4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60 #include <linux/magic.h>
62 #include <asm/processor.h>
64 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
67 * Extended scheduling parameters data structure.
69 * This is needed because the original struct sched_param can not be
70 * altered without introducing ABI issues with legacy applications
71 * (e.g., in sched_getparam()).
73 * However, the possibility of specifying more than just a priority for
74 * the tasks may be useful for a wide variety of application fields, e.g.,
75 * multimedia, streaming, automation and control, and many others.
77 * This variant (sched_attr) is meant at describing a so-called
78 * sporadic time-constrained task. In such model a task is specified by:
79 * - the activation period or minimum instance inter-arrival time;
80 * - the maximum (or average, depending on the actual scheduling
81 * discipline) computation time of all instances, a.k.a. runtime;
82 * - the deadline (relative to the actual activation time) of each
84 * Very briefly, a periodic (sporadic) task asks for the execution of
85 * some specific computation --which is typically called an instance--
86 * (at most) every period. Moreover, each instance typically lasts no more
87 * than the runtime and must be completed by time instant t equal to
88 * the instance activation time + the deadline.
90 * This is reflected by the actual fields of the sched_attr structure:
92 * @size size of the structure, for fwd/bwd compat.
94 * @sched_policy task's scheduling policy
95 * @sched_flags for customizing the scheduler behaviour
96 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
97 * @sched_priority task's static priority (SCHED_FIFO/RR)
98 * @sched_deadline representative of the task's deadline
99 * @sched_runtime representative of the task's runtime
100 * @sched_period representative of the task's period
102 * Given this task model, there are a multiplicity of scheduling algorithms
103 * and policies, that can be used to ensure all the tasks will make their
104 * timing constraints.
106 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
107 * only user of this new interface. More information about the algorithm
108 * available in the scheduling class file or in Documentation/.
116 /* SCHED_NORMAL, SCHED_BATCH */
119 /* SCHED_FIFO, SCHED_RR */
128 struct futex_pi_state;
129 struct robust_list_head;
132 struct perf_event_context;
136 #define VMACACHE_BITS 2
137 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
138 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
141 * These are the constant used to fake the fixed-point load-average
142 * counting. Some notes:
143 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
144 * a load-average precision of 10 bits integer + 11 bits fractional
145 * - if you want to count load-averages more often, you need more
146 * precision, or rounding will get you. With 2-second counting freq,
147 * the EXP_n values would be 1981, 2034 and 2043 if still using only
150 extern unsigned long avenrun[]; /* Load averages */
151 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
153 #define FSHIFT 11 /* nr of bits of precision */
154 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
155 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
156 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
157 #define EXP_5 2014 /* 1/exp(5sec/5min) */
158 #define EXP_15 2037 /* 1/exp(5sec/15min) */
160 #define CALC_LOAD(load,exp,n) \
162 load += n*(FIXED_1-exp); \
165 extern unsigned long total_forks;
166 extern int nr_threads;
167 DECLARE_PER_CPU(unsigned long, process_counts);
168 extern int nr_processes(void);
169 extern unsigned long nr_running(void);
170 extern bool single_task_running(void);
171 extern unsigned long nr_iowait(void);
172 extern unsigned long nr_iowait_cpu(int cpu);
173 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
175 extern void calc_global_load(unsigned long ticks);
177 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
178 extern void update_cpu_load_nohz(void);
180 static inline void update_cpu_load_nohz(void) { }
183 extern unsigned long get_parent_ip(unsigned long addr);
185 extern void dump_cpu_task(int cpu);
190 #ifdef CONFIG_SCHED_DEBUG
191 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
192 extern void proc_sched_set_task(struct task_struct *p);
194 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
198 * Task state bitmask. NOTE! These bits are also
199 * encoded in fs/proc/array.c: get_task_state().
201 * We have two separate sets of flags: task->state
202 * is about runnability, while task->exit_state are
203 * about the task exiting. Confusing, but this way
204 * modifying one set can't modify the other one by
207 #define TASK_RUNNING 0
208 #define TASK_INTERRUPTIBLE 1
209 #define TASK_UNINTERRUPTIBLE 2
210 #define __TASK_STOPPED 4
211 #define __TASK_TRACED 8
212 /* in tsk->exit_state */
214 #define EXIT_ZOMBIE 32
215 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
216 /* in tsk->state again */
218 #define TASK_WAKEKILL 128
219 #define TASK_WAKING 256
220 #define TASK_PARKED 512
221 #define TASK_STATE_MAX 1024
223 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
225 extern char ___assert_task_state[1 - 2*!!(
226 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
228 /* Convenience macros for the sake of set_task_state */
229 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
230 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
231 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
233 /* Convenience macros for the sake of wake_up */
234 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
235 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
237 /* get_task_state() */
238 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
239 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
240 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
242 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
243 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
244 #define task_is_stopped_or_traced(task) \
245 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
246 #define task_contributes_to_load(task) \
247 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
248 (task->flags & PF_FROZEN) == 0)
250 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
252 #define __set_task_state(tsk, state_value) \
254 (tsk)->task_state_change = _THIS_IP_; \
255 (tsk)->state = (state_value); \
257 #define set_task_state(tsk, state_value) \
259 (tsk)->task_state_change = _THIS_IP_; \
260 set_mb((tsk)->state, (state_value)); \
264 * set_current_state() includes a barrier so that the write of current->state
265 * is correctly serialised wrt the caller's subsequent test of whether to
268 * set_current_state(TASK_UNINTERRUPTIBLE);
269 * if (do_i_need_to_sleep())
272 * If the caller does not need such serialisation then use __set_current_state()
274 #define __set_current_state(state_value) \
276 current->task_state_change = _THIS_IP_; \
277 current->state = (state_value); \
279 #define set_current_state(state_value) \
281 current->task_state_change = _THIS_IP_; \
282 set_mb(current->state, (state_value)); \
287 #define __set_task_state(tsk, state_value) \
288 do { (tsk)->state = (state_value); } while (0)
289 #define set_task_state(tsk, state_value) \
290 set_mb((tsk)->state, (state_value))
293 * set_current_state() includes a barrier so that the write of current->state
294 * is correctly serialised wrt the caller's subsequent test of whether to
297 * set_current_state(TASK_UNINTERRUPTIBLE);
298 * if (do_i_need_to_sleep())
301 * If the caller does not need such serialisation then use __set_current_state()
303 #define __set_current_state(state_value) \
304 do { current->state = (state_value); } while (0)
305 #define set_current_state(state_value) \
306 set_mb(current->state, (state_value))
310 /* Task command name length */
311 #define TASK_COMM_LEN 16
313 #include <linux/spinlock.h>
316 * This serializes "schedule()" and also protects
317 * the run-queue from deletions/modifications (but
318 * _adding_ to the beginning of the run-queue has
321 extern rwlock_t tasklist_lock;
322 extern spinlock_t mmlist_lock;
326 #ifdef CONFIG_PROVE_RCU
327 extern int lockdep_tasklist_lock_is_held(void);
328 #endif /* #ifdef CONFIG_PROVE_RCU */
330 extern void sched_init(void);
331 extern void sched_init_smp(void);
332 extern asmlinkage void schedule_tail(struct task_struct *prev);
333 extern void init_idle(struct task_struct *idle, int cpu);
334 extern void init_idle_bootup_task(struct task_struct *idle);
336 extern cpumask_var_t cpu_isolated_map;
338 extern int runqueue_is_locked(int cpu);
340 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
341 extern void nohz_balance_enter_idle(int cpu);
342 extern void set_cpu_sd_state_idle(void);
343 extern int get_nohz_timer_target(int pinned);
345 static inline void nohz_balance_enter_idle(int cpu) { }
346 static inline void set_cpu_sd_state_idle(void) { }
347 static inline int get_nohz_timer_target(int pinned)
349 return smp_processor_id();
354 * Only dump TASK_* tasks. (0 for all tasks)
356 extern void show_state_filter(unsigned long state_filter);
358 static inline void show_state(void)
360 show_state_filter(0);
363 extern void show_regs(struct pt_regs *);
366 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
367 * task), SP is the stack pointer of the first frame that should be shown in the back
368 * trace (or NULL if the entire call-chain of the task should be shown).
