4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt_mask.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/signal.h>
37 #include <linux/compiler.h>
38 #include <linux/completion.h>
39 #include <linux/pid.h>
40 #include <linux/percpu.h>
41 #include <linux/topology.h>
42 #include <linux/proportions.h>
43 #include <linux/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/task_io_accounting.h>
54 #include <linux/latencytop.h>
55 #include <linux/cred.h>
56 #include <linux/llist.h>
57 #include <linux/uidgid.h>
58 #include <linux/gfp.h>
60 #include <asm/processor.h>
62 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
65 * Extended scheduling parameters data structure.
67 * This is needed because the original struct sched_param can not be
68 * altered without introducing ABI issues with legacy applications
69 * (e.g., in sched_getparam()).
71 * However, the possibility of specifying more than just a priority for
72 * the tasks may be useful for a wide variety of application fields, e.g.,
73 * multimedia, streaming, automation and control, and many others.
75 * This variant (sched_attr) is meant at describing a so-called
76 * sporadic time-constrained task. In such model a task is specified by:
77 * - the activation period or minimum instance inter-arrival time;
78 * - the maximum (or average, depending on the actual scheduling
79 * discipline) computation time of all instances, a.k.a. runtime;
80 * - the deadline (relative to the actual activation time) of each
82 * Very briefly, a periodic (sporadic) task asks for the execution of
83 * some specific computation --which is typically called an instance--
84 * (at most) every period. Moreover, each instance typically lasts no more
85 * than the runtime and must be completed by time instant t equal to
86 * the instance activation time + the deadline.
88 * This is reflected by the actual fields of the sched_attr structure:
90 * @size size of the structure, for fwd/bwd compat.
92 * @sched_policy task's scheduling policy
93 * @sched_flags for customizing the scheduler behaviour
94 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
95 * @sched_priority task's static priority (SCHED_FIFO/RR)
96 * @sched_deadline representative of the task's deadline
97 * @sched_runtime representative of the task's runtime
98 * @sched_period representative of the task's period
100 * Given this task model, there are a multiplicity of scheduling algorithms
101 * and policies, that can be used to ensure all the tasks will make their
102 * timing constraints.
104 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
105 * only user of this new interface. More information about the algorithm
106 * available in the scheduling class file or in Documentation/.
114 /* SCHED_NORMAL, SCHED_BATCH */
117 /* SCHED_FIFO, SCHED_RR */
127 struct futex_pi_state;
128 struct robust_list_head;
131 struct perf_event_context;
135 #define VMACACHE_BITS 2
136 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
137 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
140 * These are the constant used to fake the fixed-point load-average
141 * counting. Some notes:
142 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
143 * a load-average precision of 10 bits integer + 11 bits fractional
144 * - if you want to count load-averages more often, you need more
145 * precision, or rounding will get you. With 2-second counting freq,
146 * the EXP_n values would be 1981, 2034 and 2043 if still using only
149 extern unsigned long avenrun[]; /* Load averages */
150 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
152 #define FSHIFT 11 /* nr of bits of precision */
153 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
154 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
155 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
156 #define EXP_5 2014 /* 1/exp(5sec/5min) */
157 #define EXP_15 2037 /* 1/exp(5sec/15min) */
159 #define CALC_LOAD(load,exp,n) \
161 load += n*(FIXED_1-exp); \
164 extern unsigned long total_forks;
165 extern int nr_threads;
166 DECLARE_PER_CPU(unsigned long, process_counts);
167 extern int nr_processes(void);
168 extern unsigned long nr_running(void);
169 extern unsigned long nr_iowait(void);
170 extern unsigned long nr_iowait_cpu(int cpu);
171 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
173 extern void calc_global_load(unsigned long ticks);
174 extern void update_cpu_load_nohz(void);
176 extern unsigned long get_parent_ip(unsigned long addr);
178 extern void dump_cpu_task(int cpu);
183 #ifdef CONFIG_SCHED_DEBUG
184 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
185 extern void proc_sched_set_task(struct task_struct *p);
187 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
191 * Task state bitmask. NOTE! These bits are also
192 * encoded in fs/proc/array.c: get_task_state().
194 * We have two separate sets of flags: task->state
195 * is about runnability, while task->exit_state are
196 * about the task exiting. Confusing, but this way
197 * modifying one set can't modify the other one by
200 #define TASK_RUNNING 0
201 #define TASK_INTERRUPTIBLE 1
202 #define TASK_UNINTERRUPTIBLE 2
203 #define __TASK_STOPPED 4
204 #define __TASK_TRACED 8
205 /* in tsk->exit_state */
207 #define EXIT_ZOMBIE 32
208 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
209 /* in tsk->state again */
211 #define TASK_WAKEKILL 128
212 #define TASK_WAKING 256
213 #define TASK_PARKED 512
214 #define TASK_STATE_MAX 1024
216 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
218 extern char ___assert_task_state[1 - 2*!!(
219 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
221 /* Convenience macros for the sake of set_task_state */
222 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
223 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
224 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
226 /* Convenience macros for the sake of wake_up */
227 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
228 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
230 /* get_task_state() */
231 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
232 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
233 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
235 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
236 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
237 #define task_is_stopped_or_traced(task) \
238 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
239 #define task_contributes_to_load(task) \
240 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
241 (task->flags & PF_FROZEN) == 0)
243 #define __set_task_state(tsk, state_value) \
244 do { (tsk)->state = (state_value); } while (0)
245 #define set_task_state(tsk, state_value) \
246 set_mb((tsk)->state, (state_value))
249 * set_current_state() includes a barrier so that the write of current->state
250 * is correctly serialised wrt the caller's subsequent test of whether to
253 * set_current_state(TASK_UNINTERRUPTIBLE);
254 * if (do_i_need_to_sleep())
257 * If the caller does not need such serialisation then use __set_current_state()
259 #define __set_current_state(state_value) \
260 do { current->state = (state_value); } while (0)
261 #define set_current_state(state_value) \
262 set_mb(current->state, (state_value))
264 /* Task command name length */
265 #define TASK_COMM_LEN 16
267 #include <linux/spinlock.h>
270 * This serializes "schedule()" and also protects
271 * the run-queue from deletions/modifications (but
272 * _adding_ to the beginning of the run-queue has
275 extern rwlock_t tasklist_lock;
276 extern spinlock_t mmlist_lock;
280 #ifdef CONFIG_PROVE_RCU
281 extern int lockdep_tasklist_lock_is_held(void);
282 #endif /* #ifdef CONFIG_PROVE_RCU */
284 extern void sched_init(void);
285 extern void sched_init_smp(void);
286 extern asmlinkage void schedule_tail(struct task_struct *prev);
287 extern void init_idle(struct task_struct *idle, int cpu);
288 extern void init_idle_bootup_task(struct task_struct *idle);
290 extern int runqueue_is_locked(int cpu);
292 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
293 extern void nohz_balance_enter_idle(int cpu);
294 extern void set_cpu_sd_state_idle(void);
295 extern int get_nohz_timer_target(int pinned);
297 static inline void nohz_balance_enter_idle(int cpu) { }
298 static inline void set_cpu_sd_state_idle(void) { }
299 static inline int get_nohz_timer_target(int pinned)
301 return smp_processor_id();
306 * Only dump TASK_* tasks. (0 for all tasks)
308 extern void show_state_filter(unsigned long state_filter);
310 static inline void show_state(void)
312 show_state_filter(0);
315 extern void show_regs(struct pt_regs *);
318 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
319 * task), SP is the stack pointer of the first frame that should be shown in the back
320 * trace (or NULL if the entire call-chain of the task should be shown).
