4 #include <uapi/linux/sched.h>
11 #include <asm/param.h> /* for HZ */
13 #include <linux/capability.h>
14 #include <linux/threads.h>
15 #include <linux/kernel.h>
16 #include <linux/types.h>
17 #include <linux/timex.h>
18 #include <linux/jiffies.h>
19 #include <linux/rbtree.h>
20 #include <linux/thread_info.h>
21 #include <linux/cpumask.h>
22 #include <linux/errno.h>
23 #include <linux/nodemask.h>
24 #include <linux/mm_types.h>
25 #include <linux/preempt_mask.h>
28 #include <asm/ptrace.h>
29 #include <asm/cputime.h>
31 #include <linux/smp.h>
32 #include <linux/sem.h>
33 #include <linux/signal.h>
34 #include <linux/compiler.h>
35 #include <linux/completion.h>
36 #include <linux/pid.h>
37 #include <linux/percpu.h>
38 #include <linux/topology.h>
39 #include <linux/proportions.h>
40 #include <linux/seccomp.h>
41 #include <linux/rcupdate.h>
42 #include <linux/rculist.h>
43 #include <linux/rtmutex.h>
45 #include <linux/time.h>
46 #include <linux/param.h>
47 #include <linux/resource.h>
48 #include <linux/timer.h>
49 #include <linux/hrtimer.h>
50 #include <linux/task_io_accounting.h>
51 #include <linux/latencytop.h>
52 #include <linux/cred.h>
53 #include <linux/llist.h>
54 #include <linux/uidgid.h>
55 #include <linux/gfp.h>
57 #include <asm/processor.h>
59 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
62 * Extended scheduling parameters data structure.
64 * This is needed because the original struct sched_param can not be
65 * altered without introducing ABI issues with legacy applications
66 * (e.g., in sched_getparam()).
68 * However, the possibility of specifying more than just a priority for
69 * the tasks may be useful for a wide variety of application fields, e.g.,
70 * multimedia, streaming, automation and control, and many others.
72 * This variant (sched_attr) is meant at describing a so-called
73 * sporadic time-constrained task. In such model a task is specified by:
74 * - the activation period or minimum instance inter-arrival time;
75 * - the maximum (or average, depending on the actual scheduling
76 * discipline) computation time of all instances, a.k.a. runtime;
77 * - the deadline (relative to the actual activation time) of each
79 * Very briefly, a periodic (sporadic) task asks for the execution of
80 * some specific computation --which is typically called an instance--
81 * (at most) every period. Moreover, each instance typically lasts no more
82 * than the runtime and must be completed by time instant t equal to
83 * the instance activation time + the deadline.
85 * This is reflected by the actual fields of the sched_attr structure:
87 * @size size of the structure, for fwd/bwd compat.
89 * @sched_policy task's scheduling policy
90 * @sched_flags for customizing the scheduler behaviour
91 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
92 * @sched_priority task's static priority (SCHED_FIFO/RR)
93 * @sched_deadline representative of the task's deadline
94 * @sched_runtime representative of the task's runtime
95 * @sched_period representative of the task's period
97 * Given this task model, there are a multiplicity of scheduling algorithms
98 * and policies, that can be used to ensure all the tasks will make their
101 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
102 * only user of this new interface. More information about the algorithm
103 * available in the scheduling class file or in Documentation/.
111 /* SCHED_NORMAL, SCHED_BATCH */
114 /* SCHED_FIFO, SCHED_RR */
124 struct futex_pi_state;
125 struct robust_list_head;
128 struct perf_event_context;
132 * List of flags we want to share for kernel threads,
133 * if only because they are not used by them anyway.
135 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
138 * These are the constant used to fake the fixed-point load-average
139 * counting. Some notes:
140 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
141 * a load-average precision of 10 bits integer + 11 bits fractional
142 * - if you want to count load-averages more often, you need more
143 * precision, or rounding will get you. With 2-second counting freq,
144 * the EXP_n values would be 1981, 2034 and 2043 if still using only
147 extern unsigned long avenrun[]; /* Load averages */
148 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
150 #define FSHIFT 11 /* nr of bits of precision */
151 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
152 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
153 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
154 #define EXP_5 2014 /* 1/exp(5sec/5min) */
155 #define EXP_15 2037 /* 1/exp(5sec/15min) */
157 #define CALC_LOAD(load,exp,n) \
159 load += n*(FIXED_1-exp); \
162 extern unsigned long total_forks;
163 extern int nr_threads;
164 DECLARE_PER_CPU(unsigned long, process_counts);
165 extern int nr_processes(void);
166 extern unsigned long nr_running(void);
167 extern unsigned long nr_iowait(void);
168 extern unsigned long nr_iowait_cpu(int cpu);
169 extern unsigned long this_cpu_load(void);
172 extern void calc_global_load(unsigned long ticks);
173 extern void update_cpu_load_nohz(void);
175 extern unsigned long get_parent_ip(unsigned long addr);
177 extern void dump_cpu_task(int cpu);
182 #ifdef CONFIG_SCHED_DEBUG
183 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
184 extern void proc_sched_set_task(struct task_struct *p);
186 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
190 * Task state bitmask. NOTE! These bits are also
191 * encoded in fs/proc/array.c: get_task_state().
193 * We have two separate sets of flags: task->state
194 * is about runnability, while task->exit_state are
195 * about the task exiting. Confusing, but this way
196 * modifying one set can't modify the other one by
199 #define TASK_RUNNING 0
200 #define TASK_INTERRUPTIBLE 1
201 #define TASK_UNINTERRUPTIBLE 2
202 #define __TASK_STOPPED 4
203 #define __TASK_TRACED 8
204 /* in tsk->exit_state */
205 #define EXIT_ZOMBIE 16
207 /* in tsk->state again */
209 #define TASK_WAKEKILL 128
210 #define TASK_WAKING 256
211 #define TASK_PARKED 512
212 #define TASK_STATE_MAX 1024
214 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
216 extern char ___assert_task_state[1 - 2*!!(
217 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
219 /* Convenience macros for the sake of set_task_state */
220 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
221 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
222 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
224 /* Convenience macros for the sake of wake_up */
225 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
226 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
228 /* get_task_state() */
229 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
230 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
233 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
234 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
235 #define task_is_stopped_or_traced(task) \
236 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
237 #define task_contributes_to_load(task) \
238 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
239 (task->flags & PF_FROZEN) == 0)
241 #define __set_task_state(tsk, state_value) \
242 do { (tsk)->state = (state_value); } while (0)
243 #define set_task_state(tsk, state_value) \
244 set_mb((tsk)->state, (state_value))
247 * set_current_state() includes a barrier so that the write of current->state
248 * is correctly serialised wrt the caller's subsequent test of whether to
251 * set_current_state(TASK_UNINTERRUPTIBLE);
252 * if (do_i_need_to_sleep())
255 * If the caller does not need such serialisation then use __set_current_state()
257 #define __set_current_state(state_value) \
258 do { current->state = (state_value); } while (0)
259 #define set_current_state(state_value) \
260 set_mb(current->state, (state_value))
262 /* Task command name length */
263 #define TASK_COMM_LEN 16
265 #include <linux/spinlock.h>
268 * This serializes "schedule()" and also protects
269 * the run-queue from deletions/modifications (but
270 * _adding_ to the beginning of the run-queue has
273 extern rwlock_t tasklist_lock;
274 extern spinlock_t mmlist_lock;
278 #ifdef CONFIG_PROVE_RCU
279 extern int lockdep_tasklist_lock_is_held(void);
280 #endif /* #ifdef CONFIG_PROVE_RCU */
282 extern void sched_init(void);
283 extern void sched_init_smp(void);
284 extern asmlinkage void schedule_tail(struct task_struct *prev);
285 extern void init_idle(struct task_struct *idle, int cpu);
286 extern void init_idle_bootup_task(struct task_struct *idle);
288 extern int runqueue_is_locked(int cpu);
290 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
291 extern void nohz_balance_enter_idle(int cpu);
292 extern void set_cpu_sd_state_idle(void);
293 extern int get_nohz_timer_target(void);
295 static inline void nohz_balance_enter_idle(int cpu) { }
296 static inline void set_cpu_sd_state_idle(void) { }
300 * Only dump TASK_* tasks. (0 for all tasks)
302 extern void show_state_filter(unsigned long state_filter);
304 static inline void show_state(void)
306 show_state_filter(0);
309 extern void show_regs(struct pt_regs *);
312 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
313 * task), SP is the stack pointer of the first frame that should be shown in the back
314 * trace (or NULL if the entire call-chain of the task should be shown).