370 extern void show_stack(struct task_struct *task, unsigned long *sp);
372 extern void cpu_init (void);
373 extern void trap_init(void);
374 extern void update_process_times(int user);
375 extern void scheduler_tick(void);
377 extern void sched_show_task(struct task_struct *p);
379 #ifdef CONFIG_LOCKUP_DETECTOR
380 extern void touch_softlockup_watchdog(void);
381 extern void touch_softlockup_watchdog_sync(void);
382 extern void touch_all_softlockup_watchdogs(void);
383 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
385 size_t *lenp, loff_t *ppos);
386 extern unsigned int softlockup_panic;
387 void lockup_detector_init(void);
389 static inline void touch_softlockup_watchdog(void)
392 static inline void touch_softlockup_watchdog_sync(void)
395 static inline void touch_all_softlockup_watchdogs(void)
398 static inline void lockup_detector_init(void)
403 #ifdef CONFIG_DETECT_HUNG_TASK
404 void reset_hung_task_detector(void);
406 static inline void reset_hung_task_detector(void)
411 /* Attach to any functions which should be ignored in wchan output. */
412 #define __sched __attribute__((__section__(".sched.text")))
414 /* Linker adds these: start and end of __sched functions */
415 extern char __sched_text_start[], __sched_text_end[];
417 /* Is this address in the __sched functions? */
418 extern int in_sched_functions(unsigned long addr);
420 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
421 extern signed long schedule_timeout(signed long timeout);
422 extern signed long schedule_timeout_interruptible(signed long timeout);
423 extern signed long schedule_timeout_killable(signed long timeout);
424 extern signed long schedule_timeout_uninterruptible(signed long timeout);
425 asmlinkage void schedule(void);
426 extern void schedule_preempt_disabled(void);
428 extern long io_schedule_timeout(long timeout);
430 static inline void io_schedule(void)
432 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
436 struct user_namespace;
439 extern void arch_pick_mmap_layout(struct mm_struct *mm);
441 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
442 unsigned long, unsigned long);
444 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
445 unsigned long len, unsigned long pgoff,
446 unsigned long flags);
448 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
451 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
452 #define SUID_DUMP_USER 1 /* Dump as user of process */
453 #define SUID_DUMP_ROOT 2 /* Dump as root */
457 /* for SUID_DUMP_* above */
458 #define MMF_DUMPABLE_BITS 2
459 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
461 extern void set_dumpable(struct mm_struct *mm, int value);
463 * This returns the actual value of the suid_dumpable flag. For things
464 * that are using this for checking for privilege transitions, it must
465 * test against SUID_DUMP_USER rather than treating it as a boolean
468 static inline int __get_dumpable(unsigned long mm_flags)
470 return mm_flags & MMF_DUMPABLE_MASK;
473 static inline int get_dumpable(struct mm_struct *mm)
475 return __get_dumpable(mm->flags);
478 /* coredump filter bits */
479 #define MMF_DUMP_ANON_PRIVATE 2
480 #define MMF_DUMP_ANON_SHARED 3
481 #define MMF_DUMP_MAPPED_PRIVATE 4
482 #define MMF_DUMP_MAPPED_SHARED 5
483 #define MMF_DUMP_ELF_HEADERS 6
484 #define MMF_DUMP_HUGETLB_PRIVATE 7
485 #define MMF_DUMP_HUGETLB_SHARED 8
487 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
488 #define MMF_DUMP_FILTER_BITS 7
489 #define MMF_DUMP_FILTER_MASK \
490 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
491 #define MMF_DUMP_FILTER_DEFAULT \
492 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
493 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
495 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
496 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
498 # define MMF_DUMP_MASK_DEFAULT_ELF 0
500 /* leave room for more dump flags */
501 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
502 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
503 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
505 #define MMF_HAS_UPROBES 19 /* has uprobes */
506 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
508 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
510 struct sighand_struct {
512 struct k_sigaction action[_NSIG];
514 wait_queue_head_t signalfd_wqh;
517 struct pacct_struct {
520 unsigned long ac_mem;
521 cputime_t ac_utime, ac_stime;
522 unsigned long ac_minflt, ac_majflt;
533 * struct cputime - snaphsot of system and user cputime
534 * @utime: time spent in user mode
535 * @stime: time spent in system mode
537 * Gathers a generic snapshot of user and system time.
545 * struct task_cputime - collected CPU time counts
546 * @utime: time spent in user mode, in &cputime_t units
547 * @stime: time spent in kernel mode, in &cputime_t units
548 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
550 * This is an extension of struct cputime that includes the total runtime
551 * spent by the task from the scheduler point of view.
553 * As a result, this structure groups together three kinds of CPU time
554 * that are tracked for threads and thread groups. Most things considering
555 * CPU time want to group these counts together and treat all three
556 * of them in parallel.
558 struct task_cputime {
561 unsigned long long sum_exec_runtime;
563 /* Alternate field names when used to cache expirations. */
564 #define prof_exp stime
565 #define virt_exp utime
566 #define sched_exp sum_exec_runtime
568 #define INIT_CPUTIME \
569 (struct task_cputime) { \
572 .sum_exec_runtime = 0, \
576 * This is the atomic variant of task_cputime, which can be used for
577 * storing and updating task_cputime statistics without locking.
579 struct task_cputime_atomic {
582 atomic64_t sum_exec_runtime;
585 #define INIT_CPUTIME_ATOMIC \
586 (struct task_cputime_atomic) { \
587 .utime = ATOMIC64_INIT(0), \
588 .stime = ATOMIC64_INIT(0), \
589 .sum_exec_runtime = ATOMIC64_INIT(0), \
592 #ifdef CONFIG_PREEMPT_COUNT
593 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
595 #define PREEMPT_DISABLED PREEMPT_ENABLED
599 * Disable preemption until the scheduler is running.
600 * Reset by start_kernel()->sched_init()->init_idle().
602 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
603 * before the scheduler is active -- see should_resched().
605 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
608 * struct thread_group_cputimer - thread group interval timer counts
609 * @cputime_atomic: atomic thread group interval timers.
610 * @running: non-zero when there are timers running and
611 * @cputime receives updates.
613 * This structure contains the version of task_cputime, above, that is
614 * used for thread group CPU timer calculations.
616 struct thread_group_cputimer {
617 struct task_cputime_atomic cputime_atomic;
621 #include <linux/rwsem.h>
625 * NOTE! "signal_struct" does not have its own
626 * locking, because a shared signal_struct always
627 * implies a shared sighand_struct, so locking
628 * sighand_struct is always a proper superset of
629 * the locking of signal_struct.
631 struct signal_struct {
635 struct list_head thread_head;
637 wait_queue_head_t wait_chldexit; /* for wait4() */
639 /* current thread group signal load-balancing target: */
640 struct task_struct *curr_target;
642 /* shared signal handling: */
643 struct sigpending shared_pending;
645 /* thread group exit support */
648 * - notify group_exit_task when ->count is equal to notify_count
649 * - everyone except group_exit_task is stopped during signal delivery
650 * of fatal signals, group_exit_task processes the signal.
653 struct task_struct *group_exit_task;
655 /* thread group stop support, overloads group_exit_code too */
656 int group_stop_count;
657 unsigned int flags; /* see SIGNAL_* flags below */
660 * PR_SET_CHILD_SUBREAPER marks a process, like a service
661 * manager, to re-parent orphan (double-forking) child processes
662 * to this process instead of 'init'. The service manager is
663 * able to receive SIGCHLD signals and is able to investigate
664 * the process until it calls wait(). All children of this
665 * process will inherit a flag if they should look for a
666 * child_subreaper process at exit.
668 unsigned int is_child_subreaper:1;
669 unsigned int has_child_subreaper:1;
671 /* POSIX.1b Interval Timers */
673 struct list_head posix_timers;
675 /* ITIMER_REAL timer for the process */
676 struct hrtimer real_timer;
677 struct pid *leader_pid;
678 ktime_t it_real_incr;
681 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
682 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
683 * values are defined to 0 and 1 respectively
685 struct cpu_itimer it[2];
688 * Thread group totals for process CPU timers.
689 * See thread_group_cputimer(), et al, for details.
691 struct thread_group_cputimer cputimer;
693 /* Earliest-expiration cache. */
694 struct task_cputime cputime_expires;
696 struct list_head cpu_timers[3];
698 struct pid *tty_old_pgrp;
700 /* boolean value for session group leader */
703 struct tty_struct *tty; /* NULL if no tty */
705 #ifdef CONFIG_SCHED_AUTOGROUP
706 struct autogroup *autogroup;
709 * Cumulative resource counters for dead threads in the group,
710 * and for reaped dead child processes forked by this group.
711 * Live threads maintain their own counters and add to these
712 * in __exit_signal, except for the group leader.
714 seqlock_t stats_lock;
715 cputime_t utime, stime, cutime, cstime;
718 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
719 struct cputime prev_cputime;
721 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
722 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
723 unsigned long inblock, oublock, cinblock, coublock;
724 unsigned long maxrss, cmaxrss;
725 struct task_io_accounting ioac;
728 * Cumulative ns of schedule CPU time fo dead threads in the
729 * group, not including a zombie group leader, (This only differs
730 * from jiffies_to_ns(utime + stime) if sched_clock uses something
731 * other than jiffies.)
733 unsigned long long sum_sched_runtime;
736 * We don't bother to synchronize most readers of this at all,
737 * because there is no reader checking a limit that actually needs
738 * to get both rlim_cur and rlim_max atomically, and either one
739 * alone is a single word that can safely be read normally.
740 * getrlimit/setrlimit use task_lock(current->group_leader) to
741 * protect this instead of the siglock, because they really
742 * have no need to disable irqs.
744 struct rlimit rlim[RLIM_NLIMITS];
746 #ifdef CONFIG_BSD_PROCESS_ACCT
747 struct pacct_struct pacct; /* per-process accounting information */
749 #ifdef CONFIG_TASKSTATS
750 struct taskstats *stats;
754 unsigned audit_tty_log_passwd;
755 struct tty_audit_buf *tty_audit_buf;
757 #ifdef CONFIG_CGROUPS
759 * group_rwsem prevents new tasks from entering the threadgroup and
760 * member tasks from exiting,a more specifically, setting of
761 * PF_EXITING. fork and exit paths are protected with this rwsem
762 * using threadgroup_change_begin/end(). Users which require
763 * threadgroup to remain stable should use threadgroup_[un]lock()
764 * which also takes care of exec path. Currently, cgroup is the
767 struct rw_semaphore group_rwsem;
770 oom_flags_t oom_flags;
771 short oom_score_adj; /* OOM kill score adjustment */
772 short oom_score_adj_min; /* OOM kill score adjustment min value.
773 * Only settable by CAP_SYS_RESOURCE. */
775 struct mutex cred_guard_mutex; /* guard against foreign influences on
776 * credential calculations
777 * (notably. ptrace) */
781 * Bits in flags field of signal_struct.
783 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
784 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
785 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
786 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
788 * Pending notifications to parent.
