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 */
130 struct futex_pi_state;
131 struct robust_list_head;
134 struct perf_event_context;
138 #define VMACACHE_BITS 2
139 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
140 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
143 * These are the constant used to fake the fixed-point load-average
144 * counting. Some notes:
145 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
146 * a load-average precision of 10 bits integer + 11 bits fractional
147 * - if you want to count load-averages more often, you need more
148 * precision, or rounding will get you. With 2-second counting freq,
149 * the EXP_n values would be 1981, 2034 and 2043 if still using only
152 extern unsigned long avenrun[]; /* Load averages */
153 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
155 #define FSHIFT 11 /* nr of bits of precision */
156 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
157 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
158 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
159 #define EXP_5 2014 /* 1/exp(5sec/5min) */
160 #define EXP_15 2037 /* 1/exp(5sec/15min) */
162 #define CALC_LOAD(load,exp,n) \
164 load += n*(FIXED_1-exp); \
167 extern unsigned long total_forks;
168 extern int nr_threads;
169 DECLARE_PER_CPU(unsigned long, process_counts);
170 extern int nr_processes(void);
171 extern unsigned long nr_running(void);
172 extern bool single_task_running(void);
173 extern unsigned long nr_iowait(void);
174 extern unsigned long nr_iowait_cpu(int cpu);
175 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
177 extern void calc_global_load(unsigned long ticks);
178 extern void update_cpu_load_nohz(void);
180 /* Notifier for when a task gets migrated to a new CPU */
181 struct task_migration_notifier {
182 struct task_struct *task;
186 extern void register_task_migration_notifier(struct notifier_block *n);
188 extern unsigned long get_parent_ip(unsigned long addr);
190 extern void dump_cpu_task(int cpu);
195 #ifdef CONFIG_SCHED_DEBUG
196 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
197 extern void proc_sched_set_task(struct task_struct *p);
199 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
203 * Task state bitmask. NOTE! These bits are also
204 * encoded in fs/proc/array.c: get_task_state().
206 * We have two separate sets of flags: task->state
207 * is about runnability, while task->exit_state are
208 * about the task exiting. Confusing, but this way
209 * modifying one set can't modify the other one by
212 #define TASK_RUNNING 0
213 #define TASK_INTERRUPTIBLE 1
214 #define TASK_UNINTERRUPTIBLE 2
215 #define __TASK_STOPPED 4
216 #define __TASK_TRACED 8
217 /* in tsk->exit_state */
219 #define EXIT_ZOMBIE 32
220 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
221 /* in tsk->state again */
223 #define TASK_WAKEKILL 128
224 #define TASK_WAKING 256
225 #define TASK_PARKED 512
226 #define TASK_STATE_MAX 1024
228 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
230 extern char ___assert_task_state[1 - 2*!!(
231 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
233 /* Convenience macros for the sake of set_task_state */
234 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
235 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
236 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
238 /* Convenience macros for the sake of wake_up */
239 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
240 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
242 /* get_task_state() */
243 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
244 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
245 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
247 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
248 #define task_contributes_to_load(task) \
249 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
250 (task->flags & PF_FROZEN) == 0)
252 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
254 #define __set_task_state(tsk, state_value) \
256 (tsk)->task_state_change = _THIS_IP_; \
257 (tsk)->state = (state_value); \
259 #define set_task_state(tsk, state_value) \
261 (tsk)->task_state_change = _THIS_IP_; \
262 set_mb((tsk)->state, (state_value)); \
266 * set_current_state() includes a barrier so that the write of current->state
267 * is correctly serialised wrt the caller's subsequent test of whether to
270 * set_current_state(TASK_UNINTERRUPTIBLE);
271 * if (do_i_need_to_sleep())
274 * If the caller does not need such serialisation then use __set_current_state()
276 #define __set_current_state(state_value) \
278 current->task_state_change = _THIS_IP_; \
279 current->state = (state_value); \
281 #define set_current_state(state_value) \
283 current->task_state_change = _THIS_IP_; \
284 set_mb(current->state, (state_value)); \
289 #define __set_task_state(tsk, state_value) \
290 do { (tsk)->state = (state_value); } while (0)
291 #define set_task_state(tsk, state_value) \
292 set_mb((tsk)->state, (state_value))
295 * set_current_state() includes a barrier so that the write of current->state
296 * is correctly serialised wrt the caller's subsequent test of whether to
299 * set_current_state(TASK_UNINTERRUPTIBLE);
300 * if (do_i_need_to_sleep())
303 * If the caller does not need such serialisation then use __set_current_state()
305 #define __set_current_state(state_value) \
306 do { current->state = (state_value); } while (0)
307 #define set_current_state(state_value) \
308 set_mb(current->state, (state_value))
312 #define __set_current_state_no_track(state_value) \
313 do { current->state = (state_value); } while (0)
314 #define set_current_state_no_track(state_value) \
315 set_mb(current->state, (state_value))
317 /* Task command name length */
318 #define TASK_COMM_LEN 16
320 #include <linux/spinlock.h>
323 * This serializes "schedule()" and also protects
324 * the run-queue from deletions/modifications (but
325 * _adding_ to the beginning of the run-queue has
328 extern rwlock_t tasklist_lock;
329 extern spinlock_t mmlist_lock;
333 #ifdef CONFIG_PROVE_RCU
334 extern int lockdep_tasklist_lock_is_held(void);
335 #endif /* #ifdef CONFIG_PROVE_RCU */
337 extern void sched_init(void);
338 extern void sched_init_smp(void);
339 extern asmlinkage void schedule_tail(struct task_struct *prev);
340 extern void init_idle(struct task_struct *idle, int cpu);
341 extern void init_idle_bootup_task(struct task_struct *idle);
343 extern int runqueue_is_locked(int cpu);
345 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
346 extern void nohz_balance_enter_idle(int cpu);
347 extern void set_cpu_sd_state_idle(void);
348 extern int get_nohz_timer_target(int pinned);
350 static inline void nohz_balance_enter_idle(int cpu) { }
351 static inline void set_cpu_sd_state_idle(void) { }
352 static inline int get_nohz_timer_target(int pinned)
354 return smp_processor_id();
359 * Only dump TASK_* tasks. (0 for all tasks)
361 extern void show_state_filter(unsigned long state_filter);
363 static inline void show_state(void)
365 show_state_filter(0);
368 extern void show_regs(struct pt_regs *);
371 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
372 * task), SP is the stack pointer of the first frame that should be shown in the back
373 * trace (or NULL if the entire call-chain of the task should be shown).
375 extern void show_stack(struct task_struct *task, unsigned long *sp);
377 extern void cpu_init (void);
378 extern void trap_init(void);
379 extern void update_process_times(int user);
380 extern void scheduler_tick(void);
382 extern void sched_show_task(struct task_struct *p);
384 #ifdef CONFIG_LOCKUP_DETECTOR
385 extern void touch_softlockup_watchdog(void);
386 extern void touch_softlockup_watchdog_sync(void);
387 extern void touch_all_softlockup_watchdogs(void);
388 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
390 size_t *lenp, loff_t *ppos);
391 extern unsigned int softlockup_panic;
392 void lockup_detector_init(void);
394 static inline void touch_softlockup_watchdog(void)
397 static inline void touch_softlockup_watchdog_sync(void)
400 static inline void touch_all_softlockup_watchdogs(void)
403 static inline void lockup_detector_init(void)
408 #ifdef CONFIG_DETECT_HUNG_TASK
409 void reset_hung_task_detector(void);
411 static inline void reset_hung_task_detector(void)
416 /* Attach to any functions which should be ignored in wchan output. */
417 #define __sched __attribute__((__section__(".sched.text")))
419 /* Linker adds these: start and end of __sched functions */
420 extern char __sched_text_start[], __sched_text_end[];
422 /* Is this address in the __sched functions? */
423 extern int in_sched_functions(unsigned long addr);
425 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
426 extern signed long schedule_timeout(signed long timeout);
427 extern signed long schedule_timeout_interruptible(signed long timeout);
428 extern signed long schedule_timeout_killable(signed long timeout);
429 extern signed long schedule_timeout_uninterruptible(signed long timeout);
430 asmlinkage void schedule(void);
431 extern void schedule_preempt_disabled(void);
433 extern long io_schedule_timeout(long timeout);
435 static inline void io_schedule(void)
437 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
441 struct user_namespace;
444 extern void arch_pick_mmap_layout(struct mm_struct *mm);
446 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
447 unsigned long, unsigned long);
449 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
450 unsigned long len, unsigned long pgoff,
451 unsigned long flags);
453 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
456 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
457 #define SUID_DUMP_USER 1 /* Dump as user of process */
458 #define SUID_DUMP_ROOT 2 /* Dump as root */
462 /* for SUID_DUMP_* above */
463 #define MMF_DUMPABLE_BITS 2
464 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
466 extern void set_dumpable(struct mm_struct *mm, int value);
468 * This returns the actual value of the suid_dumpable flag. For things
469 * that are using this for checking for privilege transitions, it must
470 * test against SUID_DUMP_USER rather than treating it as a boolean
473 static inline int __get_dumpable(unsigned long mm_flags)
475 return mm_flags & MMF_DUMPABLE_MASK;
478 static inline int get_dumpable(struct mm_struct *mm)
480 return __get_dumpable(mm->flags);
483 /* coredump filter bits */
484 #define MMF_DUMP_ANON_PRIVATE 2
485 #define MMF_DUMP_ANON_SHARED 3
486 #define MMF_DUMP_MAPPED_PRIVATE 4
487 #define MMF_DUMP_MAPPED_SHARED 5
488 #define MMF_DUMP_ELF_HEADERS 6
489 #define MMF_DUMP_HUGETLB_PRIVATE 7
490 #define MMF_DUMP_HUGETLB_SHARED 8
492 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
493 #define MMF_DUMP_FILTER_BITS 7
494 #define MMF_DUMP_FILTER_MASK \
495 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
496 #define MMF_DUMP_FILTER_DEFAULT \
497 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
498 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
500 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
501 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
503 # define MMF_DUMP_MASK_DEFAULT_ELF 0
505 /* leave room for more dump flags */
506 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
507 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
508 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
510 #define MMF_HAS_UPROBES 19 /* has uprobes */
511 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
513 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
515 struct sighand_struct {
517 struct k_sigaction action[_NSIG];
519 wait_queue_head_t signalfd_wqh;
522 struct pacct_struct {
525 unsigned long ac_mem;
526 cputime_t ac_utime, ac_stime;
527 unsigned long ac_minflt, ac_majflt;
538 * struct cputime - snaphsot of system and user cputime
539 * @utime: time spent in user mode
540 * @stime: time spent in system mode
542 * Gathers a generic snapshot of user and system time.
