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>
27 #include <asm/ptrace.h>
28 #include <asm/cputime.h>
30 #include <linux/smp.h>
31 #include <linux/sem.h>
32 #include <linux/signal.h>
33 #include <linux/compiler.h>
34 #include <linux/completion.h>
35 #include <linux/pid.h>
36 #include <linux/percpu.h>
37 #include <linux/topology.h>
38 #include <linux/proportions.h>
39 #include <linux/seccomp.h>
40 #include <linux/rcupdate.h>
41 #include <linux/rculist.h>
42 #include <linux/rtmutex.h>
44 #include <linux/time.h>
45 #include <linux/param.h>
46 #include <linux/resource.h>
47 #include <linux/timer.h>
48 #include <linux/hrtimer.h>
49 #include <linux/task_io_accounting.h>
50 #include <linux/latencytop.h>
51 #include <linux/cred.h>
52 #include <linux/llist.h>
53 #include <linux/uidgid.h>
55 #include <asm/processor.h>
58 struct futex_pi_state;
59 struct robust_list_head;
62 struct perf_event_context;
66 * List of flags we want to share for kernel threads,
67 * if only because they are not used by them anyway.
69 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
72 * These are the constant used to fake the fixed-point load-average
73 * counting. Some notes:
74 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
75 * a load-average precision of 10 bits integer + 11 bits fractional
76 * - if you want to count load-averages more often, you need more
77 * precision, or rounding will get you. With 2-second counting freq,
78 * the EXP_n values would be 1981, 2034 and 2043 if still using only
81 extern unsigned long avenrun[]; /* Load averages */
82 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
84 #define FSHIFT 11 /* nr of bits of precision */
85 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
86 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
87 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
88 #define EXP_5 2014 /* 1/exp(5sec/5min) */
89 #define EXP_15 2037 /* 1/exp(5sec/15min) */
91 #define CALC_LOAD(load,exp,n) \
93 load += n*(FIXED_1-exp); \
96 extern unsigned long total_forks;
97 extern int nr_threads;
98 DECLARE_PER_CPU(unsigned long, process_counts);
99 extern int nr_processes(void);
100 extern unsigned long nr_running(void);
101 extern unsigned long nr_uninterruptible(void);
102 extern unsigned long nr_iowait(void);
103 extern unsigned long nr_iowait_cpu(int cpu);
104 extern unsigned long this_cpu_load(void);
107 extern void calc_global_load(unsigned long ticks);
108 extern void update_cpu_load_nohz(void);
110 extern unsigned long get_parent_ip(unsigned long addr);
112 extern void dump_cpu_task(int cpu);
117 #ifdef CONFIG_SCHED_DEBUG
118 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
119 extern void proc_sched_set_task(struct task_struct *p);
121 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
124 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
127 static inline void proc_sched_set_task(struct task_struct *p)
131 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
137 * Task state bitmask. NOTE! These bits are also
138 * encoded in fs/proc/array.c: get_task_state().
140 * We have two separate sets of flags: task->state
141 * is about runnability, while task->exit_state are
142 * about the task exiting. Confusing, but this way
143 * modifying one set can't modify the other one by
146 #define TASK_RUNNING 0
147 #define TASK_INTERRUPTIBLE 1
148 #define TASK_UNINTERRUPTIBLE 2
149 #define __TASK_STOPPED 4
150 #define __TASK_TRACED 8
151 /* in tsk->exit_state */
152 #define EXIT_ZOMBIE 16
154 /* in tsk->state again */
156 #define TASK_WAKEKILL 128
157 #define TASK_WAKING 256
158 #define TASK_STATE_MAX 512
160 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
162 extern char ___assert_task_state[1 - 2*!!(
163 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
165 /* Convenience macros for the sake of set_task_state */
166 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
167 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
168 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
170 /* Convenience macros for the sake of wake_up */
171 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
172 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
174 /* get_task_state() */
175 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
176 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
179 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
180 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
181 #define task_is_dead(task) ((task)->exit_state != 0)
182 #define task_is_stopped_or_traced(task) \
183 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
184 #define task_contributes_to_load(task) \
185 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
186 (task->flags & PF_FROZEN) == 0)
188 #define __set_task_state(tsk, state_value) \
189 do { (tsk)->state = (state_value); } while (0)
190 #define set_task_state(tsk, state_value) \
191 set_mb((tsk)->state, (state_value))
194 * set_current_state() includes a barrier so that the write of current->state
195 * is correctly serialised wrt the caller's subsequent test of whether to
198 * set_current_state(TASK_UNINTERRUPTIBLE);
199 * if (do_i_need_to_sleep())
202 * If the caller does not need such serialisation then use __set_current_state()
204 #define __set_current_state(state_value) \
205 do { current->state = (state_value); } while (0)
206 #define set_current_state(state_value) \
207 set_mb(current->state, (state_value))
209 /* Task command name length */
210 #define TASK_COMM_LEN 16
212 #include <linux/spinlock.h>
215 * This serializes "schedule()" and also protects
216 * the run-queue from deletions/modifications (but
217 * _adding_ to the beginning of the run-queue has
220 extern rwlock_t tasklist_lock;
221 extern spinlock_t mmlist_lock;
225 #ifdef CONFIG_PROVE_RCU
226 extern int lockdep_tasklist_lock_is_held(void);
227 #endif /* #ifdef CONFIG_PROVE_RCU */
229 extern void sched_init(void);
230 extern void sched_init_smp(void);
231 extern asmlinkage void schedule_tail(struct task_struct *prev);
232 extern void init_idle(struct task_struct *idle, int cpu);
233 extern void init_idle_bootup_task(struct task_struct *idle);
235 extern int runqueue_is_locked(int cpu);
237 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
238 extern void nohz_balance_enter_idle(int cpu);
239 extern void set_cpu_sd_state_idle(void);
240 extern int get_nohz_timer_target(void);
242 static inline void nohz_balance_enter_idle(int cpu) { }
243 static inline void set_cpu_sd_state_idle(void) { }
247 * Only dump TASK_* tasks. (0 for all tasks)
249 extern void show_state_filter(unsigned long state_filter);
251 static inline void show_state(void)
253 show_state_filter(0);
256 extern void show_regs(struct pt_regs *);
259 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
260 * task), SP is the stack pointer of the first frame that should be shown in the back
261 * trace (or NULL if the entire call-chain of the task should be shown).
263 extern void show_stack(struct task_struct *task, unsigned long *sp);
265 void io_schedule(void);
266 long io_schedule_timeout(long timeout);
268 extern void cpu_init (void);
269 extern void trap_init(void);
270 extern void update_process_times(int user);
271 extern void scheduler_tick(void);
273 extern void sched_show_task(struct task_struct *p);
275 #ifdef CONFIG_LOCKUP_DETECTOR
276 extern void touch_softlockup_watchdog(void);
277 extern void touch_softlockup_watchdog_sync(void);
278 extern void touch_all_softlockup_watchdogs(void);
279 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
281 size_t *lenp, loff_t *ppos);
282 extern unsigned int softlockup_panic;
283 void lockup_detector_init(void);
285 static inline void touch_softlockup_watchdog(void)
288 static inline void touch_softlockup_watchdog_sync(void)
291 static inline void touch_all_softlockup_watchdogs(void)
294 static inline void lockup_detector_init(void)
299 #ifdef CONFIG_DETECT_HUNG_TASK
300 extern unsigned int sysctl_hung_task_panic;
301 extern unsigned long sysctl_hung_task_check_count;
302 extern unsigned long sysctl_hung_task_timeout_secs;
303 extern unsigned long sysctl_hung_task_warnings;
304 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
306 size_t *lenp, loff_t *ppos);
308 /* Avoid need for ifdefs elsewhere in the code */
309 enum { sysctl_hung_task_timeout_secs = 0 };
312 /* Attach to any functions which should be ignored in wchan output. */
313 #define __sched __attribute__((__section__(".sched.text")))
315 /* Linker adds these: start and end of __sched functions */
316 extern char __sched_text_start[], __sched_text_end[];
318 /* Is this address in the __sched functions? */
319 extern int in_sched_functions(unsigned long addr);
321 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
322 extern signed long schedule_timeout(signed long timeout);
323 extern signed long schedule_timeout_interruptible(signed long timeout);
324 extern signed long schedule_timeout_killable(signed long timeout);
325 extern signed long schedule_timeout_uninterruptible(signed long timeout);
326 asmlinkage void schedule(void);
327 extern void schedule_preempt_disabled(void);
328 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
331 struct user_namespace;
334 * Default maximum number of active map areas, this limits the number of vmas
335 * per mm struct. Users can overwrite this number by sysctl but there is a
338 * When a program's coredump is generated as ELF format, a section is created
339 * per a vma. In ELF, the number of sections is represented in unsigned short.