316 extern void show_stack(struct task_struct *task, unsigned long *sp);
318 void io_schedule(void);
319 long io_schedule_timeout(long timeout);
321 extern void cpu_init (void);
322 extern void trap_init(void);
323 extern void update_process_times(int user);
324 extern void scheduler_tick(void);
326 extern void sched_show_task(struct task_struct *p);
328 #ifdef CONFIG_LOCKUP_DETECTOR
329 extern void touch_softlockup_watchdog(void);
330 extern void touch_softlockup_watchdog_sync(void);
331 extern void touch_all_softlockup_watchdogs(void);
332 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
334 size_t *lenp, loff_t *ppos);
335 extern unsigned int softlockup_panic;
336 void lockup_detector_init(void);
338 static inline void touch_softlockup_watchdog(void)
341 static inline void touch_softlockup_watchdog_sync(void)
344 static inline void touch_all_softlockup_watchdogs(void)
347 static inline void lockup_detector_init(void)
352 #ifdef CONFIG_DETECT_HUNG_TASK
353 void reset_hung_task_detector(void);
355 static inline void reset_hung_task_detector(void)
360 /* Attach to any functions which should be ignored in wchan output. */
361 #define __sched __attribute__((__section__(".sched.text")))
363 /* Linker adds these: start and end of __sched functions */
364 extern char __sched_text_start[], __sched_text_end[];
366 /* Is this address in the __sched functions? */
367 extern int in_sched_functions(unsigned long addr);
369 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
370 extern signed long schedule_timeout(signed long timeout);
371 extern signed long schedule_timeout_interruptible(signed long timeout);
372 extern signed long schedule_timeout_killable(signed long timeout);
373 extern signed long schedule_timeout_uninterruptible(signed long timeout);
374 asmlinkage void schedule(void);
375 extern void schedule_preempt_disabled(void);
378 struct user_namespace;
381 extern void arch_pick_mmap_layout(struct mm_struct *mm);
383 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
384 unsigned long, unsigned long);
386 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
387 unsigned long len, unsigned long pgoff,
388 unsigned long flags);
390 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
394 extern void set_dumpable(struct mm_struct *mm, int value);
395 extern int get_dumpable(struct mm_struct *mm);
397 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
398 #define SUID_DUMP_USER 1 /* Dump as user of process */
399 #define SUID_DUMP_ROOT 2 /* Dump as root */
403 #define MMF_DUMPABLE 0 /* core dump is permitted */
404 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
406 #define MMF_DUMPABLE_BITS 2
407 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
409 /* coredump filter bits */
410 #define MMF_DUMP_ANON_PRIVATE 2
411 #define MMF_DUMP_ANON_SHARED 3
412 #define MMF_DUMP_MAPPED_PRIVATE 4
413 #define MMF_DUMP_MAPPED_SHARED 5
414 #define MMF_DUMP_ELF_HEADERS 6
415 #define MMF_DUMP_HUGETLB_PRIVATE 7
416 #define MMF_DUMP_HUGETLB_SHARED 8
418 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
419 #define MMF_DUMP_FILTER_BITS 7
420 #define MMF_DUMP_FILTER_MASK \
421 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
422 #define MMF_DUMP_FILTER_DEFAULT \
423 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
424 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
426 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
427 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
429 # define MMF_DUMP_MASK_DEFAULT_ELF 0
431 /* leave room for more dump flags */
432 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
433 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
434 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
436 #define MMF_HAS_UPROBES 19 /* has uprobes */
437 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
439 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
441 struct sighand_struct {
443 struct k_sigaction action[_NSIG];
445 wait_queue_head_t signalfd_wqh;
448 struct pacct_struct {
451 unsigned long ac_mem;
452 cputime_t ac_utime, ac_stime;
453 unsigned long ac_minflt, ac_majflt;
464 * struct cputime - snaphsot of system and user cputime
465 * @utime: time spent in user mode
466 * @stime: time spent in system mode
468 * Gathers a generic snapshot of user and system time.
476 * struct task_cputime - collected CPU time counts
477 * @utime: time spent in user mode, in &cputime_t units
478 * @stime: time spent in kernel mode, in &cputime_t units
479 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
481 * This is an extension of struct cputime that includes the total runtime
482 * spent by the task from the scheduler point of view.
484 * As a result, this structure groups together three kinds of CPU time
485 * that are tracked for threads and thread groups. Most things considering
486 * CPU time want to group these counts together and treat all three
487 * of them in parallel.
489 struct task_cputime {
492 unsigned long long sum_exec_runtime;
494 /* Alternate field names when used to cache expirations. */
495 #define prof_exp stime
496 #define virt_exp utime
497 #define sched_exp sum_exec_runtime
499 #define INIT_CPUTIME \
500 (struct task_cputime) { \
503 .sum_exec_runtime = 0, \
506 #ifdef CONFIG_PREEMPT_COUNT
507 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
509 #define PREEMPT_DISABLED PREEMPT_ENABLED
513 * Disable preemption until the scheduler is running.
514 * Reset by start_kernel()->sched_init()->init_idle().
516 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
517 * before the scheduler is active -- see should_resched().
519 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
522 * struct thread_group_cputimer - thread group interval timer counts
523 * @cputime: thread group interval timers.
524 * @running: non-zero when there are timers running and
525 * @cputime receives updates.
526 * @lock: lock for fields in this struct.
528 * This structure contains the version of task_cputime, above, that is
529 * used for thread group CPU timer calculations.
531 struct thread_group_cputimer {
532 struct task_cputime cputime;
537 #include <linux/rwsem.h>
541 * NOTE! "signal_struct" does not have its own
542 * locking, because a shared signal_struct always
543 * implies a shared sighand_struct, so locking
544 * sighand_struct is always a proper superset of
545 * the locking of signal_struct.
547 struct signal_struct {
552 wait_queue_head_t wait_chldexit; /* for wait4() */
554 /* current thread group signal load-balancing target: */
555 struct task_struct *curr_target;
557 /* shared signal handling: */
558 struct sigpending shared_pending;
560 /* thread group exit support */
563 * - notify group_exit_task when ->count is equal to notify_count
564 * - everyone except group_exit_task is stopped during signal delivery
565 * of fatal signals, group_exit_task processes the signal.
568 struct task_struct *group_exit_task;
570 /* thread group stop support, overloads group_exit_code too */
571 int group_stop_count;
572 unsigned int flags; /* see SIGNAL_* flags below */
575 * PR_SET_CHILD_SUBREAPER marks a process, like a service
576 * manager, to re-parent orphan (double-forking) child processes
577 * to this process instead of 'init'. The service manager is
578 * able to receive SIGCHLD signals and is able to investigate
579 * the process until it calls wait(). All children of this
580 * process will inherit a flag if they should look for a
581 * child_subreaper process at exit.
583 unsigned int is_child_subreaper:1;
584 unsigned int has_child_subreaper:1;
586 /* POSIX.1b Interval Timers */
588 struct list_head posix_timers;
590 /* ITIMER_REAL timer for the process */
591 struct hrtimer real_timer;
592 struct pid *leader_pid;
593 ktime_t it_real_incr;
596 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
597 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
598 * values are defined to 0 and 1 respectively
600 struct cpu_itimer it[2];
603 * Thread group totals for process CPU timers.
604 * See thread_group_cputimer(), et al, for details.
606 struct thread_group_cputimer cputimer;
608 /* Earliest-expiration cache. */
609 struct task_cputime cputime_expires;
611 struct list_head cpu_timers[3];
613 struct pid *tty_old_pgrp;
615 /* boolean value for session group leader */
618 struct tty_struct *tty; /* NULL if no tty */
620 #ifdef CONFIG_SCHED_AUTOGROUP
621 struct autogroup *autogroup;
624 * Cumulative resource counters for dead threads in the group,
625 * and for reaped dead child processes forked by this group.
626 * Live threads maintain their own counters and add to these
627 * in __exit_signal, except for the group leader.
629 cputime_t utime, stime, cutime, cstime;
632 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
633 struct cputime prev_cputime;
635 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
636 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
637 unsigned long inblock, oublock, cinblock, coublock;
638 unsigned long maxrss, cmaxrss;
639 struct task_io_accounting ioac;
642 * Cumulative ns of schedule CPU time fo dead threads in the
643 * group, not including a zombie group leader, (This only differs
644 * from jiffies_to_ns(utime + stime) if sched_clock uses something
645 * other than jiffies.)
647 unsigned long long sum_sched_runtime;
650 * We don't bother to synchronize most readers of this at all,
651 * because there is no reader checking a limit that actually needs
652 * to get both rlim_cur and rlim_max atomically, and either one
653 * alone is a single word that can safely be read normally.
654 * getrlimit/setrlimit use task_lock(current->group_leader) to
655 * protect this instead of the siglock, because they really
656 * have no need to disable irqs.