550 * struct task_cputime - collected CPU time counts
551 * @utime: time spent in user mode, in &cputime_t units
552 * @stime: time spent in kernel mode, in &cputime_t units
553 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
555 * This is an extension of struct cputime that includes the total runtime
556 * spent by the task from the scheduler point of view.
558 * As a result, this structure groups together three kinds of CPU time
559 * that are tracked for threads and thread groups. Most things considering
560 * CPU time want to group these counts together and treat all three
561 * of them in parallel.
563 struct task_cputime {
566 unsigned long long sum_exec_runtime;
568 /* Alternate field names when used to cache expirations. */
569 #define prof_exp stime
570 #define virt_exp utime
571 #define sched_exp sum_exec_runtime
573 #define INIT_CPUTIME \
574 (struct task_cputime) { \
577 .sum_exec_runtime = 0, \
580 #ifdef CONFIG_PREEMPT_COUNT
581 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
583 #define PREEMPT_DISABLED PREEMPT_ENABLED
587 * Disable preemption until the scheduler is running.
588 * Reset by start_kernel()->sched_init()->init_idle().
590 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
591 * before the scheduler is active -- see should_resched().
593 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
596 * struct thread_group_cputimer - thread group interval timer counts
597 * @cputime: thread group interval timers.
598 * @running: non-zero when there are timers running and
599 * @cputime receives updates.
600 * @lock: lock for fields in this struct.
602 * This structure contains the version of task_cputime, above, that is
603 * used for thread group CPU timer calculations.
605 struct thread_group_cputimer {
606 struct task_cputime cputime;
611 #include <linux/rwsem.h>
615 * NOTE! "signal_struct" does not have its own
616 * locking, because a shared signal_struct always
617 * implies a shared sighand_struct, so locking
618 * sighand_struct is always a proper superset of
619 * the locking of signal_struct.
621 struct signal_struct {
625 struct list_head thread_head;
627 wait_queue_head_t wait_chldexit; /* for wait4() */
629 /* current thread group signal load-balancing target: */
630 struct task_struct *curr_target;
632 /* shared signal handling: */
633 struct sigpending shared_pending;
635 /* thread group exit support */
638 * - notify group_exit_task when ->count is equal to notify_count
639 * - everyone except group_exit_task is stopped during signal delivery
640 * of fatal signals, group_exit_task processes the signal.
643 struct task_struct *group_exit_task;
645 /* thread group stop support, overloads group_exit_code too */
646 int group_stop_count;
647 unsigned int flags; /* see SIGNAL_* flags below */
650 * PR_SET_CHILD_SUBREAPER marks a process, like a service
651 * manager, to re-parent orphan (double-forking) child processes
652 * to this process instead of 'init'. The service manager is
653 * able to receive SIGCHLD signals and is able to investigate
654 * the process until it calls wait(). All children of this
655 * process will inherit a flag if they should look for a
656 * child_subreaper process at exit.
658 unsigned int is_child_subreaper:1;
659 unsigned int has_child_subreaper:1;
661 /* POSIX.1b Interval Timers */
663 struct list_head posix_timers;
665 /* ITIMER_REAL timer for the process */
666 struct hrtimer real_timer;
667 struct pid *leader_pid;
668 ktime_t it_real_incr;
671 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
672 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
673 * values are defined to 0 and 1 respectively
675 struct cpu_itimer it[2];
678 * Thread group totals for process CPU timers.
679 * See thread_group_cputimer(), et al, for details.
681 struct thread_group_cputimer cputimer;
683 /* Earliest-expiration cache. */
684 struct task_cputime cputime_expires;
686 struct list_head cpu_timers[3];
688 struct pid *tty_old_pgrp;
690 /* boolean value for session group leader */
693 struct tty_struct *tty; /* NULL if no tty */
695 #ifdef CONFIG_SCHED_AUTOGROUP
696 struct autogroup *autogroup;
699 * Cumulative resource counters for dead threads in the group,
700 * and for reaped dead child processes forked by this group.
701 * Live threads maintain their own counters and add to these
702 * in __exit_signal, except for the group leader.
704 seqlock_t stats_lock;
705 cputime_t utime, stime, cutime, cstime;
708 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
709 struct cputime prev_cputime;
711 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
712 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
713 unsigned long inblock, oublock, cinblock, coublock;
714 unsigned long maxrss, cmaxrss;
715 struct task_io_accounting ioac;
718 * Cumulative ns of schedule CPU time fo dead threads in the
719 * group, not including a zombie group leader, (This only differs
720 * from jiffies_to_ns(utime + stime) if sched_clock uses something
721 * other than jiffies.)
723 unsigned long long sum_sched_runtime;
726 * We don't bother to synchronize most readers of this at all,
727 * because there is no reader checking a limit that actually needs
728 * to get both rlim_cur and rlim_max atomically, and either one
729 * alone is a single word that can safely be read normally.
730 * getrlimit/setrlimit use task_lock(current->group_leader) to
731 * protect this instead of the siglock, because they really
732 * have no need to disable irqs.
734 struct rlimit rlim[RLIM_NLIMITS];
736 #ifdef CONFIG_BSD_PROCESS_ACCT
737 struct pacct_struct pacct; /* per-process accounting information */
739 #ifdef CONFIG_TASKSTATS
740 struct taskstats *stats;
744 unsigned audit_tty_log_passwd;
745 struct tty_audit_buf *tty_audit_buf;
747 #ifdef CONFIG_CGROUPS
749 * group_rwsem prevents new tasks from entering the threadgroup and
750 * member tasks from exiting,a more specifically, setting of
751 * PF_EXITING. fork and exit paths are protected with this rwsem
752 * using threadgroup_change_begin/end(). Users which require
753 * threadgroup to remain stable should use threadgroup_[un]lock()
754 * which also takes care of exec path. Currently, cgroup is the
757 struct rw_semaphore group_rwsem;
760 oom_flags_t oom_flags;
761 short oom_score_adj; /* OOM kill score adjustment */
762 short oom_score_adj_min; /* OOM kill score adjustment min value.
763 * Only settable by CAP_SYS_RESOURCE. */
765 struct mutex cred_guard_mutex; /* guard against foreign influences on
766 * credential calculations
767 * (notably. ptrace) */
771 * Bits in flags field of signal_struct.
773 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
774 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
775 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
776 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
778 * Pending notifications to parent.
780 #define SIGNAL_CLD_STOPPED 0x00000010
781 #define SIGNAL_CLD_CONTINUED 0x00000020
782 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
784 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
786 /* If true, all threads except ->group_exit_task have pending SIGKILL */
787 static inline int signal_group_exit(const struct signal_struct *sig)
789 return (sig->flags & SIGNAL_GROUP_EXIT) ||
790 (sig->group_exit_task != NULL);
794 * Some day this will be a full-fledged user tracking system..
797 atomic_t __count; /* reference count */
798 atomic_t processes; /* How many processes does this user have? */
799 atomic_t sigpending; /* How many pending signals does this user have? */
800 #ifdef CONFIG_INOTIFY_USER
801 atomic_t inotify_watches; /* How many inotify watches does this user have? */
802 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
804 #ifdef CONFIG_FANOTIFY
805 atomic_t fanotify_listeners;
808 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
810 #ifdef CONFIG_POSIX_MQUEUE
811 /* protected by mq_lock */
812 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
814 unsigned long locked_shm; /* How many pages of mlocked shm ? */
817 struct key *uid_keyring; /* UID specific keyring */
818 struct key *session_keyring; /* UID's default session keyring */
821 /* Hash table maintenance information */
822 struct hlist_node uidhash_node;
825 #ifdef CONFIG_PERF_EVENTS
826 atomic_long_t locked_vm;
830 extern int uids_sysfs_init(void);
832 extern struct user_struct *find_user(kuid_t);
834 extern struct user_struct root_user;
835 #define INIT_USER (&root_user)
838 struct backing_dev_info;
839 struct reclaim_state;
841 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
843 /* cumulative counters */
844 unsigned long pcount; /* # of times run on this cpu */
845 unsigned long long run_delay; /* time spent waiting on a runqueue */
848 unsigned long long last_arrival,/* when we last ran on a cpu */
849 last_queued; /* when we were last queued to run */
851 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
853 #ifdef CONFIG_TASK_DELAY_ACCT
854 struct task_delay_info {
856 unsigned int flags; /* Private per-task flags */
858 /* For each stat XXX, add following, aligned appropriately
860 * struct timespec XXX_start, XXX_end;
864 * Atomicity of updates to XXX_delay, XXX_count protected by
865 * single lock above (split into XXX_lock if contention is an issue).
869 * XXX_count is incremented on every XXX operation, the delay
870 * associated with the operation is added to XXX_delay.
871 * XXX_delay contains the accumulated delay time in nanoseconds.