322 extern void show_stack(struct task_struct *task, unsigned long *sp);
324 void io_schedule(void);
325 long io_schedule_timeout(long timeout);
327 extern void cpu_init (void);
328 extern void trap_init(void);
329 extern void update_process_times(int user);
330 extern void scheduler_tick(void);
332 extern void sched_show_task(struct task_struct *p);
334 #ifdef CONFIG_LOCKUP_DETECTOR
335 extern void touch_softlockup_watchdog(void);
336 extern void touch_softlockup_watchdog_sync(void);
337 extern void touch_all_softlockup_watchdogs(void);
338 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
340 size_t *lenp, loff_t *ppos);
341 extern unsigned int softlockup_panic;
342 void lockup_detector_init(void);
344 static inline void touch_softlockup_watchdog(void)
347 static inline void touch_softlockup_watchdog_sync(void)
350 static inline void touch_all_softlockup_watchdogs(void)
353 static inline void lockup_detector_init(void)
358 #ifdef CONFIG_DETECT_HUNG_TASK
359 void reset_hung_task_detector(void);
361 static inline void reset_hung_task_detector(void)
366 /* Attach to any functions which should be ignored in wchan output. */
367 #define __sched __attribute__((__section__(".sched.text")))
369 /* Linker adds these: start and end of __sched functions */
370 extern char __sched_text_start[], __sched_text_end[];
372 /* Is this address in the __sched functions? */
373 extern int in_sched_functions(unsigned long addr);
375 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
376 extern signed long schedule_timeout(signed long timeout);
377 extern signed long schedule_timeout_interruptible(signed long timeout);
378 extern signed long schedule_timeout_killable(signed long timeout);
379 extern signed long schedule_timeout_uninterruptible(signed long timeout);
380 asmlinkage void schedule(void);
381 extern void schedule_preempt_disabled(void);
384 struct user_namespace;
387 extern void arch_pick_mmap_layout(struct mm_struct *mm);
389 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
390 unsigned long, unsigned long);
392 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
393 unsigned long len, unsigned long pgoff,
394 unsigned long flags);
396 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
400 #define SUID_DUMP_USER 1 /* Dump as user of process */
401 #define SUID_DUMP_ROOT 2 /* Dump as root */
405 /* for SUID_DUMP_* above */
406 #define MMF_DUMPABLE_BITS 2
407 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
409 extern void set_dumpable(struct mm_struct *mm, int value);
411 * This returns the actual value of the suid_dumpable flag. For things
412 * that are using this for checking for privilege transitions, it must
413 * test against SUID_DUMP_USER rather than treating it as a boolean
416 static inline int __get_dumpable(unsigned long mm_flags)
418 return mm_flags & MMF_DUMPABLE_MASK;
421 static inline int get_dumpable(struct mm_struct *mm)
423 return __get_dumpable(mm->flags);
426 /* coredump filter bits */
427 #define MMF_DUMP_ANON_PRIVATE 2
428 #define MMF_DUMP_ANON_SHARED 3
429 #define MMF_DUMP_MAPPED_PRIVATE 4
430 #define MMF_DUMP_MAPPED_SHARED 5
431 #define MMF_DUMP_ELF_HEADERS 6
432 #define MMF_DUMP_HUGETLB_PRIVATE 7
433 #define MMF_DUMP_HUGETLB_SHARED 8
435 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
436 #define MMF_DUMP_FILTER_BITS 7
437 #define MMF_DUMP_FILTER_MASK \
438 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
439 #define MMF_DUMP_FILTER_DEFAULT \
440 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
441 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
443 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
444 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
446 # define MMF_DUMP_MASK_DEFAULT_ELF 0
448 /* leave room for more dump flags */
449 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
450 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
451 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
453 #define MMF_HAS_UPROBES 19 /* has uprobes */
454 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
456 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
458 struct sighand_struct {
460 struct k_sigaction action[_NSIG];
462 wait_queue_head_t signalfd_wqh;
465 struct pacct_struct {
468 unsigned long ac_mem;
469 cputime_t ac_utime, ac_stime;
470 unsigned long ac_minflt, ac_majflt;
481 * struct cputime - snaphsot of system and user cputime
482 * @utime: time spent in user mode
483 * @stime: time spent in system mode
485 * Gathers a generic snapshot of user and system time.
493 * struct task_cputime - collected CPU time counts
494 * @utime: time spent in user mode, in &cputime_t units
495 * @stime: time spent in kernel mode, in &cputime_t units
496 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
498 * This is an extension of struct cputime that includes the total runtime
499 * spent by the task from the scheduler point of view.
501 * As a result, this structure groups together three kinds of CPU time
502 * that are tracked for threads and thread groups. Most things considering
503 * CPU time want to group these counts together and treat all three
504 * of them in parallel.
506 struct task_cputime {
509 unsigned long long sum_exec_runtime;
511 /* Alternate field names when used to cache expirations. */
512 #define prof_exp stime
513 #define virt_exp utime
514 #define sched_exp sum_exec_runtime
516 #define INIT_CPUTIME \
517 (struct task_cputime) { \
520 .sum_exec_runtime = 0, \
523 #ifdef CONFIG_PREEMPT_COUNT
524 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
526 #define PREEMPT_DISABLED PREEMPT_ENABLED
530 * Disable preemption until the scheduler is running.
531 * Reset by start_kernel()->sched_init()->init_idle().
533 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
534 * before the scheduler is active -- see should_resched().
536 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
539 * struct thread_group_cputimer - thread group interval timer counts
540 * @cputime: thread group interval timers.
541 * @running: non-zero when there are timers running and
542 * @cputime receives updates.
543 * @lock: lock for fields in this struct.
545 * This structure contains the version of task_cputime, above, that is
546 * used for thread group CPU timer calculations.
548 struct thread_group_cputimer {
549 struct task_cputime cputime;
554 #include <linux/rwsem.h>
558 * NOTE! "signal_struct" does not have its own
559 * locking, because a shared signal_struct always
560 * implies a shared sighand_struct, so locking
561 * sighand_struct is always a proper superset of
562 * the locking of signal_struct.
564 struct signal_struct {
568 struct list_head thread_head;
570 wait_queue_head_t wait_chldexit; /* for wait4() */
572 /* current thread group signal load-balancing target: */
573 struct task_struct *curr_target;
575 /* shared signal handling: */
576 struct sigpending shared_pending;
578 /* thread group exit support */
581 * - notify group_exit_task when ->count is equal to notify_count
582 * - everyone except group_exit_task is stopped during signal delivery
583 * of fatal signals, group_exit_task processes the signal.
586 struct task_struct *group_exit_task;
588 /* thread group stop support, overloads group_exit_code too */
589 int group_stop_count;
590 unsigned int flags; /* see SIGNAL_* flags below */
593 * PR_SET_CHILD_SUBREAPER marks a process, like a service
594 * manager, to re-parent orphan (double-forking) child processes
595 * to this process instead of 'init'. The service manager is
596 * able to receive SIGCHLD signals and is able to investigate
597 * the process until it calls wait(). All children of this
598 * process will inherit a flag if they should look for a
599 * child_subreaper process at exit.
601 unsigned int is_child_subreaper:1;
602 unsigned int has_child_subreaper:1;
604 /* POSIX.1b Interval Timers */
606 struct list_head posix_timers;
608 /* ITIMER_REAL timer for the process */
609 struct hrtimer real_timer;
610 struct pid *leader_pid;
611 ktime_t it_real_incr;
614 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
615 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
616 * values are defined to 0 and 1 respectively
618 struct cpu_itimer it[2];
621 * Thread group totals for process CPU timers.
622 * See thread_group_cputimer(), et al, for details.
624 struct thread_group_cputimer cputimer;
626 /* Earliest-expiration cache. */
627 struct task_cputime cputime_expires;
629 struct list_head cpu_timers[3];
631 struct pid *tty_old_pgrp;
633 /* boolean value for session group leader */
636 struct tty_struct *tty; /* NULL if no tty */
638 #ifdef CONFIG_SCHED_AUTOGROUP
639 struct autogroup *autogroup;
642 * Cumulative resource counters for dead threads in the group,
643 * and for reaped dead child processes forked by this group.
644 * Live threads maintain their own counters and add to these
645 * in __exit_signal, except for the group leader.
647 cputime_t utime, stime, cutime, cstime;
650 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
651 struct cputime prev_cputime;
653 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
654 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
655 unsigned long inblock, oublock, cinblock, coublock;
656 unsigned long maxrss, cmaxrss;
657 struct task_io_accounting ioac;
660 * Cumulative ns of schedule CPU time fo dead threads in the
661 * group, not including a zombie group leader, (This only differs
662 * from jiffies_to_ns(utime + stime) if sched_clock uses something
663 * other than jiffies.)
665 unsigned long long sum_sched_runtime;
668 * We don't bother to synchronize most readers of this at all,
669 * because there is no reader checking a limit that actually needs
670 * to get both rlim_cur and rlim_max atomically, and either one
671 * alone is a single word that can safely be read normally.
672 * getrlimit/setrlimit use task_lock(current->group_leader) to
673 * protect this instead of the siglock, because they really
674 * have no need to disable irqs.
676 struct rlimit rlim[RLIM_NLIMITS];
678 #ifdef CONFIG_BSD_PROCESS_ACCT
679 struct pacct_struct pacct; /* per-process accounting information */
681 #ifdef CONFIG_TASKSTATS
682 struct taskstats *stats;
686 unsigned audit_tty_log_passwd;
687 struct tty_audit_buf *tty_audit_buf;
689 #ifdef CONFIG_CGROUPS
691 * group_rwsem prevents new tasks from entering the threadgroup and
692 * member tasks from exiting,a more specifically, setting of
693 * PF_EXITING. fork and exit paths are protected with this rwsem
694 * using threadgroup_change_begin/end(). Users which require
695 * threadgroup to remain stable should use threadgroup_[un]lock()
696 * which also takes care of exec path. Currently, cgroup is the
699 struct rw_semaphore group_rwsem;
702 oom_flags_t oom_flags;
703 short oom_score_adj; /* OOM kill score adjustment */
704 short oom_score_adj_min; /* OOM kill score adjustment min value.
705 * Only settable by CAP_SYS_RESOURCE. */
707 struct mutex cred_guard_mutex; /* guard against foreign influences on
708 * credential calculations
709 * (notably. ptrace) */
713 * Bits in flags field of signal_struct.
715 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
716 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
717 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
718 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
720 * Pending notifications to parent.