658 struct rlimit rlim[RLIM_NLIMITS];
660 #ifdef CONFIG_BSD_PROCESS_ACCT
661 struct pacct_struct pacct; /* per-process accounting information */
663 #ifdef CONFIG_TASKSTATS
664 struct taskstats *stats;
668 unsigned audit_tty_log_passwd;
669 struct tty_audit_buf *tty_audit_buf;
671 #ifdef CONFIG_CGROUPS
673 * group_rwsem prevents new tasks from entering the threadgroup and
674 * member tasks from exiting,a more specifically, setting of
675 * PF_EXITING. fork and exit paths are protected with this rwsem
676 * using threadgroup_change_begin/end(). Users which require
677 * threadgroup to remain stable should use threadgroup_[un]lock()
678 * which also takes care of exec path. Currently, cgroup is the
681 struct rw_semaphore group_rwsem;
684 oom_flags_t oom_flags;
685 short oom_score_adj; /* OOM kill score adjustment */
686 short oom_score_adj_min; /* OOM kill score adjustment min value.
687 * Only settable by CAP_SYS_RESOURCE. */
689 struct mutex cred_guard_mutex; /* guard against foreign influences on
690 * credential calculations
691 * (notably. ptrace) */
695 * Bits in flags field of signal_struct.
697 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
698 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
699 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
700 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
702 * Pending notifications to parent.
704 #define SIGNAL_CLD_STOPPED 0x00000010
705 #define SIGNAL_CLD_CONTINUED 0x00000020
706 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
708 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
710 /* If true, all threads except ->group_exit_task have pending SIGKILL */
711 static inline int signal_group_exit(const struct signal_struct *sig)
713 return (sig->flags & SIGNAL_GROUP_EXIT) ||
714 (sig->group_exit_task != NULL);
718 * Some day this will be a full-fledged user tracking system..
721 atomic_t __count; /* reference count */
722 atomic_t processes; /* How many processes does this user have? */
723 atomic_t files; /* How many open files does this user have? */
724 atomic_t sigpending; /* How many pending signals does this user have? */
725 #ifdef CONFIG_INOTIFY_USER
726 atomic_t inotify_watches; /* How many inotify watches does this user have? */
727 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
729 #ifdef CONFIG_FANOTIFY
730 atomic_t fanotify_listeners;
733 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
735 #ifdef CONFIG_POSIX_MQUEUE
736 /* protected by mq_lock */
737 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
739 unsigned long locked_shm; /* How many pages of mlocked shm ? */
742 struct key *uid_keyring; /* UID specific keyring */
743 struct key *session_keyring; /* UID's default session keyring */
746 /* Hash table maintenance information */
747 struct hlist_node uidhash_node;
750 #ifdef CONFIG_PERF_EVENTS
751 atomic_long_t locked_vm;
755 extern int uids_sysfs_init(void);
757 extern struct user_struct *find_user(kuid_t);
759 extern struct user_struct root_user;
760 #define INIT_USER (&root_user)
763 struct backing_dev_info;
764 struct reclaim_state;
766 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
768 /* cumulative counters */
769 unsigned long pcount; /* # of times run on this cpu */
770 unsigned long long run_delay; /* time spent waiting on a runqueue */
773 unsigned long long last_arrival,/* when we last ran on a cpu */
774 last_queued; /* when we were last queued to run */
776 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
778 #ifdef CONFIG_TASK_DELAY_ACCT
779 struct task_delay_info {
781 unsigned int flags; /* Private per-task flags */
783 /* For each stat XXX, add following, aligned appropriately
785 * struct timespec XXX_start, XXX_end;
789 * Atomicity of updates to XXX_delay, XXX_count protected by
790 * single lock above (split into XXX_lock if contention is an issue).
794 * XXX_count is incremented on every XXX operation, the delay
795 * associated with the operation is added to XXX_delay.
796 * XXX_delay contains the accumulated delay time in nanoseconds.
798 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
799 u64 blkio_delay; /* wait for sync block io completion */
800 u64 swapin_delay; /* wait for swapin block io completion */
801 u32 blkio_count; /* total count of the number of sync block */
802 /* io operations performed */
803 u32 swapin_count; /* total count of the number of swapin block */
804 /* io operations performed */
806 struct timespec freepages_start, freepages_end;
807 u64 freepages_delay; /* wait for memory reclaim */
808 u32 freepages_count; /* total count of memory reclaim */
810 #endif /* CONFIG_TASK_DELAY_ACCT */
812 static inline int sched_info_on(void)
814 #ifdef CONFIG_SCHEDSTATS
816 #elif defined(CONFIG_TASK_DELAY_ACCT)
817 extern int delayacct_on;
832 * Increase resolution of cpu_power calculations
834 #define SCHED_POWER_SHIFT 10
835 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
838 * sched-domains (multiprocessor balancing) declarations:
841 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
842 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
843 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
844 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
845 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
846 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
847 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
848 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
849 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
850 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
851 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
852 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
853 #define SD_NUMA 0x4000 /* cross-node balancing */
855 extern int __weak arch_sd_sibiling_asym_packing(void);
857 struct sched_domain_attr {
858 int relax_domain_level;
861 #define SD_ATTR_INIT (struct sched_domain_attr) { \
862 .relax_domain_level = -1, \
865 extern int sched_domain_level_max;
869 struct sched_domain {
870 /* These fields must be setup */
871 struct sched_domain *parent; /* top domain must be null terminated */
872 struct sched_domain *child; /* bottom domain must be null terminated */
873 struct sched_group *groups; /* the balancing groups of the domain */
874 unsigned long min_interval; /* Minimum balance interval ms */
875 unsigned long max_interval; /* Maximum balance interval ms */
876 unsigned int busy_factor; /* less balancing by factor if busy */
877 unsigned int imbalance_pct; /* No balance until over watermark */
878 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
879 unsigned int busy_idx;
880 unsigned int idle_idx;
881 unsigned int newidle_idx;
882 unsigned int wake_idx;
883 unsigned int forkexec_idx;
884 unsigned int smt_gain;
886 int nohz_idle; /* NOHZ IDLE status */
887 int flags; /* See SD_* */
890 /* Runtime fields. */
891 unsigned long last_balance; /* init to jiffies. units in jiffies */
892 unsigned int balance_interval; /* initialise to 1. units in ms. */
893 unsigned int nr_balance_failed; /* initialise to 0 */
895 /* idle_balance() stats */
896 u64 max_newidle_lb_cost;
897 unsigned long next_decay_max_lb_cost;
899 #ifdef CONFIG_SCHEDSTATS
900 /* load_balance() stats */
901 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
902 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
903 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
904 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
905 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
906 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
907 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
908 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
910 /* Active load balancing */
911 unsigned int alb_count;
912 unsigned int alb_failed;
913 unsigned int alb_pushed;
915 /* SD_BALANCE_EXEC stats */
916 unsigned int sbe_count;
917 unsigned int sbe_balanced;
918 unsigned int sbe_pushed;
920 /* SD_BALANCE_FORK stats */
921 unsigned int sbf_count;
922 unsigned int sbf_balanced;
923 unsigned int sbf_pushed;
925 /* try_to_wake_up() stats */
926 unsigned int ttwu_wake_remote;
927 unsigned int ttwu_move_affine;
928 unsigned int ttwu_move_balance;
930 #ifdef CONFIG_SCHED_DEBUG
934 void *private; /* used during construction */
935 struct rcu_head rcu; /* used during destruction */
938 unsigned int span_weight;
940 * Span of all CPUs in this domain.
942 * NOTE: this field is variable length. (Allocated dynamically
943 * by attaching extra space to the end of the structure,
944 * depending on how many CPUs the kernel has booted up with)
946 unsigned long span[0];
949 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
951 return to_cpumask(sd->span);
954 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
955 struct sched_domain_attr *dattr_new);
957 /* Allocate an array of sched domains, for partition_sched_domains(). */
958 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
959 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
961 bool cpus_share_cache(int this_cpu, int that_cpu);
963 #else /* CONFIG_SMP */
965 struct sched_domain_attr;
968 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
969 struct sched_domain_attr *dattr_new)
973 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
978 #endif /* !CONFIG_SMP */
981 struct io_context; /* See blkdev.h */
984 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
985 extern void prefetch_stack(struct task_struct *t);
987 static inline void prefetch_stack(struct task_struct *t) { }
990 struct audit_context; /* See audit.c */
992 struct pipe_inode_info;
993 struct uts_namespace;
996 unsigned long weight;
1002 * These sums represent an infinite geometric series and so are bound
1003 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1004 * choices of y < 1-2^(-32)*1024.