873 u64 blkio_start; /* Shared by blkio, swapin */
874 u64 blkio_delay; /* wait for sync block io completion */
875 u64 swapin_delay; /* wait for swapin block io completion */
876 u32 blkio_count; /* total count of the number of sync block */
877 /* io operations performed */
878 u32 swapin_count; /* total count of the number of swapin block */
879 /* io operations performed */
882 u64 freepages_delay; /* wait for memory reclaim */
883 u32 freepages_count; /* total count of memory reclaim */
885 #endif /* CONFIG_TASK_DELAY_ACCT */
887 static inline int sched_info_on(void)
889 #ifdef CONFIG_SCHEDSTATS
891 #elif defined(CONFIG_TASK_DELAY_ACCT)
892 extern int delayacct_on;
907 * Increase resolution of cpu_capacity calculations
909 #define SCHED_CAPACITY_SHIFT 10
910 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
913 * Wake-queues are lists of tasks with a pending wakeup, whose
914 * callers have already marked the task as woken internally,
915 * and can thus carry on. A common use case is being able to
916 * do the wakeups once the corresponding user lock as been
919 * We hold reference to each task in the list across the wakeup,
920 * thus guaranteeing that the memory is still valid by the time
921 * the actual wakeups are performed in wake_up_q().
923 * One per task suffices, because there's never a need for a task to be
924 * in two wake queues simultaneously; it is forbidden to abandon a task
925 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
926 * already in a wake queue, the wakeup will happen soon and the second
927 * waker can just skip it.
929 * The WAKE_Q macro declares and initializes the list head.
930 * wake_up_q() does NOT reinitialize the list; it's expected to be
931 * called near the end of a function, where the fact that the queue is
932 * not used again will be easy to see by inspection.
934 * Note that this can cause spurious wakeups. schedule() callers
935 * must ensure the call is done inside a loop, confirming that the
936 * wakeup condition has in fact occurred.
939 struct wake_q_node *next;
943 struct wake_q_node *first;
944 struct wake_q_node **lastp;
947 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
949 #define WAKE_Q(name) \
950 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
952 extern void wake_q_add(struct wake_q_head *head,
953 struct task_struct *task);
954 extern void wake_up_q(struct wake_q_head *head);
957 * sched-domains (multiprocessor balancing) declarations:
960 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
961 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
962 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
963 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
964 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
965 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
966 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
967 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
968 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
969 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
970 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
971 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
972 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
973 #define SD_NUMA 0x4000 /* cross-node balancing */
975 #ifdef CONFIG_SCHED_SMT
976 static inline int cpu_smt_flags(void)
978 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
982 #ifdef CONFIG_SCHED_MC
983 static inline int cpu_core_flags(void)
985 return SD_SHARE_PKG_RESOURCES;
990 static inline int cpu_numa_flags(void)
996 struct sched_domain_attr {
997 int relax_domain_level;
1000 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1001 .relax_domain_level = -1, \
1004 extern int sched_domain_level_max;
1008 struct sched_domain {
1009 /* These fields must be setup */
1010 struct sched_domain *parent; /* top domain must be null terminated */
1011 struct sched_domain *child; /* bottom domain must be null terminated */
1012 struct sched_group *groups; /* the balancing groups of the domain */
1013 unsigned long min_interval; /* Minimum balance interval ms */
1014 unsigned long max_interval; /* Maximum balance interval ms */
1015 unsigned int busy_factor; /* less balancing by factor if busy */
1016 unsigned int imbalance_pct; /* No balance until over watermark */
1017 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1018 unsigned int busy_idx;
1019 unsigned int idle_idx;
1020 unsigned int newidle_idx;
1021 unsigned int wake_idx;
1022 unsigned int forkexec_idx;
1023 unsigned int smt_gain;
1025 int nohz_idle; /* NOHZ IDLE status */
1026 int flags; /* See SD_* */
1029 /* Runtime fields. */
1030 unsigned long last_balance; /* init to jiffies. units in jiffies */
1031 unsigned int balance_interval; /* initialise to 1. units in ms. */
1032 unsigned int nr_balance_failed; /* initialise to 0 */
1034 /* idle_balance() stats */
1035 u64 max_newidle_lb_cost;
1036 unsigned long next_decay_max_lb_cost;
1038 #ifdef CONFIG_SCHEDSTATS
1039 /* load_balance() stats */
1040 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1041 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1042 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1043 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1044 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1045 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1046 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1047 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1049 /* Active load balancing */
1050 unsigned int alb_count;
1051 unsigned int alb_failed;
1052 unsigned int alb_pushed;
1054 /* SD_BALANCE_EXEC stats */
1055 unsigned int sbe_count;
1056 unsigned int sbe_balanced;
1057 unsigned int sbe_pushed;
1059 /* SD_BALANCE_FORK stats */
1060 unsigned int sbf_count;
1061 unsigned int sbf_balanced;
1062 unsigned int sbf_pushed;
1064 /* try_to_wake_up() stats */
1065 unsigned int ttwu_wake_remote;
1066 unsigned int ttwu_move_affine;
1067 unsigned int ttwu_move_balance;
1069 #ifdef CONFIG_SCHED_DEBUG
1073 void *private; /* used during construction */
1074 struct rcu_head rcu; /* used during destruction */
1077 unsigned int span_weight;
1079 * Span of all CPUs in this domain.
1081 * NOTE: this field is variable length. (Allocated dynamically
1082 * by attaching extra space to the end of the structure,
1083 * depending on how many CPUs the kernel has booted up with)
1085 unsigned long span[0];
1088 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1090 return to_cpumask(sd->span);
1093 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1094 struct sched_domain_attr *dattr_new);
1096 /* Allocate an array of sched domains, for partition_sched_domains(). */
1097 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1098 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1100 bool cpus_share_cache(int this_cpu, int that_cpu);
1102 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1103 typedef int (*sched_domain_flags_f)(void);
1105 #define SDTL_OVERLAP 0x01
1108 struct sched_domain **__percpu sd;
1109 struct sched_group **__percpu sg;
1110 struct sched_group_capacity **__percpu sgc;
1113 struct sched_domain_topology_level {
1114 sched_domain_mask_f mask;
1115 sched_domain_flags_f sd_flags;
1118 struct sd_data data;
1119 #ifdef CONFIG_SCHED_DEBUG
1124 extern struct sched_domain_topology_level *sched_domain_topology;
1126 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1127 extern void wake_up_if_idle(int cpu);
1129 #ifdef CONFIG_SCHED_DEBUG
1130 # define SD_INIT_NAME(type) .name = #type
1132 # define SD_INIT_NAME(type)
1135 #else /* CONFIG_SMP */
1137 struct sched_domain_attr;
1140 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1141 struct sched_domain_attr *dattr_new)
1145 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1150 #endif /* !CONFIG_SMP */
1153 struct io_context; /* See blkdev.h */
1156 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1157 extern void prefetch_stack(struct task_struct *t);
1159 static inline void prefetch_stack(struct task_struct *t) { }
1162 struct audit_context; /* See audit.c */
1164 struct pipe_inode_info;
1165 struct uts_namespace;
1167 struct load_weight {
1168 unsigned long weight;
1174 * These sums represent an infinite geometric series and so are bound
1175 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1176 * choices of y < 1-2^(-32)*1024.
1178 u32 runnable_avg_sum, runnable_avg_period;
1179 u64 last_runnable_update;
1181 unsigned long load_avg_contrib;
1184 #ifdef CONFIG_SCHEDSTATS
1185 struct sched_statistics {
1195 s64 sum_sleep_runtime;
1202 u64 nr_migrations_cold;
1203 u64 nr_failed_migrations_affine;
1204 u64 nr_failed_migrations_running;
1205 u64 nr_failed_migrations_hot;
1206 u64 nr_forced_migrations;
1209 u64 nr_wakeups_sync;
1210 u64 nr_wakeups_migrate;
1211 u64 nr_wakeups_local;
1212 u64 nr_wakeups_remote;
1213 u64 nr_wakeups_affine;
1214 u64 nr_wakeups_affine_attempts;
1215 u64 nr_wakeups_passive;
1216 u64 nr_wakeups_idle;
1220 struct sched_entity {
1221 struct load_weight load; /* for load-balancing */
1222 struct rb_node run_node;
1223 struct list_head group_node;
1227 u64 sum_exec_runtime;
1229 u64 prev_sum_exec_runtime;
1233 #ifdef CONFIG_SCHEDSTATS
1234 struct sched_statistics statistics;
1237 #ifdef CONFIG_FAIR_GROUP_SCHED
1239 struct sched_entity *parent;
1240 /* rq on which this entity is (to be) queued: */
1241 struct cfs_rq *cfs_rq;
1242 /* rq "owned" by this entity/group: */
1243 struct cfs_rq *my_q;
1247 /* Per-entity load-tracking */
1248 struct sched_avg avg;
1252 struct sched_rt_entity {
1253 struct list_head run_list;
1254 unsigned long timeout;
1255 unsigned long watchdog_stamp;
1256 unsigned int time_slice;
1258 struct sched_rt_entity *back;
1259 #ifdef CONFIG_RT_GROUP_SCHED
1260 struct sched_rt_entity *parent;
1261 /* rq on which this entity is (to be) queued: */
1262 struct rt_rq *rt_rq;
1263 /* rq "owned" by this entity/group: */
1268 struct sched_dl_entity {
1269 struct rb_node rb_node;
1272 * Original scheduling parameters. Copied here from sched_attr
1273 * during sched_setattr(), they will remain the same until
1274 * the next sched_setattr().
1276 u64 dl_runtime; /* maximum runtime for each instance */
1277 u64 dl_deadline; /* relative deadline of each instance */
1278 u64 dl_period; /* separation of two instances (period) */
1279 u64 dl_bw; /* dl_runtime / dl_deadline */
1282 * Actual scheduling parameters. Initialized with the values above,
1283 * they are continously updated during task execution. Note that
1284 * the remaining runtime could be < 0 in case we are in overrun.
1286 s64 runtime; /* remaining runtime for this instance */
1287 u64 deadline; /* absolute deadline for this instance */
1288 unsigned int flags; /* specifying the scheduler behaviour */
1293 * @dl_throttled tells if we exhausted the runtime. If so, the
1294 * task has to wait for a replenishment to be performed at the
1295 * next firing of dl_timer.
