7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 #define CLONE_NEWNET 0x40000000 /* New network namespace */
30 #define CLONE_IO 0x80000000 /* Clone io context */
35 #define SCHED_NORMAL 0
39 /* SCHED_ISO: reserved but not implemented yet */
48 #include <asm/param.h> /* for HZ */
50 #include <linux/capability.h>
51 #include <linux/threads.h>
52 #include <linux/kernel.h>
53 #include <linux/types.h>
54 #include <linux/timex.h>
55 #include <linux/jiffies.h>
56 #include <linux/rbtree.h>
57 #include <linux/thread_info.h>
58 #include <linux/cpumask.h>
59 #include <linux/errno.h>
60 #include <linux/nodemask.h>
61 #include <linux/mm_types.h>
63 #include <asm/system.h>
65 #include <asm/ptrace.h>
66 #include <asm/cputime.h>
68 #include <linux/smp.h>
69 #include <linux/sem.h>
70 #include <linux/signal.h>
71 #include <linux/fs_struct.h>
72 #include <linux/compiler.h>
73 #include <linux/completion.h>
74 #include <linux/pid.h>
75 #include <linux/percpu.h>
76 #include <linux/topology.h>
77 #include <linux/proportions.h>
78 #include <linux/seccomp.h>
79 #include <linux/rcupdate.h>
80 #include <linux/rtmutex.h>
82 #include <linux/time.h>
83 #include <linux/param.h>
84 #include <linux/resource.h>
85 #include <linux/timer.h>
86 #include <linux/hrtimer.h>
87 #include <linux/task_io_accounting.h>
88 #include <linux/kobject.h>
89 #include <linux/latencytop.h>
90 #include <linux/cred.h>
92 #include <asm/processor.h>
96 struct futex_pi_state;
97 struct robust_list_head;
102 * List of flags we want to share for kernel threads,
103 * if only because they are not used by them anyway.
105 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
108 * These are the constant used to fake the fixed-point load-average
109 * counting. Some notes:
110 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
111 * a load-average precision of 10 bits integer + 11 bits fractional
112 * - if you want to count load-averages more often, you need more
113 * precision, or rounding will get you. With 2-second counting freq,
114 * the EXP_n values would be 1981, 2034 and 2043 if still using only
117 extern unsigned long avenrun[]; /* Load averages */
119 #define FSHIFT 11 /* nr of bits of precision */
120 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
121 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
122 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
123 #define EXP_5 2014 /* 1/exp(5sec/5min) */
124 #define EXP_15 2037 /* 1/exp(5sec/15min) */
126 #define CALC_LOAD(load,exp,n) \
128 load += n*(FIXED_1-exp); \
131 extern unsigned long total_forks;
132 extern int nr_threads;
133 DECLARE_PER_CPU(unsigned long, process_counts);
134 extern int nr_processes(void);
135 extern unsigned long nr_running(void);
136 extern unsigned long nr_uninterruptible(void);
137 extern unsigned long nr_active(void);
138 extern unsigned long nr_iowait(void);
143 #ifdef CONFIG_SCHED_DEBUG
144 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
145 extern void proc_sched_set_task(struct task_struct *p);
147 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
150 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
153 static inline void proc_sched_set_task(struct task_struct *p)
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
162 extern unsigned long long time_sync_thresh;
165 * Task state bitmask. NOTE! These bits are also
166 * encoded in fs/proc/array.c: get_task_state().