722 #define SIGNAL_CLD_STOPPED 0x00000010
723 #define SIGNAL_CLD_CONTINUED 0x00000020
724 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
726 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
728 /* If true, all threads except ->group_exit_task have pending SIGKILL */
729 static inline int signal_group_exit(const struct signal_struct *sig)
731 return (sig->flags & SIGNAL_GROUP_EXIT) ||
732 (sig->group_exit_task != NULL);
736 * Some day this will be a full-fledged user tracking system..
739 atomic_t __count; /* reference count */
740 atomic_t processes; /* How many processes does this user have? */
741 atomic_t sigpending; /* How many pending signals does this user have? */
742 #ifdef CONFIG_INOTIFY_USER
743 atomic_t inotify_watches; /* How many inotify watches does this user have? */
744 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
746 #ifdef CONFIG_FANOTIFY
747 atomic_t fanotify_listeners;
750 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
752 #ifdef CONFIG_POSIX_MQUEUE
753 /* protected by mq_lock */
754 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
756 unsigned long locked_shm; /* How many pages of mlocked shm ? */
759 struct key *uid_keyring; /* UID specific keyring */
760 struct key *session_keyring; /* UID's default session keyring */
763 /* Hash table maintenance information */
764 struct hlist_node uidhash_node;
767 #ifdef CONFIG_PERF_EVENTS
768 atomic_long_t locked_vm;
772 extern int uids_sysfs_init(void);
774 extern struct user_struct *find_user(kuid_t);
776 extern struct user_struct root_user;
777 #define INIT_USER (&root_user)
780 struct backing_dev_info;
781 struct reclaim_state;
783 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
785 /* cumulative counters */
786 unsigned long pcount; /* # of times run on this cpu */
787 unsigned long long run_delay; /* time spent waiting on a runqueue */
790 unsigned long long last_arrival,/* when we last ran on a cpu */
791 last_queued; /* when we were last queued to run */
793 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
795 #ifdef CONFIG_TASK_DELAY_ACCT
796 struct task_delay_info {
798 unsigned int flags; /* Private per-task flags */
800 /* For each stat XXX, add following, aligned appropriately
802 * struct timespec XXX_start, XXX_end;
806 * Atomicity of updates to XXX_delay, XXX_count protected by
807 * single lock above (split into XXX_lock if contention is an issue).
811 * XXX_count is incremented on every XXX operation, the delay
812 * associated with the operation is added to XXX_delay.
813 * XXX_delay contains the accumulated delay time in nanoseconds.
815 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
816 u64 blkio_delay; /* wait for sync block io completion */
817 u64 swapin_delay; /* wait for swapin block io completion */
818 u32 blkio_count; /* total count of the number of sync block */
819 /* io operations performed */
820 u32 swapin_count; /* total count of the number of swapin block */
821 /* io operations performed */
823 struct timespec freepages_start, freepages_end;
824 u64 freepages_delay; /* wait for memory reclaim */
825 u32 freepages_count; /* total count of memory reclaim */
827 #endif /* CONFIG_TASK_DELAY_ACCT */
829 static inline int sched_info_on(void)
831 #ifdef CONFIG_SCHEDSTATS
833 #elif defined(CONFIG_TASK_DELAY_ACCT)
834 extern int delayacct_on;
849 * Increase resolution of cpu_capacity calculations
851 #define SCHED_CAPACITY_SHIFT 10
852 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
855 * sched-domains (multiprocessor balancing) declarations:
858 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
859 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
860 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
861 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
862 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
863 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
864 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
865 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
866 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
867 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
868 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
869 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
870 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
871 #define SD_NUMA 0x4000 /* cross-node balancing */
873 #ifdef CONFIG_SCHED_SMT
874 static inline const int cpu_smt_flags(void)
876 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
880 #ifdef CONFIG_SCHED_MC
881 static inline const int cpu_core_flags(void)
883 return SD_SHARE_PKG_RESOURCES;
888 static inline const int cpu_numa_flags(void)
894 struct sched_domain_attr {
895 int relax_domain_level;
898 #define SD_ATTR_INIT (struct sched_domain_attr) { \
899 .relax_domain_level = -1, \
902 extern int sched_domain_level_max;
906 struct sched_domain {
907 /* These fields must be setup */
908 struct sched_domain *parent; /* top domain must be null terminated */
909 struct sched_domain *child; /* bottom domain must be null terminated */
910 struct sched_group *groups; /* the balancing groups of the domain */
911 unsigned long min_interval; /* Minimum balance interval ms */
912 unsigned long max_interval; /* Maximum balance interval ms */
913 unsigned int busy_factor; /* less balancing by factor if busy */
914 unsigned int imbalance_pct; /* No balance until over watermark */
915 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
916 unsigned int busy_idx;
917 unsigned int idle_idx;
918 unsigned int newidle_idx;
919 unsigned int wake_idx;
920 unsigned int forkexec_idx;
921 unsigned int smt_gain;
923 int nohz_idle; /* NOHZ IDLE status */
924 int flags; /* See SD_* */
927 /* Runtime fields. */
928 unsigned long last_balance; /* init to jiffies. units in jiffies */
929 unsigned int balance_interval; /* initialise to 1. units in ms. */
930 unsigned int nr_balance_failed; /* initialise to 0 */
932 /* idle_balance() stats */
933 u64 max_newidle_lb_cost;
934 unsigned long next_decay_max_lb_cost;
936 #ifdef CONFIG_SCHEDSTATS
937 /* load_balance() stats */
938 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
939 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
940 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
947 /* Active load balancing */
948 unsigned int alb_count;
949 unsigned int alb_failed;
950 unsigned int alb_pushed;
952 /* SD_BALANCE_EXEC stats */
953 unsigned int sbe_count;
954 unsigned int sbe_balanced;
955 unsigned int sbe_pushed;
957 /* SD_BALANCE_FORK stats */
958 unsigned int sbf_count;
959 unsigned int sbf_balanced;
960 unsigned int sbf_pushed;
962 /* try_to_wake_up() stats */
963 unsigned int ttwu_wake_remote;
964 unsigned int ttwu_move_affine;
965 unsigned int ttwu_move_balance;
967 #ifdef CONFIG_SCHED_DEBUG
971 void *private; /* used during construction */
972 struct rcu_head rcu; /* used during destruction */
975 unsigned int span_weight;
977 * Span of all CPUs in this domain.
979 * NOTE: this field is variable length. (Allocated dynamically
980 * by attaching extra space to the end of the structure,
981 * depending on how many CPUs the kernel has booted up with)
983 unsigned long span[0];
986 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
988 return to_cpumask(sd->span);
991 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
992 struct sched_domain_attr *dattr_new);
994 /* Allocate an array of sched domains, for partition_sched_domains(). */
995 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
996 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
998 bool cpus_share_cache(int this_cpu, int that_cpu);
1000 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1001 typedef const int (*sched_domain_flags_f)(void);
1003 #define SDTL_OVERLAP 0x01
1006 struct sched_domain **__percpu sd;
1007 struct sched_group **__percpu sg;
1008 struct sched_group_capacity **__percpu sgc;
1011 struct sched_domain_topology_level {
1012 sched_domain_mask_f mask;
1013 sched_domain_flags_f sd_flags;
1016 struct sd_data data;
1017 #ifdef CONFIG_SCHED_DEBUG
1022 extern struct sched_domain_topology_level *sched_domain_topology;
1024 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1026 #ifdef CONFIG_SCHED_DEBUG
1027 # define SD_INIT_NAME(type) .name = #type
1029 # define SD_INIT_NAME(type)
1032 #else /* CONFIG_SMP */
1034 struct sched_domain_attr;
1037 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1038 struct sched_domain_attr *dattr_new)
1042 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1047 #endif /* !CONFIG_SMP */
1050 struct io_context; /* See blkdev.h */
1053 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1054 extern void prefetch_stack(struct task_struct *t);
1056 static inline void prefetch_stack(struct task_struct *t) { }
1059 struct audit_context; /* See audit.c */
1061 struct pipe_inode_info;
1062 struct uts_namespace;
1064 struct load_weight {
1065 unsigned long weight;
1071 * These sums represent an infinite geometric series and so are bound
1072 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1073 * choices of y < 1-2^(-32)*1024.