790 #define SIGNAL_CLD_STOPPED 0x00000010
791 #define SIGNAL_CLD_CONTINUED 0x00000020
792 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
794 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
796 /* If true, all threads except ->group_exit_task have pending SIGKILL */
797 static inline int signal_group_exit(const struct signal_struct *sig)
799 return (sig->flags & SIGNAL_GROUP_EXIT) ||
800 (sig->group_exit_task != NULL);
804 * Some day this will be a full-fledged user tracking system..
807 atomic_t __count; /* reference count */
808 atomic_t processes; /* How many processes does this user have? */
809 atomic_t sigpending; /* How many pending signals does this user have? */
810 #ifdef CONFIG_INOTIFY_USER
811 atomic_t inotify_watches; /* How many inotify watches does this user have? */
812 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
814 #ifdef CONFIG_FANOTIFY
815 atomic_t fanotify_listeners;
818 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
820 #ifdef CONFIG_POSIX_MQUEUE
821 /* protected by mq_lock */
822 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
824 unsigned long locked_shm; /* How many pages of mlocked shm ? */
827 struct key *uid_keyring; /* UID specific keyring */
828 struct key *session_keyring; /* UID's default session keyring */
831 /* Hash table maintenance information */
832 struct hlist_node uidhash_node;
835 #ifdef CONFIG_PERF_EVENTS
836 atomic_long_t locked_vm;
840 extern int uids_sysfs_init(void);
842 extern struct user_struct *find_user(kuid_t);
844 extern struct user_struct root_user;
845 #define INIT_USER (&root_user)
848 struct backing_dev_info;
849 struct reclaim_state;
851 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
853 /* cumulative counters */
854 unsigned long pcount; /* # of times run on this cpu */
855 unsigned long long run_delay; /* time spent waiting on a runqueue */
858 unsigned long long last_arrival,/* when we last ran on a cpu */
859 last_queued; /* when we were last queued to run */
861 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
863 #ifdef CONFIG_TASK_DELAY_ACCT
864 struct task_delay_info {
866 unsigned int flags; /* Private per-task flags */
868 /* For each stat XXX, add following, aligned appropriately
870 * struct timespec XXX_start, XXX_end;
874 * Atomicity of updates to XXX_delay, XXX_count protected by
875 * single lock above (split into XXX_lock if contention is an issue).
879 * XXX_count is incremented on every XXX operation, the delay
880 * associated with the operation is added to XXX_delay.
881 * XXX_delay contains the accumulated delay time in nanoseconds.
883 u64 blkio_start; /* Shared by blkio, swapin */
884 u64 blkio_delay; /* wait for sync block io completion */
885 u64 swapin_delay; /* wait for swapin block io completion */
886 u32 blkio_count; /* total count of the number of sync block */
887 /* io operations performed */
888 u32 swapin_count; /* total count of the number of swapin block */
889 /* io operations performed */
892 u64 freepages_delay; /* wait for memory reclaim */
893 u32 freepages_count; /* total count of memory reclaim */
895 #endif /* CONFIG_TASK_DELAY_ACCT */
897 static inline int sched_info_on(void)
899 #ifdef CONFIG_SCHEDSTATS
901 #elif defined(CONFIG_TASK_DELAY_ACCT)
902 extern int delayacct_on;
917 * Increase resolution of cpu_capacity calculations
919 #define SCHED_CAPACITY_SHIFT 10
920 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
923 * Wake-queues are lists of tasks with a pending wakeup, whose
924 * callers have already marked the task as woken internally,
925 * and can thus carry on. A common use case is being able to
926 * do the wakeups once the corresponding user lock as been
929 * We hold reference to each task in the list across the wakeup,
930 * thus guaranteeing that the memory is still valid by the time
931 * the actual wakeups are performed in wake_up_q().
933 * One per task suffices, because there's never a need for a task to be
934 * in two wake queues simultaneously; it is forbidden to abandon a task
935 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
936 * already in a wake queue, the wakeup will happen soon and the second
937 * waker can just skip it.
939 * The WAKE_Q macro declares and initializes the list head.
940 * wake_up_q() does NOT reinitialize the list; it's expected to be
941 * called near the end of a function, where the fact that the queue is
942 * not used again will be easy to see by inspection.
944 * Note that this can cause spurious wakeups. schedule() callers
945 * must ensure the call is done inside a loop, confirming that the
946 * wakeup condition has in fact occurred.
949 struct wake_q_node *next;
953 struct wake_q_node *first;
954 struct wake_q_node **lastp;
957 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
959 #define WAKE_Q(name) \
960 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
962 extern void wake_q_add(struct wake_q_head *head,
963 struct task_struct *task);
964 extern void wake_up_q(struct wake_q_head *head);
967 * sched-domains (multiprocessor balancing) declarations:
970 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
971 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
972 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
973 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
974 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
975 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
976 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
977 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
978 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
979 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
980 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
981 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
982 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
983 #define SD_NUMA 0x4000 /* cross-node balancing */
985 #ifdef CONFIG_SCHED_SMT
986 static inline int cpu_smt_flags(void)
988 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
992 #ifdef CONFIG_SCHED_MC
993 static inline int cpu_core_flags(void)
995 return SD_SHARE_PKG_RESOURCES;
1000 static inline int cpu_numa_flags(void)
1006 struct sched_domain_attr {
1007 int relax_domain_level;
1010 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1011 .relax_domain_level = -1, \
1014 extern int sched_domain_level_max;
1018 struct sched_domain {
1019 /* These fields must be setup */
1020 struct sched_domain *parent; /* top domain must be null terminated */
1021 struct sched_domain *child; /* bottom domain must be null terminated */
1022 struct sched_group *groups; /* the balancing groups of the domain */
1023 unsigned long min_interval; /* Minimum balance interval ms */
1024 unsigned long max_interval; /* Maximum balance interval ms */
1025 unsigned int busy_factor; /* less balancing by factor if busy */
1026 unsigned int imbalance_pct; /* No balance until over watermark */
1027 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1028 unsigned int busy_idx;
1029 unsigned int idle_idx;
1030 unsigned int newidle_idx;
1031 unsigned int wake_idx;
1032 unsigned int forkexec_idx;
1033 unsigned int smt_gain;
1035 int nohz_idle; /* NOHZ IDLE status */
1036 int flags; /* See SD_* */
1039 /* Runtime fields. */
1040 unsigned long last_balance; /* init to jiffies. units in jiffies */
1041 unsigned int balance_interval; /* initialise to 1. units in ms. */
1042 unsigned int nr_balance_failed; /* initialise to 0 */
1044 /* idle_balance() stats */
1045 u64 max_newidle_lb_cost;
1046 unsigned long next_decay_max_lb_cost;
1048 #ifdef CONFIG_SCHEDSTATS
1049 /* load_balance() stats */
1050 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1051 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1052 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1053 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1054 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1055 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1056 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1057 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1059 /* Active load balancing */
1060 unsigned int alb_count;
1061 unsigned int alb_failed;
1062 unsigned int alb_pushed;
1064 /* SD_BALANCE_EXEC stats */
1065 unsigned int sbe_count;
1066 unsigned int sbe_balanced;
1067 unsigned int sbe_pushed;
1069 /* SD_BALANCE_FORK stats */
1070 unsigned int sbf_count;
1071 unsigned int sbf_balanced;
1072 unsigned int sbf_pushed;
1074 /* try_to_wake_up() stats */
1075 unsigned int ttwu_wake_remote;
1076 unsigned int ttwu_move_affine;
1077 unsigned int ttwu_move_balance;
1079 #ifdef CONFIG_SCHED_DEBUG
1083 void *private; /* used during construction */
1084 struct rcu_head rcu; /* used during destruction */
1087 unsigned int span_weight;
1089 * Span of all CPUs in this domain.
1091 * NOTE: this field is variable length. (Allocated dynamically
1092 * by attaching extra space to the end of the structure,
1093 * depending on how many CPUs the kernel has booted up with)
1095 unsigned long span[0];
1098 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1100 return to_cpumask(sd->span);
1103 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1104 struct sched_domain_attr *dattr_new);
1106 /* Allocate an array of sched domains, for partition_sched_domains(). */
1107 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1108 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1110 bool cpus_share_cache(int this_cpu, int that_cpu);
1112 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1113 typedef int (*sched_domain_flags_f)(void);
1115 #define SDTL_OVERLAP 0x01
1118 struct sched_domain **__percpu sd;
1119 struct sched_group **__percpu sg;
1120 struct sched_group_capacity **__percpu sgc;
1123 struct sched_domain_topology_level {
1124 sched_domain_mask_f mask;
1125 sched_domain_flags_f sd_flags;
1128 struct sd_data data;
1129 #ifdef CONFIG_SCHED_DEBUG
1134 extern struct sched_domain_topology_level *sched_domain_topology;
1136 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1137 extern void wake_up_if_idle(int cpu);
1139 #ifdef CONFIG_SCHED_DEBUG
1140 # define SD_INIT_NAME(type) .name = #type
1142 # define SD_INIT_NAME(type)
1145 #else /* CONFIG_SMP */
1147 struct sched_domain_attr;
1150 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1151 struct sched_domain_attr *dattr_new)
1155 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1160 #endif /* !CONFIG_SMP */
1163 struct io_context; /* See blkdev.h */
1166 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1167 extern void prefetch_stack(struct task_struct *t);
1169 static inline void prefetch_stack(struct task_struct *t) { }
1172 struct audit_context; /* See audit.c */
1174 struct pipe_inode_info;
1175 struct uts_namespace;
1177 struct load_weight {
1178 unsigned long weight;
1183 u64 last_runnable_update;
1186 * utilization_avg_contrib describes the amount of time that a
1187 * sched_entity is running on a CPU. It is based on running_avg_sum
1188 * and is scaled in the range [0..SCHED_LOAD_SCALE].