168 * We have two separate sets of flags: task->state
169 * is about runnability, while task->exit_state are
170 * about the task exiting. Confusing, but this way
171 * modifying one set can't modify the other one by
174 #define TASK_RUNNING 0
175 #define TASK_INTERRUPTIBLE 1
176 #define TASK_UNINTERRUPTIBLE 2
177 #define __TASK_STOPPED 4
178 #define __TASK_TRACED 8
179 /* in tsk->exit_state */
180 #define EXIT_ZOMBIE 16
182 /* in tsk->state again */
184 #define TASK_WAKEKILL 128
186 /* Convenience macros for the sake of set_task_state */
187 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
188 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
189 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
191 /* Convenience macros for the sake of wake_up */
192 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
193 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
195 /* get_task_state() */
196 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
197 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
200 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
201 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
202 #define task_is_stopped_or_traced(task) \
203 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
204 #define task_contributes_to_load(task) \
205 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
207 #define __set_task_state(tsk, state_value) \
208 do { (tsk)->state = (state_value); } while (0)
209 #define set_task_state(tsk, state_value) \
210 set_mb((tsk)->state, (state_value))
213 * set_current_state() includes a barrier so that the write of current->state
214 * is correctly serialised wrt the caller's subsequent test of whether to
217 * set_current_state(TASK_UNINTERRUPTIBLE);
218 * if (do_i_need_to_sleep())
221 * If the caller does not need such serialisation then use __set_current_state()
223 #define __set_current_state(state_value) \
224 do { current->state = (state_value); } while (0)
225 #define set_current_state(state_value) \
226 set_mb(current->state, (state_value))
228 /* Task command name length */
229 #define TASK_COMM_LEN 16
231 #include <linux/spinlock.h>
234 * This serializes "schedule()" and also protects
235 * the run-queue from deletions/modifications (but
236 * _adding_ to the beginning of the run-queue has
239 extern rwlock_t tasklist_lock;
240 extern spinlock_t mmlist_lock;
244 extern void sched_init(void);
245 extern void sched_init_smp(void);
246 extern asmlinkage void schedule_tail(struct task_struct *prev);
247 extern void init_idle(struct task_struct *idle, int cpu);
248 extern void init_idle_bootup_task(struct task_struct *idle);
250 extern int runqueue_is_locked(void);
251 extern void task_rq_unlock_wait(struct task_struct *p);
253 extern cpumask_var_t nohz_cpu_mask;
254 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
255 extern int select_nohz_load_balancer(int cpu);
257 static inline int select_nohz_load_balancer(int cpu)
264 * Only dump TASK_* tasks. (0 for all tasks)
266 extern void show_state_filter(unsigned long state_filter);
268 static inline void show_state(void)
270 show_state_filter(0);
273 extern void show_regs(struct pt_regs *);
276 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
277 * task), SP is the stack pointer of the first frame that should be shown in the back
278 * trace (or NULL if the entire call-chain of the task should be shown).
280 extern void show_stack(struct task_struct *task, unsigned long *sp);
282 void io_schedule(void);
283 long io_schedule_timeout(long timeout);
285 extern void cpu_init (void);
286 extern void trap_init(void);
287 extern void update_process_times(int user);
288 extern void scheduler_tick(void);
290 extern void sched_show_task(struct task_struct *p);
292 #ifdef CONFIG_DETECT_SOFTLOCKUP
293 extern void softlockup_tick(void);
294 extern void touch_softlockup_watchdog(void);
295 extern void touch_all_softlockup_watchdogs(void);
296 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
297 struct file *filp, void __user *buffer,
298 size_t *lenp, loff_t *ppos);
299 extern unsigned int softlockup_panic;
300 extern unsigned long sysctl_hung_task_check_count;
301 extern unsigned long sysctl_hung_task_timeout_secs;
302 extern unsigned long sysctl_hung_task_warnings;
303 extern int softlockup_thresh;
305 static inline void softlockup_tick(void)
308 static inline void spawn_softlockup_task(void)
311 static inline void touch_softlockup_watchdog(void)
314 static inline void touch_all_softlockup_watchdogs(void)
320 /* Attach to any functions which should be ignored in wchan output. */
321 #define __sched __attribute__((__section__(".sched.text")))
323 /* Linker adds these: start and end of __sched functions */
324 extern char __sched_text_start[], __sched_text_end[];
326 /* Is this address in the __sched functions? */
327 extern int in_sched_functions(unsigned long addr);
329 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
330 extern signed long schedule_timeout(signed long timeout);
331 extern signed long schedule_timeout_interruptible(signed long timeout);
332 extern signed long schedule_timeout_killable(signed long timeout);
333 extern signed long schedule_timeout_uninterruptible(signed long timeout);
334 asmlinkage void __schedule(void);
335 asmlinkage void schedule(void);
336 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
339 struct user_namespace;
341 /* Maximum number of active map areas.. This is a random (large) number */
342 #define DEFAULT_MAX_MAP_COUNT 65536
344 extern int sysctl_max_map_count;
346 #include <linux/aio.h>
349 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
350 unsigned long, unsigned long);
352 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
353 unsigned long len, unsigned long pgoff,
354 unsigned long flags);
355 extern void arch_unmap_area(struct mm_struct *, unsigned long);
356 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
358 #if USE_SPLIT_PTLOCKS
360 * The mm counters are not protected by its page_table_lock,
361 * so must be incremented atomically.