1006 u32 runnable_avg_sum, runnable_avg_period;
1007 u64 last_runnable_update;
1009 unsigned long load_avg_contrib;
1012 #ifdef CONFIG_SCHEDSTATS
1013 struct sched_statistics {
1023 s64 sum_sleep_runtime;
1030 u64 nr_migrations_cold;
1031 u64 nr_failed_migrations_affine;
1032 u64 nr_failed_migrations_running;
1033 u64 nr_failed_migrations_hot;
1034 u64 nr_forced_migrations;
1037 u64 nr_wakeups_sync;
1038 u64 nr_wakeups_migrate;
1039 u64 nr_wakeups_local;
1040 u64 nr_wakeups_remote;
1041 u64 nr_wakeups_affine;
1042 u64 nr_wakeups_affine_attempts;
1043 u64 nr_wakeups_passive;
1044 u64 nr_wakeups_idle;
1048 struct sched_entity {
1049 struct load_weight load; /* for load-balancing */
1050 struct rb_node run_node;
1051 struct list_head group_node;
1055 u64 sum_exec_runtime;
1057 u64 prev_sum_exec_runtime;
1061 #ifdef CONFIG_SCHEDSTATS
1062 struct sched_statistics statistics;
1065 #ifdef CONFIG_FAIR_GROUP_SCHED
1066 struct sched_entity *parent;
1067 /* rq on which this entity is (to be) queued: */
1068 struct cfs_rq *cfs_rq;
1069 /* rq "owned" by this entity/group: */
1070 struct cfs_rq *my_q;
1074 /* Per-entity load-tracking */
1075 struct sched_avg avg;
1079 struct sched_rt_entity {
1080 struct list_head run_list;
1081 unsigned long timeout;
1082 unsigned long watchdog_stamp;
1083 unsigned int time_slice;
1085 struct sched_rt_entity *back;
1086 #ifdef CONFIG_RT_GROUP_SCHED
1087 struct sched_rt_entity *parent;
1088 /* rq on which this entity is (to be) queued: */
1089 struct rt_rq *rt_rq;
1090 /* rq "owned" by this entity/group: */
1095 struct sched_dl_entity {
1096 struct rb_node rb_node;
1099 * Original scheduling parameters. Copied here from sched_attr
1100 * during sched_setscheduler2(), they will remain the same until
1101 * the next sched_setscheduler2().
1103 u64 dl_runtime; /* maximum runtime for each instance */
1104 u64 dl_deadline; /* relative deadline of each instance */
1105 u64 dl_period; /* separation of two instances (period) */
1108 * Actual scheduling parameters. Initialized with the values above,
1109 * they are continously updated during task execution. Note that
1110 * the remaining runtime could be < 0 in case we are in overrun.
1112 s64 runtime; /* remaining runtime for this instance */
1113 u64 deadline; /* absolute deadline for this instance */
1114 unsigned int flags; /* specifying the scheduler behaviour */
1119 * @dl_throttled tells if we exhausted the runtime. If so, the
1120 * task has to wait for a replenishment to be performed at the
1121 * next firing of dl_timer.
1123 * @dl_new tells if a new instance arrived. If so we must
1124 * start executing it with full runtime and reset its absolute
1127 int dl_throttled, dl_new;
1130 * Bandwidth enforcement timer. Each -deadline task has its
1131 * own bandwidth to be enforced, thus we need one timer per task.
1133 struct hrtimer dl_timer;
1138 enum perf_event_task_context {
1139 perf_invalid_context = -1,
1140 perf_hw_context = 0,
1142 perf_nr_task_contexts,
1145 struct task_struct {
1146 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1149 unsigned int flags; /* per process flags, defined below */
1150 unsigned int ptrace;
1153 struct llist_node wake_entry;
1155 struct task_struct *last_wakee;
1156 unsigned long wakee_flips;
1157 unsigned long wakee_flip_decay_ts;
1163 int prio, static_prio, normal_prio;
1164 unsigned int rt_priority;
1165 const struct sched_class *sched_class;
1166 struct sched_entity se;
1167 struct sched_rt_entity rt;
1168 #ifdef CONFIG_CGROUP_SCHED
1169 struct task_group *sched_task_group;
1171 struct sched_dl_entity dl;
1173 #ifdef CONFIG_PREEMPT_NOTIFIERS
1174 /* list of struct preempt_notifier: */
1175 struct hlist_head preempt_notifiers;
1178 #ifdef CONFIG_BLK_DEV_IO_TRACE
1179 unsigned int btrace_seq;
1182 unsigned int policy;
1183 int nr_cpus_allowed;
1184 cpumask_t cpus_allowed;
1186 #ifdef CONFIG_PREEMPT_RCU
1187 int rcu_read_lock_nesting;
1188 char rcu_read_unlock_special;
1189 struct list_head rcu_node_entry;
1190 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1191 #ifdef CONFIG_TREE_PREEMPT_RCU
1192 struct rcu_node *rcu_blocked_node;
1193 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1194 #ifdef CONFIG_RCU_BOOST
1195 struct rt_mutex *rcu_boost_mutex;
1196 #endif /* #ifdef CONFIG_RCU_BOOST */
1198 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1199 struct sched_info sched_info;
1202 struct list_head tasks;
1204 struct plist_node pushable_tasks;
1205 struct rb_node pushable_dl_tasks;
1208 struct mm_struct *mm, *active_mm;
1209 #ifdef CONFIG_COMPAT_BRK
1210 unsigned brk_randomized:1;
1212 #if defined(SPLIT_RSS_COUNTING)
1213 struct task_rss_stat rss_stat;
1217 int exit_code, exit_signal;
1218 int pdeath_signal; /* The signal sent when the parent dies */
1219 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1221 /* Used for emulating ABI behavior of previous Linux versions */
1222 unsigned int personality;
1224 unsigned did_exec:1;
1225 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1227 unsigned in_iowait:1;
1229 /* task may not gain privileges */
1230 unsigned no_new_privs:1;
1232 /* Revert to default priority/policy when forking */
1233 unsigned sched_reset_on_fork:1;
1234 unsigned sched_contributes_to_load:1;
1239 #ifdef CONFIG_CC_STACKPROTECTOR
1240 /* Canary value for the -fstack-protector gcc feature */
1241 unsigned long stack_canary;
1244 * pointers to (original) parent process, youngest child, younger sibling,
1245 * older sibling, respectively. (p->father can be replaced with
1246 * p->real_parent->pid)
1248 struct task_struct __rcu *real_parent; /* real parent process */
1249 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1251 * children/sibling forms the list of my natural children
1253 struct list_head children; /* list of my children */
1254 struct list_head sibling; /* linkage in my parent's children list */
1255 struct task_struct *group_leader; /* threadgroup leader */
1258 * ptraced is the list of tasks this task is using ptrace on.
1259 * This includes both natural children and PTRACE_ATTACH targets.
1260 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1262 struct list_head ptraced;
1263 struct list_head ptrace_entry;
1265 /* PID/PID hash table linkage. */
1266 struct pid_link pids[PIDTYPE_MAX];
1267 struct list_head thread_group;
1269 struct completion *vfork_done; /* for vfork() */
1270 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1271 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1273 cputime_t utime, stime, utimescaled, stimescaled;
1275 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1276 struct cputime prev_cputime;
1278 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1279 seqlock_t vtime_seqlock;
1280 unsigned long long vtime_snap;
1285 } vtime_snap_whence;
1287 unsigned long nvcsw, nivcsw; /* context switch counts */
1288 struct timespec start_time; /* monotonic time */
1289 struct timespec real_start_time; /* boot based time */
1290 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1291 unsigned long min_flt, maj_flt;
1293 struct task_cputime cputime_expires;
1294 struct list_head cpu_timers[3];
1296 /* process credentials */
1297 const struct cred __rcu *real_cred; /* objective and real subjective task
1298 * credentials (COW) */
1299 const struct cred __rcu *cred; /* effective (overridable) subjective task
1300 * credentials (COW) */
1301 char comm[TASK_COMM_LEN]; /* executable name excluding path
1302 - access with [gs]et_task_comm (which lock
1303 it with task_lock())
1304 - initialized normally by setup_new_exec */
1305 /* file system info */
1306 int link_count, total_link_count;
1307 #ifdef CONFIG_SYSVIPC
1309 struct sysv_sem sysvsem;
1311 #ifdef CONFIG_DETECT_HUNG_TASK
1312 /* hung task detection */
1313 unsigned long last_switch_count;
1315 /* CPU-specific state of this task */
1316 struct thread_struct thread;
1317 /* filesystem information */
1318 struct fs_struct *fs;
1319 /* open file information */
1320 struct files_struct *files;
1322 struct nsproxy *nsproxy;
1323 /* signal handlers */
1324 struct signal_struct *signal;
1325 struct sighand_struct *sighand;
1327 sigset_t blocked, real_blocked;
1328 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1329 struct sigpending pending;
1331 unsigned long sas_ss_sp;
1333 int (*notifier)(void *priv);
1334 void *notifier_data;
1335 sigset_t *notifier_mask;
1336 struct callback_head *task_works;
1338 struct audit_context *audit_context;
1339 #ifdef CONFIG_AUDITSYSCALL
1341 unsigned int sessionid;
1343 struct seccomp seccomp;
1345 /* Thread group tracking */
1348 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1350 spinlock_t alloc_lock;
1352 /* Protection of the PI data structures: */
1353 raw_spinlock_t pi_lock;
1355 #ifdef CONFIG_RT_MUTEXES
1356 /* PI waiters blocked on a rt_mutex held by this task */
1357 struct plist_head pi_waiters;
1358 /* Deadlock detection and priority inheritance handling */
1359 struct rt_mutex_waiter *pi_blocked_on;
1362 #ifdef CONFIG_DEBUG_MUTEXES
1363 /* mutex deadlock detection */
1364 struct mutex_waiter *blocked_on;
1366 #ifdef CONFIG_TRACE_IRQFLAGS