1189 * load_avg_contrib described the amount of time that a sched_entity
1190 * is runnable on a rq. It is based on both runnable_avg_sum and the
1191 * weight of the task.
1193 unsigned long load_avg_contrib, utilization_avg_contrib;
1195 * These sums represent an infinite geometric series and so are bound
1196 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1197 * choices of y < 1-2^(-32)*1024.
1198 * running_avg_sum reflects the time that the sched_entity is
1199 * effectively running on the CPU.
1200 * runnable_avg_sum represents the amount of time a sched_entity is on
1201 * a runqueue which includes the running time that is monitored by
1204 u32 runnable_avg_sum, avg_period, running_avg_sum;
1207 #ifdef CONFIG_SCHEDSTATS
1208 struct sched_statistics {
1218 s64 sum_sleep_runtime;
1225 u64 nr_migrations_cold;
1226 u64 nr_failed_migrations_affine;
1227 u64 nr_failed_migrations_running;
1228 u64 nr_failed_migrations_hot;
1229 u64 nr_forced_migrations;
1232 u64 nr_wakeups_sync;
1233 u64 nr_wakeups_migrate;
1234 u64 nr_wakeups_local;
1235 u64 nr_wakeups_remote;
1236 u64 nr_wakeups_affine;
1237 u64 nr_wakeups_affine_attempts;
1238 u64 nr_wakeups_passive;
1239 u64 nr_wakeups_idle;
1243 struct sched_entity {
1244 struct load_weight load; /* for load-balancing */
1245 struct rb_node run_node;
1246 struct list_head group_node;
1250 u64 sum_exec_runtime;
1252 u64 prev_sum_exec_runtime;
1256 #ifdef CONFIG_SCHEDSTATS
1257 struct sched_statistics statistics;
1260 #ifdef CONFIG_FAIR_GROUP_SCHED
1262 struct sched_entity *parent;
1263 /* rq on which this entity is (to be) queued: */
1264 struct cfs_rq *cfs_rq;
1265 /* rq "owned" by this entity/group: */
1266 struct cfs_rq *my_q;
1270 /* Per-entity load-tracking */
1271 struct sched_avg avg;
1275 struct sched_rt_entity {
1276 struct list_head run_list;
1277 unsigned long timeout;
1278 unsigned long watchdog_stamp;
1279 unsigned int time_slice;
1281 struct sched_rt_entity *back;
1282 #ifdef CONFIG_RT_GROUP_SCHED
1283 struct sched_rt_entity *parent;
1284 /* rq on which this entity is (to be) queued: */
1285 struct rt_rq *rt_rq;
1286 /* rq "owned" by this entity/group: */
1291 struct sched_dl_entity {
1292 struct rb_node rb_node;
1295 * Original scheduling parameters. Copied here from sched_attr
1296 * during sched_setattr(), they will remain the same until
1297 * the next sched_setattr().
1299 u64 dl_runtime; /* maximum runtime for each instance */
1300 u64 dl_deadline; /* relative deadline of each instance */
1301 u64 dl_period; /* separation of two instances (period) */
1302 u64 dl_bw; /* dl_runtime / dl_deadline */
1305 * Actual scheduling parameters. Initialized with the values above,
1306 * they are continously updated during task execution. Note that
1307 * the remaining runtime could be < 0 in case we are in overrun.
1309 s64 runtime; /* remaining runtime for this instance */
1310 u64 deadline; /* absolute deadline for this instance */
1311 unsigned int flags; /* specifying the scheduler behaviour */
1316 * @dl_throttled tells if we exhausted the runtime. If so, the
1317 * task has to wait for a replenishment to be performed at the
1318 * next firing of dl_timer.
1320 * @dl_new tells if a new instance arrived. If so we must
1321 * start executing it with full runtime and reset its absolute
1324 * @dl_boosted tells if we are boosted due to DI. If so we are
1325 * outside bandwidth enforcement mechanism (but only until we
1326 * exit the critical section);
1328 * @dl_yielded tells if task gave up the cpu before consuming
1329 * all its available runtime during the last job.
1331 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1334 * Bandwidth enforcement timer. Each -deadline task has its
1335 * own bandwidth to be enforced, thus we need one timer per task.
1337 struct hrtimer dl_timer;
1349 enum perf_event_task_context {
1350 perf_invalid_context = -1,
1351 perf_hw_context = 0,
1353 perf_nr_task_contexts,
1356 struct task_struct {
1357 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1360 unsigned int flags; /* per process flags, defined below */
1361 unsigned int ptrace;
1364 struct llist_node wake_entry;
1366 struct task_struct *last_wakee;
1367 unsigned long wakee_flips;
1368 unsigned long wakee_flip_decay_ts;
1374 int prio, static_prio, normal_prio;
1375 unsigned int rt_priority;
1376 const struct sched_class *sched_class;
1377 struct sched_entity se;
1378 struct sched_rt_entity rt;
1379 #ifdef CONFIG_CGROUP_SCHED
1380 struct task_group *sched_task_group;
1382 struct sched_dl_entity dl;
1384 #ifdef CONFIG_PREEMPT_NOTIFIERS
1385 /* list of struct preempt_notifier: */
1386 struct hlist_head preempt_notifiers;
1389 #ifdef CONFIG_BLK_DEV_IO_TRACE
1390 unsigned int btrace_seq;
1393 unsigned int policy;
1394 int nr_cpus_allowed;
1395 cpumask_t cpus_allowed;
1397 #ifdef CONFIG_PREEMPT_RCU
1398 int rcu_read_lock_nesting;
1399 union rcu_special rcu_read_unlock_special;
1400 struct list_head rcu_node_entry;
1401 struct rcu_node *rcu_blocked_node;
1402 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1403 #ifdef CONFIG_TASKS_RCU
1404 unsigned long rcu_tasks_nvcsw;
1405 bool rcu_tasks_holdout;
1406 struct list_head rcu_tasks_holdout_list;
1407 int rcu_tasks_idle_cpu;
1408 #endif /* #ifdef CONFIG_TASKS_RCU */
1410 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1411 struct sched_info sched_info;
1414 struct list_head tasks;
1416 struct plist_node pushable_tasks;
1417 struct rb_node pushable_dl_tasks;
1420 struct mm_struct *mm, *active_mm;
1421 #ifdef CONFIG_COMPAT_BRK
1422 unsigned brk_randomized:1;
1424 /* per-thread vma caching */
1425 u32 vmacache_seqnum;
1426 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1427 #if defined(SPLIT_RSS_COUNTING)
1428 struct task_rss_stat rss_stat;
1432 int exit_code, exit_signal;
1433 int pdeath_signal; /* The signal sent when the parent dies */
1434 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1436 /* Used for emulating ABI behavior of previous Linux versions */
1437 unsigned int personality;
1439 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1441 unsigned in_iowait:1;
1443 /* Revert to default priority/policy when forking */
1444 unsigned sched_reset_on_fork:1;
1445 unsigned sched_contributes_to_load:1;
1447 #ifdef CONFIG_MEMCG_KMEM
1448 unsigned memcg_kmem_skip_account:1;
1451 unsigned long atomic_flags; /* Flags needing atomic access. */
1453 struct restart_block restart_block;
1458 #ifdef CONFIG_CC_STACKPROTECTOR
1459 /* Canary value for the -fstack-protector gcc feature */
1460 unsigned long stack_canary;
1463 * pointers to (original) parent process, youngest child, younger sibling,
1464 * older sibling, respectively. (p->father can be replaced with
1465 * p->real_parent->pid)
1467 struct task_struct __rcu *real_parent; /* real parent process */
1468 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1470 * children/sibling forms the list of my natural children
1472 struct list_head children; /* list of my children */
1473 struct list_head sibling; /* linkage in my parent's children list */
1474 struct task_struct *group_leader; /* threadgroup leader */
1477 * ptraced is the list of tasks this task is using ptrace on.
1478 * This includes both natural children and PTRACE_ATTACH targets.