363 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
364 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
365 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
366 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
367 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
369 #else /* !USE_SPLIT_PTLOCKS */
371 * The mm counters are protected by its page_table_lock,
372 * so can be incremented directly.
374 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
375 #define get_mm_counter(mm, member) ((mm)->_##member)
376 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
377 #define inc_mm_counter(mm, member) (mm)->_##member++
378 #define dec_mm_counter(mm, member) (mm)->_##member--
380 #endif /* !USE_SPLIT_PTLOCKS */
382 #define get_mm_rss(mm) \
383 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
384 #define update_hiwater_rss(mm) do { \
385 unsigned long _rss = get_mm_rss(mm); \
386 if ((mm)->hiwater_rss < _rss) \
387 (mm)->hiwater_rss = _rss; \
389 #define update_hiwater_vm(mm) do { \
390 if ((mm)->hiwater_vm < (mm)->total_vm) \
391 (mm)->hiwater_vm = (mm)->total_vm; \
394 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
396 return max(mm->hiwater_rss, get_mm_rss(mm));
399 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
401 return max(mm->hiwater_vm, mm->total_vm);
404 extern void set_dumpable(struct mm_struct *mm, int value);
405 extern int get_dumpable(struct mm_struct *mm);
409 #define MMF_DUMPABLE 0 /* core dump is permitted */
410 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
411 #define MMF_DUMPABLE_BITS 2
413 /* coredump filter bits */
414 #define MMF_DUMP_ANON_PRIVATE 2
415 #define MMF_DUMP_ANON_SHARED 3
416 #define MMF_DUMP_MAPPED_PRIVATE 4
417 #define MMF_DUMP_MAPPED_SHARED 5
418 #define MMF_DUMP_ELF_HEADERS 6
419 #define MMF_DUMP_HUGETLB_PRIVATE 7
420 #define MMF_DUMP_HUGETLB_SHARED 8
421 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
422 #define MMF_DUMP_FILTER_BITS 7
423 #define MMF_DUMP_FILTER_MASK \
424 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
425 #define MMF_DUMP_FILTER_DEFAULT \
426 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
427 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
429 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
430 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
432 # define MMF_DUMP_MASK_DEFAULT_ELF 0
435 struct sighand_struct {
437 struct k_sigaction action[_NSIG];
439 wait_queue_head_t signalfd_wqh;
442 struct pacct_struct {
445 unsigned long ac_mem;
446 cputime_t ac_utime, ac_stime;
447 unsigned long ac_minflt, ac_majflt;
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 structure groups together three kinds of CPU time that are
457 * tracked for threads and thread groups. Most things considering
458 * CPU time want to group these counts together and treat all three
459 * of them in parallel.
461 struct task_cputime {
464 unsigned long long sum_exec_runtime;
466 /* Alternate field names when used to cache expirations. */
467 #define prof_exp stime
468 #define virt_exp utime
469 #define sched_exp sum_exec_runtime
471 #define INIT_CPUTIME \
472 (struct task_cputime) { \
473 .utime = cputime_zero, \
474 .stime = cputime_zero, \
475 .sum_exec_runtime = 0, \
479 * struct thread_group_cputimer - thread group interval timer counts
480 * @cputime: thread group interval timers.
481 * @running: non-zero when there are timers running and
482 * @cputime receives updates.
483 * @lock: lock for fields in this struct.
485 * This structure contains the version of task_cputime, above, that is
486 * used for thread group CPU timer calculations.
488 struct thread_group_cputimer {
489 struct task_cputime cputime;
495 * NOTE! "signal_struct" does not have it's own
496 * locking, because a shared signal_struct always
497 * implies a shared sighand_struct, so locking
498 * sighand_struct is always a proper superset of
499 * the locking of signal_struct.
501 struct signal_struct {
505 wait_queue_head_t wait_chldexit; /* for wait4() */
507 /* current thread group signal load-balancing target: */
508 struct task_struct *curr_target;
510 /* shared signal handling: */
511 struct sigpending shared_pending;
513 /* thread group exit support */
516 * - notify group_exit_task when ->count is equal to notify_count
517 * - everyone except group_exit_task is stopped during signal delivery
518 * of fatal signals, group_exit_task processes the signal.