340 * This means the number of sections should be smaller than 65535 at coredump.
341 * Because the kernel adds some informative sections to a image of program at
342 * generating coredump, we need some margin. The number of extra sections is
343 * 1-3 now and depends on arch. We use "5" as safe margin, here.
345 #define MAPCOUNT_ELF_CORE_MARGIN (5)
346 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
348 extern int sysctl_max_map_count;
350 #include <linux/aio.h>
353 extern void arch_pick_mmap_layout(struct mm_struct *mm);
355 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
356 unsigned long, unsigned long);
358 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
359 unsigned long len, unsigned long pgoff,
360 unsigned long flags);
361 extern void arch_unmap_area(struct mm_struct *, unsigned long);
362 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
364 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
368 extern void set_dumpable(struct mm_struct *mm, int value);
369 extern int get_dumpable(struct mm_struct *mm);
371 /* get/set_dumpable() values */
372 #define SUID_DUMPABLE_DISABLED 0
373 #define SUID_DUMPABLE_ENABLED 1
374 #define SUID_DUMPABLE_SAFE 2
378 #define MMF_DUMPABLE 0 /* core dump is permitted */
379 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
381 #define MMF_DUMPABLE_BITS 2
382 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
384 /* coredump filter bits */
385 #define MMF_DUMP_ANON_PRIVATE 2
386 #define MMF_DUMP_ANON_SHARED 3
387 #define MMF_DUMP_MAPPED_PRIVATE 4
388 #define MMF_DUMP_MAPPED_SHARED 5
389 #define MMF_DUMP_ELF_HEADERS 6
390 #define MMF_DUMP_HUGETLB_PRIVATE 7
391 #define MMF_DUMP_HUGETLB_SHARED 8
393 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
394 #define MMF_DUMP_FILTER_BITS 7
395 #define MMF_DUMP_FILTER_MASK \
396 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
397 #define MMF_DUMP_FILTER_DEFAULT \
398 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
399 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
401 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
402 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
404 # define MMF_DUMP_MASK_DEFAULT_ELF 0
406 /* leave room for more dump flags */
407 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
408 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
409 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
411 #define MMF_HAS_UPROBES 19 /* has uprobes */
412 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
414 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
416 struct sighand_struct {
418 struct k_sigaction action[_NSIG];
420 wait_queue_head_t signalfd_wqh;
423 struct pacct_struct {
426 unsigned long ac_mem;
427 cputime_t ac_utime, ac_stime;
428 unsigned long ac_minflt, ac_majflt;
439 * struct cputime - snaphsot of system and user cputime
440 * @utime: time spent in user mode
441 * @stime: time spent in system mode
443 * Gathers a generic snapshot of user and system time.
451 * struct task_cputime - collected CPU time counts
452 * @utime: time spent in user mode, in &cputime_t units
453 * @stime: time spent in kernel mode, in &cputime_t units
454 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
456 * This is an extension of struct cputime that includes the total runtime
457 * spent by the task from the scheduler point of view.
459 * As a result, this structure groups together three kinds of CPU time
460 * that are tracked for threads and thread groups. Most things considering
461 * CPU time want to group these counts together and treat all three
462 * of them in parallel.
464 struct task_cputime {
467 unsigned long long sum_exec_runtime;
469 /* Alternate field names when used to cache expirations. */
470 #define prof_exp stime
471 #define virt_exp utime
472 #define sched_exp sum_exec_runtime
474 #define INIT_CPUTIME \
475 (struct task_cputime) { \
478 .sum_exec_runtime = 0, \
482 * Disable preemption until the scheduler is running.
483 * Reset by start_kernel()->sched_init()->init_idle().
485 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
486 * before the scheduler is active -- see should_resched().
488 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
491 * struct thread_group_cputimer - thread group interval timer counts
492 * @cputime: thread group interval timers.
493 * @running: non-zero when there are timers running and
494 * @cputime receives updates.
495 * @lock: lock for fields in this struct.
497 * This structure contains the version of task_cputime, above, that is
498 * used for thread group CPU timer calculations.
500 struct thread_group_cputimer {
501 struct task_cputime cputime;
506 #include <linux/rwsem.h>
510 * NOTE! "signal_struct" does not have its own
511 * locking, because a shared signal_struct always
512 * implies a shared sighand_struct, so locking
513 * sighand_struct is always a proper superset of
514 * the locking of signal_struct.
516 struct signal_struct {
521 wait_queue_head_t wait_chldexit; /* for wait4() */
523 /* current thread group signal load-balancing target: */
524 struct task_struct *curr_target;
526 /* shared signal handling: */
527 struct sigpending shared_pending;
529 /* thread group exit support */
532 * - notify group_exit_task when ->count is equal to notify_count
533 * - everyone except group_exit_task is stopped during signal delivery
534 * of fatal signals, group_exit_task processes the signal.
537 struct task_struct *group_exit_task;
539 /* thread group stop support, overloads group_exit_code too */
540 int group_stop_count;
541 unsigned int flags; /* see SIGNAL_* flags below */
544 * PR_SET_CHILD_SUBREAPER marks a process, like a service
545 * manager, to re-parent orphan (double-forking) child processes
546 * to this process instead of 'init'. The service manager is
547 * able to receive SIGCHLD signals and is able to investigate
548 * the process until it calls wait(). All children of this
549 * process will inherit a flag if they should look for a
550 * child_subreaper process at exit.
552 unsigned int is_child_subreaper:1;
553 unsigned int has_child_subreaper:1;
555 /* POSIX.1b Interval Timers */
556 struct list_head posix_timers;
558 /* ITIMER_REAL timer for the process */
559 struct hrtimer real_timer;
560 struct pid *leader_pid;
561 ktime_t it_real_incr;
564 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
565 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
566 * values are defined to 0 and 1 respectively
568 struct cpu_itimer it[2];
571 * Thread group totals for process CPU timers.
572 * See thread_group_cputimer(), et al, for details.
574 struct thread_group_cputimer cputimer;
576 /* Earliest-expiration cache. */
577 struct task_cputime cputime_expires;
579 struct list_head cpu_timers[3];
581 struct pid *tty_old_pgrp;
583 /* boolean value for session group leader */
586 struct tty_struct *tty; /* NULL if no tty */
588 #ifdef CONFIG_SCHED_AUTOGROUP
589 struct autogroup *autogroup;
592 * Cumulative resource counters for dead threads in the group,
593 * and for reaped dead child processes forked by this group.
594 * Live threads maintain their own counters and add to these
595 * in __exit_signal, except for the group leader.
597 cputime_t utime, stime, cutime, cstime;
600 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
601 struct cputime prev_cputime;
603 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
604 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
605 unsigned long inblock, oublock, cinblock, coublock;
606 unsigned long maxrss, cmaxrss;
607 struct task_io_accounting ioac;
610 * Cumulative ns of schedule CPU time fo dead threads in the
611 * group, not including a zombie group leader, (This only differs
612 * from jiffies_to_ns(utime + stime) if sched_clock uses something
613 * other than jiffies.)
615 unsigned long long sum_sched_runtime;
618 * We don't bother to synchronize most readers of this at all,
619 * because there is no reader checking a limit that actually needs
620 * to get both rlim_cur and rlim_max atomically, and either one
621 * alone is a single word that can safely be read normally.
622 * getrlimit/setrlimit use task_lock(current->group_leader) to
623 * protect this instead of the siglock, because they really
624 * have no need to disable irqs.
626 struct rlimit rlim[RLIM_NLIMITS];
628 #ifdef CONFIG_BSD_PROCESS_ACCT
629 struct pacct_struct pacct; /* per-process accounting information */
631 #ifdef CONFIG_TASKSTATS
632 struct taskstats *stats;
636 struct tty_audit_buf *tty_audit_buf;
638 #ifdef CONFIG_CGROUPS
640 * group_rwsem prevents new tasks from entering the threadgroup and
641 * member tasks from exiting,a more specifically, setting of
642 * PF_EXITING. fork and exit paths are protected with this rwsem
643 * using threadgroup_change_begin/end(). Users which require
644 * threadgroup to remain stable should use threadgroup_[un]lock()
645 * which also takes care of exec path. Currently, cgroup is the
648 struct rw_semaphore group_rwsem;
651 oom_flags_t oom_flags;
652 short oom_score_adj; /* OOM kill score adjustment */
653 short oom_score_adj_min; /* OOM kill score adjustment min value.
654 * Only settable by CAP_SYS_RESOURCE. */
656 struct mutex cred_guard_mutex; /* guard against foreign influences on
657 * credential calculations
658 * (notably. ptrace) */
662 * Bits in flags field of signal_struct.
664 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
665 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
666 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
668 * Pending notifications to parent.