1297 * @dl_new tells if a new instance arrived. If so we must
1298 * start executing it with full runtime and reset its absolute
1301 * @dl_boosted tells if we are boosted due to DI. If so we are
1302 * outside bandwidth enforcement mechanism (but only until we
1303 * exit the critical section);
1305 * @dl_yielded tells if task gave up the cpu before consuming
1306 * all its available runtime during the last job.
1308 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1311 * Bandwidth enforcement timer. Each -deadline task has its
1312 * own bandwidth to be enforced, thus we need one timer per task.
1314 struct hrtimer dl_timer;
1326 enum perf_event_task_context {
1327 perf_invalid_context = -1,
1328 perf_hw_context = 0,
1330 perf_nr_task_contexts,
1333 struct task_struct {
1334 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1335 volatile long saved_state; /* saved state for "spinlock sleepers" */
1338 unsigned int flags; /* per process flags, defined below */
1339 unsigned int ptrace;
1342 struct llist_node wake_entry;
1344 struct task_struct *last_wakee;
1345 unsigned long wakee_flips;
1346 unsigned long wakee_flip_decay_ts;
1352 int prio, static_prio, normal_prio;
1353 unsigned int rt_priority;
1354 const struct sched_class *sched_class;
1355 struct sched_entity se;
1356 struct sched_rt_entity rt;
1357 #ifdef CONFIG_CGROUP_SCHED
1358 struct task_group *sched_task_group;
1360 struct sched_dl_entity dl;
1362 #ifdef CONFIG_PREEMPT_NOTIFIERS
1363 /* list of struct preempt_notifier: */
1364 struct hlist_head preempt_notifiers;
1367 #ifdef CONFIG_BLK_DEV_IO_TRACE
1368 unsigned int btrace_seq;
1371 unsigned int policy;
1372 #ifdef CONFIG_PREEMPT_RT_FULL
1373 int migrate_disable;
1374 # ifdef CONFIG_SCHED_DEBUG
1375 int migrate_disable_atomic;
1378 int nr_cpus_allowed;
1379 cpumask_t cpus_allowed;
1381 #ifdef CONFIG_PREEMPT_RCU
1382 int rcu_read_lock_nesting;
1383 union rcu_special rcu_read_unlock_special;
1384 struct list_head rcu_node_entry;
1385 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1386 #ifdef CONFIG_PREEMPT_RCU
1387 struct rcu_node *rcu_blocked_node;
1388 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1389 #ifdef CONFIG_TASKS_RCU
1390 unsigned long rcu_tasks_nvcsw;
1391 bool rcu_tasks_holdout;
1392 struct list_head rcu_tasks_holdout_list;
1393 int rcu_tasks_idle_cpu;
1394 #endif /* #ifdef CONFIG_TASKS_RCU */
1396 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1397 struct sched_info sched_info;
1400 struct list_head tasks;
1402 struct plist_node pushable_tasks;
1403 struct rb_node pushable_dl_tasks;
1406 struct mm_struct *mm, *active_mm;
1407 #ifdef CONFIG_COMPAT_BRK
1408 unsigned brk_randomized:1;
1410 /* per-thread vma caching */
1411 u32 vmacache_seqnum;
1412 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1413 #if defined(SPLIT_RSS_COUNTING)
1414 struct task_rss_stat rss_stat;
1418 int exit_code, exit_signal;
1419 int pdeath_signal; /* The signal sent when the parent dies */
1420 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1422 /* Used for emulating ABI behavior of previous Linux versions */
1423 unsigned int personality;
1425 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1427 unsigned in_iowait:1;
1429 /* Revert to default priority/policy when forking */
1430 unsigned sched_reset_on_fork:1;
1431 unsigned sched_contributes_to_load:1;
1433 #ifdef CONFIG_MEMCG_KMEM
1434 unsigned memcg_kmem_skip_account:1;
1437 unsigned long atomic_flags; /* Flags needing atomic access. */
1439 struct restart_block restart_block;
1444 #ifdef CONFIG_CC_STACKPROTECTOR
1445 /* Canary value for the -fstack-protector gcc feature */
1446 unsigned long stack_canary;
1449 * pointers to (original) parent process, youngest child, younger sibling,
1450 * older sibling, respectively. (p->father can be replaced with
1451 * p->real_parent->pid)
1453 struct task_struct __rcu *real_parent; /* real parent process */
1454 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1456 * children/sibling forms the list of my natural children
1458 struct list_head children; /* list of my children */
1459 struct list_head sibling; /* linkage in my parent's children list */
1460 struct task_struct *group_leader; /* threadgroup leader */
1463 * ptraced is the list of tasks this task is using ptrace on.
1464 * This includes both natural children and PTRACE_ATTACH targets.
1465 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1467 struct list_head ptraced;
1468 struct list_head ptrace_entry;
1470 /* PID/PID hash table linkage. */
1471 struct pid_link pids[PIDTYPE_MAX];
1472 struct list_head thread_group;
1473 struct list_head thread_node;
1475 struct completion *vfork_done; /* for vfork() */
1476 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1477 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1479 cputime_t utime, stime, utimescaled, stimescaled;
1481 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1482 struct cputime prev_cputime;
1484 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1485 raw_spinlock_t vtime_lock;
1486 seqcount_t vtime_seq;
1487 unsigned long long vtime_snap;
1492 } vtime_snap_whence;
1494 unsigned long nvcsw, nivcsw; /* context switch counts */
1495 u64 start_time; /* monotonic time in nsec */
1496 u64 real_start_time; /* boot based time in nsec */
1497 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1498 unsigned long min_flt, maj_flt;
1500 struct task_cputime cputime_expires;
1501 struct list_head cpu_timers[3];
1502 #ifdef CONFIG_PREEMPT_RT_BASE
1503 struct task_struct *posix_timer_list;
1506 /* process credentials */
1507 const struct cred __rcu *real_cred; /* objective and real subjective task
1508 * credentials (COW) */
1509 const struct cred __rcu *cred; /* effective (overridable) subjective task
1510 * credentials (COW) */
1511 char comm[TASK_COMM_LEN]; /* executable name excluding path
1512 - access with [gs]et_task_comm (which lock
1513 it with task_lock())
1514 - initialized normally by setup_new_exec */
1515 /* file system info */
1516 int link_count, total_link_count;
1517 #ifdef CONFIG_SYSVIPC
1519 struct sysv_sem sysvsem;
1520 struct sysv_shm sysvshm;
1522 #ifdef CONFIG_DETECT_HUNG_TASK
1523 /* hung task detection */
1524 unsigned long last_switch_count;
1526 /* CPU-specific state of this task */
1527 struct thread_struct thread;
1528 /* filesystem information */
1529 struct fs_struct *fs;
1530 /* open file information */
1531 struct files_struct *files;
1533 struct nsproxy *nsproxy;
1534 /* signal handlers */
1535 struct signal_struct *signal;
1536 struct sighand_struct *sighand;
1537 struct sigqueue *sigqueue_cache;
1539 sigset_t blocked, real_blocked;
1540 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1541 struct sigpending pending;
1542 #ifdef CONFIG_PREEMPT_RT_FULL
1543 /* TODO: move me into ->restart_block ? */
1544 struct siginfo forced_info;
1547 unsigned long sas_ss_sp;
1549 int (*notifier)(void *priv);
1550 void *notifier_data;
1551 sigset_t *notifier_mask;
1552 struct callback_head *task_works;
1554 struct audit_context *audit_context;
1555 #ifdef CONFIG_AUDITSYSCALL
1557 unsigned int sessionid;
1559 struct seccomp seccomp;
1561 /* Thread group tracking */
1564 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1566 spinlock_t alloc_lock;
1568 /* Protection of the PI data structures: */
1569 raw_spinlock_t pi_lock;
1571 struct wake_q_node wake_q;
1573 #ifdef CONFIG_RT_MUTEXES
1574 /* PI waiters blocked on a rt_mutex held by this task */
1575 struct rb_root pi_waiters;
1576 struct rb_node *pi_waiters_leftmost;
1577 /* Deadlock detection and priority inheritance handling */
1578 struct rt_mutex_waiter *pi_blocked_on;
1581 #ifdef CONFIG_DEBUG_MUTEXES
1582 /* mutex deadlock detection */
1583 struct mutex_waiter *blocked_on;
1585 #ifdef CONFIG_TRACE_IRQFLAGS
1586 unsigned int irq_events;
1587 unsigned long hardirq_enable_ip;
1588 unsigned long hardirq_disable_ip;
1589 unsigned int hardirq_enable_event;
1590 unsigned int hardirq_disable_event;
1591 int hardirqs_enabled;
1592 int hardirq_context;
1593 unsigned long softirq_disable_ip;
1594 unsigned long softirq_enable_ip;
1595 unsigned int softirq_disable_event;
1596 unsigned int softirq_enable_event;
1597 int softirqs_enabled;
1598 int softirq_context;
1600 #ifdef CONFIG_LOCKDEP
1601 # define MAX_LOCK_DEPTH 48UL
1604 unsigned int lockdep_recursion;
1605 struct held_lock held_locks[MAX_LOCK_DEPTH];
1606 gfp_t lockdep_reclaim_gfp;
1609 /* journalling filesystem info */
1612 /* stacked block device info */
1613 struct bio_list *bio_list;
1616 /* stack plugging */
1617 struct blk_plug *plug;
1621 struct reclaim_state *reclaim_state;
1623 struct backing_dev_info *backing_dev_info;
1625 struct io_context *io_context;
1627 unsigned long ptrace_message;
1628 siginfo_t *last_siginfo; /* For ptrace use. */
1629 struct task_io_accounting ioac;
1630 #if defined(CONFIG_TASK_XACCT)
1631 u64 acct_rss_mem1; /* accumulated rss usage */
1632 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1633 cputime_t acct_timexpd; /* stime + utime since last update */
1635 #ifdef CONFIG_CPUSETS
1636 nodemask_t mems_allowed; /* Protected by alloc_lock */
1637 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1638 int cpuset_mem_spread_rotor;
1639 int cpuset_slab_spread_rotor;
1641 #ifdef CONFIG_CGROUPS
1642 /* Control Group info protected by css_set_lock */
1643 struct css_set __rcu *cgroups;
1644 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1645 struct list_head cg_list;
1648 struct robust_list_head __user *robust_list;
1649 #ifdef CONFIG_COMPAT
1650 struct compat_robust_list_head __user *compat_robust_list;
1652 struct list_head pi_state_list;
1653 struct futex_pi_state *pi_state_cache;
1655 #ifdef CONFIG_PERF_EVENTS
1656 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1657 struct mutex perf_event_mutex;
1658 struct list_head perf_event_list;
1660 #ifdef CONFIG_DEBUG_PREEMPT
1661 unsigned long preempt_disable_ip;
1664 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1666 short pref_node_fork;
1668 #ifdef CONFIG_NUMA_BALANCING
1670 unsigned int numa_scan_period;
1671 unsigned int numa_scan_period_max;
1672 int numa_preferred_nid;
1673 unsigned long numa_migrate_retry;
1674 u64 node_stamp; /* migration stamp */
1675 u64 last_task_numa_placement;
1676 u64 last_sum_exec_runtime;
1677 struct callback_head numa_work;
1679 struct list_head numa_entry;
1680 struct numa_group *numa_group;
1683 * numa_faults is an array split into four regions:
1684 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1685 * in this precise order.