1367 unsigned int irq_events;
1368 unsigned long hardirq_enable_ip;
1369 unsigned long hardirq_disable_ip;
1370 unsigned int hardirq_enable_event;
1371 unsigned int hardirq_disable_event;
1372 int hardirqs_enabled;
1373 int hardirq_context;
1374 unsigned long softirq_disable_ip;
1375 unsigned long softirq_enable_ip;
1376 unsigned int softirq_disable_event;
1377 unsigned int softirq_enable_event;
1378 int softirqs_enabled;
1379 int softirq_context;
1381 #ifdef CONFIG_LOCKDEP
1382 # define MAX_LOCK_DEPTH 48UL
1385 unsigned int lockdep_recursion;
1386 struct held_lock held_locks[MAX_LOCK_DEPTH];
1387 gfp_t lockdep_reclaim_gfp;
1390 /* journalling filesystem info */
1393 /* stacked block device info */
1394 struct bio_list *bio_list;
1397 /* stack plugging */
1398 struct blk_plug *plug;
1402 struct reclaim_state *reclaim_state;
1404 struct backing_dev_info *backing_dev_info;
1406 struct io_context *io_context;
1408 unsigned long ptrace_message;
1409 siginfo_t *last_siginfo; /* For ptrace use. */
1410 struct task_io_accounting ioac;
1411 #if defined(CONFIG_TASK_XACCT)
1412 u64 acct_rss_mem1; /* accumulated rss usage */
1413 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1414 cputime_t acct_timexpd; /* stime + utime since last update */
1416 #ifdef CONFIG_CPUSETS
1417 nodemask_t mems_allowed; /* Protected by alloc_lock */
1418 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1419 int cpuset_mem_spread_rotor;
1420 int cpuset_slab_spread_rotor;
1422 #ifdef CONFIG_CGROUPS
1423 /* Control Group info protected by css_set_lock */
1424 struct css_set __rcu *cgroups;
1425 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1426 struct list_head cg_list;
1429 struct robust_list_head __user *robust_list;
1430 #ifdef CONFIG_COMPAT
1431 struct compat_robust_list_head __user *compat_robust_list;
1433 struct list_head pi_state_list;
1434 struct futex_pi_state *pi_state_cache;
1436 #ifdef CONFIG_PERF_EVENTS
1437 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1438 struct mutex perf_event_mutex;
1439 struct list_head perf_event_list;
1442 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1444 short pref_node_fork;
1446 #ifdef CONFIG_NUMA_BALANCING
1448 unsigned int numa_scan_period;
1449 unsigned int numa_scan_period_max;
1450 int numa_preferred_nid;
1451 int numa_migrate_deferred;
1452 unsigned long numa_migrate_retry;
1453 u64 node_stamp; /* migration stamp */
1454 struct callback_head numa_work;
1456 struct list_head numa_entry;
1457 struct numa_group *numa_group;
1460 * Exponential decaying average of faults on a per-node basis.
1461 * Scheduling placement decisions are made based on the these counts.
1462 * The values remain static for the duration of a PTE scan
1464 unsigned long *numa_faults;
1465 unsigned long total_numa_faults;
1468 * numa_faults_buffer records faults per node during the current
1469 * scan window. When the scan completes, the counts in numa_faults
1470 * decay and these values are copied.
1472 unsigned long *numa_faults_buffer;
1475 * numa_faults_locality tracks if faults recorded during the last
1476 * scan window were remote/local. The task scan period is adapted
1477 * based on the locality of the faults with different weights
1478 * depending on whether they were shared or private faults
1480 unsigned long numa_faults_locality[2];
1482 unsigned long numa_pages_migrated;
1483 #endif /* CONFIG_NUMA_BALANCING */
1485 struct rcu_head rcu;
1488 * cache last used pipe for splice
1490 struct pipe_inode_info *splice_pipe;
1492 struct page_frag task_frag;
1494 #ifdef CONFIG_TASK_DELAY_ACCT
1495 struct task_delay_info *delays;
1497 #ifdef CONFIG_FAULT_INJECTION
1501 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1502 * balance_dirty_pages() for some dirty throttling pause
1505 int nr_dirtied_pause;
1506 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1508 #ifdef CONFIG_LATENCYTOP
1509 int latency_record_count;
1510 struct latency_record latency_record[LT_SAVECOUNT];
1513 * time slack values; these are used to round up poll() and
1514 * select() etc timeout values. These are in nanoseconds.
1516 unsigned long timer_slack_ns;
1517 unsigned long default_timer_slack_ns;
1519 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1520 /* Index of current stored address in ret_stack */
1522 /* Stack of return addresses for return function tracing */
1523 struct ftrace_ret_stack *ret_stack;
1524 /* time stamp for last schedule */
1525 unsigned long long ftrace_timestamp;
1527 * Number of functions that haven't been traced
1528 * because of depth overrun.
1530 atomic_t trace_overrun;
1531 /* Pause for the tracing */
1532 atomic_t tracing_graph_pause;
1534 #ifdef CONFIG_TRACING
1535 /* state flags for use by tracers */
1536 unsigned long trace;
1537 /* bitmask and counter of trace recursion */
1538 unsigned long trace_recursion;
1539 #endif /* CONFIG_TRACING */
1540 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1541 struct memcg_batch_info {
1542 int do_batch; /* incremented when batch uncharge started */
1543 struct mem_cgroup *memcg; /* target memcg of uncharge */
1544 unsigned long nr_pages; /* uncharged usage */
1545 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1547 unsigned int memcg_kmem_skip_account;
1548 struct memcg_oom_info {
1549 struct mem_cgroup *memcg;
1552 unsigned int may_oom:1;
1555 #ifdef CONFIG_UPROBES
1556 struct uprobe_task *utask;
1558 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1559 unsigned int sequential_io;
1560 unsigned int sequential_io_avg;
1564 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1565 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1567 #define TNF_MIGRATED 0x01
1568 #define TNF_NO_GROUP 0x02
1569 #define TNF_SHARED 0x04
1570 #define TNF_FAULT_LOCAL 0x08
1572 #ifdef CONFIG_NUMA_BALANCING
1573 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1574 extern pid_t task_numa_group_id(struct task_struct *p);
1575 extern void set_numabalancing_state(bool enabled);
1576 extern void task_numa_free(struct task_struct *p);
1578 extern unsigned int sysctl_numa_balancing_migrate_deferred;
1580 static inline void task_numa_fault(int last_node, int node, int pages,
1584 static inline pid_t task_numa_group_id(struct task_struct *p)
1588 static inline void set_numabalancing_state(bool enabled)
1591 static inline void task_numa_free(struct task_struct *p)
1596 static inline struct pid *task_pid(struct task_struct *task)
1598 return task->pids[PIDTYPE_PID].pid;
1601 static inline struct pid *task_tgid(struct task_struct *task)
1603 return task->group_leader->pids[PIDTYPE_PID].pid;
1607 * Without tasklist or rcu lock it is not safe to dereference
1608 * the result of task_pgrp/task_session even if task == current,
1609 * we can race with another thread doing sys_setsid/sys_setpgid.
1611 static inline struct pid *task_pgrp(struct task_struct *task)
1613 return task->group_leader->pids[PIDTYPE_PGID].pid;
1616 static inline struct pid *task_session(struct task_struct *task)
1618 return task->group_leader->pids[PIDTYPE_SID].pid;
1621 struct pid_namespace;
1624 * the helpers to get the task's different pids as they are seen
1625 * from various namespaces
1627 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1628 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1630 * task_xid_nr_ns() : id seen from the ns specified;
1632 * set_task_vxid() : assigns a virtual id to a task;
1634 * see also pid_nr() etc in include/linux/pid.h
1636 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1637 struct pid_namespace *ns);
1639 static inline pid_t task_pid_nr(struct task_struct *tsk)
1644 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1645 struct pid_namespace *ns)
1647 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1650 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1652 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1656 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1661 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1663 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1665 return pid_vnr(task_tgid(tsk));
1669 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1670 struct pid_namespace *ns)
1672 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1675 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1677 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1681 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1682 struct pid_namespace *ns)
1684 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1687 static inline pid_t task_session_vnr(struct task_struct *tsk)
1689 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1692 /* obsolete, do not use */
1693 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1695 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1699 * pid_alive - check that a task structure is not stale
1700 * @p: Task structure to be checked.
1702 * Test if a process is not yet dead (at most zombie state)
1703 * If pid_alive fails, then pointers within the task structure
1704 * can be stale and must not be dereferenced.
1706 * Return: 1 if the process is alive. 0 otherwise.