670 #define SIGNAL_CLD_STOPPED 0x00000010
671 #define SIGNAL_CLD_CONTINUED 0x00000020
672 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
674 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
676 /* If true, all threads except ->group_exit_task have pending SIGKILL */
677 static inline int signal_group_exit(const struct signal_struct *sig)
679 return (sig->flags & SIGNAL_GROUP_EXIT) ||
680 (sig->group_exit_task != NULL);
684 * Some day this will be a full-fledged user tracking system..
687 atomic_t __count; /* reference count */
688 atomic_t processes; /* How many processes does this user have? */
689 atomic_t files; /* How many open files does this user have? */
690 atomic_t sigpending; /* How many pending signals does this user have? */
691 #ifdef CONFIG_INOTIFY_USER
692 atomic_t inotify_watches; /* How many inotify watches does this user have? */
693 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
695 #ifdef CONFIG_FANOTIFY
696 atomic_t fanotify_listeners;
699 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
701 #ifdef CONFIG_POSIX_MQUEUE
702 /* protected by mq_lock */
703 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
705 unsigned long locked_shm; /* How many pages of mlocked shm ? */
708 struct key *uid_keyring; /* UID specific keyring */
709 struct key *session_keyring; /* UID's default session keyring */
712 /* Hash table maintenance information */
713 struct hlist_node uidhash_node;
716 #ifdef CONFIG_PERF_EVENTS
717 atomic_long_t locked_vm;
721 extern int uids_sysfs_init(void);
723 extern struct user_struct *find_user(kuid_t);
725 extern struct user_struct root_user;
726 #define INIT_USER (&root_user)
729 struct backing_dev_info;
730 struct reclaim_state;
732 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
734 /* cumulative counters */
735 unsigned long pcount; /* # of times run on this cpu */
736 unsigned long long run_delay; /* time spent waiting on a runqueue */
739 unsigned long long last_arrival,/* when we last ran on a cpu */
740 last_queued; /* when we were last queued to run */
742 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
744 #ifdef CONFIG_TASK_DELAY_ACCT
745 struct task_delay_info {
747 unsigned int flags; /* Private per-task flags */
749 /* For each stat XXX, add following, aligned appropriately
751 * struct timespec XXX_start, XXX_end;
755 * Atomicity of updates to XXX_delay, XXX_count protected by
756 * single lock above (split into XXX_lock if contention is an issue).
760 * XXX_count is incremented on every XXX operation, the delay
761 * associated with the operation is added to XXX_delay.
762 * XXX_delay contains the accumulated delay time in nanoseconds.
764 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
765 u64 blkio_delay; /* wait for sync block io completion */
766 u64 swapin_delay; /* wait for swapin block io completion */
767 u32 blkio_count; /* total count of the number of sync block */
768 /* io operations performed */
769 u32 swapin_count; /* total count of the number of swapin block */
770 /* io operations performed */
772 struct timespec freepages_start, freepages_end;
773 u64 freepages_delay; /* wait for memory reclaim */
774 u32 freepages_count; /* total count of memory reclaim */
776 #endif /* CONFIG_TASK_DELAY_ACCT */
778 static inline int sched_info_on(void)
780 #ifdef CONFIG_SCHEDSTATS
782 #elif defined(CONFIG_TASK_DELAY_ACCT)
783 extern int delayacct_on;
798 * Increase resolution of nice-level calculations for 64-bit architectures.
799 * The extra resolution improves shares distribution and load balancing of
800 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
801 * hierarchies, especially on larger systems. This is not a user-visible change
802 * and does not change the user-interface for setting shares/weights.
804 * We increase resolution only if we have enough bits to allow this increased
805 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
806 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
809 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
810 # define SCHED_LOAD_RESOLUTION 10
811 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
812 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
814 # define SCHED_LOAD_RESOLUTION 0
815 # define scale_load(w) (w)
816 # define scale_load_down(w) (w)
819 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
820 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
823 * Increase resolution of cpu_power calculations
825 #define SCHED_POWER_SHIFT 10
826 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
829 * sched-domains (multiprocessor balancing) declarations:
832 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
833 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
834 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
835 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
836 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
837 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
838 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
839 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
840 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
841 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
842 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
843 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
845 extern int __weak arch_sd_sibiling_asym_packing(void);
847 struct sched_group_power {
850 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
853 unsigned int power, power_orig;
854 unsigned long next_update;
856 * Number of busy cpus in this group.
858 atomic_t nr_busy_cpus;
860 unsigned long cpumask[0]; /* iteration mask */
864 struct sched_group *next; /* Must be a circular list */
867 unsigned int group_weight;
868 struct sched_group_power *sgp;
871 * The CPUs this group covers.
873 * NOTE: this field is variable length. (Allocated dynamically
874 * by attaching extra space to the end of the structure,
875 * depending on how many CPUs the kernel has booted up with)
877 unsigned long cpumask[0];
880 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
882 return to_cpumask(sg->cpumask);
886 * cpumask masking which cpus in the group are allowed to iterate up the domain
889 static inline struct cpumask *sched_group_mask(struct sched_group *sg)
891 return to_cpumask(sg->sgp->cpumask);
895 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
896 * @group: The group whose first cpu is to be returned.
898 static inline unsigned int group_first_cpu(struct sched_group *group)
900 return cpumask_first(sched_group_cpus(group));
903 struct sched_domain_attr {
904 int relax_domain_level;
907 #define SD_ATTR_INIT (struct sched_domain_attr) { \
908 .relax_domain_level = -1, \
911 extern int sched_domain_level_max;
913 struct sched_domain {
914 /* These fields must be setup */
915 struct sched_domain *parent; /* top domain must be null terminated */
916 struct sched_domain *child; /* bottom domain must be null terminated */
917 struct sched_group *groups; /* the balancing groups of the domain */
918 unsigned long min_interval; /* Minimum balance interval ms */
919 unsigned long max_interval; /* Maximum balance interval ms */
920 unsigned int busy_factor; /* less balancing by factor if busy */
921 unsigned int imbalance_pct; /* No balance until over watermark */
922 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
923 unsigned int busy_idx;
924 unsigned int idle_idx;
925 unsigned int newidle_idx;
926 unsigned int wake_idx;
927 unsigned int forkexec_idx;
928 unsigned int smt_gain;
929 int flags; /* See SD_* */
932 /* Runtime fields. */
933 unsigned long last_balance; /* init to jiffies. units in jiffies */
934 unsigned int balance_interval; /* initialise to 1. units in ms. */
935 unsigned int nr_balance_failed; /* initialise to 0 */
939 #ifdef CONFIG_SCHEDSTATS
940 /* load_balance() stats */
941 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
947 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
948 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
950 /* Active load balancing */
951 unsigned int alb_count;
952 unsigned int alb_failed;
953 unsigned int alb_pushed;
955 /* SD_BALANCE_EXEC stats */
956 unsigned int sbe_count;
957 unsigned int sbe_balanced;
958 unsigned int sbe_pushed;
960 /* SD_BALANCE_FORK stats */
961 unsigned int sbf_count;
962 unsigned int sbf_balanced;
963 unsigned int sbf_pushed;
965 /* try_to_wake_up() stats */
966 unsigned int ttwu_wake_remote;
967 unsigned int ttwu_move_affine;
968 unsigned int ttwu_move_balance;
970 #ifdef CONFIG_SCHED_DEBUG
974 void *private; /* used during construction */
975 struct rcu_head rcu; /* used during destruction */
978 unsigned int span_weight;
980 * Span of all CPUs in this domain.
