return unlikely(tif_need_resched());
3122 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3123 static inline void __current_set_polling(void) { }
3124 static inline void __current_clr_polling(void) { }
3126 static inline bool __must_check current_set_polling_and_test(void)
3128 return unlikely(tif_need_resched());
3130 static inline bool __must_check current_clr_polling_and_test(void)
3132 return unlikely(tif_need_resched());
3136 static inline void current_clr_polling(void)
3138 __current_clr_polling();
3141 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3142 * Once the bit is cleared, we'll get IPIs with every new
3143 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3146 smp_mb(); /* paired with resched_curr() */
3148 preempt_fold_need_resched();
3151 static __always_inline bool need_resched(void)
3153 return unlikely(tif_need_resched());
3157 * Thread group CPU time accounting.
3159 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3160 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3162 static inline void thread_group_cputime_init(struct signal_struct *sig)
3164 raw_spin_lock_init(&sig->cputimer.lock);
3168 * Reevaluate whether the task has signals pending delivery.
3169 * Wake the task if so.
3170 * This is required every time the blocked sigset_t changes.
3171 * callers must hold sighand->siglock.
3173 extern void recalc_sigpending_and_wake(struct task_struct *t);
3174 extern void recalc_sigpending(void);
3176 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3178 static inline void signal_wake_up(struct task_struct *t, bool resume)
3180 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3182 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3184 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3188 * Wrappers for p->thread_info->cpu access. No-op on UP.
3192 static inline unsigned int task_cpu(const struct task_struct *p)
3194 return task_thread_info(p)->cpu;
3197 static inline int task_node(const struct task_struct *p)
3199 return cpu_to_node(task_cpu(p));
3202 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3206 static inline unsigned int task_cpu(const struct task_struct *p)
3211 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3215 #endif /* CONFIG_SMP */
3217 static inline int __migrate_disabled(struct task_struct *p)
3219 #ifdef CONFIG_PREEMPT_RT_FULL
3220 return p->migrate_disable;
3226 /* Future-safe accessor for struct task_struct's cpus_allowed. */
3227 static inline const struct cpumask *tsk_cpus_allowed(struct task_struct *p)
3229 #ifdef CONFIG_PREEMPT_RT_FULL
3230 if (p->migrate_disable)
3231 return cpumask_of(task_cpu(p));
3234 return &p->cpus_allowed;
3237 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3238 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3240 #ifdef CONFIG_CGROUP_SCHED
3241 extern struct task_group root_task_group;
3242 #endif /* CONFIG_CGROUP_SCHED */
3244 extern int task_can_switch_user(struct user_struct *up,
3245 struct task_struct *tsk);
3247 #ifdef CONFIG_TASK_XACCT
3248 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3250 tsk->ioac.rchar += amt;
3253 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3255 tsk->ioac.wchar += amt;
3258 static inline void inc_syscr(struct task_struct *tsk)
3263 static inline void inc_syscw(struct task_struct *tsk)
3268 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3272 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3276 static inline void inc_syscr(struct task_struct *tsk)
3280 static inline void inc_syscw(struct task_struct *tsk)
3285 #ifndef TASK_SIZE_OF
3286 #define TASK_SIZE_OF(tsk) TASK_SIZE
3290 extern void mm_update_next_owner(struct mm_struct *mm);
3292 static inline void mm_update_next_owner(struct mm_struct *mm)
3295 #endif /* CONFIG_MEMCG */
3297 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3300 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
3303 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3306 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
3309 static inline unsigned long rlimit(unsigned int limit)
3311 return task_rlimit(current, limit);
3314 static inline unsigned long rlimit_max(unsigned int limit)
3316 return task_rlimit_max(current, limit);