1687 * faults_memory: Exponential decaying average of faults on a per-node
1688 * basis. Scheduling placement decisions are made based on these
1689 * counts. The values remain static for the duration of a PTE scan.
1690 * faults_cpu: Track the nodes the process was running on when a NUMA
1691 * hinting fault was incurred.
1692 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1693 * during the current scan window. When the scan completes, the counts
1694 * in faults_memory and faults_cpu decay and these values are copied.
1696 unsigned long *numa_faults;
1697 unsigned long total_numa_faults;
1700 * numa_faults_locality tracks if faults recorded during the last
1701 * scan window were remote/local or failed to migrate. The task scan
1702 * period is adapted based on the locality of the faults with different
1703 * weights depending on whether they were shared or private faults
1705 unsigned long numa_faults_locality[3];
1707 unsigned long numa_pages_migrated;
1708 #endif /* CONFIG_NUMA_BALANCING */
1710 struct rcu_head rcu;
1713 * cache last used pipe for splice
1715 struct pipe_inode_info *splice_pipe;
1717 struct page_frag task_frag;
1719 #ifdef CONFIG_TASK_DELAY_ACCT
1720 struct task_delay_info *delays;
1722 #ifdef CONFIG_FAULT_INJECTION
1726 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1727 * balance_dirty_pages() for some dirty throttling pause
1730 int nr_dirtied_pause;
1731 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1733 #ifdef CONFIG_LATENCYTOP
1734 int latency_record_count;
1735 struct latency_record latency_record[LT_SAVECOUNT];
1738 * time slack values; these are used to round up poll() and
1739 * select() etc timeout values. These are in nanoseconds.
1741 unsigned long timer_slack_ns;
1742 unsigned long default_timer_slack_ns;
1745 unsigned int kasan_depth;
1747 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1748 /* Index of current stored address in ret_stack */
1750 /* Stack of return addresses for return function tracing */
1751 struct ftrace_ret_stack *ret_stack;
1752 /* time stamp for last schedule */
1753 unsigned long long ftrace_timestamp;
1755 * Number of functions that haven't been traced
1756 * because of depth overrun.
1758 atomic_t trace_overrun;
1759 /* Pause for the tracing */
1760 atomic_t tracing_graph_pause;
1762 #ifdef CONFIG_TRACING
1763 /* state flags for use by tracers */
1764 unsigned long trace;
1765 /* bitmask and counter of trace recursion */
1766 unsigned long trace_recursion;
1767 #ifdef CONFIG_WAKEUP_LATENCY_HIST
1768 u64 preempt_timestamp_hist;
1769 #ifdef CONFIG_MISSED_TIMER_OFFSETS_HIST
1773 #endif /* CONFIG_TRACING */
1775 struct memcg_oom_info {
1776 struct mem_cgroup *memcg;
1779 unsigned int may_oom:1;
1782 #ifdef CONFIG_UPROBES
1783 struct uprobe_task *utask;
1785 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1786 unsigned int sequential_io;
1787 unsigned int sequential_io_avg;
1789 #ifdef CONFIG_PREEMPT_RT_BASE
1790 struct rcu_head put_rcu;
1791 int softirq_nestcnt;
1793 #ifdef CONFIG_PREEMPT_RT_FULL
1794 # if defined CONFIG_HIGHMEM || defined CONFIG_X86_32
1796 pte_t kmap_pte[KM_TYPE_NR];
1799 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1800 unsigned long task_state_change;
1802 int pagefault_disabled;
1805 #define TNF_MIGRATED 0x01
1806 #define TNF_NO_GROUP 0x02
1807 #define TNF_SHARED 0x04
1808 #define TNF_FAULT_LOCAL 0x08
1809 #define TNF_MIGRATE_FAIL 0x10
1811 #ifdef CONFIG_NUMA_BALANCING
1812 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1813 extern pid_t task_numa_group_id(struct task_struct *p);
1814 extern void set_numabalancing_state(bool enabled);
1815 extern void task_numa_free(struct task_struct *p);
1816 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1817 int src_nid, int dst_cpu);
1819 static inline void task_numa_fault(int last_node, int node, int pages,
1823 static inline pid_t task_numa_group_id(struct task_struct *p)
1827 static inline void set_numabalancing_state(bool enabled)
1830 static inline void task_numa_free(struct task_struct *p)
1833 static inline bool should_numa_migrate_memory(struct task_struct *p,
1834 struct page *page, int src_nid, int dst_cpu)
1840 static inline struct pid *task_pid(struct task_struct *task)
1842 return task->pids[PIDTYPE_PID].pid;
1845 static inline struct pid *task_tgid(struct task_struct *task)
1847 return task->group_leader->pids[PIDTYPE_PID].pid;
1851 * Without tasklist or rcu lock it is not safe to dereference
1852 * the result of task_pgrp/task_session even if task == current,
1853 * we can race with another thread doing sys_setsid/sys_setpgid.
1855 static inline struct pid *task_pgrp(struct task_struct *task)
1857 return task->group_leader->pids[PIDTYPE_PGID].pid;
1860 static inline struct pid *task_session(struct task_struct *task)
1862 return task->group_leader->pids[PIDTYPE_SID].pid;
1865 struct pid_namespace;
1868 * the helpers to get the task's different pids as they are seen
1869 * from various namespaces
1871 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1872 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1874 * task_xid_nr_ns() : id seen from the ns specified;
1876 * set_task_vxid() : assigns a virtual id to a task;
1878 * see also pid_nr() etc in include/linux/pid.h
1880 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1881 struct pid_namespace *ns);
1883 static inline pid_t task_pid_nr(struct task_struct *tsk)
1888 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1889 struct pid_namespace *ns)
1891 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1894 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1896 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1900 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1905 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1907 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1909 return pid_vnr(task_tgid(tsk));
1913 static inline int pid_alive(const struct task_struct *p);
1914 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1920 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1926 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1928 return task_ppid_nr_ns(tsk, &init_pid_ns);
1931 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1932 struct pid_namespace *ns)
1934 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1937 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1939 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1943 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1944 struct pid_namespace *ns)
1946 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1949 static inline pid_t task_session_vnr(struct task_struct *tsk)
1951 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1954 /* obsolete, do not use */
1955 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1957 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1961 * pid_alive - check that a task structure is not stale
1962 * @p: Task structure to be checked.
1964 * Test if a process is not yet dead (at most zombie state)
1965 * If pid_alive fails, then pointers within the task structure
1966 * can be stale and must not be dereferenced.
1968 * Return: 1 if the process is alive. 0 otherwise.
1970 static inline int pid_alive(const struct task_struct *p)
1972 return p->pids[PIDTYPE_PID].pid != NULL;
1976 * is_global_init - check if a task structure is init
1977 * @tsk: Task structure to be checked.
1979 * Check if a task structure is the first user space task the kernel created.
1981 * Return: 1 if the task structure is init. 0 otherwise.
1983 static inline int is_global_init(struct task_struct *tsk)
1985 return tsk->pid == 1;
1988 extern struct pid *cad_pid;
1990 extern void free_task(struct task_struct *tsk);
1991 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1993 #ifdef CONFIG_PREEMPT_RT_BASE
1994 extern void __put_task_struct_cb(struct rcu_head *rhp);
1996 static inline void put_task_struct(struct task_struct *t)
1998 if (atomic_dec_and_test(&t->usage))
1999 call_rcu(&t->put_rcu, __put_task_struct_cb);
2002 extern void __put_task_struct(struct task_struct *t);
2004 static inline void put_task_struct(struct task_struct *t)
2006 if (atomic_dec_and_test(&t->usage))
2007 __put_task_struct(t);
2011 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2012 extern void task_cputime(struct task_struct *t,
2013 cputime_t *utime, cputime_t *stime);
2014 extern void task_cputime_scaled(struct task_struct *t,
2015 cputime_t *utimescaled, cputime_t *stimescaled);
2016 extern cputime_t task_gtime(struct task_struct *t);
2018 static inline void task_cputime(struct task_struct *t,
2019 cputime_t *utime, cputime_t *stime)
2027 static inline void task_cputime_scaled(struct task_struct *t,
2028 cputime_t *utimescaled,
2029 cputime_t *stimescaled)
2032 *utimescaled = t->utimescaled;
2034 *stimescaled = t->stimescaled;
2037 static inline cputime_t task_gtime(struct task_struct *t)
2042 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2043 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2048 #define PF_EXITING 0x00000004 /* getting shut down */
2049 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2050 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2051 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2052 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2053 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2054 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2055 #define PF_DUMPCORE 0x00000200 /* dumped core */
2056 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2057 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2058 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2059 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2060 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2061 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2062 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2063 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2064 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2065 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2066 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2067 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2068 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2069 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2070 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2071 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2072 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2073 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2074 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2077 * Only the _current_ task can read/write to tsk->flags, but other
2078 * tasks can access tsk->flags in readonly mode for example
2079 * with tsk_used_math (like during threaded core dumping).