982 * NOTE: this field is variable length. (Allocated dynamically
983 * by attaching extra space to the end of the structure,
984 * depending on how many CPUs the kernel has booted up with)
986 unsigned long span[0];
989 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
991 return to_cpumask(sd->span);
994 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
995 struct sched_domain_attr *dattr_new);
997 /* Allocate an array of sched domains, for partition_sched_domains(). */
998 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
999 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1001 /* Test a flag in parent sched domain */
1002 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1004 if (sd->parent && (sd->parent->flags & flag))
1010 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1011 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1013 bool cpus_share_cache(int this_cpu, int that_cpu);
1015 #else /* CONFIG_SMP */
1017 struct sched_domain_attr;
1020 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1021 struct sched_domain_attr *dattr_new)
1025 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1030 #endif /* !CONFIG_SMP */
1033 struct io_context; /* See blkdev.h */
1036 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1037 extern void prefetch_stack(struct task_struct *t);
1039 static inline void prefetch_stack(struct task_struct *t) { }
1042 struct audit_context; /* See audit.c */
1044 struct pipe_inode_info;
1045 struct uts_namespace;
1048 struct sched_domain;
1053 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1054 #define WF_FORK 0x02 /* child wakeup after fork */
1055 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1057 #define ENQUEUE_WAKEUP 1
1058 #define ENQUEUE_HEAD 2
1060 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1062 #define ENQUEUE_WAKING 0
1065 #define DEQUEUE_SLEEP 1
1067 struct sched_class {
1068 const struct sched_class *next;
1070 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1071 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1072 void (*yield_task) (struct rq *rq);
1073 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1075 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1077 struct task_struct * (*pick_next_task) (struct rq *rq);
1078 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1081 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1082 void (*migrate_task_rq)(struct task_struct *p, int next_cpu);
1084 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1085 void (*post_schedule) (struct rq *this_rq);
1086 void (*task_waking) (struct task_struct *task);
1087 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1089 void (*set_cpus_allowed)(struct task_struct *p,
1090 const struct cpumask *newmask);
1092 void (*rq_online)(struct rq *rq);
1093 void (*rq_offline)(struct rq *rq);
1096 void (*set_curr_task) (struct rq *rq);
1097 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1098 void (*task_fork) (struct task_struct *p);
1100 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1101 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1102 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1105 unsigned int (*get_rr_interval) (struct rq *rq,
1106 struct task_struct *task);
1108 #ifdef CONFIG_FAIR_GROUP_SCHED
1109 void (*task_move_group) (struct task_struct *p, int on_rq);
1113 struct load_weight {
1114 unsigned long weight, inv_weight;
1119 * These sums represent an infinite geometric series and so are bound
1120 * above by 1024/(1-y). Thus we only need a u32 to store them for for all
1121 * choices of y < 1-2^(-32)*1024.
1123 u32 runnable_avg_sum, runnable_avg_period;
1124 u64 last_runnable_update;
1126 unsigned long load_avg_contrib;
1129 #ifdef CONFIG_SCHEDSTATS
1130 struct sched_statistics {
1140 s64 sum_sleep_runtime;
1147 u64 nr_migrations_cold;
1148 u64 nr_failed_migrations_affine;
1149 u64 nr_failed_migrations_running;
1150 u64 nr_failed_migrations_hot;
1151 u64 nr_forced_migrations;
1154 u64 nr_wakeups_sync;
1155 u64 nr_wakeups_migrate;
1156 u64 nr_wakeups_local;
1157 u64 nr_wakeups_remote;
1158 u64 nr_wakeups_affine;
1159 u64 nr_wakeups_affine_attempts;
1160 u64 nr_wakeups_passive;
1161 u64 nr_wakeups_idle;
1165 struct sched_entity {
1166 struct load_weight load; /* for load-balancing */
1167 struct rb_node run_node;
1168 struct list_head group_node;
1172 u64 sum_exec_runtime;
1174 u64 prev_sum_exec_runtime;
1178 #ifdef CONFIG_SCHEDSTATS
1179 struct sched_statistics statistics;
1182 #ifdef CONFIG_FAIR_GROUP_SCHED
1183 struct sched_entity *parent;
1184 /* rq on which this entity is (to be) queued: */
1185 struct cfs_rq *cfs_rq;
1186 /* rq "owned" by this entity/group: */
1187 struct cfs_rq *my_q;
1190 * Load-tracking only depends on SMP, FAIR_GROUP_SCHED dependency below may be
1191 * removed when useful for applications beyond shares distribution (e.g.
1194 #if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)
1195 /* Per-entity load-tracking */
1196 struct sched_avg avg;
1200 struct sched_rt_entity {
1201 struct list_head run_list;
1202 unsigned long timeout;
1203 unsigned int time_slice;
1205 struct sched_rt_entity *back;
1206 #ifdef CONFIG_RT_GROUP_SCHED
1207 struct sched_rt_entity *parent;
1208 /* rq on which this entity is (to be) queued: */
1209 struct rt_rq *rt_rq;
1210 /* rq "owned" by this entity/group: */
1216 * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1217 * Timeslices get refilled after they expire.
1219 #define RR_TIMESLICE (100 * HZ / 1000)
1223 enum perf_event_task_context {
1224 perf_invalid_context = -1,
1225 perf_hw_context = 0,
1227 perf_nr_task_contexts,
1230 struct task_struct {
1231 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1234 unsigned int flags; /* per process flags, defined below */
1235 unsigned int ptrace;
1238 struct llist_node wake_entry;
1243 int prio, static_prio, normal_prio;
1244 unsigned int rt_priority;
1245 const struct sched_class *sched_class;
1246 struct sched_entity se;
1247 struct sched_rt_entity rt;
1248 #ifdef CONFIG_CGROUP_SCHED
1249 struct task_group *sched_task_group;
1252 #ifdef CONFIG_PREEMPT_NOTIFIERS
1253 /* list of struct preempt_notifier: */
1254 struct hlist_head preempt_notifiers;
1258 * fpu_counter contains the number of consecutive context switches
1259 * that the FPU is used. If this is over a threshold, the lazy fpu
1260 * saving becomes unlazy to save the trap. This is an unsigned char
1261 * so that after 256 times the counter wraps and the behavior turns
1262 * lazy again; this to deal with bursty apps that only use FPU for
1265 unsigned char fpu_counter;
1266 #ifdef CONFIG_BLK_DEV_IO_TRACE
1267 unsigned int btrace_seq;
1270 unsigned int policy;
1271 int nr_cpus_allowed;
1272 cpumask_t cpus_allowed;
1274 #ifdef CONFIG_PREEMPT_RCU
1275 int rcu_read_lock_nesting;
1276 char rcu_read_unlock_special;
1277 struct list_head rcu_node_entry;
1278 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1279 #ifdef CONFIG_TREE_PREEMPT_RCU
1280 struct rcu_node *rcu_blocked_node;
1281 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1282 #ifdef CONFIG_RCU_BOOST
1283 struct rt_mutex *rcu_boost_mutex;
1284 #endif /* #ifdef CONFIG_RCU_BOOST */
1286 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1287 struct sched_info sched_info;
1290 struct list_head tasks;
1292 struct plist_node pushable_tasks;
1295 struct mm_struct *mm, *active_mm;
1296 #ifdef CONFIG_COMPAT_BRK
1297 unsigned brk_randomized:1;
1299 #if defined(SPLIT_RSS_COUNTING)
1300 struct task_rss_stat rss_stat;
1304 int exit_code, exit_signal;
1305 int pdeath_signal; /* The signal sent when the parent dies */
1306 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1308 unsigned int personality;
1309 unsigned did_exec:1;
1310 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1312 unsigned in_iowait:1;
1314 /* task may not gain privileges */
1315 unsigned no_new_privs:1;
1317 /* Revert to default priority/policy when forking */
1318 unsigned sched_reset_on_fork:1;
1319 unsigned sched_contributes_to_load:1;
1324 #ifdef CONFIG_CC_STACKPROTECTOR
1325 /* Canary value for the -fstack-protector gcc feature */
1326 unsigned long stack_canary;
1329 * pointers to (original) parent process, youngest child, younger sibling,
1330 * older sibling, respectively. (p->father can be replaced with
1331 * p->real_parent->pid)
1333 struct task_struct __rcu *real_parent; /* real parent process */
1334 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1336 * children/sibling forms the list of my natural children
1338 struct list_head children; /* list of my children */
1339 struct list_head sibling; /* linkage in my parent's children list */
1340 struct task_struct *group_leader; /* threadgroup leader */
1343 * ptraced is the list of tasks this task is using ptrace on.
1344 * This includes both natural children and PTRACE_ATTACH targets.