1479 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1481 struct list_head ptraced;
1482 struct list_head ptrace_entry;
1484 /* PID/PID hash table linkage. */
1485 struct pid_link pids[PIDTYPE_MAX];
1486 struct list_head thread_group;
1487 struct list_head thread_node;
1489 struct completion *vfork_done; /* for vfork() */
1490 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1491 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1493 cputime_t utime, stime, utimescaled, stimescaled;
1495 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1496 struct cputime prev_cputime;
1498 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1499 seqlock_t vtime_seqlock;
1500 unsigned long long vtime_snap;
1505 } vtime_snap_whence;
1507 unsigned long nvcsw, nivcsw; /* context switch counts */
1508 u64 start_time; /* monotonic time in nsec */
1509 u64 real_start_time; /* boot based time in nsec */
1510 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1511 unsigned long min_flt, maj_flt;
1513 struct task_cputime cputime_expires;
1514 struct list_head cpu_timers[3];
1516 /* process credentials */
1517 const struct cred __rcu *real_cred; /* objective and real subjective task
1518 * credentials (COW) */
1519 const struct cred __rcu *cred; /* effective (overridable) subjective task
1520 * credentials (COW) */
1521 char comm[TASK_COMM_LEN]; /* executable name excluding path
1522 - access with [gs]et_task_comm (which lock
1523 it with task_lock())
1524 - initialized normally by setup_new_exec */
1525 /* file system info */
1526 int link_count, total_link_count;
1527 #ifdef CONFIG_SYSVIPC
1529 struct sysv_sem sysvsem;
1530 struct sysv_shm sysvshm;
1532 #ifdef CONFIG_DETECT_HUNG_TASK
1533 /* hung task detection */
1534 unsigned long last_switch_count;
1536 /* CPU-specific state of this task */
1537 struct thread_struct thread;
1538 /* filesystem information */
1539 struct fs_struct *fs;
1540 /* open file information */
1541 struct files_struct *files;
1543 struct nsproxy *nsproxy;
1544 /* signal handlers */
1545 struct signal_struct *signal;
1546 struct sighand_struct *sighand;
1548 sigset_t blocked, real_blocked;
1549 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1550 struct sigpending pending;
1552 unsigned long sas_ss_sp;
1554 int (*notifier)(void *priv);
1555 void *notifier_data;
1556 sigset_t *notifier_mask;
1557 struct callback_head *task_works;
1559 struct audit_context *audit_context;
1560 #ifdef CONFIG_AUDITSYSCALL
1562 unsigned int sessionid;
1564 struct seccomp seccomp;
1566 /* Thread group tracking */
1569 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1571 spinlock_t alloc_lock;
1573 /* Protection of the PI data structures: */
1574 raw_spinlock_t pi_lock;
1576 struct wake_q_node wake_q;
1578 #ifdef CONFIG_RT_MUTEXES
1579 /* PI waiters blocked on a rt_mutex held by this task */
1580 struct rb_root pi_waiters;
1581 struct rb_node *pi_waiters_leftmost;
1582 /* Deadlock detection and priority inheritance handling */
1583 struct rt_mutex_waiter *pi_blocked_on;
1586 #ifdef CONFIG_DEBUG_MUTEXES
1587 /* mutex deadlock detection */
1588 struct mutex_waiter *blocked_on;
1590 #ifdef CONFIG_TRACE_IRQFLAGS
1591 unsigned int irq_events;
1592 unsigned long hardirq_enable_ip;
1593 unsigned long hardirq_disable_ip;
1594 unsigned int hardirq_enable_event;
1595 unsigned int hardirq_disable_event;
1596 int hardirqs_enabled;
1597 int hardirq_context;
1598 unsigned long softirq_disable_ip;
1599 unsigned long softirq_enable_ip;
1600 unsigned int softirq_disable_event;
1601 unsigned int softirq_enable_event;
1602 int softirqs_enabled;
1603 int softirq_context;
1605 #ifdef CONFIG_LOCKDEP
1606 # define MAX_LOCK_DEPTH 48UL
1609 unsigned int lockdep_recursion;
1610 struct held_lock held_locks[MAX_LOCK_DEPTH];
1611 gfp_t lockdep_reclaim_gfp;
1614 /* journalling filesystem info */
1617 /* stacked block device info */
1618 struct bio_list *bio_list;
1621 /* stack plugging */
1622 struct blk_plug *plug;
1626 struct reclaim_state *reclaim_state;
1628 struct backing_dev_info *backing_dev_info;
1630 struct io_context *io_context;
1632 unsigned long ptrace_message;
1633 siginfo_t *last_siginfo; /* For ptrace use. */
1634 struct task_io_accounting ioac;
1635 #if defined(CONFIG_TASK_XACCT)
1636 u64 acct_rss_mem1; /* accumulated rss usage */
1637 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1638 cputime_t acct_timexpd; /* stime + utime since last update */
1640 #ifdef CONFIG_CPUSETS
1641 nodemask_t mems_allowed; /* Protected by alloc_lock */
1642 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1643 int cpuset_mem_spread_rotor;
1644 int cpuset_slab_spread_rotor;
1646 #ifdef CONFIG_CGROUPS
1647 /* Control Group info protected by css_set_lock */
1648 struct css_set __rcu *cgroups;
1649 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1650 struct list_head cg_list;
1653 struct robust_list_head __user *robust_list;
1654 #ifdef CONFIG_COMPAT
1655 struct compat_robust_list_head __user *compat_robust_list;
1657 struct list_head pi_state_list;
1658 struct futex_pi_state *pi_state_cache;
1660 #ifdef CONFIG_PERF_EVENTS
1661 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1662 struct mutex perf_event_mutex;
1663 struct list_head perf_event_list;
1665 #ifdef CONFIG_DEBUG_PREEMPT
1666 unsigned long preempt_disable_ip;
1669 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1671 short pref_node_fork;
1673 #ifdef CONFIG_NUMA_BALANCING
1675 unsigned int numa_scan_period;
1676 unsigned int numa_scan_period_max;
1677 int numa_preferred_nid;
1678 unsigned long numa_migrate_retry;
1679 u64 node_stamp; /* migration stamp */
1680 u64 last_task_numa_placement;
1681 u64 last_sum_exec_runtime;
1682 struct callback_head numa_work;
1684 struct list_head numa_entry;
1685 struct numa_group *numa_group;
1688 * numa_faults is an array split into four regions:
1689 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1690 * in this precise order.
1692 * faults_memory: Exponential decaying average of faults on a per-node
1693 * basis. Scheduling placement decisions are made based on these
1694 * counts. The values remain static for the duration of a PTE scan.
1695 * faults_cpu: Track the nodes the process was running on when a NUMA
1696 * hinting fault was incurred.
1697 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1698 * during the current scan window. When the scan completes, the counts
1699 * in faults_memory and faults_cpu decay and these values are copied.
1701 unsigned long *numa_faults;
1702 unsigned long total_numa_faults;
1705 * numa_faults_locality tracks if faults recorded during the last
1706 * scan window were remote/local or failed to migrate. The task scan
1707 * period is adapted based on the locality of the faults with different
1708 * weights depending on whether they were shared or private faults
1710 unsigned long numa_faults_locality[3];
1712 unsigned long numa_pages_migrated;
1713 #endif /* CONFIG_NUMA_BALANCING */
1715 struct rcu_head rcu;
1718 * cache last used pipe for splice
1720 struct pipe_inode_info *splice_pipe;
1722 struct page_frag task_frag;
1724 #ifdef CONFIG_TASK_DELAY_ACCT
1725 struct task_delay_info *delays;
1727 #ifdef CONFIG_FAULT_INJECTION
1731 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1732 * balance_dirty_pages() for some dirty throttling pause
1735 int nr_dirtied_pause;
1736 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1738 #ifdef CONFIG_LATENCYTOP
1739 int latency_record_count;
1740 struct latency_record latency_record[LT_SAVECOUNT];
1743 * time slack values; these are used to round up poll() and
1744 * select() etc timeout values. These are in nanoseconds.
1746 unsigned long timer_slack_ns;
1747 unsigned long default_timer_slack_ns;
1750 unsigned int kasan_depth;
1752 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1753 /* Index of current stored address in ret_stack */
1755 /* Stack of return addresses for return function tracing */
1756 struct ftrace_ret_stack *ret_stack;
1757 /* time stamp for last schedule */
1758 unsigned long long ftrace_timestamp;
1760 * Number of functions that haven't been traced
1761 * because of depth overrun.
1763 atomic_t trace_overrun;
1764 /* Pause for the tracing */
1765 atomic_t tracing_graph_pause;
1767 #ifdef CONFIG_TRACING
1768 /* state flags for use by tracers */
1769 unsigned long trace;
1770 /* bitmask and counter of trace recursion */
1771 unsigned long trace_recursion;
1772 #endif /* CONFIG_TRACING */
1774 struct memcg_oom_info {
1775 struct mem_cgroup *memcg;
1778 unsigned int may_oom:1;
1781 #ifdef CONFIG_UPROBES
1782 struct uprobe_task *utask;
1784 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1785 unsigned int sequential_io;
1786 unsigned int sequential_io_avg;
1788 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1789 unsigned long task_state_change;
1793 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1794 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1796 #define TNF_MIGRATED 0x01
1797 #define TNF_NO_GROUP 0x02
1798 #define TNF_SHARED 0x04
1799 #define TNF_FAULT_LOCAL 0x08
1800 #define TNF_MIGRATE_FAIL 0x10
1802 #ifdef CONFIG_NUMA_BALANCING
1803 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1804 extern pid_t task_numa_group_id(struct task_struct *p);
1805 extern void set_numabalancing_state(bool enabled);
1806 extern void task_numa_free(struct task_struct *p);
1807 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1808 int src_nid, int dst_cpu);
1810 static inline void task_numa_fault(int last_node, int node, int pages,
1814 static inline pid_t task_numa_group_id(struct task_struct *p)
1818 static inline void set_numabalancing_state(bool enabled)
1821 static inline void task_numa_free(struct task_struct *p)
1824 static inline bool should_numa_migrate_memory(struct task_struct *p,
1825 struct page *page, int src_nid, int dst_cpu)
1831 static inline struct pid *task_pid(struct task_struct *task)
1833 return task->pids[PIDTYPE_PID].pid;
1836 static inline struct pid *task_tgid(struct task_struct *task)
1838 return task->group_leader->pids[PIDTYPE_PID].pid;
1842 * Without tasklist or rcu lock it is not safe to dereference
1843 * the result of task_pgrp/task_session even if task == current,
1844 * we can race with another thread doing sys_setsid/sys_setpgid.