1708 static inline int pid_alive(struct task_struct *p)
1710 return p->pids[PIDTYPE_PID].pid != NULL;
1714 * is_global_init - check if a task structure is init
1715 * @tsk: Task structure to be checked.
1717 * Check if a task structure is the first user space task the kernel created.
1719 * Return: 1 if the task structure is init. 0 otherwise.
1721 static inline int is_global_init(struct task_struct *tsk)
1723 return tsk->pid == 1;
1726 extern struct pid *cad_pid;
1728 extern void free_task(struct task_struct *tsk);
1729 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1731 extern void __put_task_struct(struct task_struct *t);
1733 static inline void put_task_struct(struct task_struct *t)
1735 if (atomic_dec_and_test(&t->usage))
1736 __put_task_struct(t);
1739 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1740 extern void task_cputime(struct task_struct *t,
1741 cputime_t *utime, cputime_t *stime);
1742 extern void task_cputime_scaled(struct task_struct *t,
1743 cputime_t *utimescaled, cputime_t *stimescaled);
1744 extern cputime_t task_gtime(struct task_struct *t);
1746 static inline void task_cputime(struct task_struct *t,
1747 cputime_t *utime, cputime_t *stime)
1755 static inline void task_cputime_scaled(struct task_struct *t,
1756 cputime_t *utimescaled,
1757 cputime_t *stimescaled)
1760 *utimescaled = t->utimescaled;
1762 *stimescaled = t->stimescaled;
1765 static inline cputime_t task_gtime(struct task_struct *t)
1770 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1771 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1776 #define PF_EXITING 0x00000004 /* getting shut down */
1777 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1778 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1779 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1780 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1781 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1782 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1783 #define PF_DUMPCORE 0x00000200 /* dumped core */
1784 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1785 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1786 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1787 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1788 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1789 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1790 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1791 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1792 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1793 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1794 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1795 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1796 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1797 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1798 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1799 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1800 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1801 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1802 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1803 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1804 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1805 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1808 * Only the _current_ task can read/write to tsk->flags, but other
1809 * tasks can access tsk->flags in readonly mode for example
1810 * with tsk_used_math (like during threaded core dumping).
1811 * There is however an exception to this rule during ptrace
1812 * or during fork: the ptracer task is allowed to write to the
1813 * child->flags of its traced child (same goes for fork, the parent
1814 * can write to the child->flags), because we're guaranteed the
1815 * child is not running and in turn not changing child->flags
1816 * at the same time the parent does it.
1818 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1819 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1820 #define clear_used_math() clear_stopped_child_used_math(current)
1821 #define set_used_math() set_stopped_child_used_math(current)
1822 #define conditional_stopped_child_used_math(condition, child) \
1823 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1824 #define conditional_used_math(condition) \
1825 conditional_stopped_child_used_math(condition, current)
1826 #define copy_to_stopped_child_used_math(child) \
1827 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1828 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1829 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1830 #define used_math() tsk_used_math(current)
1832 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1833 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1835 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1840 static inline unsigned int memalloc_noio_save(void)
1842 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1843 current->flags |= PF_MEMALLOC_NOIO;
1847 static inline void memalloc_noio_restore(unsigned int flags)
1849 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1853 * task->jobctl flags
1855 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1857 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1858 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1859 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1860 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1861 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1862 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1863 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1865 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1866 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1867 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1868 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1869 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1870 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1871 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1873 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1874 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1876 extern bool task_set_jobctl_pending(struct task_struct *task,
1878 extern void task_clear_jobctl_trapping(struct task_struct *task);
1879 extern void task_clear_jobctl_pending(struct task_struct *task,
1882 #ifdef CONFIG_PREEMPT_RCU
1884 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1885 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1887 static inline void rcu_copy_process(struct task_struct *p)
1889 p->rcu_read_lock_nesting = 0;
1890 p->rcu_read_unlock_special = 0;
1891 #ifdef CONFIG_TREE_PREEMPT_RCU
1892 p->rcu_blocked_node = NULL;
1893 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1894 #ifdef CONFIG_RCU_BOOST
1895 p->rcu_boost_mutex = NULL;
1896 #endif /* #ifdef CONFIG_RCU_BOOST */
1897 INIT_LIST_HEAD(&p->rcu_node_entry);
1902 static inline void rcu_copy_process(struct task_struct *p)
1908 static inline void tsk_restore_flags(struct task_struct *task,
1909 unsigned long orig_flags, unsigned long flags)
1911 task->flags &= ~flags;
1912 task->flags |= orig_flags & flags;
1916 extern void do_set_cpus_allowed(struct task_struct *p,
1917 const struct cpumask *new_mask);
1919 extern int set_cpus_allowed_ptr(struct task_struct *p,
1920 const struct cpumask *new_mask);
1922 static inline void do_set_cpus_allowed(struct task_struct *p,
1923 const struct cpumask *new_mask)
1926 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1927 const struct cpumask *new_mask)
1929 if (!cpumask_test_cpu(0, new_mask))
1935 #ifdef CONFIG_NO_HZ_COMMON
1936 void calc_load_enter_idle(void);
1937 void calc_load_exit_idle(void);
1939 static inline void calc_load_enter_idle(void) { }
1940 static inline void calc_load_exit_idle(void) { }
1941 #endif /* CONFIG_NO_HZ_COMMON */
1943 #ifndef CONFIG_CPUMASK_OFFSTACK
1944 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1946 return set_cpus_allowed_ptr(p, &new_mask);
1951 * Do not use outside of architecture code which knows its limitations.
1953 * sched_clock() has no promise of monotonicity or bounded drift between
1954 * CPUs, use (which you should not) requires disabling IRQs.
1956 * Please use one of the three interfaces below.
1958 extern unsigned long long notrace sched_clock(void);
1960 * See the comment in kernel/sched/clock.c
1962 extern u64 cpu_clock(int cpu);
1963 extern u64 local_clock(void);
1964 extern u64 sched_clock_cpu(int cpu);
1967 extern void sched_clock_init(void);
1969 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1970 static inline void sched_clock_tick(void)
1974 static inline void sched_clock_idle_sleep_event(void)
1978 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1983 * Architectures can set this to 1 if they have specified
1984 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1985 * but then during bootup it turns out that sched_clock()
1986 * is reliable after all:
1988 extern int sched_clock_stable;
1990 extern void sched_clock_tick(void);
1991 extern void sched_clock_idle_sleep_event(void);
1992 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1995 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1997 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1998 * The reason for this explicit opt-in is not to have perf penalty with
1999 * slow sched_clocks.
2001 extern void enable_sched_clock_irqtime(void);
2002 extern void disable_sched_clock_irqtime(void);
2004 static inline void enable_sched_clock_irqtime(void) {}
2005 static inline void disable_sched_clock_irqtime(void) {}
2008 extern unsigned long long
2009 task_sched_runtime(struct task_struct *task);
2011 /* sched_exec is called by processes performing an exec */
2013 extern void sched_exec(void);
2015 #define sched_exec() {}
2018 extern void sched_clock_idle_sleep_event(void);
2019 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2021 #ifdef CONFIG_HOTPLUG_CPU
2022 extern void idle_task_exit(void);
2024 static inline void idle_task_exit(void) {}
2027 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2028 extern void wake_up_nohz_cpu(int cpu);
2030 static inline void wake_up_nohz_cpu(int cpu) { }
2033 #ifdef CONFIG_NO_HZ_FULL
2034 extern bool sched_can_stop_tick(void);
2035 extern u64 scheduler_tick_max_deferment(void);
2037 static inline bool sched_can_stop_tick(void) { return false; }
2040 #ifdef CONFIG_SCHED_AUTOGROUP
2041 extern void sched_autogroup_create_attach(struct task_struct *p);
2042 extern void sched_autogroup_detach(struct task_struct *p);
2043 extern void sched_autogroup_fork(struct signal_struct *sig);
2044 extern void sched_autogroup_exit(struct signal_struct *sig);
2045 #ifdef CONFIG_PROC_FS
2046 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2047 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2050 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2051 static inline void sched_autogroup_detach(struct task_struct *p) { }
2052 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2053 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2056 extern bool yield_to(struct task_struct *p, bool preempt);
2057 extern void set_user_nice(struct task_struct *p, long nice);
2058 extern int task_prio(const struct task_struct *p);
2059 extern int task_nice(const struct task_struct *p);
2060 extern int can_nice(const struct task_struct *p, const int nice);
2061 extern int task_curr(const struct task_struct *p);
2062 extern int idle_cpu(int cpu);
2063 extern int sched_setscheduler(struct task_struct *, int,
2064 const struct sched_param *);
2065 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2066 const struct sched_param *);
2067 extern int sched_setattr(struct task_struct *,
2068 const struct sched_attr *);
2069 extern struct task_struct *idle_task(int cpu);
2071 * is_idle_task - is the specified task an idle task?
2072 * @p: the task in question.
2074 * Return: 1 if @p is an idle task. 0 otherwise.