1345 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1347 struct list_head ptraced;
1348 struct list_head ptrace_entry;
1350 /* PID/PID hash table linkage. */
1351 struct pid_link pids[PIDTYPE_MAX];
1352 struct list_head thread_group;
1354 struct completion *vfork_done; /* for vfork() */
1355 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1356 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1358 cputime_t utime, stime, utimescaled, stimescaled;
1360 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1361 struct cputime prev_cputime;
1363 unsigned long nvcsw, nivcsw; /* context switch counts */
1364 struct timespec start_time; /* monotonic time */
1365 struct timespec real_start_time; /* boot based time */
1366 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1367 unsigned long min_flt, maj_flt;
1369 struct task_cputime cputime_expires;
1370 struct list_head cpu_timers[3];
1372 /* process credentials */
1373 const struct cred __rcu *real_cred; /* objective and real subjective task
1374 * credentials (COW) */
1375 const struct cred __rcu *cred; /* effective (overridable) subjective task
1376 * credentials (COW) */
1377 char comm[TASK_COMM_LEN]; /* executable name excluding path
1378 - access with [gs]et_task_comm (which lock
1379 it with task_lock())
1380 - initialized normally by setup_new_exec */
1381 /* file system info */
1382 int link_count, total_link_count;
1383 #ifdef CONFIG_SYSVIPC
1385 struct sysv_sem sysvsem;
1387 #ifdef CONFIG_DETECT_HUNG_TASK
1388 /* hung task detection */
1389 unsigned long last_switch_count;
1391 /* CPU-specific state of this task */
1392 struct thread_struct thread;
1393 /* filesystem information */
1394 struct fs_struct *fs;
1395 /* open file information */
1396 struct files_struct *files;
1398 struct nsproxy *nsproxy;
1399 /* signal handlers */
1400 struct signal_struct *signal;
1401 struct sighand_struct *sighand;
1403 sigset_t blocked, real_blocked;
1404 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1405 struct sigpending pending;
1407 unsigned long sas_ss_sp;
1409 int (*notifier)(void *priv);
1410 void *notifier_data;
1411 sigset_t *notifier_mask;
1412 struct callback_head *task_works;
1414 struct audit_context *audit_context;
1415 #ifdef CONFIG_AUDITSYSCALL
1417 unsigned int sessionid;
1419 struct seccomp seccomp;
1421 /* Thread group tracking */
1424 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1426 spinlock_t alloc_lock;
1428 /* Protection of the PI data structures: */
1429 raw_spinlock_t pi_lock;
1431 #ifdef CONFIG_RT_MUTEXES
1432 /* PI waiters blocked on a rt_mutex held by this task */
1433 struct plist_head pi_waiters;
1434 /* Deadlock detection and priority inheritance handling */
1435 struct rt_mutex_waiter *pi_blocked_on;
1438 #ifdef CONFIG_DEBUG_MUTEXES
1439 /* mutex deadlock detection */
1440 struct mutex_waiter *blocked_on;
1442 #ifdef CONFIG_TRACE_IRQFLAGS
1443 unsigned int irq_events;
1444 unsigned long hardirq_enable_ip;
1445 unsigned long hardirq_disable_ip;
1446 unsigned int hardirq_enable_event;
1447 unsigned int hardirq_disable_event;
1448 int hardirqs_enabled;
1449 int hardirq_context;
1450 unsigned long softirq_disable_ip;
1451 unsigned long softirq_enable_ip;
1452 unsigned int softirq_disable_event;
1453 unsigned int softirq_enable_event;
1454 int softirqs_enabled;
1455 int softirq_context;
1457 #ifdef CONFIG_LOCKDEP
1458 # define MAX_LOCK_DEPTH 48UL
1461 unsigned int lockdep_recursion;
1462 struct held_lock held_locks[MAX_LOCK_DEPTH];
1463 gfp_t lockdep_reclaim_gfp;
1466 /* journalling filesystem info */
1469 /* stacked block device info */
1470 struct bio_list *bio_list;
1473 /* stack plugging */
1474 struct blk_plug *plug;
1478 struct reclaim_state *reclaim_state;
1480 struct backing_dev_info *backing_dev_info;
1482 struct io_context *io_context;
1484 unsigned long ptrace_message;
1485 siginfo_t *last_siginfo; /* For ptrace use. */
1486 struct task_io_accounting ioac;
1487 #if defined(CONFIG_TASK_XACCT)
1488 u64 acct_rss_mem1; /* accumulated rss usage */
1489 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1490 cputime_t acct_timexpd; /* stime + utime since last update */
1492 #ifdef CONFIG_CPUSETS
1493 nodemask_t mems_allowed; /* Protected by alloc_lock */
1494 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1495 int cpuset_mem_spread_rotor;
1496 int cpuset_slab_spread_rotor;
1498 #ifdef CONFIG_CGROUPS
1499 /* Control Group info protected by css_set_lock */
1500 struct css_set __rcu *cgroups;
1501 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1502 struct list_head cg_list;
1505 struct robust_list_head __user *robust_list;
1506 #ifdef CONFIG_COMPAT
1507 struct compat_robust_list_head __user *compat_robust_list;
1509 struct list_head pi_state_list;
1510 struct futex_pi_state *pi_state_cache;
1512 #ifdef CONFIG_PERF_EVENTS
1513 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1514 struct mutex perf_event_mutex;
1515 struct list_head perf_event_list;
1518 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1520 short pref_node_fork;
1522 struct rcu_head rcu;
1525 * cache last used pipe for splice
1527 struct pipe_inode_info *splice_pipe;
1529 struct page_frag task_frag;
1531 #ifdef CONFIG_TASK_DELAY_ACCT
1532 struct task_delay_info *delays;
1534 #ifdef CONFIG_FAULT_INJECTION
1538 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1539 * balance_dirty_pages() for some dirty throttling pause
1542 int nr_dirtied_pause;
1543 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1545 #ifdef CONFIG_LATENCYTOP
1546 int latency_record_count;
1547 struct latency_record latency_record[LT_SAVECOUNT];
1550 * time slack values; these are used to round up poll() and
1551 * select() etc timeout values. These are in nanoseconds.
1553 unsigned long timer_slack_ns;
1554 unsigned long default_timer_slack_ns;
1556 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1557 /* Index of current stored address in ret_stack */
1559 /* Stack of return addresses for return function tracing */
1560 struct ftrace_ret_stack *ret_stack;
1561 /* time stamp for last schedule */
1562 unsigned long long ftrace_timestamp;
1564 * Number of functions that haven't been traced
1565 * because of depth overrun.
1567 atomic_t trace_overrun;
1568 /* Pause for the tracing */
1569 atomic_t tracing_graph_pause;
1571 #ifdef CONFIG_TRACING
1572 /* state flags for use by tracers */
1573 unsigned long trace;
1574 /* bitmask and counter of trace recursion */
1575 unsigned long trace_recursion;
1576 #endif /* CONFIG_TRACING */
1577 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1578 struct memcg_batch_info {
1579 int do_batch; /* incremented when batch uncharge started */
1580 struct mem_cgroup *memcg; /* target memcg of uncharge */
1581 unsigned long nr_pages; /* uncharged usage */
1582 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1585 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1586 atomic_t ptrace_bp_refcnt;
1588 #ifdef CONFIG_UPROBES
1589 struct uprobe_task *utask;
1593 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1594 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1597 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1598 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1599 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1600 * values are inverted: lower p->prio value means higher priority.
1602 * The MAX_USER_RT_PRIO value allows the actual maximum
1603 * RT priority to be separate from the value exported to
1604 * user-space. This allows kernel threads to set their
1605 * priority to a value higher than any user task. Note:
1606 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1609 #define MAX_USER_RT_PRIO 100
1610 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1612 #define MAX_PRIO (MAX_RT_PRIO + 40)
1613 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1615 static inline int rt_prio(int prio)
1617 if (unlikely(prio < MAX_RT_PRIO))
1622 static inline int rt_task(struct task_struct *p)
1624 return rt_prio(p->prio);
1627 static inline struct pid *task_pid(struct task_struct *task)
1629 return task->pids[PIDTYPE_PID].pid;
1632 static inline struct pid *task_tgid(struct task_struct *task)
1634 return task->group_leader->pids[PIDTYPE_PID].pid;
1638 * Without tasklist or rcu lock it is not safe to dereference
1639 * the result of task_pgrp/task_session even if task == current,
1640 * we can race with another thread doing sys_setsid/sys_setpgid.
1642 static inline struct pid *task_pgrp(struct task_struct *task)
1644 return task->group_leader->pids[PIDTYPE_PGID].pid;
1647 static inline struct pid *task_session(struct task_struct *task)
1649 return task->group_leader->pids[PIDTYPE_SID].pid;
1652 struct pid_namespace;
1655 * the helpers to get the task's different pids as they are seen
1656 * from various namespaces
1658 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1659 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1661 * task_xid_nr_ns() : id seen from the ns specified;
1663 * set_task_vxid() : assigns a virtual id to a task;
1665 * see also pid_nr() etc in include/linux/pid.h
1667 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1668 struct pid_namespace *ns);
1670 static inline pid_t task_pid_nr(struct task_struct *tsk)
1675 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1676 struct pid_namespace *ns)
1678 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1681 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1683 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1687 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1692 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1694 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1696 return pid_vnr(task_tgid(tsk));
1700 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1701 struct pid_namespace *ns)
1703 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1706 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1708 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1712 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1713 struct pid_namespace *ns)
1715 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1718 static inline pid_t task_session_vnr(struct task_struct *tsk)
1720 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1723 /* obsolete, do not use */
1724 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1726 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1730 * pid_alive - check that a task structure is not stale
1731 * @p: Task structure to be checked.
1733 * Test if a process is not yet dead (at most zombie state)
1734 * If pid_alive fails, then pointers within the task structure
1735 * can be stale and must not be dereferenced.
1737 static inline int pid_alive(struct task_struct *p)
1739 return p->pids[PIDTYPE_PID].pid != NULL;
1743 * is_global_init - check if a task structure is init
1744 * @tsk: Task structure to be checked.