2080 * There is however an exception to this rule during ptrace
2081 * or during fork: the ptracer task is allowed to write to the
2082 * child->flags of its traced child (same goes for fork, the parent
2083 * can write to the child->flags), because we're guaranteed the
2084 * child is not running and in turn not changing child->flags
2085 * at the same time the parent does it.
2087 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2088 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2089 #define clear_used_math() clear_stopped_child_used_math(current)
2090 #define set_used_math() set_stopped_child_used_math(current)
2091 #define conditional_stopped_child_used_math(condition, child) \
2092 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2093 #define conditional_used_math(condition) \
2094 conditional_stopped_child_used_math(condition, current)
2095 #define copy_to_stopped_child_used_math(child) \
2096 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2097 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2098 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2099 #define used_math() tsk_used_math(current)
2101 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2102 * __GFP_FS is also cleared as it implies __GFP_IO.
2104 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2106 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2107 flags &= ~(__GFP_IO | __GFP_FS);
2111 static inline unsigned int memalloc_noio_save(void)
2113 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2114 current->flags |= PF_MEMALLOC_NOIO;
2118 static inline void memalloc_noio_restore(unsigned int flags)
2120 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2123 /* Per-process atomic flags. */
2124 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2125 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2126 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2129 #define TASK_PFA_TEST(name, func) \
2130 static inline bool task_##func(struct task_struct *p) \
2131 { return test_bit(PFA_##name, &p->atomic_flags); }
2132 #define TASK_PFA_SET(name, func) \
2133 static inline void task_set_##func(struct task_struct *p) \
2134 { set_bit(PFA_##name, &p->atomic_flags); }
2135 #define TASK_PFA_CLEAR(name, func) \
2136 static inline void task_clear_##func(struct task_struct *p) \
2137 { clear_bit(PFA_##name, &p->atomic_flags); }
2139 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2140 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2142 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2143 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2144 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2146 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2147 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2148 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2151 * task->jobctl flags
2153 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2155 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2156 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2157 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2158 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2159 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2160 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2161 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2163 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
2164 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
2165 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
2166 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
2167 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
2168 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
2169 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
2171 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2172 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2174 extern bool task_set_jobctl_pending(struct task_struct *task,
2176 extern void task_clear_jobctl_trapping(struct task_struct *task);
2177 extern void task_clear_jobctl_pending(struct task_struct *task,
2180 static inline void rcu_copy_process(struct task_struct *p)
2182 #ifdef CONFIG_PREEMPT_RCU
2183 p->rcu_read_lock_nesting = 0;
2184 p->rcu_read_unlock_special.s = 0;
2185 p->rcu_blocked_node = NULL;
2186 INIT_LIST_HEAD(&p->rcu_node_entry);
2187 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2188 #ifdef CONFIG_TASKS_RCU
2189 p->rcu_tasks_holdout = false;
2190 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2191 p->rcu_tasks_idle_cpu = -1;
2192 #endif /* #ifdef CONFIG_TASKS_RCU */
2195 static inline void tsk_restore_flags(struct task_struct *task,
2196 unsigned long orig_flags, unsigned long flags)
2198 task->flags &= ~flags;
2199 task->flags |= orig_flags & flags;
2202 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2203 const struct cpumask *trial);
2204 extern int task_can_attach(struct task_struct *p,
2205 const struct cpumask *cs_cpus_allowed);
2207 extern void do_set_cpus_allowed(struct task_struct *p,
2208 const struct cpumask *new_mask);
2210 extern int set_cpus_allowed_ptr(struct task_struct *p,
2211 const struct cpumask *new_mask);
2212 int migrate_me(void);
2213 void tell_sched_cpu_down_begin(int cpu);
2214 void tell_sched_cpu_down_done(int cpu);
2217 static inline void do_set_cpus_allowed(struct task_struct *p,
2218 const struct cpumask *new_mask)
2221 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2222 const struct cpumask *new_mask)
2224 if (!cpumask_test_cpu(0, new_mask))
2228 static inline int migrate_me(void) { return 0; }
2229 static inline void tell_sched_cpu_down_begin(int cpu) { }
2230 static inline void tell_sched_cpu_down_done(int cpu) { }
2233 #ifdef CONFIG_NO_HZ_COMMON
2234 void calc_load_enter_idle(void);
2235 void calc_load_exit_idle(void);
2237 static inline void calc_load_enter_idle(void) { }
2238 static inline void calc_load_exit_idle(void) { }
2239 #endif /* CONFIG_NO_HZ_COMMON */
2241 #ifndef CONFIG_CPUMASK_OFFSTACK
2242 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2244 return set_cpus_allowed_ptr(p, &new_mask);
2249 * Do not use outside of architecture code which knows its limitations.
2251 * sched_clock() has no promise of monotonicity or bounded drift between
2252 * CPUs, use (which you should not) requires disabling IRQs.
2254 * Please use one of the three interfaces below.
2256 extern unsigned long long notrace sched_clock(void);
2258 * See the comment in kernel/sched/clock.c
2260 extern u64 cpu_clock(int cpu);
2261 extern u64 local_clock(void);
2262 extern u64 running_clock(void);
2263 extern u64 sched_clock_cpu(int cpu);
2266 extern void sched_clock_init(void);
2268 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2269 static inline void sched_clock_tick(void)
2273 static inline void sched_clock_idle_sleep_event(void)
2277 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2282 * Architectures can set this to 1 if they have specified
2283 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2284 * but then during bootup it turns out that sched_clock()
2285 * is reliable after all:
2287 extern int sched_clock_stable(void);
2288 extern void set_sched_clock_stable(void);
2289 extern void clear_sched_clock_stable(void);
2291 extern void sched_clock_tick(void);
2292 extern void sched_clock_idle_sleep_event(void);
2293 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2296 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2298 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2299 * The reason for this explicit opt-in is not to have perf penalty with
2300 * slow sched_clocks.
2302 extern void enable_sched_clock_irqtime(void);
2303 extern void disable_sched_clock_irqtime(void);
2305 static inline void enable_sched_clock_irqtime(void) {}
2306 static inline void disable_sched_clock_irqtime(void) {}
2309 extern unsigned long long
2310 task_sched_runtime(struct task_struct *task);
2312 /* sched_exec is called by processes performing an exec */
2314 extern void sched_exec(void);
2316 #define sched_exec() {}
2319 extern void sched_clock_idle_sleep_event(void);
2320 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2322 #ifdef CONFIG_HOTPLUG_CPU
2323 extern void idle_task_exit(void);
2325 static inline void idle_task_exit(void) {}
2328 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2329 extern void wake_up_nohz_cpu(int cpu);
2331 static inline void wake_up_nohz_cpu(int cpu) { }
2334 #ifdef CONFIG_NO_HZ_FULL
2335 extern bool sched_can_stop_tick(void);
2336 extern u64 scheduler_tick_max_deferment(void);
2338 static inline bool sched_can_stop_tick(void) { return false; }
2341 #ifdef CONFIG_SCHED_AUTOGROUP
2342 extern void sched_autogroup_create_attach(struct task_struct *p);
2343 extern void sched_autogroup_detach(struct task_struct *p);
2344 extern void sched_autogroup_fork(struct signal_struct *sig);
2345 extern void sched_autogroup_exit(struct signal_struct *sig);
2346 #ifdef CONFIG_PROC_FS
2347 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2348 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2351 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2352 static inline void sched_autogroup_detach(struct task_struct *p) { }
2353 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2354 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2357 extern int yield_to(struct task_struct *p, bool preempt);
2358 extern void set_user_nice(struct task_struct *p, long nice);
2359 extern int task_prio(const struct task_struct *p);
2361 * task_nice - return the nice value of a given task.
2362 * @p: the task in question.
2364 * Return: The nice value [ -20 ... 0 ... 19 ].
2366 static inline int task_nice(const struct task_struct *p)
2368 return PRIO_TO_NICE((p)->static_prio);
2370 extern int can_nice(const struct task_struct *p, const int nice);
2371 extern int task_curr(const struct task_struct *p);
2372 extern int idle_cpu(int cpu);
2373 extern int sched_setscheduler(struct task_struct *, int,
2374 const struct sched_param *);
2375 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2376 const struct sched_param *);
2377 extern int sched_setattr(struct task_struct *,
2378 const struct sched_attr *);
2379 extern struct task_struct *idle_task(int cpu);
2381 * is_idle_task - is the specified task an idle task?
2382 * @p: the task in question.
2384 * Return: 1 if @p is an idle task. 0 otherwise.
2386 static inline bool is_idle_task(const struct task_struct *p)
2390 extern struct task_struct *curr_task(int cpu);
2391 extern void set_curr_task(int cpu, struct task_struct *p);
2396 * The default (Linux) execution domain.