2076 static inline bool is_idle_task(const struct task_struct *p)
2080 extern struct task_struct *curr_task(int cpu);
2081 extern void set_curr_task(int cpu, struct task_struct *p);
2086 * The default (Linux) execution domain.
2088 extern struct exec_domain default_exec_domain;
2090 union thread_union {
2091 struct thread_info thread_info;
2092 unsigned long stack[THREAD_SIZE/sizeof(long)];
2095 #ifndef __HAVE_ARCH_KSTACK_END
2096 static inline int kstack_end(void *addr)
2098 /* Reliable end of stack detection:
2099 * Some APM bios versions misalign the stack
2101 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2105 extern union thread_union init_thread_union;
2106 extern struct task_struct init_task;
2108 extern struct mm_struct init_mm;
2110 extern struct pid_namespace init_pid_ns;
2113 * find a task by one of its numerical ids
2115 * find_task_by_pid_ns():
2116 * finds a task by its pid in the specified namespace
2117 * find_task_by_vpid():
2118 * finds a task by its virtual pid
2120 * see also find_vpid() etc in include/linux/pid.h
2123 extern struct task_struct *find_task_by_vpid(pid_t nr);
2124 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2125 struct pid_namespace *ns);
2127 /* per-UID process charging. */
2128 extern struct user_struct * alloc_uid(kuid_t);
2129 static inline struct user_struct *get_uid(struct user_struct *u)
2131 atomic_inc(&u->__count);
2134 extern void free_uid(struct user_struct *);
2136 #include <asm/current.h>
2138 extern void xtime_update(unsigned long ticks);
2140 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2141 extern int wake_up_process(struct task_struct *tsk);
2142 extern void wake_up_new_task(struct task_struct *tsk);
2144 extern void kick_process(struct task_struct *tsk);
2146 static inline void kick_process(struct task_struct *tsk) { }
2148 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2149 extern void sched_dead(struct task_struct *p);
2151 extern void proc_caches_init(void);
2152 extern void flush_signals(struct task_struct *);
2153 extern void __flush_signals(struct task_struct *);
2154 extern void ignore_signals(struct task_struct *);
2155 extern void flush_signal_handlers(struct task_struct *, int force_default);
2156 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2158 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2160 unsigned long flags;
2163 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2164 ret = dequeue_signal(tsk, mask, info);
2165 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2170 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2172 extern void unblock_all_signals(void);
2173 extern void release_task(struct task_struct * p);
2174 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2175 extern int force_sigsegv(int, struct task_struct *);
2176 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2177 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2178 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2179 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2180 const struct cred *, u32);
2181 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2182 extern int kill_pid(struct pid *pid, int sig, int priv);
2183 extern int kill_proc_info(int, struct siginfo *, pid_t);
2184 extern __must_check bool do_notify_parent(struct task_struct *, int);
2185 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2186 extern void force_sig(int, struct task_struct *);
2187 extern int send_sig(int, struct task_struct *, int);
2188 extern int zap_other_threads(struct task_struct *p);
2189 extern struct sigqueue *sigqueue_alloc(void);
2190 extern void sigqueue_free(struct sigqueue *);
2191 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2192 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2194 static inline void restore_saved_sigmask(void)
2196 if (test_and_clear_restore_sigmask())
2197 __set_current_blocked(¤t->saved_sigmask);
2200 static inline sigset_t *sigmask_to_save(void)
2202 sigset_t *res = ¤t->blocked;
2203 if (unlikely(test_restore_sigmask()))
2204 res = ¤t->saved_sigmask;
2208 static inline int kill_cad_pid(int sig, int priv)
2210 return kill_pid(cad_pid, sig, priv);
2213 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2214 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2215 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2216 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2219 * True if we are on the alternate signal stack.
2221 static inline int on_sig_stack(unsigned long sp)
2223 #ifdef CONFIG_STACK_GROWSUP
2224 return sp >= current->sas_ss_sp &&
2225 sp - current->sas_ss_sp < current->sas_ss_size;
2227 return sp > current->sas_ss_sp &&
2228 sp - current->sas_ss_sp <= current->sas_ss_size;
2232 static inline int sas_ss_flags(unsigned long sp)
2234 return (current->sas_ss_size == 0 ? SS_DISABLE
2235 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2238 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2240 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2241 #ifdef CONFIG_STACK_GROWSUP
2242 return current->sas_ss_sp;
2244 return current->sas_ss_sp + current->sas_ss_size;
2250 * Routines for handling mm_structs
2252 extern struct mm_struct * mm_alloc(void);
2254 /* mmdrop drops the mm and the page tables */
2255 extern void __mmdrop(struct mm_struct *);
2256 static inline void mmdrop(struct mm_struct * mm)
2258 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2262 /* mmput gets rid of the mappings and all user-space */
2263 extern void mmput(struct mm_struct *);
2264 /* Grab a reference to a task's mm, if it is not already going away */
2265 extern struct mm_struct *get_task_mm(struct task_struct *task);
2267 * Grab a reference to a task's mm, if it is not already going away
2268 * and ptrace_may_access with the mode parameter passed to it
2271 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2272 /* Remove the current tasks stale references to the old mm_struct */
2273 extern void mm_release(struct task_struct *, struct mm_struct *);
2274 /* Allocate a new mm structure and copy contents from tsk->mm */
2275 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2277 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2278 struct task_struct *);
2279 extern void flush_thread(void);
2280 extern void exit_thread(void);
2282 extern void exit_files(struct task_struct *);
2283 extern void __cleanup_sighand(struct sighand_struct *);
2285 extern void exit_itimers(struct signal_struct *);
2286 extern void flush_itimer_signals(void);
2288 extern void do_group_exit(int);
2290 extern int allow_signal(int);
2291 extern int disallow_signal(int);
2293 extern int do_execve(const char *,
2294 const char __user * const __user *,
2295 const char __user * const __user *);
2296 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2297 struct task_struct *fork_idle(int);
2298 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2300 extern void set_task_comm(struct task_struct *tsk, char *from);
2301 extern char *get_task_comm(char *to, struct task_struct *tsk);
2304 void scheduler_ipi(void);
2305 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2307 static inline void scheduler_ipi(void) { }
2308 static inline unsigned long wait_task_inactive(struct task_struct *p,
2315 #define next_task(p) \
2316 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2318 #define for_each_process(p) \
2319 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2321 extern bool current_is_single_threaded(void);
2324 * Careful: do_each_thread/while_each_thread is a double loop so
2325 * 'break' will not work as expected - use goto instead.
2327 #define do_each_thread(g, t) \
2328 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2330 #define while_each_thread(g, t) \
2331 while ((t = next_thread(t)) != g)
2333 static inline int get_nr_threads(struct task_struct *tsk)
2335 return tsk->signal->nr_threads;
2338 static inline bool thread_group_leader(struct task_struct *p)
2340 return p->exit_signal >= 0;
2343 /* Do to the insanities of de_thread it is possible for a process
2344 * to have the pid of the thread group leader without actually being
2345 * the thread group leader. For iteration through the pids in proc
2346 * all we care about is that we have a task with the appropriate
2347 * pid, we don't actually care if we have the right task.
2349 static inline bool has_group_leader_pid(struct task_struct *p)
2351 return task_pid(p) == p->signal->leader_pid;
2355 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2357 return p1->signal == p2->signal;
2360 static inline struct task_struct *next_thread(const struct task_struct *p)
2362 return list_entry_rcu(p->thread_group.next,
2363 struct task_struct, thread_group);
2366 static inline int thread_group_empty(struct task_struct *p)
2368 return list_empty(&p->thread_group);
2371 #define delay_group_leader(p) \
2372 (thread_group_leader(p) && !thread_group_empty(p))
2375 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2376 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2377 * pins the final release of task.io_context. Also protects ->cpuset and
2378 * ->cgroup.subsys[]. And ->vfork_done.
2380 * Nests both inside and outside of read_lock(&tasklist_lock).
2381 * It must not be nested with write_lock_irq(&tasklist_lock),
2382 * neither inside nor outside.
2384 static inline void task_lock(struct task_struct *p)
2386 spin_lock(&p->alloc_lock);
2389 static inline void task_unlock(struct task_struct *p)
2391 spin_unlock(&p->alloc_lock);
2394 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2395 unsigned long *flags);
2397 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2398 unsigned long *flags)
2400 struct sighand_struct *ret;
2402 ret = __lock_task_sighand(tsk, flags);
2403 (void)__cond_lock(&tsk->sighand->siglock, ret);
2407 static inline void unlock_task_sighand(struct task_struct *tsk,
2408 unsigned long *flags)
2410 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2413 #ifdef CONFIG_CGROUPS
2414 static inline void threadgroup_change_begin(struct task_struct *tsk)
2416 down_read(&tsk->signal->group_rwsem);
2418 static inline void threadgroup_change_end(struct task_struct *tsk)
2420 up_read(&tsk->signal->group_rwsem);
2424 * threadgroup_lock - lock threadgroup
2425 * @tsk: member task of the threadgroup to lock
2427 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2428 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2429 * change ->group_leader/pid. This is useful for cases where the threadgroup
2430 * needs to stay stable across blockable operations.