1846 static inline struct pid *task_pgrp(struct task_struct *task)
1848 return task->group_leader->pids[PIDTYPE_PGID].pid;
1851 static inline struct pid *task_session(struct task_struct *task)
1853 return task->group_leader->pids[PIDTYPE_SID].pid;
1856 struct pid_namespace;
1859 * the helpers to get the task's different pids as they are seen
1860 * from various namespaces
1862 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1863 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1865 * task_xid_nr_ns() : id seen from the ns specified;
1867 * set_task_vxid() : assigns a virtual id to a task;
1869 * see also pid_nr() etc in include/linux/pid.h
1871 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1872 struct pid_namespace *ns);
1874 static inline pid_t task_pid_nr(struct task_struct *tsk)
1879 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1880 struct pid_namespace *ns)
1882 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1885 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1887 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1891 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1896 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1898 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1900 return pid_vnr(task_tgid(tsk));
1904 static inline int pid_alive(const struct task_struct *p);
1905 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1911 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1917 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1919 return task_ppid_nr_ns(tsk, &init_pid_ns);
1922 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1923 struct pid_namespace *ns)
1925 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1928 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1930 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1934 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1935 struct pid_namespace *ns)
1937 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1940 static inline pid_t task_session_vnr(struct task_struct *tsk)
1942 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1945 /* obsolete, do not use */
1946 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1948 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1952 * pid_alive - check that a task structure is not stale
1953 * @p: Task structure to be checked.
1955 * Test if a process is not yet dead (at most zombie state)
1956 * If pid_alive fails, then pointers within the task structure
1957 * can be stale and must not be dereferenced.
1959 * Return: 1 if the process is alive. 0 otherwise.
1961 static inline int pid_alive(const struct task_struct *p)
1963 return p->pids[PIDTYPE_PID].pid != NULL;
1967 * is_global_init - check if a task structure is init
1968 * @tsk: Task structure to be checked.
1970 * Check if a task structure is the first user space task the kernel created.
1972 * Return: 1 if the task structure is init. 0 otherwise.
1974 static inline int is_global_init(struct task_struct *tsk)
1976 return tsk->pid == 1;
1979 extern struct pid *cad_pid;
1981 extern void free_task(struct task_struct *tsk);
1982 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1984 extern void __put_task_struct(struct task_struct *t);
1986 static inline void put_task_struct(struct task_struct *t)
1988 if (atomic_dec_and_test(&t->usage))
1989 __put_task_struct(t);
1992 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1993 extern void task_cputime(struct task_struct *t,
1994 cputime_t *utime, cputime_t *stime);
1995 extern void task_cputime_scaled(struct task_struct *t,
1996 cputime_t *utimescaled, cputime_t *stimescaled);
1997 extern cputime_t task_gtime(struct task_struct *t);
1999 static inline void task_cputime(struct task_struct *t,
2000 cputime_t *utime, cputime_t *stime)
2008 static inline void task_cputime_scaled(struct task_struct *t,
2009 cputime_t *utimescaled,
2010 cputime_t *stimescaled)
2013 *utimescaled = t->utimescaled;
2015 *stimescaled = t->stimescaled;
2018 static inline cputime_t task_gtime(struct task_struct *t)
2023 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2024 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2029 #define PF_EXITING 0x00000004 /* getting shut down */
2030 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2031 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2032 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2033 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2034 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2035 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2036 #define PF_DUMPCORE 0x00000200 /* dumped core */
2037 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2038 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2039 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2040 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2041 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2042 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2043 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2044 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2045 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2046 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2047 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2048 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2049 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2050 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2051 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2052 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2053 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2054 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2055 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2058 * Only the _current_ task can read/write to tsk->flags, but other
2059 * tasks can access tsk->flags in readonly mode for example
2060 * with tsk_used_math (like during threaded core dumping).
2061 * There is however an exception to this rule during ptrace
2062 * or during fork: the ptracer task is allowed to write to the
2063 * child->flags of its traced child (same goes for fork, the parent
2064 * can write to the child->flags), because we're guaranteed the
2065 * child is not running and in turn not changing child->flags
2066 * at the same time the parent does it.
2068 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2069 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2070 #define clear_used_math() clear_stopped_child_used_math(current)
2071 #define set_used_math() set_stopped_child_used_math(current)
2072 #define conditional_stopped_child_used_math(condition, child) \
2073 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2074 #define conditional_used_math(condition) \
2075 conditional_stopped_child_used_math(condition, current)
2076 #define copy_to_stopped_child_used_math(child) \
2077 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2078 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2079 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2080 #define used_math() tsk_used_math(current)
2082 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2083 * __GFP_FS is also cleared as it implies __GFP_IO.
2085 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2087 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2088 flags &= ~(__GFP_IO | __GFP_FS);
2092 static inline unsigned int memalloc_noio_save(void)
2094 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2095 current->flags |= PF_MEMALLOC_NOIO;
2099 static inline void memalloc_noio_restore(unsigned int flags)
2101 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2104 /* Per-process atomic flags. */
2105 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2106 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2107 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2110 #define TASK_PFA_TEST(name, func) \
2111 static inline bool task_##func(struct task_struct *p) \
2112 { return test_bit(PFA_##name, &p->atomic_flags); }
2113 #define TASK_PFA_SET(name, func) \
2114 static inline void task_set_##func(struct task_struct *p) \
2115 { set_bit(PFA_##name, &p->atomic_flags); }
2116 #define TASK_PFA_CLEAR(name, func) \
2117 static inline void task_clear_##func(struct task_struct *p) \
2118 { clear_bit(PFA_##name, &p->atomic_flags); }
2120 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2121 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2123 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2124 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2125 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2127 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2128 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2129 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2132 * task->jobctl flags
2134 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2136 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2137 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2138 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2139 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2140 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2141 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2142 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2144 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2145 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2146 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2147 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2148 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2149 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2150 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2152 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2153 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2155 extern bool task_set_jobctl_pending(struct task_struct *task,
2156 unsigned long mask);
2157 extern void task_clear_jobctl_trapping(struct task_struct *task);
2158 extern void task_clear_jobctl_pending(struct task_struct *task,
2159 unsigned long mask);
2161 static inline void rcu_copy_process(struct task_struct *p)
2163 #ifdef CONFIG_PREEMPT_RCU
2164 p->rcu_read_lock_nesting = 0;
2165 p->rcu_read_unlock_special.s = 0;
2166 p->rcu_blocked_node = NULL;
2167 INIT_LIST_HEAD(&p->rcu_node_entry);
2168 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2169 #ifdef CONFIG_TASKS_RCU
2170 p->rcu_tasks_holdout = false;
2171 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2172 p->rcu_tasks_idle_cpu = -1;
2173 #endif /* #ifdef CONFIG_TASKS_RCU */
2176 static inline void tsk_restore_flags(struct task_struct *task,
2177 unsigned long orig_flags, unsigned long flags)
2179 task->flags &= ~flags;
2180 task->flags |= orig_flags & flags;
2183 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2184 const struct cpumask *trial);
2185 extern int task_can_attach(struct task_struct *p,
2186 const struct cpumask *cs_cpus_allowed);
2188 extern void do_set_cpus_allowed(struct task_struct *p,
2189 const struct cpumask *new_mask);
2191 extern int set_cpus_allowed_ptr(struct task_struct *p,
2192 const struct cpumask *new_mask);
2194 static inline void do_set_cpus_allowed(struct task_struct *p,
2195 const struct cpumask *new_mask)
2198 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2199 const struct cpumask *new_mask)
2201 if (!cpumask_test_cpu(0, new_mask))
2207 #ifdef CONFIG_NO_HZ_COMMON
2208 void calc_load_enter_idle(void);
2209 void calc_load_exit_idle(void);
2211 static inline void calc_load_enter_idle(void) { }
2212 static inline void calc_load_exit_idle(void) { }
2213 #endif /* CONFIG_NO_HZ_COMMON */
2215 #ifndef CONFIG_CPUMASK_OFFSTACK
2216 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2218 return set_cpus_allowed_ptr(p, &new_mask);
2223 * Do not use outside of architecture code which knows its limitations.
2225 * sched_clock() has no promise of monotonicity or bounded drift between
2226 * CPUs, use (which you should not) requires disabling IRQs.
2228 * Please use one of the three interfaces below.
2230 extern unsigned long long notrace sched_clock(void);
2232 * See the comment in kernel/sched/clock.c
2234 extern u64 cpu_clock(int cpu);
2235 extern u64 local_clock(void);
2236 extern u64 running_clock(void);
2237 extern u64 sched_clock_cpu(int cpu);
2240 extern void sched_clock_init(void);
2242 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2243 static inline void sched_clock_tick(void)
2247 static inline void sched_clock_idle_sleep_event(void)
2251 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2256 * Architectures can set this to 1 if they have specified
2257 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2258 * but then during bootup it turns out that sched_clock()
2259 * is reliable after all:
2261 extern int sched_clock_stable(void);
2262 extern void set_sched_clock_stable(void);
2263 extern void clear_sched_clock_stable(void);
2265 extern void sched_clock_tick(void);
2266 extern void sched_clock_idle_sleep_event(void);
2267 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2270 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2272 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2273 * The reason for this explicit opt-in is not to have perf penalty with
2274 * slow sched_clocks.
2276 extern void enable_sched_clock_irqtime(void);
2277 extern void disable_sched_clock_irqtime(void);
2279 static inline void enable_sched_clock_irqtime(void) {}
2280 static inline void disable_sched_clock_irqtime(void) {}
2283 extern unsigned long long
2284 task_sched_runtime(struct task_struct *task);
2286 /* sched_exec is called by processes performing an exec */
2288 extern void sched_exec(void);
2290 #define sched_exec() {}
2293 extern void sched_clock_idle_sleep_event(void);
2294 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2296 #ifdef CONFIG_HOTPLUG_CPU
2297 extern void idle_task_exit(void);
2299 static inline void idle_task_exit(void) {}
2302 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2303 extern void wake_up_nohz_cpu(int cpu);
2305 static inline void wake_up_nohz_cpu(int cpu) { }
2308 #ifdef CONFIG_NO_HZ_FULL
2309 extern bool sched_can_stop_tick(void);
2310 extern u64 scheduler_tick_max_deferment(void);
2312 static inline bool sched_can_stop_tick(void) { return false; }
2315 #ifdef CONFIG_SCHED_AUTOGROUP
2316 extern void sched_autogroup_create_attach(struct task_struct *p);
2317 extern void sched_autogroup_detach(struct task_struct *p);
2318 extern void sched_autogroup_fork(struct signal_struct *sig);
2319 extern void sched_autogroup_exit(struct signal_struct *sig);
2320 #ifdef CONFIG_PROC_FS
2321 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2322 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2325 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2326 static inline void sched_autogroup_detach(struct task_struct *p) { }
2327 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2328 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2331 extern int yield_to(struct task_struct *p, bool preempt);
2332 extern void set_user_nice(struct task_struct *p, long nice);
2333 extern int task_prio(const struct task_struct *p);
2335 * task_nice - return the nice value of a given task.