521 struct task_struct *group_exit_task;
523 /* thread group stop support, overloads group_exit_code too */
524 int group_stop_count;
525 unsigned int flags; /* see SIGNAL_* flags below */
527 /* POSIX.1b Interval Timers */
528 struct list_head posix_timers;
530 /* ITIMER_REAL timer for the process */
531 struct hrtimer real_timer;
532 struct pid *leader_pid;
533 ktime_t it_real_incr;
535 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
536 cputime_t it_prof_expires, it_virt_expires;
537 cputime_t it_prof_incr, it_virt_incr;
540 * Thread group totals for process CPU timers.
541 * See thread_group_cputimer(), et al, for details.
543 struct thread_group_cputimer cputimer;
545 /* Earliest-expiration cache. */
546 struct task_cputime cputime_expires;
548 struct list_head cpu_timers[3];
550 /* job control IDs */
553 * pgrp and session fields are deprecated.
554 * use the task_session_Xnr and task_pgrp_Xnr routines below
558 pid_t pgrp __deprecated;
562 struct pid *tty_old_pgrp;
565 pid_t session __deprecated;
569 /* boolean value for session group leader */
572 struct tty_struct *tty; /* NULL if no tty */
575 * Cumulative resource counters for dead threads in the group,
576 * and for reaped dead child processes forked by this group.
577 * Live threads maintain their own counters and add to these
578 * in __exit_signal, except for the group leader.
580 cputime_t utime, stime, cutime, cstime;
583 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
584 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
585 unsigned long inblock, oublock, cinblock, coublock;
586 struct task_io_accounting ioac;
589 * Cumulative ns of schedule CPU time fo dead threads in the
590 * group, not including a zombie group leader, (This only differs
591 * from jiffies_to_ns(utime + stime) if sched_clock uses something
592 * other than jiffies.)
594 unsigned long long sum_sched_runtime;
597 * We don't bother to synchronize most readers of this at all,
598 * because there is no reader checking a limit that actually needs
599 * to get both rlim_cur and rlim_max atomically, and either one
600 * alone is a single word that can safely be read normally.
601 * getrlimit/setrlimit use task_lock(current->group_leader) to
602 * protect this instead of the siglock, because they really
603 * have no need to disable irqs.
605 struct rlimit rlim[RLIM_NLIMITS];
607 #ifdef CONFIG_BSD_PROCESS_ACCT
608 struct pacct_struct pacct; /* per-process accounting information */
610 #ifdef CONFIG_TASKSTATS
611 struct taskstats *stats;
615 struct tty_audit_buf *tty_audit_buf;
619 /* Context switch must be unlocked if interrupts are to be enabled */
620 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
621 # define __ARCH_WANT_UNLOCKED_CTXSW
625 * Bits in flags field of signal_struct.
627 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
628 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
629 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
630 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
632 * Pending notifications to parent.
634 #define SIGNAL_CLD_STOPPED 0x00000010
635 #define SIGNAL_CLD_CONTINUED 0x00000020
636 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
638 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
640 /* If true, all threads except ->group_exit_task have pending SIGKILL */
641 static inline int signal_group_exit(const struct signal_struct *sig)
643 return (sig->flags & SIGNAL_GROUP_EXIT) ||
644 (sig->group_exit_task != NULL);
648 * Some day this will be a full-fledged user tracking system..