1746 * Check if a task structure is the first user space task the kernel created.
1748 static inline int is_global_init(struct task_struct *tsk)
1750 return tsk->pid == 1;
1754 * is_container_init:
1755 * check whether in the task is init in its own pid namespace.
1757 extern int is_container_init(struct task_struct *tsk);
1759 extern struct pid *cad_pid;
1761 extern void free_task(struct task_struct *tsk);
1762 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1764 extern void __put_task_struct(struct task_struct *t);
1766 static inline void put_task_struct(struct task_struct *t)
1768 if (atomic_dec_and_test(&t->usage))
1769 __put_task_struct(t);
1772 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1773 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1778 #define PF_EXITING 0x00000004 /* getting shut down */
1779 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1780 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1781 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1782 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1783 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1784 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1785 #define PF_DUMPCORE 0x00000200 /* dumped core */
1786 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1787 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1788 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1789 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1790 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1791 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1792 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1793 #define PF_KSWAPD 0x00040000 /* I am kswapd */
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_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
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 */
1807 * Only the _current_ task can read/write to tsk->flags, but other
1808 * tasks can access tsk->flags in readonly mode for example
1809 * with tsk_used_math (like during threaded core dumping).
1810 * There is however an exception to this rule during ptrace
1811 * or during fork: the ptracer task is allowed to write to the
1812 * child->flags of its traced child (same goes for fork, the parent
1813 * can write to the child->flags), because we're guaranteed the
1814 * child is not running and in turn not changing child->flags
1815 * at the same time the parent does it.
1817 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1818 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1819 #define clear_used_math() clear_stopped_child_used_math(current)
1820 #define set_used_math() set_stopped_child_used_math(current)
1821 #define conditional_stopped_child_used_math(condition, child) \
1822 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1823 #define conditional_used_math(condition) \
1824 conditional_stopped_child_used_math(condition, current)
1825 #define copy_to_stopped_child_used_math(child) \
1826 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1827 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1828 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1829 #define used_math() tsk_used_math(current)
1832 * task->jobctl flags
1834 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1836 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1837 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1838 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1839 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1840 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1841 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1842 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1844 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1845 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1846 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1847 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1848 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1849 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1850 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1852 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1853 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1855 extern bool task_set_jobctl_pending(struct task_struct *task,
1857 extern void task_clear_jobctl_trapping(struct task_struct *task);
1858 extern void task_clear_jobctl_pending(struct task_struct *task,
1861 #ifdef CONFIG_PREEMPT_RCU
1863 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1864 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1866 static inline void rcu_copy_process(struct task_struct *p)
1868 p->rcu_read_lock_nesting = 0;
1869 p->rcu_read_unlock_special = 0;
1870 #ifdef CONFIG_TREE_PREEMPT_RCU
1871 p->rcu_blocked_node = NULL;
1872 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1873 #ifdef CONFIG_RCU_BOOST
1874 p->rcu_boost_mutex = NULL;
1875 #endif /* #ifdef CONFIG_RCU_BOOST */
1876 INIT_LIST_HEAD(&p->rcu_node_entry);
1881 static inline void rcu_copy_process(struct task_struct *p)
1887 static inline void tsk_restore_flags(struct task_struct *task,
1888 unsigned long orig_flags, unsigned long flags)
1890 task->flags &= ~flags;
1891 task->flags |= orig_flags & flags;
1895 extern void do_set_cpus_allowed(struct task_struct *p,
1896 const struct cpumask *new_mask);
1898 extern int set_cpus_allowed_ptr(struct task_struct *p,
1899 const struct cpumask *new_mask);
1901 static inline void do_set_cpus_allowed(struct task_struct *p,
1902 const struct cpumask *new_mask)
1905 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1906 const struct cpumask *new_mask)
1908 if (!cpumask_test_cpu(0, new_mask))
1915 void calc_load_enter_idle(void);
1916 void calc_load_exit_idle(void);
1918 static inline void calc_load_enter_idle(void) { }
1919 static inline void calc_load_exit_idle(void) { }
1920 #endif /* CONFIG_NO_HZ */
1922 #ifndef CONFIG_CPUMASK_OFFSTACK
1923 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1925 return set_cpus_allowed_ptr(p, &new_mask);
1930 * Do not use outside of architecture code which knows its limitations.
1932 * sched_clock() has no promise of monotonicity or bounded drift between
1933 * CPUs, use (which you should not) requires disabling IRQs.
1935 * Please use one of the three interfaces below.
1937 extern unsigned long long notrace sched_clock(void);
1939 * See the comment in kernel/sched/clock.c
1941 extern u64 cpu_clock(int cpu);
1942 extern u64 local_clock(void);
1943 extern u64 sched_clock_cpu(int cpu);
1946 extern void sched_clock_init(void);
1948 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1949 static inline void sched_clock_tick(void)
1953 static inline void sched_clock_idle_sleep_event(void)
1957 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1962 * Architectures can set this to 1 if they have specified
1963 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1964 * but then during bootup it turns out that sched_clock()
1965 * is reliable after all:
1967 extern int sched_clock_stable;
1969 extern void sched_clock_tick(void);
1970 extern void sched_clock_idle_sleep_event(void);
1971 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1974 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1976 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1977 * The reason for this explicit opt-in is not to have perf penalty with
1978 * slow sched_clocks.
1980 extern void enable_sched_clock_irqtime(void);
1981 extern void disable_sched_clock_irqtime(void);
1983 static inline void enable_sched_clock_irqtime(void) {}
1984 static inline void disable_sched_clock_irqtime(void) {}
1987 extern unsigned long long
1988 task_sched_runtime(struct task_struct *task);
1990 /* sched_exec is called by processes performing an exec */
1992 extern void sched_exec(void);
1994 #define sched_exec() {}
1997 extern void sched_clock_idle_sleep_event(void);
1998 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2000 #ifdef CONFIG_HOTPLUG_CPU
2001 extern void idle_task_exit(void);
2003 static inline void idle_task_exit(void) {}
2006 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2007 extern void wake_up_idle_cpu(int cpu);
2009 static inline void wake_up_idle_cpu(int cpu) { }
2012 extern unsigned int sysctl_sched_latency;
2013 extern unsigned int sysctl_sched_min_granularity;
2014 extern unsigned int sysctl_sched_wakeup_granularity;
2015 extern unsigned int sysctl_sched_child_runs_first;
2017 enum sched_tunable_scaling {
2018 SCHED_TUNABLESCALING_NONE,
2019 SCHED_TUNABLESCALING_LOG,
2020 SCHED_TUNABLESCALING_LINEAR,
2021 SCHED_TUNABLESCALING_END,
2023 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2025 #ifdef CONFIG_SCHED_DEBUG
2026 extern unsigned int sysctl_sched_migration_cost;
2027 extern unsigned int sysctl_sched_nr_migrate;
2028 extern unsigned int sysctl_sched_time_avg;
2029 extern unsigned int sysctl_timer_migration;
2030 extern unsigned int sysctl_sched_shares_window;
2032 int sched_proc_update_handler(struct ctl_table *table, int write,
2033 void __user *buffer, size_t *length,
2036 #ifdef CONFIG_SCHED_DEBUG
2037 static inline unsigned int get_sysctl_timer_migration(void)
2039 return sysctl_timer_migration;
2042 static inline unsigned int get_sysctl_timer_migration(void)
2047 extern unsigned int sysctl_sched_rt_period;
2048 extern int sysctl_sched_rt_runtime;
2050 int sched_rt_handler(struct ctl_table *table, int write,
2051 void __user *buffer, size_t *lenp,
2054 #ifdef CONFIG_SCHED_AUTOGROUP
2055 extern unsigned int sysctl_sched_autogroup_enabled;
2057 extern void sched_autogroup_create_attach(struct task_struct *p);
2058 extern void sched_autogroup_detach(struct task_struct *p);
2059 extern void sched_autogroup_fork(struct signal_struct *sig);
2060 extern void sched_autogroup_exit(struct signal_struct *sig);
2061 #ifdef CONFIG_PROC_FS
2062 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2063 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2066 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2067 static inline void sched_autogroup_detach(struct task_struct *p) { }
2068 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2069 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2072 #ifdef CONFIG_CFS_BANDWIDTH
2073 extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2076 #ifdef CONFIG_RT_MUTEXES
2077 extern int rt_mutex_getprio(struct task_struct *p);
2078 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2079 extern void rt_mutex_adjust_pi(struct task_struct *p);
2080 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2082 return tsk->pi_blocked_on != NULL;
2085 static inline int rt_mutex_getprio(struct task_struct *p)
2087 return p->normal_prio;
2089 # define rt_mutex_adjust_pi(p) do { } while (0)
2090 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2096 extern bool yield_to(struct task_struct *p, bool preempt);
2097 extern void set_user_nice(struct task_struct *p, long nice);
2098 extern int task_prio(const struct task_struct *p);
2099 extern int task_nice(const struct task_struct *p);
2100 extern int can_nice(const struct task_struct *p, const int nice);
2101 extern int task_curr(const struct task_struct *p);
2102 extern int idle_cpu(int cpu);
2103 extern int sched_setscheduler(struct task_struct *, int,
2104 const struct sched_param *);
2105 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2106 const struct sched_param *);
2107 extern struct task_struct *idle_task(int cpu);
2109 * is_idle_task - is the specified task an idle task?