2336 * @p: the task in question.
2338 * Return: The nice value [ -20 ... 0 ... 19 ].
2340 static inline int task_nice(const struct task_struct *p)
2342 return PRIO_TO_NICE((p)->static_prio);
2344 extern int can_nice(const struct task_struct *p, const int nice);
2345 extern int task_curr(const struct task_struct *p);
2346 extern int idle_cpu(int cpu);
2347 extern int sched_setscheduler(struct task_struct *, int,
2348 const struct sched_param *);
2349 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2350 const struct sched_param *);
2351 extern int sched_setattr(struct task_struct *,
2352 const struct sched_attr *);
2353 extern struct task_struct *idle_task(int cpu);
2355 * is_idle_task - is the specified task an idle task?
2356 * @p: the task in question.
2358 * Return: 1 if @p is an idle task. 0 otherwise.
2360 static inline bool is_idle_task(const struct task_struct *p)
2364 extern struct task_struct *curr_task(int cpu);
2365 extern void set_curr_task(int cpu, struct task_struct *p);
2369 union thread_union {
2370 struct thread_info thread_info;
2371 unsigned long stack[THREAD_SIZE/sizeof(long)];
2374 #ifndef __HAVE_ARCH_KSTACK_END
2375 static inline int kstack_end(void *addr)
2377 /* Reliable end of stack detection:
2378 * Some APM bios versions misalign the stack
2380 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2384 extern union thread_union init_thread_union;
2385 extern struct task_struct init_task;
2387 extern struct mm_struct init_mm;
2389 extern struct pid_namespace init_pid_ns;
2392 * find a task by one of its numerical ids
2394 * find_task_by_pid_ns():
2395 * finds a task by its pid in the specified namespace
2396 * find_task_by_vpid():
2397 * finds a task by its virtual pid
2399 * see also find_vpid() etc in include/linux/pid.h
2402 extern struct task_struct *find_task_by_vpid(pid_t nr);
2403 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2404 struct pid_namespace *ns);
2406 /* per-UID process charging. */
2407 extern struct user_struct * alloc_uid(kuid_t);
2408 static inline struct user_struct *get_uid(struct user_struct *u)
2410 atomic_inc(&u->__count);
2413 extern void free_uid(struct user_struct *);
2415 #include <asm/current.h>
2417 extern void xtime_update(unsigned long ticks);
2419 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2420 extern int wake_up_process(struct task_struct *tsk);
2421 extern void wake_up_new_task(struct task_struct *tsk);
2423 extern void kick_process(struct task_struct *tsk);
2425 static inline void kick_process(struct task_struct *tsk) { }
2427 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2428 extern void sched_dead(struct task_struct *p);
2430 extern void proc_caches_init(void);
2431 extern void flush_signals(struct task_struct *);
2432 extern void __flush_signals(struct task_struct *);
2433 extern void ignore_signals(struct task_struct *);
2434 extern void flush_signal_handlers(struct task_struct *, int force_default);
2435 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2437 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2439 unsigned long flags;
2442 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2443 ret = dequeue_signal(tsk, mask, info);
2444 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2449 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2451 extern void unblock_all_signals(void);
2452 extern void release_task(struct task_struct * p);
2453 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2454 extern int force_sigsegv(int, struct task_struct *);
2455 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2456 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2457 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2458 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2459 const struct cred *, u32);
2460 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2461 extern int kill_pid(struct pid *pid, int sig, int priv);
2462 extern int kill_proc_info(int, struct siginfo *, pid_t);
2463 extern __must_check bool do_notify_parent(struct task_struct *, int);
2464 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2465 extern void force_sig(int, struct task_struct *);
2466 extern int send_sig(int, struct task_struct *, int);
2467 extern int zap_other_threads(struct task_struct *p);
2468 extern struct sigqueue *sigqueue_alloc(void);
2469 extern void sigqueue_free(struct sigqueue *);
2470 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2471 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2473 static inline void restore_saved_sigmask(void)
2475 if (test_and_clear_restore_sigmask())
2476 __set_current_blocked(¤t->saved_sigmask);
2479 static inline sigset_t *sigmask_to_save(void)
2481 sigset_t *res = ¤t->blocked;
2482 if (unlikely(test_restore_sigmask()))
2483 res = ¤t->saved_sigmask;
2487 static inline int kill_cad_pid(int sig, int priv)
2489 return kill_pid(cad_pid, sig, priv);
2492 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2493 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2494 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2495 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2498 * True if we are on the alternate signal stack.
2500 static inline int on_sig_stack(unsigned long sp)
2502 #ifdef CONFIG_STACK_GROWSUP
2503 return sp >= current->sas_ss_sp &&
2504 sp - current->sas_ss_sp < current->sas_ss_size;
2506 return sp > current->sas_ss_sp &&
2507 sp - current->sas_ss_sp <= current->sas_ss_size;
2511 static inline int sas_ss_flags(unsigned long sp)
2513 if (!current->sas_ss_size)
2516 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2519 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2521 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2522 #ifdef CONFIG_STACK_GROWSUP
2523 return current->sas_ss_sp;
2525 return current->sas_ss_sp + current->sas_ss_size;
2531 * Routines for handling mm_structs
2533 extern struct mm_struct * mm_alloc(void);
2535 /* mmdrop drops the mm and the page tables */
2536 extern void __mmdrop(struct mm_struct *);
2537 static inline void mmdrop(struct mm_struct * mm)
2539 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2543 /* mmput gets rid of the mappings and all user-space */
2544 extern void mmput(struct mm_struct *);
2545 /* Grab a reference to a task's mm, if it is not already going away */
2546 extern struct mm_struct *get_task_mm(struct task_struct *task);
2548 * Grab a reference to a task's mm, if it is not already going away
2549 * and ptrace_may_access with the mode parameter passed to it
2552 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2553 /* Remove the current tasks stale references to the old mm_struct */
2554 extern void mm_release(struct task_struct *, struct mm_struct *);
2556 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2557 struct task_struct *);
2558 extern void flush_thread(void);
2559 extern void exit_thread(void);
2561 extern void exit_files(struct task_struct *);
2562 extern void __cleanup_sighand(struct sighand_struct *);
2564 extern void exit_itimers(struct signal_struct *);
2565 extern void flush_itimer_signals(void);
2567 extern void do_group_exit(int);
2569 extern int do_execve(struct filename *,
2570 const char __user * const __user *,
2571 const char __user * const __user *);
2572 extern int do_execveat(int, struct filename *,
2573 const char __user * const __user *,
2574 const char __user * const __user *,
2576 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2577 struct task_struct *fork_idle(int);
2578 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2580 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2581 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2583 __set_task_comm(tsk, from, false);
2585 extern char *get_task_comm(char *to, struct task_struct *tsk);
2588 void scheduler_ipi(void);
2589 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2591 static inline void scheduler_ipi(void) { }
2592 static inline unsigned long wait_task_inactive(struct task_struct *p,
2599 #define next_task(p) \
2600 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2602 #define for_each_process(p) \
2603 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2605 extern bool current_is_single_threaded(void);
2608 * Careful: do_each_thread/while_each_thread is a double loop so
2609 * 'break' will not work as expected - use goto instead.
2611 #define do_each_thread(g, t) \
2612 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2614 #define while_each_thread(g, t) \
2615 while ((t = next_thread(t)) != g)
2617 #define __for_each_thread(signal, t) \
2618 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2620 #define for_each_thread(p, t) \
2621 __for_each_thread((p)->signal, t)
2623 /* Careful: this is a double loop, 'break' won't work as expected. */
2624 #define for_each_process_thread(p, t) \
2625 for_each_process(p) for_each_thread(p, t)
2627 static inline int get_nr_threads(struct task_struct *tsk)
2629 return tsk->signal->nr_threads;
2632 static inline bool thread_group_leader(struct task_struct *p)
2634 return p->exit_signal >= 0;
2637 /* Do to the insanities of de_thread it is possible for a process
2638 * to have the pid of the thread group leader without actually being
2639 * the thread group leader. For iteration through the pids in proc
2640 * all we care about is that we have a task with the appropriate
2641 * pid, we don't actually care if we have the right task.
2643 static inline bool has_group_leader_pid(struct task_struct *p)
2645 return task_pid(p) == p->signal->leader_pid;
2649 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2651 return p1->signal == p2->signal;
2654 static inline struct task_struct *next_thread(const struct task_struct *p)
2656 return list_entry_rcu(p->thread_group.next,
2657 struct task_struct, thread_group);
2660 static inline int thread_group_empty(struct task_struct *p)
2662 return list_empty(&p->thread_group);
2665 #define delay_group_leader(p) \
2666 (thread_group_leader(p) && !thread_group_empty(p))
2669 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2670 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2671 * pins the final release of task.io_context. Also protects ->cpuset and
2672 * ->cgroup.subsys[]. And ->vfork_done.