651 atomic_t __count; /* reference count */
652 atomic_t processes; /* How many processes does this user have? */
653 atomic_t files; /* How many open files does this user have? */
654 atomic_t sigpending; /* How many pending signals does this user have? */
655 #ifdef CONFIG_INOTIFY_USER
656 atomic_t inotify_watches; /* How many inotify watches does this user have? */
657 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
660 atomic_t epoll_watches; /* The number of file descriptors currently watched */
662 #ifdef CONFIG_POSIX_MQUEUE
663 /* protected by mq_lock */
664 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
666 unsigned long locked_shm; /* How many pages of mlocked shm ? */
669 struct key *uid_keyring; /* UID specific keyring */
670 struct key *session_keyring; /* UID's default session keyring */
673 /* Hash table maintenance information */
674 struct hlist_node uidhash_node;
676 struct user_namespace *user_ns;
678 #ifdef CONFIG_USER_SCHED
679 struct task_group *tg;
682 struct work_struct work;
687 extern int uids_sysfs_init(void);
689 extern struct user_struct *find_user(uid_t);
691 extern struct user_struct root_user;
692 #define INIT_USER (&root_user)
695 struct backing_dev_info;
696 struct reclaim_state;
698 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
700 /* cumulative counters */
701 unsigned long pcount; /* # of times run on this cpu */
702 unsigned long long run_delay; /* time spent waiting on a runqueue */
705 unsigned long long last_arrival,/* when we last ran on a cpu */
706 last_queued; /* when we were last queued to run */
707 #ifdef CONFIG_SCHEDSTATS
709 unsigned int bkl_count;
712 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
714 #ifdef CONFIG_TASK_DELAY_ACCT
715 struct task_delay_info {
717 unsigned int flags; /* Private per-task flags */
719 /* For each stat XXX, add following, aligned appropriately
721 * struct timespec XXX_start, XXX_end;
725 * Atomicity of updates to XXX_delay, XXX_count protected by
726 * single lock above (split into XXX_lock if contention is an issue).
730 * XXX_count is incremented on every XXX operation, the delay
731 * associated with the operation is added to XXX_delay.
732 * XXX_delay contains the accumulated delay time in nanoseconds.
734 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
735 u64 blkio_delay; /* wait for sync block io completion */
736 u64 swapin_delay; /* wait for swapin block io completion */
737 u32 blkio_count; /* total count of the number of sync block */
738 /* io operations performed */
739 u32 swapin_count; /* total count of the number of swapin block */
740 /* io operations performed */
742 struct timespec freepages_start, freepages_end;
743 u64 freepages_delay; /* wait for memory reclaim */
744 u32 freepages_count; /* total count of memory reclaim */
746 #endif /* CONFIG_TASK_DELAY_ACCT */
748 static inline int sched_info_on(void)
750 #ifdef CONFIG_SCHEDSTATS
752 #elif defined(CONFIG_TASK_DELAY_ACCT)
753 extern int delayacct_on;
768 * sched-domains (multiprocessor balancing) declarations:
772 * Increase resolution of nice-level calculations:
774 #define SCHED_LOAD_SHIFT 10
775 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
777 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
780 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
781 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
782 #define SD_BALANCE_EXEC 4 /* Balance on exec */
783 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
784 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
785 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
786 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
787 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
788 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
789 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
790 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
791 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
793 enum powersavings_balance_level {
794 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
795 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
796 * first for long running threads
798 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
799 * cpu package for power savings
801 MAX_POWERSAVINGS_BALANCE_LEVELS
804 extern int sched_mc_power_savings, sched_smt_power_savings;
806 static inline int sd_balance_for_mc_power(void)
808 if (sched_smt_power_savings)
809 return SD_POWERSAVINGS_BALANCE;
814 static inline int sd_balance_for_package_power(void)
816 if (sched_mc_power_savings | sched_smt_power_savings)
817 return SD_POWERSAVINGS_BALANCE;
823 * Optimise SD flags for power savings:
824 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
825 * Keep default SD flags if sched_{smt,mc}_power_saving=0
828 static inline int sd_power_saving_flags(void)
830 if (sched_mc_power_savings | sched_smt_power_savings)
831 return SD_BALANCE_NEWIDLE;
837 struct sched_group *next; /* Must be a circular list */
840 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
841 * single CPU. This is read only (except for setup, hotplug CPU).