1075 u32 runnable_avg_sum, runnable_avg_period;
1076 u64 last_runnable_update;
1078 unsigned long load_avg_contrib;
1081 #ifdef CONFIG_SCHEDSTATS
1082 struct sched_statistics {
1092 s64 sum_sleep_runtime;
1099 u64 nr_migrations_cold;
1100 u64 nr_failed_migrations_affine;
1101 u64 nr_failed_migrations_running;
1102 u64 nr_failed_migrations_hot;
1103 u64 nr_forced_migrations;
1106 u64 nr_wakeups_sync;
1107 u64 nr_wakeups_migrate;
1108 u64 nr_wakeups_local;
1109 u64 nr_wakeups_remote;
1110 u64 nr_wakeups_affine;
1111 u64 nr_wakeups_affine_attempts;
1112 u64 nr_wakeups_passive;
1113 u64 nr_wakeups_idle;
1117 struct sched_entity {
1118 struct load_weight load; /* for load-balancing */
1119 struct rb_node run_node;
1120 struct list_head group_node;
1124 u64 sum_exec_runtime;
1126 u64 prev_sum_exec_runtime;
1130 #ifdef CONFIG_SCHEDSTATS
1131 struct sched_statistics statistics;
1134 #ifdef CONFIG_FAIR_GROUP_SCHED
1136 struct sched_entity *parent;
1137 /* rq on which this entity is (to be) queued: */
1138 struct cfs_rq *cfs_rq;
1139 /* rq "owned" by this entity/group: */
1140 struct cfs_rq *my_q;
1144 /* Per-entity load-tracking */
1145 struct sched_avg avg;
1149 struct sched_rt_entity {
1150 struct list_head run_list;
1151 unsigned long timeout;
1152 unsigned long watchdog_stamp;
1153 unsigned int time_slice;
1155 struct sched_rt_entity *back;
1156 #ifdef CONFIG_RT_GROUP_SCHED
1157 struct sched_rt_entity *parent;
1158 /* rq on which this entity is (to be) queued: */
1159 struct rt_rq *rt_rq;
1160 /* rq "owned" by this entity/group: */
1165 struct sched_dl_entity {
1166 struct rb_node rb_node;
1169 * Original scheduling parameters. Copied here from sched_attr
1170 * during sched_setattr(), they will remain the same until
1171 * the next sched_setattr().
1173 u64 dl_runtime; /* maximum runtime for each instance */
1174 u64 dl_deadline; /* relative deadline of each instance */
1175 u64 dl_period; /* separation of two instances (period) */
1176 u64 dl_bw; /* dl_runtime / dl_deadline */
1179 * Actual scheduling parameters. Initialized with the values above,
1180 * they are continously updated during task execution. Note that
1181 * the remaining runtime could be < 0 in case we are in overrun.
1183 s64 runtime; /* remaining runtime for this instance */
1184 u64 deadline; /* absolute deadline for this instance */
1185 unsigned int flags; /* specifying the scheduler behaviour */
1190 * @dl_throttled tells if we exhausted the runtime. If so, the
1191 * task has to wait for a replenishment to be performed at the
1192 * next firing of dl_timer.
1194 * @dl_new tells if a new instance arrived. If so we must
1195 * start executing it with full runtime and reset its absolute
1198 * @dl_boosted tells if we are boosted due to DI. If so we are
1199 * outside bandwidth enforcement mechanism (but only until we
1200 * exit the critical section);
1202 * @dl_yielded tells if task gave up the cpu before consuming
1203 * all its available runtime during the last job.
1205 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1208 * Bandwidth enforcement timer. Each -deadline task has its
1209 * own bandwidth to be enforced, thus we need one timer per task.
1211 struct hrtimer dl_timer;
1216 enum perf_event_task_context {
1217 perf_invalid_context = -1,
1218 perf_hw_context = 0,
1220 perf_nr_task_contexts,
1223 struct task_struct {
1224 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1227 unsigned int flags; /* per process flags, defined below */
1228 unsigned int ptrace;
1231 struct llist_node wake_entry;
1233 struct task_struct *last_wakee;
1234 unsigned long wakee_flips;
1235 unsigned long wakee_flip_decay_ts;
1241 int prio, static_prio, normal_prio;
1242 unsigned int rt_priority;
1243 const struct sched_class *sched_class;
1244 struct sched_entity se;
1245 struct sched_rt_entity rt;
1246 #ifdef CONFIG_CGROUP_SCHED
1247 struct task_group *sched_task_group;
1249 struct sched_dl_entity dl;
1251 #ifdef CONFIG_PREEMPT_NOTIFIERS
1252 /* list of struct preempt_notifier: */
1253 struct hlist_head preempt_notifiers;
1256 #ifdef CONFIG_BLK_DEV_IO_TRACE
1257 unsigned int btrace_seq;
1260 unsigned int policy;
1261 int nr_cpus_allowed;
1262 cpumask_t cpus_allowed;
1264 #ifdef CONFIG_PREEMPT_RCU
1265 int rcu_read_lock_nesting;
1266 char rcu_read_unlock_special;
1267 struct list_head rcu_node_entry;
1268 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1269 #ifdef CONFIG_TREE_PREEMPT_RCU
1270 struct rcu_node *rcu_blocked_node;
1271 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1272 #ifdef CONFIG_RCU_BOOST
1273 struct rt_mutex *rcu_boost_mutex;
1274 #endif /* #ifdef CONFIG_RCU_BOOST */
1276 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1277 struct sched_info sched_info;
1280 struct list_head tasks;
1282 struct plist_node pushable_tasks;
1283 struct rb_node pushable_dl_tasks;
1286 struct mm_struct *mm, *active_mm;
1287 #ifdef CONFIG_COMPAT_BRK
1288 unsigned brk_randomized:1;
1290 /* per-thread vma caching */
1291 u32 vmacache_seqnum;
1292 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1293 #if defined(SPLIT_RSS_COUNTING)
1294 struct task_rss_stat rss_stat;
1298 int exit_code, exit_signal;
1299 int pdeath_signal; /* The signal sent when the parent dies */
1300 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1302 /* Used for emulating ABI behavior of previous Linux versions */
1303 unsigned int personality;
1305 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1307 unsigned in_iowait:1;
1309 /* task may not gain privileges */
1310 unsigned no_new_privs:1;
1312 /* Revert to default priority/policy when forking */
1313 unsigned sched_reset_on_fork:1;
1314 unsigned sched_contributes_to_load:1;
1319 #ifdef CONFIG_CC_STACKPROTECTOR
1320 /* Canary value for the -fstack-protector gcc feature */
1321 unsigned long stack_canary;
1324 * pointers to (original) parent process, youngest child, younger sibling,
1325 * older sibling, respectively. (p->father can be replaced with
1326 * p->real_parent->pid)
1328 struct task_struct __rcu *real_parent; /* real parent process */
1329 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1331 * children/sibling forms the list of my natural children
1333 struct list_head children; /* list of my children */
1334 struct list_head sibling; /* linkage in my parent's children list */
1335 struct task_struct *group_leader; /* threadgroup leader */
1338 * ptraced is the list of tasks this task is using ptrace on.
1339 * This includes both natural children and PTRACE_ATTACH targets.
1340 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1342 struct list_head ptraced;
1343 struct list_head ptrace_entry;
1345 /* PID/PID hash table linkage. */
1346 struct pid_link pids[PIDTYPE_MAX];
1347 struct list_head thread_group;
1348 struct list_head thread_node;
1350 struct completion *vfork_done; /* for vfork() */
1351 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1352 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1354 cputime_t utime, stime, utimescaled, stimescaled;
1356 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1357 struct cputime prev_cputime;
1359 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1360 seqlock_t vtime_seqlock;
1361 unsigned long long vtime_snap;
1366 } vtime_snap_whence;
1368 unsigned long nvcsw, nivcsw; /* context switch counts */
1369 struct timespec start_time; /* monotonic time */
1370 struct timespec real_start_time; /* boot based time */
1371 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1372 unsigned long min_flt, maj_flt;
1374 struct task_cputime cputime_expires;
1375 struct list_head cpu_timers[3];
1377 /* process credentials */
1378 const struct cred __rcu *real_cred; /* objective and real subjective task
1379 * credentials (COW) */
1380 const struct cred __rcu *cred; /* effective (overridable) subjective task
1381 * credentials (COW) */
1382 char comm[TASK_COMM_LEN]; /* executable name excluding path
1383 - access with [gs]et_task_comm (which lock
1384 it with task_lock())
1385 - initialized normally by setup_new_exec */
1386 /* file system info */
1387 int link_count, total_link_count;
1388 #ifdef CONFIG_SYSVIPC
1390 struct sysv_sem sysvsem;
1392 #ifdef CONFIG_DETECT_HUNG_TASK
1393 /* hung task detection */
1394 unsigned long last_switch_count;
1396 /* CPU-specific state of this task */
1397 struct thread_struct thread;
1398 /* filesystem information */
1399 struct fs_struct *fs;
1400 /* open file information */
1401 struct files_struct *files;
1403 struct nsproxy *nsproxy;
1404 /* signal handlers */
1405 struct signal_struct *signal;
1406 struct sighand_struct *sighand;
1408 sigset_t blocked, real_blocked;
1409 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1410 struct sigpending pending;
1412 unsigned long sas_ss_sp;
1414 int (*notifier)(void *priv);
1415 void *notifier_data;
1416 sigset_t *notifier_mask;
1417 struct callback_head *task_works;
1419 struct audit_context *audit_context;
1420 #ifdef CONFIG_AUDITSYSCALL
1422 unsigned int sessionid;
1424 struct seccomp seccomp;
1426 /* Thread group tracking */
1429 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1431 spinlock_t alloc_lock;
1433 /* Protection of the PI data structures: */
1434 raw_spinlock_t pi_lock;
1436 #ifdef CONFIG_RT_MUTEXES
1437 /* PI waiters blocked on a rt_mutex held by this task */
1438 struct rb_root pi_waiters;
1439 struct rb_node *pi_waiters_leftmost;
1440 /* Deadlock detection and priority inheritance handling */
1441 struct rt_mutex_waiter *pi_blocked_on;
1442 /* Top pi_waiters task */
1443 struct task_struct *pi_top_task;
1446 #ifdef CONFIG_DEBUG_MUTEXES
1447 /* mutex deadlock detection */