2110 * @p: the task in question.
2112 static inline bool is_idle_task(const struct task_struct *p)
2116 extern struct task_struct *curr_task(int cpu);
2117 extern void set_curr_task(int cpu, struct task_struct *p);
2122 * The default (Linux) execution domain.
2124 extern struct exec_domain default_exec_domain;
2126 union thread_union {
2127 struct thread_info thread_info;
2128 unsigned long stack[THREAD_SIZE/sizeof(long)];
2131 #ifndef __HAVE_ARCH_KSTACK_END
2132 static inline int kstack_end(void *addr)
2134 /* Reliable end of stack detection:
2135 * Some APM bios versions misalign the stack
2137 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2141 extern union thread_union init_thread_union;
2142 extern struct task_struct init_task;
2144 extern struct mm_struct init_mm;
2146 extern struct pid_namespace init_pid_ns;
2149 * find a task by one of its numerical ids
2151 * find_task_by_pid_ns():
2152 * finds a task by its pid in the specified namespace
2153 * find_task_by_vpid():
2154 * finds a task by its virtual pid
2156 * see also find_vpid() etc in include/linux/pid.h
2159 extern struct task_struct *find_task_by_vpid(pid_t nr);
2160 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2161 struct pid_namespace *ns);
2163 extern void __set_special_pids(struct pid *pid);
2165 /* per-UID process charging. */
2166 extern struct user_struct * alloc_uid(kuid_t);
2167 static inline struct user_struct *get_uid(struct user_struct *u)
2169 atomic_inc(&u->__count);
2172 extern void free_uid(struct user_struct *);
2174 #include <asm/current.h>
2176 extern void xtime_update(unsigned long ticks);
2178 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2179 extern int wake_up_process(struct task_struct *tsk);
2180 extern void wake_up_new_task(struct task_struct *tsk);
2182 extern void kick_process(struct task_struct *tsk);
2184 static inline void kick_process(struct task_struct *tsk) { }
2186 extern void sched_fork(struct task_struct *p);
2187 extern void sched_dead(struct task_struct *p);
2189 extern void proc_caches_init(void);
2190 extern void flush_signals(struct task_struct *);
2191 extern void __flush_signals(struct task_struct *);
2192 extern void ignore_signals(struct task_struct *);
2193 extern void flush_signal_handlers(struct task_struct *, int force_default);
2194 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2196 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2198 unsigned long flags;
2201 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2202 ret = dequeue_signal(tsk, mask, info);
2203 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2208 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2210 extern void unblock_all_signals(void);
2211 extern void release_task(struct task_struct * p);
2212 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2213 extern int force_sigsegv(int, struct task_struct *);
2214 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2215 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2216 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2217 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2218 const struct cred *, u32);
2219 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2220 extern int kill_pid(struct pid *pid, int sig, int priv);
2221 extern int kill_proc_info(int, struct siginfo *, pid_t);
2222 extern __must_check bool do_notify_parent(struct task_struct *, int);
2223 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2224 extern void force_sig(int, struct task_struct *);
2225 extern int send_sig(int, struct task_struct *, int);
2226 extern int zap_other_threads(struct task_struct *p);
2227 extern struct sigqueue *sigqueue_alloc(void);
2228 extern void sigqueue_free(struct sigqueue *);
2229 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2230 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2231 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2233 static inline void restore_saved_sigmask(void)
2235 if (test_and_clear_restore_sigmask())
2236 __set_current_blocked(¤t->saved_sigmask);
2239 static inline sigset_t *sigmask_to_save(void)
2241 sigset_t *res = ¤t->blocked;
2242 if (unlikely(test_restore_sigmask()))
2243 res = ¤t->saved_sigmask;
2247 static inline int kill_cad_pid(int sig, int priv)
2249 return kill_pid(cad_pid, sig, priv);
2252 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2253 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2254 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2255 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2258 * True if we are on the alternate signal stack.
2260 static inline int on_sig_stack(unsigned long sp)
2262 #ifdef CONFIG_STACK_GROWSUP
2263 return sp >= current->sas_ss_sp &&
2264 sp - current->sas_ss_sp < current->sas_ss_size;
2266 return sp > current->sas_ss_sp &&
2267 sp - current->sas_ss_sp <= current->sas_ss_size;
2271 static inline int sas_ss_flags(unsigned long sp)
2273 return (current->sas_ss_size == 0 ? SS_DISABLE
2274 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2278 * Routines for handling mm_structs
2280 extern struct mm_struct * mm_alloc(void);
2282 /* mmdrop drops the mm and the page tables */
2283 extern void __mmdrop(struct mm_struct *);
2284 static inline void mmdrop(struct mm_struct * mm)
2286 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2290 /* mmput gets rid of the mappings and all user-space */
2291 extern void mmput(struct mm_struct *);
2292 /* Grab a reference to a task's mm, if it is not already going away */
2293 extern struct mm_struct *get_task_mm(struct task_struct *task);
2295 * Grab a reference to a task's mm, if it is not already going away
2296 * and ptrace_may_access with the mode parameter passed to it
2299 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2300 /* Remove the current tasks stale references to the old mm_struct */
2301 extern void mm_release(struct task_struct *, struct mm_struct *);
2302 /* Allocate a new mm structure and copy contents from tsk->mm */
2303 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2305 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2306 struct task_struct *);
2307 extern void flush_thread(void);
2308 extern void exit_thread(void);
2310 extern void exit_files(struct task_struct *);
2311 extern void __cleanup_sighand(struct sighand_struct *);
2313 extern void exit_itimers(struct signal_struct *);
2314 extern void flush_itimer_signals(void);
2316 extern void do_group_exit(int);
2318 extern int allow_signal(int);
2319 extern int disallow_signal(int);
2321 extern int do_execve(const char *,
2322 const char __user * const __user *,
2323 const char __user * const __user *);
2324 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2325 struct task_struct *fork_idle(int);
2326 #ifdef CONFIG_GENERIC_KERNEL_THREAD
2327 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2330 extern void set_task_comm(struct task_struct *tsk, char *from);
2331 extern char *get_task_comm(char *to, struct task_struct *tsk);
2334 void scheduler_ipi(void);
2335 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2337 static inline void scheduler_ipi(void) { }
2338 static inline unsigned long wait_task_inactive(struct task_struct *p,
2345 #define next_task(p) \
2346 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2348 #define for_each_process(p) \
2349 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2351 extern bool current_is_single_threaded(void);
2354 * Careful: do_each_thread/while_each_thread is a double loop so
2355 * 'break' will not work as expected - use goto instead.
2357 #define do_each_thread(g, t) \
2358 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2360 #define while_each_thread(g, t) \
2361 while ((t = next_thread(t)) != g)
2363 static inline int get_nr_threads(struct task_struct *tsk)
2365 return tsk->signal->nr_threads;
2368 static inline bool thread_group_leader(struct task_struct *p)
2370 return p->exit_signal >= 0;
2373 /* Do to the insanities of de_thread it is possible for a process
2374 * to have the pid of the thread group leader without actually being
2375 * the thread group leader. For iteration through the pids in proc
2376 * all we care about is that we have a task with the appropriate
2377 * pid, we don't actually care if we have the right task.
2379 static inline int has_group_leader_pid(struct task_struct *p)
2381 return p->pid == p->tgid;
2385 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2387 return p1->tgid == p2->tgid;
2390 static inline struct task_struct *next_thread(const struct task_struct *p)
2392 return list_entry_rcu(p->thread_group.next,
2393 struct task_struct, thread_group);
2396 static inline int thread_group_empty(struct task_struct *p)
2398 return list_empty(&p->thread_group);
2401 #define delay_group_leader(p) \
2402 (thread_group_leader(p) && !thread_group_empty(p))
2405 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2406 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2407 * pins the final release of task.io_context. Also protects ->cpuset and
2408 * ->cgroup.subsys[]. And ->vfork_done.
2410 * Nests both inside and outside of read_lock(&tasklist_lock).
2411 * It must not be nested with write_lock_irq(&tasklist_lock),
2412 * neither inside nor outside.