2674 * Nests both inside and outside of read_lock(&tasklist_lock).
2675 * It must not be nested with write_lock_irq(&tasklist_lock),
2676 * neither inside nor outside.
2678 static inline void task_lock(struct task_struct *p)
2680 spin_lock(&p->alloc_lock);
2683 static inline void task_unlock(struct task_struct *p)
2685 spin_unlock(&p->alloc_lock);
2688 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2689 unsigned long *flags);
2691 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2692 unsigned long *flags)
2694 struct sighand_struct *ret;
2696 ret = __lock_task_sighand(tsk, flags);
2697 (void)__cond_lock(&tsk->sighand->siglock, ret);
2701 static inline void unlock_task_sighand(struct task_struct *tsk,
2702 unsigned long *flags)
2704 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2707 #ifdef CONFIG_CGROUPS
2708 static inline void threadgroup_change_begin(struct task_struct *tsk)
2710 down_read(&tsk->signal->group_rwsem);
2712 static inline void threadgroup_change_end(struct task_struct *tsk)
2714 up_read(&tsk->signal->group_rwsem);
2718 * threadgroup_lock - lock threadgroup
2719 * @tsk: member task of the threadgroup to lock
2721 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2722 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2723 * change ->group_leader/pid. This is useful for cases where the threadgroup
2724 * needs to stay stable across blockable operations.
2726 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2727 * synchronization. While held, no new task will be added to threadgroup
2728 * and no existing live task will have its PF_EXITING set.
2730 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2731 * sub-thread becomes a new leader.
2733 static inline void threadgroup_lock(struct task_struct *tsk)
2735 down_write(&tsk->signal->group_rwsem);
2739 * threadgroup_unlock - unlock threadgroup
2740 * @tsk: member task of the threadgroup to unlock
2742 * Reverse threadgroup_lock().
2744 static inline void threadgroup_unlock(struct task_struct *tsk)
2746 up_write(&tsk->signal->group_rwsem);
2749 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2750 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2751 static inline void threadgroup_lock(struct task_struct *tsk) {}
2752 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2755 #ifndef __HAVE_THREAD_FUNCTIONS
2757 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2758 #define task_stack_page(task) ((task)->stack)
2760 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2762 *task_thread_info(p) = *task_thread_info(org);
2763 task_thread_info(p)->task = p;
2767 * Return the address of the last usable long on the stack.
2769 * When the stack grows down, this is just above the thread
2770 * info struct. Going any lower will corrupt the threadinfo.
2772 * When the stack grows up, this is the highest address.
2773 * Beyond that position, we corrupt data on the next page.
2775 static inline unsigned long *end_of_stack(struct task_struct *p)
2777 #ifdef CONFIG_STACK_GROWSUP
2778 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2780 return (unsigned long *)(task_thread_info(p) + 1);
2785 #define task_stack_end_corrupted(task) \
2786 (*(end_of_stack(task)) != STACK_END_MAGIC)
2788 static inline int object_is_on_stack(void *obj)
2790 void *stack = task_stack_page(current);
2792 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2795 extern void thread_info_cache_init(void);
2797 #ifdef CONFIG_DEBUG_STACK_USAGE
2798 static inline unsigned long stack_not_used(struct task_struct *p)
2800 unsigned long *n = end_of_stack(p);
2802 do { /* Skip over canary */
2806 return (unsigned long)n - (unsigned long)end_of_stack(p);
2809 extern void set_task_stack_end_magic(struct task_struct *tsk);
2811 /* set thread flags in other task's structures
2812 * - see asm/thread_info.h for TIF_xxxx flags available
2814 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2816 set_ti_thread_flag(task_thread_info(tsk), flag);
2819 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2821 clear_ti_thread_flag(task_thread_info(tsk), flag);
2824 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2826 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2829 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2831 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2834 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2836 return test_ti_thread_flag(task_thread_info(tsk), flag);
2839 static inline void set_tsk_need_resched(struct task_struct *tsk)
2841 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2844 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2846 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2849 static inline int test_tsk_need_resched(struct task_struct *tsk)
2851 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2854 static inline int restart_syscall(void)
2856 set_tsk_thread_flag(current, TIF_SIGPENDING);
2857 return -ERESTARTNOINTR;
2860 static inline int signal_pending(struct task_struct *p)
2862 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2865 static inline int __fatal_signal_pending(struct task_struct *p)
2867 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2870 static inline int fatal_signal_pending(struct task_struct *p)
2872 return signal_pending(p) && __fatal_signal_pending(p);
2875 static inline int signal_pending_state(long state, struct task_struct *p)
2877 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2879 if (!signal_pending(p))
2882 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2886 * cond_resched() and cond_resched_lock(): latency reduction via
2887 * explicit rescheduling in places that are safe. The return
2888 * value indicates whether a reschedule was done in fact.
2889 * cond_resched_lock() will drop the spinlock before scheduling,
2890 * cond_resched_softirq() will enable bhs before scheduling.
2892 extern int _cond_resched(void);
2894 #define cond_resched() ({ \
2895 ___might_sleep(__FILE__, __LINE__, 0); \
2899 extern int __cond_resched_lock(spinlock_t *lock);
2901 #ifdef CONFIG_PREEMPT_COUNT
2902 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2904 #define PREEMPT_LOCK_OFFSET 0
2907 #define cond_resched_lock(lock) ({ \
2908 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2909 __cond_resched_lock(lock); \
2912 extern int __cond_resched_softirq(void);
2914 #define cond_resched_softirq() ({ \
2915 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2916 __cond_resched_softirq(); \
2919 static inline void cond_resched_rcu(void)
2921 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2929 * Does a critical section need to be broken due to another
2930 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2931 * but a general need for low latency)
2933 static inline int spin_needbreak(spinlock_t *lock)
2935 #ifdef CONFIG_PREEMPT
2936 return spin_is_contended(lock);
2943 * Idle thread specific functions to determine the need_resched
2946 #ifdef TIF_POLLING_NRFLAG
2947 static inline int tsk_is_polling(struct task_struct *p)
2949 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2952 static inline void __current_set_polling(void)
2954 set_thread_flag(TIF_POLLING_NRFLAG);
2957 static inline bool __must_check current_set_polling_and_test(void)
2959 __current_set_polling();
2962 * Polling state must be visible before we test NEED_RESCHED,
2963 * paired by resched_curr()
2965 smp_mb__after_atomic();
2967 return unlikely(tif_need_resched());
2970 static inline void __current_clr_polling(void)
2972 clear_thread_flag(TIF_POLLING_NRFLAG);
2975 static inline bool __must_check current_clr_polling_and_test(void)
2977 __current_clr_polling();
2980 * Polling state must be visible before we test NEED_RESCHED,
2981 * paired by resched_curr()
2983 smp_mb__after_atomic();
2985 return unlikely(tif_need_resched());
2989 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2990 static inline void __current_set_polling(void) { }
2991 static inline void __current_clr_polling(void) { }
2993 static inline bool __must_check current_set_polling_and_test(void)
2995 return unlikely(tif_need_resched());
2997 static inline bool __must_check current_clr_polling_and_test(void)
2999 return unlikely(tif_need_resched());
3003 static inline void current_clr_polling(void)
3005 __current_clr_polling();
3008 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3009 * Once the bit is cleared, we'll get IPIs with every new
3010 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3013 smp_mb(); /* paired with resched_curr() */
3015 preempt_fold_need_resched();
3018 static __always_inline bool need_resched(void)
3020 return unlikely(tif_need_resched());
3024 * Thread group CPU time accounting.
3026 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3027 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3030 * Reevaluate whether the task has signals pending delivery.
3031 * Wake the task if so.
3032 * This is required every time the blocked sigset_t changes.
3033 * callers must hold sighand->siglock.
3035 extern void recalc_sigpending_and_wake(struct task_struct *t);
3036 extern void recalc_sigpending(void);
3038 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3040 static inline void signal_wake_up(struct task_struct *t, bool resume)
3042 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3044 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3046 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3050 * Wrappers for p->thread_info->cpu access. No-op on UP.
3054 static inline unsigned int task_cpu(const struct task_struct *p)
3056 return task_thread_info(p)->cpu;
3059 static inline int task_node(const struct task_struct *p)
3061 return cpu_to_node(task_cpu(p));
3064 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3068 static inline unsigned int task_cpu(const struct task_struct *p)
3073 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3077 #endif /* CONFIG_SMP */
3079 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3080 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3082 #ifdef CONFIG_CGROUP_SCHED
3083 extern struct task_group root_task_group;
3084 #endif /* CONFIG_CGROUP_SCHED */
3086 extern int task_can_switch_user(struct user_struct *up,
3087 struct task_struct *tsk);
3089 #ifdef CONFIG_TASK_XACCT
3090 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3092 tsk->ioac.rchar += amt;
3095 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3097 tsk->ioac.wchar += amt;
3100 static inline void inc_syscr(struct task_struct *tsk)
3105 static inline void inc_syscw(struct task_struct *tsk)
3110 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3114 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3118 static inline void inc_syscr(struct task_struct *tsk)
3122 static inline void inc_syscw(struct task_struct *tsk)
3127 #ifndef TASK_SIZE_OF
3128 #define TASK_SIZE_OF(tsk) TASK_SIZE
3132 extern void mm_update_next_owner(struct mm_struct *mm);
3134 static inline void mm_update_next_owner(struct mm_struct *mm)
3137 #endif /* CONFIG_MEMCG */
3139 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3142 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3145 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3148 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3151 static inline unsigned long rlimit(unsigned int limit)
3153 return task_rlimit(current, limit);
3156 static inline unsigned long rlimit_max(unsigned int limit)
3158 return task_rlimit_max(current, limit);