1448 struct mutex_waiter *blocked_on;
1450 #ifdef CONFIG_TRACE_IRQFLAGS
1451 unsigned int irq_events;
1452 unsigned long hardirq_enable_ip;
1453 unsigned long hardirq_disable_ip;
1454 unsigned int hardirq_enable_event;
1455 unsigned int hardirq_disable_event;
1456 int hardirqs_enabled;
1457 int hardirq_context;
1458 unsigned long softirq_disable_ip;
1459 unsigned long softirq_enable_ip;
1460 unsigned int softirq_disable_event;
1461 unsigned int softirq_enable_event;
1462 int softirqs_enabled;
1463 int softirq_context;
1465 #ifdef CONFIG_LOCKDEP
1466 # define MAX_LOCK_DEPTH 48UL
1469 unsigned int lockdep_recursion;
1470 struct held_lock held_locks[MAX_LOCK_DEPTH];
1471 gfp_t lockdep_reclaim_gfp;
1474 /* journalling filesystem info */
1477 /* stacked block device info */
1478 struct bio_list *bio_list;
1481 /* stack plugging */
1482 struct blk_plug *plug;
1486 struct reclaim_state *reclaim_state;
1488 struct backing_dev_info *backing_dev_info;
1490 struct io_context *io_context;
1492 unsigned long ptrace_message;
1493 siginfo_t *last_siginfo; /* For ptrace use. */
1494 struct task_io_accounting ioac;
1495 #if defined(CONFIG_TASK_XACCT)
1496 u64 acct_rss_mem1; /* accumulated rss usage */
1497 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1498 cputime_t acct_timexpd; /* stime + utime since last update */
1500 #ifdef CONFIG_CPUSETS
1501 nodemask_t mems_allowed; /* Protected by alloc_lock */
1502 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1503 int cpuset_mem_spread_rotor;
1504 int cpuset_slab_spread_rotor;
1506 #ifdef CONFIG_CGROUPS
1507 /* Control Group info protected by css_set_lock */
1508 struct css_set __rcu *cgroups;
1509 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1510 struct list_head cg_list;
1513 struct robust_list_head __user *robust_list;
1514 #ifdef CONFIG_COMPAT
1515 struct compat_robust_list_head __user *compat_robust_list;
1517 struct list_head pi_state_list;
1518 struct futex_pi_state *pi_state_cache;
1520 #ifdef CONFIG_PERF_EVENTS
1521 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1522 struct mutex perf_event_mutex;
1523 struct list_head perf_event_list;
1525 #ifdef CONFIG_DEBUG_PREEMPT
1526 unsigned long preempt_disable_ip;
1529 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1531 short pref_node_fork;
1533 #ifdef CONFIG_NUMA_BALANCING
1535 unsigned int numa_scan_period;
1536 unsigned int numa_scan_period_max;
1537 int numa_preferred_nid;
1538 unsigned long numa_migrate_retry;
1539 u64 node_stamp; /* migration stamp */
1540 u64 last_task_numa_placement;
1541 u64 last_sum_exec_runtime;
1542 struct callback_head numa_work;
1544 struct list_head numa_entry;
1545 struct numa_group *numa_group;
1548 * Exponential decaying average of faults on a per-node basis.
1549 * Scheduling placement decisions are made based on the these counts.
1550 * The values remain static for the duration of a PTE scan
1552 unsigned long *numa_faults_memory;
1553 unsigned long total_numa_faults;
1556 * numa_faults_buffer records faults per node during the current
1557 * scan window. When the scan completes, the counts in
1558 * numa_faults_memory decay and these values are copied.
1560 unsigned long *numa_faults_buffer_memory;
1563 * Track the nodes the process was running on when a NUMA hinting
1564 * fault was incurred.
1566 unsigned long *numa_faults_cpu;
1567 unsigned long *numa_faults_buffer_cpu;
1570 * numa_faults_locality tracks if faults recorded during the last
1571 * scan window were remote/local. The task scan period is adapted
1572 * based on the locality of the faults with different weights
1573 * depending on whether they were shared or private faults
1575 unsigned long numa_faults_locality[2];
1577 unsigned long numa_pages_migrated;
1578 #endif /* CONFIG_NUMA_BALANCING */
1580 struct rcu_head rcu;
1583 * cache last used pipe for splice
1585 struct pipe_inode_info *splice_pipe;
1587 struct page_frag task_frag;
1589 #ifdef CONFIG_TASK_DELAY_ACCT
1590 struct task_delay_info *delays;
1592 #ifdef CONFIG_FAULT_INJECTION
1596 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1597 * balance_dirty_pages() for some dirty throttling pause
1600 int nr_dirtied_pause;
1601 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1603 #ifdef CONFIG_LATENCYTOP
1604 int latency_record_count;
1605 struct latency_record latency_record[LT_SAVECOUNT];
1608 * time slack values; these are used to round up poll() and
1609 * select() etc timeout values. These are in nanoseconds.
1611 unsigned long timer_slack_ns;
1612 unsigned long default_timer_slack_ns;
1614 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1615 /* Index of current stored address in ret_stack */
1617 /* Stack of return addresses for return function tracing */
1618 struct ftrace_ret_stack *ret_stack;
1619 /* time stamp for last schedule */
1620 unsigned long long ftrace_timestamp;
1622 * Number of functions that haven't been traced
1623 * because of depth overrun.
1625 atomic_t trace_overrun;
1626 /* Pause for the tracing */
1627 atomic_t tracing_graph_pause;
1629 #ifdef CONFIG_TRACING
1630 /* state flags for use by tracers */
1631 unsigned long trace;
1632 /* bitmask and counter of trace recursion */
1633 unsigned long trace_recursion;
1634 #endif /* CONFIG_TRACING */
1635 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1636 struct memcg_batch_info {
1637 int do_batch; /* incremented when batch uncharge started */
1638 struct mem_cgroup *memcg; /* target memcg of uncharge */
1639 unsigned long nr_pages; /* uncharged usage */
1640 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1642 unsigned int memcg_kmem_skip_account;
1643 struct memcg_oom_info {
1644 struct mem_cgroup *memcg;
1647 unsigned int may_oom:1;
1650 #ifdef CONFIG_UPROBES
1651 struct uprobe_task *utask;
1653 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1654 unsigned int sequential_io;
1655 unsigned int sequential_io_avg;
1659 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1660 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1662 #define TNF_MIGRATED 0x01
1663 #define TNF_NO_GROUP 0x02
1664 #define TNF_SHARED 0x04
1665 #define TNF_FAULT_LOCAL 0x08
1667 #ifdef CONFIG_NUMA_BALANCING
1668 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1669 extern pid_t task_numa_group_id(struct task_struct *p);
1670 extern void set_numabalancing_state(bool enabled);
1671 extern void task_numa_free(struct task_struct *p);
1672 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1673 int src_nid, int dst_cpu);
1675 static inline void task_numa_fault(int last_node, int node, int pages,
1679 static inline pid_t task_numa_group_id(struct task_struct *p)
1683 static inline void set_numabalancing_state(bool enabled)
1686 static inline void task_numa_free(struct task_struct *p)
1689 static inline bool should_numa_migrate_memory(struct task_struct *p,
1690 struct page *page, int src_nid, int dst_cpu)
1696 static inline struct pid *task_pid(struct task_struct *task)
1698 return task->pids[PIDTYPE_PID].pid;
1701 static inline struct pid *task_tgid(struct task_struct *task)
1703 return task->group_leader->pids[PIDTYPE_PID].pid;
1707 * Without tasklist or rcu lock it is not safe to dereference
1708 * the result of task_pgrp/task_session even if task == current,
1709 * we can race with another thread doing sys_setsid/sys_setpgid.
1711 static inline struct pid *task_pgrp(struct task_struct *task)
1713 return task->group_leader->pids[PIDTYPE_PGID].pid;
1716 static inline struct pid *task_session(struct task_struct *task)
1718 return task->group_leader->pids[PIDTYPE_SID].pid;
1721 struct pid_namespace;
1724 * the helpers to get the task's different pids as they are seen
1725 * from various namespaces
1727 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1728 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1730 * task_xid_nr_ns() : id seen from the ns specified;
1732 * set_task_vxid() : assigns a virtual id to a task;
1734 * see also pid_nr() etc in include/linux/pid.h
1736 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1737 struct pid_namespace *ns);
1739 static inline pid_t task_pid_nr(struct task_struct *tsk)
1744 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1745 struct pid_namespace *ns)
1747 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1750 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1752 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1756 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1761 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1763 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1765 return pid_vnr(task_tgid(tsk));
1769 static inline int pid_alive(const struct task_struct *p);
1770 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1776 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1782 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1784 return task_ppid_nr_ns(tsk, &init_pid_ns);
1787 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1788 struct pid_namespace *ns)
1790 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1793 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1795 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1799 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1800 struct pid_namespace *ns)
1802 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1805 static inline pid_t task_session_vnr(struct task_struct *tsk)
1807 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1810 /* obsolete, do not use */
1811 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1813 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1817 * pid_alive - check that a task structure is not stale
1818 * @p: Task structure to be checked.