2414 static inline void task_lock(struct task_struct *p)
2416 spin_lock(&p->alloc_lock);
2419 static inline void task_unlock(struct task_struct *p)
2421 spin_unlock(&p->alloc_lock);
2424 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2425 unsigned long *flags);
2427 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2428 unsigned long *flags)
2430 struct sighand_struct *ret;
2432 ret = __lock_task_sighand(tsk, flags);
2433 (void)__cond_lock(&tsk->sighand->siglock, ret);
2437 static inline void unlock_task_sighand(struct task_struct *tsk,
2438 unsigned long *flags)
2440 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2443 #ifdef CONFIG_CGROUPS
2444 static inline void threadgroup_change_begin(struct task_struct *tsk)
2446 down_read(&tsk->signal->group_rwsem);
2448 static inline void threadgroup_change_end(struct task_struct *tsk)
2450 up_read(&tsk->signal->group_rwsem);
2454 * threadgroup_lock - lock threadgroup
2455 * @tsk: member task of the threadgroup to lock
2457 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2458 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2459 * perform exec. This is useful for cases where the threadgroup needs to
2460 * stay stable across blockable operations.
2462 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2463 * synchronization. While held, no new task will be added to threadgroup
2464 * and no existing live task will have its PF_EXITING set.
2466 * During exec, a task goes and puts its thread group through unusual
2467 * changes. After de-threading, exclusive access is assumed to resources
2468 * which are usually shared by tasks in the same group - e.g. sighand may
2469 * be replaced with a new one. Also, the exec'ing task takes over group
2470 * leader role including its pid. Exclude these changes while locked by
2471 * grabbing cred_guard_mutex which is used to synchronize exec path.
2473 static inline void threadgroup_lock(struct task_struct *tsk)
2476 * exec uses exit for de-threading nesting group_rwsem inside
2477 * cred_guard_mutex. Grab cred_guard_mutex first.
2479 mutex_lock(&tsk->signal->cred_guard_mutex);
2480 down_write(&tsk->signal->group_rwsem);
2484 * threadgroup_unlock - unlock threadgroup
2485 * @tsk: member task of the threadgroup to unlock
2487 * Reverse threadgroup_lock().
2489 static inline void threadgroup_unlock(struct task_struct *tsk)
2491 up_write(&tsk->signal->group_rwsem);
2492 mutex_unlock(&tsk->signal->cred_guard_mutex);
2495 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2496 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2497 static inline void threadgroup_lock(struct task_struct *tsk) {}
2498 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2501 #ifndef __HAVE_THREAD_FUNCTIONS
2503 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2504 #define task_stack_page(task) ((task)->stack)
2506 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2508 *task_thread_info(p) = *task_thread_info(org);
2509 task_thread_info(p)->task = p;
2512 static inline unsigned long *end_of_stack(struct task_struct *p)
2514 return (unsigned long *)(task_thread_info(p) + 1);
2519 static inline int object_is_on_stack(void *obj)
2521 void *stack = task_stack_page(current);
2523 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2526 extern void thread_info_cache_init(void);
2528 #ifdef CONFIG_DEBUG_STACK_USAGE
2529 static inline unsigned long stack_not_used(struct task_struct *p)
2531 unsigned long *n = end_of_stack(p);
2533 do { /* Skip over canary */
2537 return (unsigned long)n - (unsigned long)end_of_stack(p);
2541 /* set thread flags in other task's structures
2542 * - see asm/thread_info.h for TIF_xxxx flags available
2544 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2546 set_ti_thread_flag(task_thread_info(tsk), flag);
2549 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2551 clear_ti_thread_flag(task_thread_info(tsk), flag);
2554 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2556 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2559 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2561 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2564 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2566 return test_ti_thread_flag(task_thread_info(tsk), flag);
2569 static inline void set_tsk_need_resched(struct task_struct *tsk)
2571 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2574 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2576 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2579 static inline int test_tsk_need_resched(struct task_struct *tsk)
2581 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2584 static inline int restart_syscall(void)
2586 set_tsk_thread_flag(current, TIF_SIGPENDING);
2587 return -ERESTARTNOINTR;
2590 static inline int signal_pending(struct task_struct *p)
2592 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2595 static inline int __fatal_signal_pending(struct task_struct *p)
2597 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2600 static inline int fatal_signal_pending(struct task_struct *p)
2602 return signal_pending(p) && __fatal_signal_pending(p);
2605 static inline int signal_pending_state(long state, struct task_struct *p)
2607 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2609 if (!signal_pending(p))
2612 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2615 static inline int need_resched(void)
2617 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2621 * cond_resched() and cond_resched_lock(): latency reduction via
2622 * explicit rescheduling in places that are safe. The return
2623 * value indicates whether a reschedule was done in fact.
2624 * cond_resched_lock() will drop the spinlock before scheduling,
2625 * cond_resched_softirq() will enable bhs before scheduling.
2627 extern int _cond_resched(void);
2629 #define cond_resched() ({ \
2630 __might_sleep(__FILE__, __LINE__, 0); \
2634 extern int __cond_resched_lock(spinlock_t *lock);
2636 #ifdef CONFIG_PREEMPT_COUNT
2637 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2639 #define PREEMPT_LOCK_OFFSET 0
2642 #define cond_resched_lock(lock) ({ \
2643 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2644 __cond_resched_lock(lock); \
2647 extern int __cond_resched_softirq(void);
2649 #define cond_resched_softirq() ({ \
2650 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2651 __cond_resched_softirq(); \
2655 * Does a critical section need to be broken due to another
2656 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2657 * but a general need for low latency)
2659 static inline int spin_needbreak(spinlock_t *lock)
2661 #ifdef CONFIG_PREEMPT
2662 return spin_is_contended(lock);
2669 * Thread group CPU time accounting.
2671 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2672 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2674 static inline void thread_group_cputime_init(struct signal_struct *sig)
2676 raw_spin_lock_init(&sig->cputimer.lock);
2680 * Reevaluate whether the task has signals pending delivery.
2681 * Wake the task if so.
2682 * This is required every time the blocked sigset_t changes.
2683 * callers must hold sighand->siglock.
2685 extern void recalc_sigpending_and_wake(struct task_struct *t);
2686 extern void recalc_sigpending(void);
2688 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2691 * Wrappers for p->thread_info->cpu access. No-op on UP.
2695 static inline unsigned int task_cpu(const struct task_struct *p)
2697 return task_thread_info(p)->cpu;
2700 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2704 static inline unsigned int task_cpu(const struct task_struct *p)
2709 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2713 #endif /* CONFIG_SMP */
2715 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2716 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2718 extern void normalize_rt_tasks(void);
2720 #ifdef CONFIG_CGROUP_SCHED
2722 extern struct task_group root_task_group;
2724 extern struct task_group *sched_create_group(struct task_group *parent);
2725 extern void sched_destroy_group(struct task_group *tg);
2726 extern void sched_move_task(struct task_struct *tsk);
2727 #ifdef CONFIG_FAIR_GROUP_SCHED
2728 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2729 extern unsigned long sched_group_shares(struct task_group *tg);
2731 #ifdef CONFIG_RT_GROUP_SCHED
2732 extern int sched_group_set_rt_runtime(struct task_group *tg,
2733 long rt_runtime_us);
2734 extern long sched_group_rt_runtime(struct task_group *tg);
2735 extern int sched_group_set_rt_period(struct task_group *tg,
2737 extern long sched_group_rt_period(struct task_group *tg);
2738 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2740 #endif /* CONFIG_CGROUP_SCHED */
2742 extern int task_can_switch_user(struct user_struct *up,
2743 struct task_struct *tsk);
2745 #ifdef CONFIG_TASK_XACCT
2746 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2748 tsk->ioac.rchar += amt;
2751 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2753 tsk->ioac.wchar += amt;
2756 static inline void inc_syscr(struct task_struct *tsk)
2761 static inline void inc_syscw(struct task_struct *tsk)
2766 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2770 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2774 static inline void inc_syscr(struct task_struct *tsk)
2778 static inline void inc_syscw(struct task_struct *tsk)
2783 #ifndef TASK_SIZE_OF
2784 #define TASK_SIZE_OF(tsk) TASK_SIZE
2787 #ifdef CONFIG_MM_OWNER
2788 extern void mm_update_next_owner(struct mm_struct *mm);
2789 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2791 static inline void mm_update_next_owner(struct mm_struct *mm)
2795 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2798 #endif /* CONFIG_MM_OWNER */
2800 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2803 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2806 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2809 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2812 static inline unsigned long rlimit(unsigned int limit)
2814 return task_rlimit(current, limit);
2817 static inline unsigned long rlimit_max(unsigned int limit)
2819 return task_rlimit_max(current, limit);