2432 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2433 * synchronization. While held, no new task will be added to threadgroup
2434 * and no existing live task will have its PF_EXITING set.
2436 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2437 * sub-thread becomes a new leader.
2439 static inline void threadgroup_lock(struct task_struct *tsk)
2441 down_write(&tsk->signal->group_rwsem);
2445 * threadgroup_unlock - unlock threadgroup
2446 * @tsk: member task of the threadgroup to unlock
2448 * Reverse threadgroup_lock().
2450 static inline void threadgroup_unlock(struct task_struct *tsk)
2452 up_write(&tsk->signal->group_rwsem);
2455 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2456 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2457 static inline void threadgroup_lock(struct task_struct *tsk) {}
2458 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2461 #ifndef __HAVE_THREAD_FUNCTIONS
2463 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2464 #define task_stack_page(task) ((task)->stack)
2466 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2468 *task_thread_info(p) = *task_thread_info(org);
2469 task_thread_info(p)->task = p;
2472 static inline unsigned long *end_of_stack(struct task_struct *p)
2474 return (unsigned long *)(task_thread_info(p) + 1);
2479 static inline int object_is_on_stack(void *obj)
2481 void *stack = task_stack_page(current);
2483 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2486 extern void thread_info_cache_init(void);
2488 #ifdef CONFIG_DEBUG_STACK_USAGE
2489 static inline unsigned long stack_not_used(struct task_struct *p)
2491 unsigned long *n = end_of_stack(p);
2493 do { /* Skip over canary */
2497 return (unsigned long)n - (unsigned long)end_of_stack(p);
2501 /* set thread flags in other task's structures
2502 * - see asm/thread_info.h for TIF_xxxx flags available
2504 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2506 set_ti_thread_flag(task_thread_info(tsk), flag);
2509 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2511 clear_ti_thread_flag(task_thread_info(tsk), flag);
2514 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2516 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2519 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2521 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2524 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2526 return test_ti_thread_flag(task_thread_info(tsk), flag);
2529 static inline void set_tsk_need_resched(struct task_struct *tsk)
2531 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2534 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2536 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2539 static inline int test_tsk_need_resched(struct task_struct *tsk)
2541 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2544 static inline int restart_syscall(void)
2546 set_tsk_thread_flag(current, TIF_SIGPENDING);
2547 return -ERESTARTNOINTR;
2550 static inline int signal_pending(struct task_struct *p)
2552 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2555 static inline int __fatal_signal_pending(struct task_struct *p)
2557 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2560 static inline int fatal_signal_pending(struct task_struct *p)
2562 return signal_pending(p) && __fatal_signal_pending(p);
2565 static inline int signal_pending_state(long state, struct task_struct *p)
2567 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2569 if (!signal_pending(p))
2572 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2576 * cond_resched() and cond_resched_lock(): latency reduction via
2577 * explicit rescheduling in places that are safe. The return
2578 * value indicates whether a reschedule was done in fact.
2579 * cond_resched_lock() will drop the spinlock before scheduling,
2580 * cond_resched_softirq() will enable bhs before scheduling.
2582 extern int _cond_resched(void);
2584 #define cond_resched() ({ \
2585 __might_sleep(__FILE__, __LINE__, 0); \
2589 extern int __cond_resched_lock(spinlock_t *lock);
2591 #ifdef CONFIG_PREEMPT_COUNT
2592 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2594 #define PREEMPT_LOCK_OFFSET 0
2597 #define cond_resched_lock(lock) ({ \
2598 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2599 __cond_resched_lock(lock); \
2602 extern int __cond_resched_softirq(void);
2604 #define cond_resched_softirq() ({ \
2605 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2606 __cond_resched_softirq(); \
2609 static inline void cond_resched_rcu(void)
2611 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2619 * Does a critical section need to be broken due to another
2620 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2621 * but a general need for low latency)
2623 static inline int spin_needbreak(spinlock_t *lock)
2625 #ifdef CONFIG_PREEMPT
2626 return spin_is_contended(lock);
2633 * Idle thread specific functions to determine the need_resched
2634 * polling state. We have two versions, one based on TS_POLLING in
2635 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2639 static inline int tsk_is_polling(struct task_struct *p)
2641 return task_thread_info(p)->status & TS_POLLING;
2643 static inline void __current_set_polling(void)
2645 current_thread_info()->status |= TS_POLLING;
2648 static inline bool __must_check current_set_polling_and_test(void)
2650 __current_set_polling();
2653 * Polling state must be visible before we test NEED_RESCHED,
2654 * paired by resched_task()
2658 return unlikely(tif_need_resched());
2661 static inline void __current_clr_polling(void)
2663 current_thread_info()->status &= ~TS_POLLING;
2666 static inline bool __must_check current_clr_polling_and_test(void)
2668 __current_clr_polling();
2671 * Polling state must be visible before we test NEED_RESCHED,
2672 * paired by resched_task()
2676 return unlikely(tif_need_resched());
2678 #elif defined(TIF_POLLING_NRFLAG)
2679 static inline int tsk_is_polling(struct task_struct *p)
2681 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2684 static inline void __current_set_polling(void)
2686 set_thread_flag(TIF_POLLING_NRFLAG);
2689 static inline bool __must_check current_set_polling_and_test(void)
2691 __current_set_polling();
2694 * Polling state must be visible before we test NEED_RESCHED,
2695 * paired by resched_task()
2697 * XXX: assumes set/clear bit are identical barrier wise.
2699 smp_mb__after_clear_bit();
2701 return unlikely(tif_need_resched());
2704 static inline void __current_clr_polling(void)
2706 clear_thread_flag(TIF_POLLING_NRFLAG);
2709 static inline bool __must_check current_clr_polling_and_test(void)
2711 __current_clr_polling();
2714 * Polling state must be visible before we test NEED_RESCHED,
2715 * paired by resched_task()
2717 smp_mb__after_clear_bit();
2719 return unlikely(tif_need_resched());
2723 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2724 static inline void __current_set_polling(void) { }
2725 static inline void __current_clr_polling(void) { }
2727 static inline bool __must_check current_set_polling_and_test(void)
2729 return unlikely(tif_need_resched());
2731 static inline bool __must_check current_clr_polling_and_test(void)
2733 return unlikely(tif_need_resched());
2737 static __always_inline bool need_resched(void)
2739 return unlikely(tif_need_resched());
2743 * Thread group CPU time accounting.
2745 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2746 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2748 static inline void thread_group_cputime_init(struct signal_struct *sig)
2750 raw_spin_lock_init(&sig->cputimer.lock);
2754 * Reevaluate whether the task has signals pending delivery.
2755 * Wake the task if so.
2756 * This is required every time the blocked sigset_t changes.
2757 * callers must hold sighand->siglock.
2759 extern void recalc_sigpending_and_wake(struct task_struct *t);
2760 extern void recalc_sigpending(void);
2762 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2764 static inline void signal_wake_up(struct task_struct *t, bool resume)
2766 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2768 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2770 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2774 * Wrappers for p->thread_info->cpu access. No-op on UP.
2778 static inline unsigned int task_cpu(const struct task_struct *p)
2780 return task_thread_info(p)->cpu;
2783 static inline int task_node(const struct task_struct *p)
2785 return cpu_to_node(task_cpu(p));
2788 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2792 static inline unsigned int task_cpu(const struct task_struct *p)
2797 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2801 #endif /* CONFIG_SMP */
2803 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2804 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2806 #ifdef CONFIG_CGROUP_SCHED
2807 extern struct task_group root_task_group;
2808 #endif /* CONFIG_CGROUP_SCHED */
2810 extern int task_can_switch_user(struct user_struct *up,
2811 struct task_struct *tsk);
2813 #ifdef CONFIG_TASK_XACCT
2814 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2816 tsk->ioac.rchar += amt;
2819 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2821 tsk->ioac.wchar += amt;
2824 static inline void inc_syscr(struct task_struct *tsk)
2829 static inline void inc_syscw(struct task_struct *tsk)
2834 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2838 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2842 static inline void inc_syscr(struct task_struct *tsk)
2846 static inline void inc_syscw(struct task_struct *tsk)
2851 #ifndef TASK_SIZE_OF
2852 #define TASK_SIZE_OF(tsk) TASK_SIZE
2855 #ifdef CONFIG_MM_OWNER
2856 extern void mm_update_next_owner(struct mm_struct *mm);
2857 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2859 static inline void mm_update_next_owner(struct mm_struct *mm)
2863 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2866 #endif /* CONFIG_MM_OWNER */
2868 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2871 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2874 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2877 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2880 static inline unsigned long rlimit(unsigned int limit)
2882 return task_rlimit(current, limit);
2885 static inline unsigned long rlimit_max(unsigned int limit)
2887 return task_rlimit_max(current, limit);