2398 extern struct exec_domain default_exec_domain;
2400 union thread_union {
2401 struct thread_info thread_info;
2402 unsigned long stack[THREAD_SIZE/sizeof(long)];
2405 #ifndef __HAVE_ARCH_KSTACK_END
2406 static inline int kstack_end(void *addr)
2408 /* Reliable end of stack detection:
2409 * Some APM bios versions misalign the stack
2411 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2415 extern union thread_union init_thread_union;
2416 extern struct task_struct init_task;
2418 extern struct mm_struct init_mm;
2420 extern struct pid_namespace init_pid_ns;
2423 * find a task by one of its numerical ids
2425 * find_task_by_pid_ns():
2426 * finds a task by its pid in the specified namespace
2427 * find_task_by_vpid():
2428 * finds a task by its virtual pid
2430 * see also find_vpid() etc in include/linux/pid.h
2433 extern struct task_struct *find_task_by_vpid(pid_t nr);
2434 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2435 struct pid_namespace *ns);
2437 /* per-UID process charging. */
2438 extern struct user_struct * alloc_uid(kuid_t);
2439 static inline struct user_struct *get_uid(struct user_struct *u)
2441 atomic_inc(&u->__count);
2444 extern void free_uid(struct user_struct *);
2446 #include <asm/current.h>
2448 extern void xtime_update(unsigned long ticks);
2450 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2451 extern int wake_up_process(struct task_struct *tsk);
2452 extern int wake_up_lock_sleeper(struct task_struct * tsk);
2453 extern void wake_up_new_task(struct task_struct *tsk);
2455 extern void kick_process(struct task_struct *tsk);
2457 static inline void kick_process(struct task_struct *tsk) { }
2459 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2460 extern void sched_dead(struct task_struct *p);
2462 extern void proc_caches_init(void);
2463 extern void flush_signals(struct task_struct *);
2464 extern void __flush_signals(struct task_struct *);
2465 extern void ignore_signals(struct task_struct *);
2466 extern void flush_signal_handlers(struct task_struct *, int force_default);
2467 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2469 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2471 unsigned long flags;
2474 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2475 ret = dequeue_signal(tsk, mask, info);
2476 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2481 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2483 extern void unblock_all_signals(void);
2484 extern void release_task(struct task_struct * p);
2485 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2486 extern int force_sigsegv(int, struct task_struct *);
2487 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2488 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2489 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2490 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2491 const struct cred *, u32);
2492 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2493 extern int kill_pid(struct pid *pid, int sig, int priv);
2494 extern int kill_proc_info(int, struct siginfo *, pid_t);
2495 extern __must_check bool do_notify_parent(struct task_struct *, int);
2496 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2497 extern void force_sig(int, struct task_struct *);
2498 extern int send_sig(int, struct task_struct *, int);
2499 extern int zap_other_threads(struct task_struct *p);
2500 extern struct sigqueue *sigqueue_alloc(void);
2501 extern void sigqueue_free(struct sigqueue *);
2502 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2503 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2505 static inline void restore_saved_sigmask(void)
2507 if (test_and_clear_restore_sigmask())
2508 __set_current_blocked(¤t->saved_sigmask);
2511 static inline sigset_t *sigmask_to_save(void)
2513 sigset_t *res = ¤t->blocked;
2514 if (unlikely(test_restore_sigmask()))
2515 res = ¤t->saved_sigmask;
2519 static inline int kill_cad_pid(int sig, int priv)
2521 return kill_pid(cad_pid, sig, priv);
2524 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2525 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2526 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2527 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2530 * True if we are on the alternate signal stack.
2532 static inline int on_sig_stack(unsigned long sp)
2534 #ifdef CONFIG_STACK_GROWSUP
2535 return sp >= current->sas_ss_sp &&
2536 sp - current->sas_ss_sp < current->sas_ss_size;
2538 return sp > current->sas_ss_sp &&
2539 sp - current->sas_ss_sp <= current->sas_ss_size;
2543 static inline int sas_ss_flags(unsigned long sp)
2545 if (!current->sas_ss_size)
2548 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2551 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2553 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2554 #ifdef CONFIG_STACK_GROWSUP
2555 return current->sas_ss_sp;
2557 return current->sas_ss_sp + current->sas_ss_size;
2563 * Routines for handling mm_structs
2565 extern struct mm_struct * mm_alloc(void);
2567 /* mmdrop drops the mm and the page tables */
2568 extern void __mmdrop(struct mm_struct *);
2570 static inline void mmdrop(struct mm_struct * mm)
2572 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2576 #ifdef CONFIG_PREEMPT_RT_BASE
2577 extern void __mmdrop_delayed(struct rcu_head *rhp);
2578 static inline void mmdrop_delayed(struct mm_struct *mm)
2580 if (atomic_dec_and_test(&mm->mm_count))
2581 call_rcu(&mm->delayed_drop, __mmdrop_delayed);
2584 # define mmdrop_delayed(mm) mmdrop(mm)
2587 /* mmput gets rid of the mappings and all user-space */
2588 extern void mmput(struct mm_struct *);
2589 /* Grab a reference to a task's mm, if it is not already going away */
2590 extern struct mm_struct *get_task_mm(struct task_struct *task);
2592 * Grab a reference to a task's mm, if it is not already going away
2593 * and ptrace_may_access with the mode parameter passed to it
2596 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2597 /* Remove the current tasks stale references to the old mm_struct */
2598 extern void mm_release(struct task_struct *, struct mm_struct *);
2600 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2601 struct task_struct *);
2602 extern void flush_thread(void);
2603 extern void exit_thread(void);
2605 extern void exit_files(struct task_struct *);
2606 extern void __cleanup_sighand(struct sighand_struct *);
2608 extern void exit_itimers(struct signal_struct *);
2609 extern void flush_itimer_signals(void);
2611 extern void do_group_exit(int);
2613 extern int do_execve(struct filename *,
2614 const char __user * const __user *,
2615 const char __user * const __user *);
2616 extern int do_execveat(int, struct filename *,
2617 const char __user * const __user *,
2618 const char __user * const __user *,
2620 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2621 struct task_struct *fork_idle(int);
2622 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2624 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2625 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2627 __set_task_comm(tsk, from, false);
2629 extern char *get_task_comm(char *to, struct task_struct *tsk);
2632 void scheduler_ipi(void);
2633 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2635 static inline void scheduler_ipi(void) { }
2636 static inline unsigned long wait_task_inactive(struct task_struct *p,
2643 #define next_task(p) \
2644 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2646 #define for_each_process(p) \
2647 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2649 extern bool current_is_single_threaded(void);
2652 * Careful: do_each_thread/while_each_thread is a double loop so
2653 * 'break' will not work as expected - use goto instead.
2655 #define do_each_thread(g, t) \
2656 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2658 #define while_each_thread(g, t) \
2659 while ((t = next_thread(t)) != g)
2661 #define __for_each_thread(signal, t) \
2662 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2664 #define for_each_thread(p, t) \
2665 __for_each_thread((p)->signal, t)
2667 /* Careful: this is a double loop, 'break' won't work as expected. */
2668 #define for_each_process_thread(p, t) \
2669 for_each_process(p) for_each_thread(p, t)
2671 static inline int get_nr_threads(struct task_struct *tsk)
2673 return tsk->signal->nr_threads;
2676 static inline bool thread_group_leader(struct task_struct *p)
2678 return p->exit_signal >= 0;
2681 /* Do to the insanities of de_thread it is possible for a process
2682 * to have the pid of the thread group leader without actually being
2683 * the thread group leader. For iteration through the pids in proc
2684 * all we care about is that we have a task with the appropriate
2685 * pid, we don't actually care if we have the right task.
2687 static inline bool has_group_leader_pid(struct task_struct *p)
2689 return task_pid(p) == p->signal->leader_pid;
2693 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2695 return p1->signal == p2->signal;
2698 static inline struct task_struct *next_thread(const struct task_struct *p)
2700 return list_entry_rcu(p->thread_group.next,
2701 struct task_struct, thread_group);
2704 static inline int thread_group_empty(struct task_struct *p)
2706 return list_empty(&p->thread_group);
2709 #define delay_group_leader(p) \
2710 (thread_group_leader(p) && !thread_group_empty(p))
2713 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2714 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2715 * pins the final release of task.io_context. Also protects ->cpuset and
2716 * ->cgroup.subsys[]. And ->vfork_done.
2718 * Nests both inside and outside of read_lock(&tasklist_lock).
2719 * It must not be nested with write_lock_irq(&tasklist_lock),
2720 * neither inside nor outside.
2722 static inline void task_lock(struct task_struct *p)
2724 spin_lock(&p->alloc_lock);
2727 static inline void task_unlock(struct task_struct *p)
2729 spin_unlock(&p->alloc_lock);
2732 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2733 unsigned long *flags);
2735 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2736 unsigned long *flags)
2738 struct sighand_struct *ret;
2740 ret = __lock_task_sighand(tsk, flags);
2741 (void)__cond_lock(&tsk->sighand->siglock, ret);
2745 static inline void unlock_task_sighand(struct task_struct *tsk,
2746 unsigned long *flags)
2748 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2751 #ifdef CONFIG_CGROUPS
2752 static inline void threadgroup_change_begin(struct task_struct *tsk)
2754 down_read(&tsk->signal->group_rwsem);
2756 static inline void threadgroup_change_end(struct task_struct *tsk)
2758 up_read(&tsk->signal->group_rwsem);
2762 * threadgroup_lock - lock threadgroup
2763 * @tsk: member task of the threadgroup to lock
2765 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2766 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2767 * change ->group_leader/pid. This is useful for cases where the threadgroup
2768 * needs to stay stable across blockable operations.
2770 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2771 * synchronization. While held, no new task will be added to threadgroup
2772 * and no existing live task will have its PF_EXITING set.
2774 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2775 * sub-thread becomes a new leader.
2777 static inline void threadgroup_lock(struct task_struct *tsk)
2779 down_write(&tsk->signal->group_rwsem);
2783 * threadgroup_unlock - unlock threadgroup
2784 * @tsk: member task of the threadgroup to unlock
2786 * Reverse threadgroup_lock().