1820 * Test if a process is not yet dead (at most zombie state)
1821 * If pid_alive fails, then pointers within the task structure
1822 * can be stale and must not be dereferenced.
1824 * Return: 1 if the process is alive. 0 otherwise.
1826 static inline int pid_alive(const struct task_struct *p)
1828 return p->pids[PIDTYPE_PID].pid != NULL;
1832 * is_global_init - check if a task structure is init
1833 * @tsk: Task structure to be checked.
1835 * Check if a task structure is the first user space task the kernel created.
1837 * Return: 1 if the task structure is init. 0 otherwise.
1839 static inline int is_global_init(struct task_struct *tsk)
1841 return tsk->pid == 1;
1844 extern struct pid *cad_pid;
1846 extern void free_task(struct task_struct *tsk);
1847 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1849 extern void __put_task_struct(struct task_struct *t);
1851 static inline void put_task_struct(struct task_struct *t)
1853 if (atomic_dec_and_test(&t->usage))
1854 __put_task_struct(t);
1857 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1858 extern void task_cputime(struct task_struct *t,
1859 cputime_t *utime, cputime_t *stime);
1860 extern void task_cputime_scaled(struct task_struct *t,
1861 cputime_t *utimescaled, cputime_t *stimescaled);
1862 extern cputime_t task_gtime(struct task_struct *t);
1864 static inline void task_cputime(struct task_struct *t,
1865 cputime_t *utime, cputime_t *stime)
1873 static inline void task_cputime_scaled(struct task_struct *t,
1874 cputime_t *utimescaled,
1875 cputime_t *stimescaled)
1878 *utimescaled = t->utimescaled;
1880 *stimescaled = t->stimescaled;
1883 static inline cputime_t task_gtime(struct task_struct *t)
1888 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1889 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1894 #define PF_EXITING 0x00000004 /* getting shut down */
1895 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1896 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1897 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1898 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1899 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1900 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1901 #define PF_DUMPCORE 0x00000200 /* dumped core */
1902 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1903 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1904 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1905 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1906 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1907 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1908 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1909 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1910 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1911 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1912 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1913 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1914 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1915 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1916 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1917 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1918 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1919 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1920 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1921 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1922 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1925 * Only the _current_ task can read/write to tsk->flags, but other
1926 * tasks can access tsk->flags in readonly mode for example
1927 * with tsk_used_math (like during threaded core dumping).
1928 * There is however an exception to this rule during ptrace
1929 * or during fork: the ptracer task is allowed to write to the
1930 * child->flags of its traced child (same goes for fork, the parent
1931 * can write to the child->flags), because we're guaranteed the
1932 * child is not running and in turn not changing child->flags
1933 * at the same time the parent does it.
1935 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1936 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1937 #define clear_used_math() clear_stopped_child_used_math(current)
1938 #define set_used_math() set_stopped_child_used_math(current)
1939 #define conditional_stopped_child_used_math(condition, child) \
1940 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1941 #define conditional_used_math(condition) \
1942 conditional_stopped_child_used_math(condition, current)
1943 #define copy_to_stopped_child_used_math(child) \
1944 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1945 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1946 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1947 #define used_math() tsk_used_math(current)
1949 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1950 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1952 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1957 static inline unsigned int memalloc_noio_save(void)
1959 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1960 current->flags |= PF_MEMALLOC_NOIO;
1964 static inline void memalloc_noio_restore(unsigned int flags)
1966 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1970 * task->jobctl flags
1972 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1974 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1975 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1976 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1977 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1978 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1979 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1980 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1982 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1983 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1984 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1985 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1986 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1987 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1988 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1990 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1991 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1993 extern bool task_set_jobctl_pending(struct task_struct *task,
1995 extern void task_clear_jobctl_trapping(struct task_struct *task);
1996 extern void task_clear_jobctl_pending(struct task_struct *task,
1999 #ifdef CONFIG_PREEMPT_RCU
2001 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
2002 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
2004 static inline void rcu_copy_process(struct task_struct *p)
2006 p->rcu_read_lock_nesting = 0;
2007 p->rcu_read_unlock_special = 0;
2008 #ifdef CONFIG_TREE_PREEMPT_RCU
2009 p->rcu_blocked_node = NULL;
2010 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
2011 #ifdef CONFIG_RCU_BOOST
2012 p->rcu_boost_mutex = NULL;
2013 #endif /* #ifdef CONFIG_RCU_BOOST */
2014 INIT_LIST_HEAD(&p->rcu_node_entry);
2019 static inline void rcu_copy_process(struct task_struct *p)
2025 static inline void tsk_restore_flags(struct task_struct *task,
2026 unsigned long orig_flags, unsigned long flags)
2028 task->flags &= ~flags;
2029 task->flags |= orig_flags & flags;
2033 extern void do_set_cpus_allowed(struct task_struct *p,
2034 const struct cpumask *new_mask);
2036 extern int set_cpus_allowed_ptr(struct task_struct *p,
2037 const struct cpumask *new_mask);
2039 static inline void do_set_cpus_allowed(struct task_struct *p,
2040 const struct cpumask *new_mask)
2043 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2044 const struct cpumask *new_mask)
2046 if (!cpumask_test_cpu(0, new_mask))
2052 #ifdef CONFIG_NO_HZ_COMMON
2053 void calc_load_enter_idle(void);
2054 void calc_load_exit_idle(void);
2056 static inline void calc_load_enter_idle(void) { }
2057 static inline void calc_load_exit_idle(void) { }
2058 #endif /* CONFIG_NO_HZ_COMMON */
2060 #ifndef CONFIG_CPUMASK_OFFSTACK
2061 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2063 return set_cpus_allowed_ptr(p, &new_mask);
2068 * Do not use outside of architecture code which knows its limitations.
2070 * sched_clock() has no promise of monotonicity or bounded drift between
2071 * CPUs, use (which you should not) requires disabling IRQs.
2073 * Please use one of the three interfaces below.
2075 extern unsigned long long notrace sched_clock(void);
2077 * See the comment in kernel/sched/clock.c
2079 extern u64 cpu_clock(int cpu);
2080 extern u64 local_clock(void);
2081 extern u64 sched_clock_cpu(int cpu);
2084 extern void sched_clock_init(void);
2086 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2087 static inline void sched_clock_tick(void)
2091 static inline void sched_clock_idle_sleep_event(void)
2095 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2100 * Architectures can set this to 1 if they have specified
2101 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2102 * but then during bootup it turns out that sched_clock()
2103 * is reliable after all:
2105 extern int sched_clock_stable(void);
2106 extern void set_sched_clock_stable(void);
2107 extern void clear_sched_clock_stable(void);
2109 extern void sched_clock_tick(void);
2110 extern void sched_clock_idle_sleep_event(void);
2111 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2114 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2116 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2117 * The reason for this explicit opt-in is not to have perf penalty with
2118 * slow sched_clocks.
2120 extern void enable_sched_clock_irqtime(void);
2121 extern void disable_sched_clock_irqtime(void);
2123 static inline void enable_sched_clock_irqtime(void) {}
2124 static inline void disable_sched_clock_irqtime(void) {}
2127 extern unsigned long long
2128 task_sched_runtime(struct task_struct *task);
2130 /* sched_exec is called by processes performing an exec */
2132 extern void sched_exec(void);
2134 #define sched_exec() {}
2137 extern void sched_clock_idle_sleep_event(void);
2138 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2140 #ifdef CONFIG_HOTPLUG_CPU
2141 extern void idle_task_exit(void);
2143 static inline void idle_task_exit(void) {}
2146 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2147 extern void wake_up_nohz_cpu(int cpu);
2149 static inline void wake_up_nohz_cpu(int cpu) { }
2152 #ifdef CONFIG_NO_HZ_FULL
2153 extern bool sched_can_stop_tick(void);
2154 extern u64 scheduler_tick_max_deferment(void);
2156 static inline bool sched_can_stop_tick(void) { return false; }
2159 #ifdef CONFIG_SCHED_AUTOGROUP
2160 extern void sched_autogroup_create_attach(struct task_struct *p);
2161 extern void sched_autogroup_detach(struct task_struct *p);
2162 extern void sched_autogroup_fork(struct signal_struct *sig);
2163 extern void sched_autogroup_exit(struct signal_struct *sig);
2164 #ifdef CONFIG_PROC_FS
2165 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2166 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2169 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2170 static inline void sched_autogroup_detach(struct task_struct *p) { }
2171 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2172 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2175 extern int yield_to(struct task_struct *p, bool preempt);
2176 extern void set_user_nice(struct task_struct *p, long nice);
2177 extern int task_prio(const struct task_struct *p);
2179 * task_nice - return the nice value of a given task.