842 * Note : Never change cpu_power without recompute its reciprocal
844 unsigned int __cpu_power;
846 * reciprocal value of cpu_power to avoid expensive divides
847 * (see include/linux/reciprocal_div.h)
849 u32 reciprocal_cpu_power;
851 unsigned long cpumask[];
854 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
856 return to_cpumask(sg->cpumask);
859 enum sched_domain_level {
869 struct sched_domain_attr {
870 int relax_domain_level;
873 #define SD_ATTR_INIT (struct sched_domain_attr) { \
874 .relax_domain_level = -1, \
877 struct sched_domain {
878 /* These fields must be setup */
879 struct sched_domain *parent; /* top domain must be null terminated */
880 struct sched_domain *child; /* bottom domain must be null terminated */
881 struct sched_group *groups; /* the balancing groups of the domain */
882 unsigned long min_interval; /* Minimum balance interval ms */
883 unsigned long max_interval; /* Maximum balance interval ms */
884 unsigned int busy_factor; /* less balancing by factor if busy */
885 unsigned int imbalance_pct; /* No balance until over watermark */
886 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
887 unsigned int busy_idx;
888 unsigned int idle_idx;
889 unsigned int newidle_idx;
890 unsigned int wake_idx;
891 unsigned int forkexec_idx;
892 int flags; /* See SD_* */
893 enum sched_domain_level level;
895 /* Runtime fields. */
896 unsigned long last_balance; /* init to jiffies. units in jiffies */
897 unsigned int balance_interval; /* initialise to 1. units in ms. */
898 unsigned int nr_balance_failed; /* initialise to 0 */
902 #ifdef CONFIG_SCHEDSTATS
903 /* load_balance() stats */
904 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
905 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
906 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
907 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
908 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
909 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
910 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
911 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
913 /* Active load balancing */
914 unsigned int alb_count;
915 unsigned int alb_failed;
916 unsigned int alb_pushed;
918 /* SD_BALANCE_EXEC stats */
919 unsigned int sbe_count;
920 unsigned int sbe_balanced;
921 unsigned int sbe_pushed;
923 /* SD_BALANCE_FORK stats */
924 unsigned int sbf_count;
925 unsigned int sbf_balanced;
926 unsigned int sbf_pushed;
928 /* try_to_wake_up() stats */
929 unsigned int ttwu_wake_remote;
930 unsigned int ttwu_move_affine;
931 unsigned int ttwu_move_balance;
933 #ifdef CONFIG_SCHED_DEBUG
937 /* span of all CPUs in this domain */
938 unsigned long span[];
941 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
943 return to_cpumask(sd->span);
946 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
947 struct sched_domain_attr *dattr_new);
949 /* Test a flag in parent sched domain */
950 static inline int test_sd_parent(struct sched_domain *sd, int flag)
952 if (sd->parent && (sd->parent->flags & flag))
958 #else /* CONFIG_SMP */
960 struct sched_domain_attr;
963 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
964 struct sched_domain_attr *dattr_new)
967 #endif /* !CONFIG_SMP */
969 struct io_context; /* See blkdev.h */
972 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
973 extern void prefetch_stack(struct task_struct *t);
975 static inline void prefetch_stack(struct task_struct *t) { }
978 struct audit_context; /* See audit.c */
980 struct pipe_inode_info;
981 struct uts_namespace;
987 const struct sched_class *next;
989 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
990 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
991 void (*yield_task) (struct rq *rq);
993 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
995 struct task_struct * (*pick_next_task) (struct rq *rq);
996 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
999 int (*select_task_rq)(struct task_struct *p, int sync);
1001 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1002 struct rq *busiest, unsigned long max_load_move,
1003 struct sched_domain *sd, enum cpu_idle_type idle,
1004 int *all_pinned, int *this_best_prio);
1006 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1007 struct rq *busiest, struct sched_domain *sd,
1008 enum cpu_idle_type idle);
1009 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1010 int (*needs_post_schedule) (struct rq *this_rq);
1011 void (*post_schedule) (struct rq *this_rq);
1012 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1014 void (*set_cpus_allowed)(struct task_struct *p,
1015 const struct cpumask *newmask);
1017 void (*rq_online)(struct rq *rq);
1018 void (*rq_offline)(struct rq *rq);