2788 static inline void threadgroup_unlock(struct task_struct *tsk)
2790 up_write(&tsk->signal->group_rwsem);
2793 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2794 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2795 static inline void threadgroup_lock(struct task_struct *tsk) {}
2796 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2799 #ifndef __HAVE_THREAD_FUNCTIONS
2801 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2802 #define task_stack_page(task) ((task)->stack)
2804 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2806 *task_thread_info(p) = *task_thread_info(org);
2807 task_thread_info(p)->task = p;
2811 * Return the address of the last usable long on the stack.
2813 * When the stack grows down, this is just above the thread
2814 * info struct. Going any lower will corrupt the threadinfo.
2816 * When the stack grows up, this is the highest address.
2817 * Beyond that position, we corrupt data on the next page.
2819 static inline unsigned long *end_of_stack(struct task_struct *p)
2821 #ifdef CONFIG_STACK_GROWSUP
2822 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2824 return (unsigned long *)(task_thread_info(p) + 1);
2829 #define task_stack_end_corrupted(task) \
2830 (*(end_of_stack(task)) != STACK_END_MAGIC)
2832 static inline int object_is_on_stack(void *obj)
2834 void *stack = task_stack_page(current);
2836 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2839 extern void thread_info_cache_init(void);
2841 #ifdef CONFIG_DEBUG_STACK_USAGE
2842 static inline unsigned long stack_not_used(struct task_struct *p)
2844 unsigned long *n = end_of_stack(p);
2846 do { /* Skip over canary */
2850 return (unsigned long)n - (unsigned long)end_of_stack(p);
2853 extern void set_task_stack_end_magic(struct task_struct *tsk);
2855 /* set thread flags in other task's structures
2856 * - see asm/thread_info.h for TIF_xxxx flags available
2858 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2860 set_ti_thread_flag(task_thread_info(tsk), flag);
2863 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2865 clear_ti_thread_flag(task_thread_info(tsk), flag);
2868 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2870 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2873 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2875 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2878 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2880 return test_ti_thread_flag(task_thread_info(tsk), flag);
2883 static inline void set_tsk_need_resched(struct task_struct *tsk)
2885 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2888 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2890 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2893 static inline int test_tsk_need_resched(struct task_struct *tsk)
2895 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2898 static inline int restart_syscall(void)
2900 set_tsk_thread_flag(current, TIF_SIGPENDING);
2901 return -ERESTARTNOINTR;
2904 static inline int signal_pending(struct task_struct *p)
2906 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2909 static inline int __fatal_signal_pending(struct task_struct *p)
2911 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2914 static inline int fatal_signal_pending(struct task_struct *p)
2916 return signal_pending(p) && __fatal_signal_pending(p);
2919 static inline int signal_pending_state(long state, struct task_struct *p)
2921 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2923 if (!signal_pending(p))
2926 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2929 static inline bool __task_is_stopped_or_traced(struct task_struct *task)
2931 if (task->state & (__TASK_STOPPED | __TASK_TRACED))
2933 #ifdef CONFIG_PREEMPT_RT_FULL
2934 if (task->saved_state & (__TASK_STOPPED | __TASK_TRACED))
2940 static inline bool task_is_stopped_or_traced(struct task_struct *task)
2942 bool traced_stopped;
2944 #ifdef CONFIG_PREEMPT_RT_FULL
2945 unsigned long flags;
2947 raw_spin_lock_irqsave(&task->pi_lock, flags);
2948 traced_stopped = __task_is_stopped_or_traced(task);
2949 raw_spin_unlock_irqrestore(&task->pi_lock, flags);
2951 traced_stopped = __task_is_stopped_or_traced(task);
2953 return traced_stopped;
2956 static inline bool task_is_traced(struct task_struct *task)
2958 bool traced = false;
2960 if (task->state & __TASK_TRACED)
2962 #ifdef CONFIG_PREEMPT_RT_FULL
2963 /* in case the task is sleeping on tasklist_lock */
2964 raw_spin_lock_irq(&task->pi_lock);
2965 if (task->state & __TASK_TRACED)
2967 else if (task->saved_state & __TASK_TRACED)
2969 raw_spin_unlock_irq(&task->pi_lock);
2975 * cond_resched() and cond_resched_lock(): latency reduction via
2976 * explicit rescheduling in places that are safe. The return
2977 * value indicates whether a reschedule was done in fact.
2978 * cond_resched_lock() will drop the spinlock before scheduling,
2979 * cond_resched_softirq() will enable bhs before scheduling.
2981 extern int _cond_resched(void);
2983 #define cond_resched() ({ \
2984 ___might_sleep(__FILE__, __LINE__, 0); \
2988 extern int __cond_resched_lock(spinlock_t *lock);
2990 #if defined(CONFIG_PREEMPT_COUNT) && !defined(CONFIG_PREEMPT_RT_FULL)
2991 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2993 #define PREEMPT_LOCK_OFFSET 0
2996 #define cond_resched_lock(lock) ({ \
2997 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2998 __cond_resched_lock(lock); \
3001 #ifndef CONFIG_PREEMPT_RT_FULL
3002 extern int __cond_resched_softirq(void);
3004 #define cond_resched_softirq() ({ \
3005 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3006 __cond_resched_softirq(); \
3009 # define cond_resched_softirq() cond_resched()
3012 static inline void cond_resched_rcu(void)
3014 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3022 * Does a critical section need to be broken due to another
3023 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3024 * but a general need for low latency)
3026 static inline int spin_needbreak(spinlock_t *lock)
3028 #ifdef CONFIG_PREEMPT
3029 return spin_is_contended(lock);
3036 * Idle thread specific functions to determine the need_resched
3039 #ifdef TIF_POLLING_NRFLAG
3040 static inline int tsk_is_polling(struct task_struct *p)
3042 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3045 static inline void __current_set_polling(void)
3047 set_thread_flag(TIF_POLLING_NRFLAG);
3050 static inline bool __must_check current_set_polling_and_test(void)
3052 __current_set_polling();
3055 * Polling state must be visible before we test NEED_RESCHED,
3056 * paired by resched_curr()
3058 smp_mb__after_atomic();
3060 return unlikely(tif_need_resched());
3063 static inline void __current_clr_polling(void)
3065 clear_thread_flag(TIF_POLLING_NRFLAG);
3068 static inline bool __must_check current_clr_polling_and_test(void)
3070 __current_clr_polling();
3073 * Polling state must be visible before we test NEED_RESCHED,
3074 * paired by resched_curr()
3076 smp_mb__after_atomic();
3078 return unlikely(tif_need_resched());
3082 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3083 static inline void __current_set_polling(void) { }
3084 static inline void __current_clr_polling(void) { }
3086 static inline bool __must_check current_set_polling_and_test(void)
3088 return unlikely(tif_need_resched());
3090 static inline bool __must_check current_clr_polling_and_test(void)
3092 return unlikely(tif_need_resched());
3096 static inline void current_clr_polling(void)
3098 __current_clr_polling();
3101 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3102 * Once the bit is cleared, we'll get IPIs with every new
3103 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3106 smp_mb(); /* paired with resched_curr() */
3108 preempt_fold_need_resched();
3111 static __always_inline bool need_resched(void)
3113 return unlikely(tif_need_resched());
3117 * Thread group CPU time accounting.
3119 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3120 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3122 static inline void thread_group_cputime_init(struct signal_struct *sig)
3124 raw_spin_lock_init(&sig->cputimer.lock);
3128 * Reevaluate whether the task has signals pending delivery.
3129 * Wake the task if so.
3130 * This is required every time the blocked sigset_t changes.
3131 * callers must hold sighand->siglock.
3133 extern void recalc_sigpending_and_wake(struct task_struct *t);
3134 extern void recalc_sigpending(void);
3136 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3138 static inline void signal_wake_up(struct task_struct *t, bool resume)
3140 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3142 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3144 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3148 * Wrappers for p->thread_info->cpu access. No-op on UP.
3152 static inline unsigned int task_cpu(const struct task_struct *p)
3154 return task_thread_info(p)->cpu;
3157 static inline int task_node(const struct task_struct *p)
3159 return cpu_to_node(task_cpu(p));
3162 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3166 static inline unsigned int task_cpu(const struct task_struct *p)
3171 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3175 #endif /* CONFIG_SMP */
3177 static inline int __migrate_disabled(struct task_struct *p)
3179 #ifdef CONFIG_PREEMPT_RT_FULL
3180 return p->migrate_disable;
3186 /* Future-safe accessor for struct task_struct's cpus_allowed. */
3187 static inline const struct cpumask *tsk_cpus_allowed(struct task_struct *p)
3189 #ifdef CONFIG_PREEMPT_RT_FULL
3190 if (p->migrate_disable)
3191 return cpumask_of(task_cpu(p));
3194 return &p->cpus_allowed;
3197 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3198 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3200 #ifdef CONFIG_CGROUP_SCHED
3201 extern struct task_group root_task_group;
3202 #endif /* CONFIG_CGROUP_SCHED */
3204 extern int task_can_switch_user(struct user_struct *up,
3205 struct task_struct *tsk);
3207 #ifdef CONFIG_TASK_XACCT
3208 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3210 tsk->ioac.rchar += amt;
3213 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3215 tsk->ioac.wchar += amt;
3218 static inline void inc_syscr(struct task_struct *tsk)
3223 static inline void inc_syscw(struct task_struct *tsk)
3228 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3232 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3236 static inline void inc_syscr(struct task_struct *tsk)
3240 static inline void inc_syscw(struct task_struct *tsk)
3245 #ifndef TASK_SIZE_OF
3246 #define TASK_SIZE_OF(tsk) TASK_SIZE
3250 extern void mm_update_next_owner(struct mm_struct *mm);
3252 static inline void mm_update_next_owner(struct mm_struct *mm)
3255 #endif /* CONFIG_MEMCG */
3257 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3260 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
3263 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3266 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
3269 static inline unsigned long rlimit(unsigned int limit)
3271 return task_rlimit(current, limit);
3274 static inline unsigned long rlimit_max(unsigned int limit)
3276 return task_rlimit_max(current, limit);