2180 * @p: the task in question.
2182 * Return: The nice value [ -20 ... 0 ... 19 ].
2184 static inline int task_nice(const struct task_struct *p)
2186 return PRIO_TO_NICE((p)->static_prio);
2188 extern int can_nice(const struct task_struct *p, const int nice);
2189 extern int task_curr(const struct task_struct *p);
2190 extern int idle_cpu(int cpu);
2191 extern int sched_setscheduler(struct task_struct *, int,
2192 const struct sched_param *);
2193 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2194 const struct sched_param *);
2195 extern int sched_setattr(struct task_struct *,
2196 const struct sched_attr *);
2197 extern struct task_struct *idle_task(int cpu);
2199 * is_idle_task - is the specified task an idle task?
2200 * @p: the task in question.
2202 * Return: 1 if @p is an idle task. 0 otherwise.
2204 static inline bool is_idle_task(const struct task_struct *p)
2208 extern struct task_struct *curr_task(int cpu);
2209 extern void set_curr_task(int cpu, struct task_struct *p);
2214 * The default (Linux) execution domain.
2216 extern struct exec_domain default_exec_domain;
2218 union thread_union {
2219 struct thread_info thread_info;
2220 unsigned long stack[THREAD_SIZE/sizeof(long)];
2223 #ifndef __HAVE_ARCH_KSTACK_END
2224 static inline int kstack_end(void *addr)
2226 /* Reliable end of stack detection:
2227 * Some APM bios versions misalign the stack
2229 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2233 extern union thread_union init_thread_union;
2234 extern struct task_struct init_task;
2236 extern struct mm_struct init_mm;
2238 extern struct pid_namespace init_pid_ns;
2241 * find a task by one of its numerical ids
2243 * find_task_by_pid_ns():
2244 * finds a task by its pid in the specified namespace
2245 * find_task_by_vpid():
2246 * finds a task by its virtual pid
2248 * see also find_vpid() etc in include/linux/pid.h
2251 extern struct task_struct *find_task_by_vpid(pid_t nr);
2252 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2253 struct pid_namespace *ns);
2255 /* per-UID process charging. */
2256 extern struct user_struct * alloc_uid(kuid_t);
2257 static inline struct user_struct *get_uid(struct user_struct *u)
2259 atomic_inc(&u->__count);
2262 extern void free_uid(struct user_struct *);
2264 #include <asm/current.h>
2266 extern void xtime_update(unsigned long ticks);
2268 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2269 extern int wake_up_process(struct task_struct *tsk);
2270 extern void wake_up_new_task(struct task_struct *tsk);
2272 extern void kick_process(struct task_struct *tsk);
2274 static inline void kick_process(struct task_struct *tsk) { }
2276 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2277 extern void sched_dead(struct task_struct *p);
2279 extern void proc_caches_init(void);
2280 extern void flush_signals(struct task_struct *);
2281 extern void __flush_signals(struct task_struct *);
2282 extern void ignore_signals(struct task_struct *);
2283 extern void flush_signal_handlers(struct task_struct *, int force_default);
2284 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2286 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2288 unsigned long flags;
2291 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2292 ret = dequeue_signal(tsk, mask, info);
2293 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2298 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2300 extern void unblock_all_signals(void);
2301 extern void release_task(struct task_struct * p);
2302 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2303 extern int force_sigsegv(int, struct task_struct *);
2304 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2305 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2306 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2307 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2308 const struct cred *, u32);
2309 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2310 extern int kill_pid(struct pid *pid, int sig, int priv);
2311 extern int kill_proc_info(int, struct siginfo *, pid_t);
2312 extern __must_check bool do_notify_parent(struct task_struct *, int);
2313 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2314 extern void force_sig(int, struct task_struct *);
2315 extern int send_sig(int, struct task_struct *, int);
2316 extern int zap_other_threads(struct task_struct *p);
2317 extern struct sigqueue *sigqueue_alloc(void);
2318 extern void sigqueue_free(struct sigqueue *);
2319 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2320 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2322 static inline void restore_saved_sigmask(void)
2324 if (test_and_clear_restore_sigmask())
2325 __set_current_blocked(¤t->saved_sigmask);
2328 static inline sigset_t *sigmask_to_save(void)
2330 sigset_t *res = ¤t->blocked;
2331 if (unlikely(test_restore_sigmask()))
2332 res = ¤t->saved_sigmask;
2336 static inline int kill_cad_pid(int sig, int priv)
2338 return kill_pid(cad_pid, sig, priv);
2341 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2342 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2343 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2344 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2347 * True if we are on the alternate signal stack.
2349 static inline int on_sig_stack(unsigned long sp)
2351 #ifdef CONFIG_STACK_GROWSUP
2352 return sp >= current->sas_ss_sp &&
2353 sp - current->sas_ss_sp < current->sas_ss_size;
2355 return sp > current->sas_ss_sp &&
2356 sp - current->sas_ss_sp <= current->sas_ss_size;
2360 static inline int sas_ss_flags(unsigned long sp)
2362 return (current->sas_ss_size == 0 ? SS_DISABLE
2363 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2366 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2368 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2369 #ifdef CONFIG_STACK_GROWSUP
2370 return current->sas_ss_sp;
2372 return current->sas_ss_sp + current->sas_ss_size;
2378 * Routines for handling mm_structs
2380 extern struct mm_struct * mm_alloc(void);
2382 /* mmdrop drops the mm and the page tables */
2383 extern void __mmdrop(struct mm_struct *);
2384 static inline void mmdrop(struct mm_struct * mm)
2386 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2390 /* mmput gets rid of the mappings and all user-space */
2391 extern void mmput(struct mm_struct *);
2392 /* Grab a reference to a task's mm, if it is not already going away */
2393 extern struct mm_struct *get_task_mm(struct task_struct *task);
2395 * Grab a reference to a task's mm, if it is not already going away
2396 * and ptrace_may_access with the mode parameter passed to it
2399 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2400 /* Remove the current tasks stale references to the old mm_struct */
2401 extern void mm_release(struct task_struct *, struct mm_struct *);
2403 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2404 struct task_struct *);
2405 extern void flush_thread(void);
2406 extern void exit_thread(void);
2408 extern void exit_files(struct task_struct *);
2409 extern void __cleanup_sighand(struct sighand_struct *);
2411 extern void exit_itimers(struct signal_struct *);
2412 extern void flush_itimer_signals(void);
2414 extern void do_group_exit(int);
2416 extern int do_execve(struct filename *,
2417 const char __user * const __user *,
2418 const char __user * const __user *);
2419 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2420 struct task_struct *fork_idle(int);
2421 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2423 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2424 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2426 __set_task_comm(tsk, from, false);
2428 extern char *get_task_comm(char *to, struct task_struct *tsk);
2431 void scheduler_ipi(void);
2432 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2434 static inline void scheduler_ipi(void) { }
2435 static inline unsigned long wait_task_inactive(struct task_struct *p,
2442 #define next_task(p) \
2443 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2445 #define for_each_process(p) \
2446 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2448 extern bool current_is_single_threaded(void);
2451 * Careful: do_each_thread/while_each_thread is a double loop so
2452 * 'break' will not work as expected - use goto instead.
2454 #define do_each_thread(g, t) \
2455 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2457 #define while_each_thread(g, t) \
2458 while ((t = next_thread(t)) != g)
2460 #define __for_each_thread(signal, t) \
2461 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2463 #define for_each_thread(p, t) \
2464 __for_each_thread((p)->signal, t)
2466 /* Careful: this is a double loop, 'break' won't work as expected. */
2467 #define for_each_process_thread(p, t) \
2468 for_each_process(p) for_each_thread(p, t)
2470 static inline int get_nr_threads(struct task_struct *tsk)
2472 return tsk->signal->nr_threads;
2475 static inline bool thread_group_leader(struct task_struct *p)
2477 return p->exit_signal >= 0;
2480 /* Do to the insanities of de_thread it is possible for a process
2481 * to have the pid of the thread group leader without actually being
2482 * the thread group leader. For iteration through the pids in proc
2483 * all we care about is that we have a task with the appropriate
2484 * pid, we don't actually care if we have the right task.
2486 static inline bool has_group_leader_pid(struct task_struct *p)
2488 return task_pid(p) == p->signal->leader_pid;
2492 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2494 return p1->signal == p2->signal;
2497 static inline struct task_struct *next_thread(const struct task_struct *p)
2499 return list_entry_rcu(p->thread_group.next,
2500 struct task_struct, thread_group);
2503 static inline int thread_group_empty(struct task_struct *p)
2505 return list_empty(&p->thread_group);
2508 #define delay_group_leader(p) \
2509 (thread_group_leader(p) && !thread_group_empty(p))
2512 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2513 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2514 * pins the final release of task.io_context. Also protects ->cpuset and
2515 * ->cgroup.subsys[]. And ->vfork_done.
2517 * Nests both inside and outside of read_lock(&tasklist_lock).
2518 * It must not be nested with write_lock_irq(&tasklist_lock),
2519 * neither inside nor outside.