1021 void (*set_curr_task) (struct rq *rq);
1022 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1023 void (*task_new) (struct rq *rq, struct task_struct *p);
1025 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1027 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1029 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1030 int oldprio, int running);
1032 #ifdef CONFIG_FAIR_GROUP_SCHED
1033 void (*moved_group) (struct task_struct *p);
1037 struct load_weight {
1038 unsigned long weight, inv_weight;
1042 * CFS stats for a schedulable entity (task, task-group etc)
1044 * Current field usage histogram:
1051 struct sched_entity {
1052 struct load_weight load; /* for load-balancing */
1053 struct rb_node run_node;
1054 struct list_head group_node;
1058 u64 sum_exec_runtime;
1060 u64 prev_sum_exec_runtime;
1069 #ifdef CONFIG_SCHEDSTATS
1077 s64 sum_sleep_runtime;
1084 u64 nr_migrations_cold;
1085 u64 nr_failed_migrations_affine;
1086 u64 nr_failed_migrations_running;
1087 u64 nr_failed_migrations_hot;
1088 u64 nr_forced_migrations;
1089 u64 nr_forced2_migrations;
1092 u64 nr_wakeups_sync;
1093 u64 nr_wakeups_migrate;
1094 u64 nr_wakeups_local;
1095 u64 nr_wakeups_remote;
1096 u64 nr_wakeups_affine;
1097 u64 nr_wakeups_affine_attempts;
1098 u64 nr_wakeups_passive;
1099 u64 nr_wakeups_idle;
1102 #ifdef CONFIG_FAIR_GROUP_SCHED
1103 struct sched_entity *parent;
1104 /* rq on which this entity is (to be) queued: */
1105 struct cfs_rq *cfs_rq;
1106 /* rq "owned" by this entity/group: */
1107 struct cfs_rq *my_q;
1111 struct sched_rt_entity {
1112 struct list_head run_list;
1113 unsigned long timeout;
1114 unsigned int time_slice;
1115 int nr_cpus_allowed;
1117 struct sched_rt_entity *back;
1118 #ifdef CONFIG_RT_GROUP_SCHED
1119 struct sched_rt_entity *parent;
1120 /* rq on which this entity is (to be) queued: */
1121 struct rt_rq *rt_rq;
1122 /* rq "owned" by this entity/group: */
1127 struct task_struct {
1128 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1131 unsigned int flags; /* per process flags, defined below */
1132 unsigned int ptrace;
1134 int lock_depth; /* BKL lock depth */
1137 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1142 int prio, static_prio, normal_prio;
1143 unsigned int rt_priority;
1144 const struct sched_class *sched_class;
1145 struct sched_entity se;
1146 struct sched_rt_entity rt;
1148 #ifdef CONFIG_PREEMPT_NOTIFIERS
1149 /* list of struct preempt_notifier: */
1150 struct hlist_head preempt_notifiers;
1154 * fpu_counter contains the number of consecutive context switches
1155 * that the FPU is used. If this is over a threshold, the lazy fpu
1156 * saving becomes unlazy to save the trap. This is an unsigned char
1157 * so that after 256 times the counter wraps and the behavior turns
1158 * lazy again; this to deal with bursty apps that only use FPU for
1161 unsigned char fpu_counter;
1162 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1163 #ifdef CONFIG_BLK_DEV_IO_TRACE
1164 unsigned int btrace_seq;
1167 unsigned int policy;
1168 cpumask_t cpus_allowed;
1170 #ifdef CONFIG_PREEMPT_RCU
1171 int rcu_read_lock_nesting;
1172 int rcu_flipctr_idx;
1173 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1175 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1176 struct sched_info sched_info;
1179 struct list_head tasks;
1180 struct plist_node pushable_tasks;
1182 struct mm_struct *mm, *active_mm;
1185 struct linux_binfmt *binfmt;
1187 int exit_code, exit_signal;
1188 int pdeath_signal; /* The signal sent when the parent dies */
1190 unsigned int personality;
1191 unsigned did_exec:1;
1192 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1197 /* Canary value for the -fstack-protector gcc feature */
1198 unsigned long stack_canary;
1201 * pointers to (original) parent process, youngest child, younger sibling,
1202 * older sibling, respectively. (p->father can be replaced with
1203 * p->real_parent->pid)
1205 struct task_struct *real_parent; /* real parent process */
1206 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1208 * children/sibling forms the list of my natural children
1210 struct list_head children; /* list of my children */
1211 struct list_head sibling; /* linkage in my parent's children list */
1212 struct task_struct *group_leader; /* threadgroup leader */
1215 * ptraced is the list of tasks this task is using ptrace on.
1216 * This includes both natural children and PTRACE_ATTACH targets.
1217 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1219 struct list_head ptraced;
1220 struct list_head ptrace_entry;
1222 #ifdef CONFIG_X86_PTRACE_BTS
1224 * This is the tracer handle for the ptrace BTS extension.
1225 * This field actually belongs to the ptracer task.