2521 static inline void task_lock(struct task_struct *p)
2523 spin_lock(&p->alloc_lock);
2526 static inline void task_unlock(struct task_struct *p)
2528 spin_unlock(&p->alloc_lock);
2531 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2532 unsigned long *flags);
2534 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2535 unsigned long *flags)
2537 struct sighand_struct *ret;
2539 ret = __lock_task_sighand(tsk, flags);
2540 (void)__cond_lock(&tsk->sighand->siglock, ret);
2544 static inline void unlock_task_sighand(struct task_struct *tsk,
2545 unsigned long *flags)
2547 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2550 #ifdef CONFIG_CGROUPS
2551 static inline void threadgroup_change_begin(struct task_struct *tsk)
2553 down_read(&tsk->signal->group_rwsem);
2555 static inline void threadgroup_change_end(struct task_struct *tsk)
2557 up_read(&tsk->signal->group_rwsem);
2561 * threadgroup_lock - lock threadgroup
2562 * @tsk: member task of the threadgroup to lock
2564 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2565 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2566 * change ->group_leader/pid. This is useful for cases where the threadgroup
2567 * needs to stay stable across blockable operations.
2569 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2570 * synchronization. While held, no new task will be added to threadgroup
2571 * and no existing live task will have its PF_EXITING set.
2573 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2574 * sub-thread becomes a new leader.
2576 static inline void threadgroup_lock(struct task_struct *tsk)
2578 down_write(&tsk->signal->group_rwsem);
2582 * threadgroup_unlock - unlock threadgroup
2583 * @tsk: member task of the threadgroup to unlock
2585 * Reverse threadgroup_lock().
2587 static inline void threadgroup_unlock(struct task_struct *tsk)
2589 up_write(&tsk->signal->group_rwsem);
2592 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2593 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2594 static inline void threadgroup_lock(struct task_struct *tsk) {}
2595 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2598 #ifndef __HAVE_THREAD_FUNCTIONS
2600 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2601 #define task_stack_page(task) ((task)->stack)
2603 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2605 *task_thread_info(p) = *task_thread_info(org);
2606 task_thread_info(p)->task = p;
2609 static inline unsigned long *end_of_stack(struct task_struct *p)
2611 return (unsigned long *)(task_thread_info(p) + 1);
2616 static inline int object_is_on_stack(void *obj)
2618 void *stack = task_stack_page(current);
2620 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2623 extern void thread_info_cache_init(void);
2625 #ifdef CONFIG_DEBUG_STACK_USAGE
2626 static inline unsigned long stack_not_used(struct task_struct *p)
2628 unsigned long *n = end_of_stack(p);
2630 do { /* Skip over canary */
2634 return (unsigned long)n - (unsigned long)end_of_stack(p);
2638 /* set thread flags in other task's structures
2639 * - see asm/thread_info.h for TIF_xxxx flags available
2641 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2643 set_ti_thread_flag(task_thread_info(tsk), flag);
2646 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2648 clear_ti_thread_flag(task_thread_info(tsk), flag);
2651 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2653 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2656 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2658 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2661 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2663 return test_ti_thread_flag(task_thread_info(tsk), flag);
2666 static inline void set_tsk_need_resched(struct task_struct *tsk)
2668 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2671 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2673 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2676 static inline int test_tsk_need_resched(struct task_struct *tsk)
2678 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2681 static inline int restart_syscall(void)
2683 set_tsk_thread_flag(current, TIF_SIGPENDING);
2684 return -ERESTARTNOINTR;
2687 static inline int signal_pending(struct task_struct *p)
2689 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2692 static inline int __fatal_signal_pending(struct task_struct *p)
2694 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2697 static inline int fatal_signal_pending(struct task_struct *p)
2699 return signal_pending(p) && __fatal_signal_pending(p);
2702 static inline int signal_pending_state(long state, struct task_struct *p)
2704 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2706 if (!signal_pending(p))
2709 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2713 * cond_resched() and cond_resched_lock(): latency reduction via
2714 * explicit rescheduling in places that are safe. The return
2715 * value indicates whether a reschedule was done in fact.
2716 * cond_resched_lock() will drop the spinlock before scheduling,
2717 * cond_resched_softirq() will enable bhs before scheduling.
2719 extern int _cond_resched(void);
2721 #define cond_resched() ({ \
2722 __might_sleep(__FILE__, __LINE__, 0); \
2726 extern int __cond_resched_lock(spinlock_t *lock);
2728 #ifdef CONFIG_PREEMPT_COUNT
2729 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2731 #define PREEMPT_LOCK_OFFSET 0
2734 #define cond_resched_lock(lock) ({ \
2735 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2736 __cond_resched_lock(lock); \
2739 extern int __cond_resched_softirq(void);
2741 #define cond_resched_softirq() ({ \
2742 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2743 __cond_resched_softirq(); \
2746 static inline void cond_resched_rcu(void)
2748 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2756 * Does a critical section need to be broken due to another
2757 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2758 * but a general need for low latency)
2760 static inline int spin_needbreak(spinlock_t *lock)
2762 #ifdef CONFIG_PREEMPT
2763 return spin_is_contended(lock);
2770 * Idle thread specific functions to determine the need_resched
2773 #ifdef TIF_POLLING_NRFLAG
2774 static inline int tsk_is_polling(struct task_struct *p)
2776 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2779 static inline void __current_set_polling(void)
2781 set_thread_flag(TIF_POLLING_NRFLAG);
2784 static inline bool __must_check current_set_polling_and_test(void)
2786 __current_set_polling();
2789 * Polling state must be visible before we test NEED_RESCHED,
2790 * paired by resched_task()
2792 smp_mb__after_atomic();
2794 return unlikely(tif_need_resched());
2797 static inline void __current_clr_polling(void)
2799 clear_thread_flag(TIF_POLLING_NRFLAG);
2802 static inline bool __must_check current_clr_polling_and_test(void)
2804 __current_clr_polling();
2807 * Polling state must be visible before we test NEED_RESCHED,
2808 * paired by resched_task()
2810 smp_mb__after_atomic();
2812 return unlikely(tif_need_resched());
2816 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2817 static inline void __current_set_polling(void) { }
2818 static inline void __current_clr_polling(void) { }
2820 static inline bool __must_check current_set_polling_and_test(void)
2822 return unlikely(tif_need_resched());
2824 static inline bool __must_check current_clr_polling_and_test(void)
2826 return unlikely(tif_need_resched());
2830 static inline void current_clr_polling(void)
2832 __current_clr_polling();
2835 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2836 * Once the bit is cleared, we'll get IPIs with every new
2837 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2840 smp_mb(); /* paired with resched_task() */
2842 preempt_fold_need_resched();
2845 static __always_inline bool need_resched(void)
2847 return unlikely(tif_need_resched());
2851 * Thread group CPU time accounting.
2853 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2854 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2856 static inline void thread_group_cputime_init(struct signal_struct *sig)
2858 raw_spin_lock_init(&sig->cputimer.lock);
2862 * Reevaluate whether the task has signals pending delivery.
2863 * Wake the task if so.
2864 * This is required every time the blocked sigset_t changes.
2865 * callers must hold sighand->siglock.
2867 extern void recalc_sigpending_and_wake(struct task_struct *t);
2868 extern void recalc_sigpending(void);
2870 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2872 static inline void signal_wake_up(struct task_struct *t, bool resume)
2874 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2876 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2878 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2882 * Wrappers for p->thread_info->cpu access. No-op on UP.
2886 static inline unsigned int task_cpu(const struct task_struct *p)
2888 return task_thread_info(p)->cpu;
2891 static inline int task_node(const struct task_struct *p)
2893 return cpu_to_node(task_cpu(p));
2896 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2900 static inline unsigned int task_cpu(const struct task_struct *p)
2905 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2909 #endif /* CONFIG_SMP */
2911 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2912 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2914 #ifdef CONFIG_CGROUP_SCHED
2915 extern struct task_group root_task_group;
2916 #endif /* CONFIG_CGROUP_SCHED */
2918 extern int task_can_switch_user(struct user_struct *up,
2919 struct task_struct *tsk);
2921 #ifdef CONFIG_TASK_XACCT
2922 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2924 tsk->ioac.rchar += amt;
2927 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2929 tsk->ioac.wchar += amt;
2932 static inline void inc_syscr(struct task_struct *tsk)
2937 static inline void inc_syscw(struct task_struct *tsk)
2942 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2946 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2950 static inline void inc_syscr(struct task_struct *tsk)
2954 static inline void inc_syscw(struct task_struct *tsk)
2959 #ifndef TASK_SIZE_OF
2960 #define TASK_SIZE_OF(tsk) TASK_SIZE
2964 extern void mm_update_next_owner(struct mm_struct *mm);
2965 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2967 static inline void mm_update_next_owner(struct mm_struct *mm)
2971 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2974 #endif /* CONFIG_MEMCG */
2976 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2979 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2982 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2985 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2988 static inline unsigned long rlimit(unsigned int limit)
2990 return task_rlimit(current, limit);
2993 static inline unsigned long rlimit_max(unsigned int limit)
2995 return task_rlimit_max(current, limit);