1227 struct bts_tracer *bts;
1229 * The buffer to hold the BTS data.
1233 #endif /* CONFIG_X86_PTRACE_BTS */
1235 /* PID/PID hash table linkage. */
1236 struct pid_link pids[PIDTYPE_MAX];
1237 struct list_head thread_group;
1239 struct completion *vfork_done; /* for vfork() */
1240 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1241 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1243 cputime_t utime, stime, utimescaled, stimescaled;
1245 cputime_t prev_utime, prev_stime;
1246 unsigned long nvcsw, nivcsw; /* context switch counts */
1247 struct timespec start_time; /* monotonic time */
1248 struct timespec real_start_time; /* boot based time */
1249 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1250 unsigned long min_flt, maj_flt;
1252 struct task_cputime cputime_expires;
1253 struct list_head cpu_timers[3];
1255 /* process credentials */
1256 const struct cred *real_cred; /* objective and real subjective task
1257 * credentials (COW) */
1258 const struct cred *cred; /* effective (overridable) subjective task
1259 * credentials (COW) */
1260 struct mutex cred_exec_mutex; /* execve vs ptrace cred calculation mutex */
1262 char comm[TASK_COMM_LEN]; /* executable name excluding path
1263 - access with [gs]et_task_comm (which lock
1264 it with task_lock())
1265 - initialized normally by flush_old_exec */
1266 /* file system info */
1267 int link_count, total_link_count;
1268 #ifdef CONFIG_SYSVIPC
1270 struct sysv_sem sysvsem;
1272 #ifdef CONFIG_DETECT_SOFTLOCKUP
1273 /* hung task detection */
1274 unsigned long last_switch_timestamp;
1275 unsigned long last_switch_count;
1277 /* CPU-specific state of this task */
1278 struct thread_struct thread;
1279 /* filesystem information */
1280 struct fs_struct *fs;
1281 /* open file information */
1282 struct files_struct *files;
1284 struct nsproxy *nsproxy;
1285 /* signal handlers */
1286 struct signal_struct *signal;
1287 struct sighand_struct *sighand;
1289 sigset_t blocked, real_blocked;
1290 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1291 struct sigpending pending;
1293 unsigned long sas_ss_sp;
1295 int (*notifier)(void *priv);
1296 void *notifier_data;
1297 sigset_t *notifier_mask;
1298 struct audit_context *audit_context;
1299 #ifdef CONFIG_AUDITSYSCALL
1301 unsigned int sessionid;
1305 /* Thread group tracking */
1308 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1309 spinlock_t alloc_lock;
1311 /* Protection of the PI data structures: */
1314 #ifdef CONFIG_RT_MUTEXES
1315 /* PI waiters blocked on a rt_mutex held by this task */
1316 struct plist_head pi_waiters;
1317 /* Deadlock detection and priority inheritance handling */
1318 struct rt_mutex_waiter *pi_blocked_on;
1321 #ifdef CONFIG_DEBUG_MUTEXES
1322 /* mutex deadlock detection */
1323 struct mutex_waiter *blocked_on;
1325 #ifdef CONFIG_TRACE_IRQFLAGS
1326 unsigned int irq_events;
1327 int hardirqs_enabled;
1328 unsigned long hardirq_enable_ip;
1329 unsigned int hardirq_enable_event;
1330 unsigned long hardirq_disable_ip;
1331 unsigned int hardirq_disable_event;
1332 int softirqs_enabled;
1333 unsigned long softirq_disable_ip;
1334 unsigned int softirq_disable_event;
1335 unsigned long softirq_enable_ip;
1336 unsigned int softirq_enable_event;
1337 int hardirq_context;
1338 int softirq_context;
1340 #ifdef CONFIG_LOCKDEP
1341 # define MAX_LOCK_DEPTH 48UL
1344 unsigned int lockdep_recursion;
1345 struct held_lock held_locks[MAX_LOCK_DEPTH];
1346 gfp_t lockdep_reclaim_gfp;
1349 /* journalling filesystem info */
1352 /* stacked block device info */
1353 struct bio *bio_list, **bio_tail;
1356 struct reclaim_state *reclaim_state;
1358 struct backing_dev_info *backing_dev_info;
1360 struct io_context *io_context;
1362 unsigned long ptrace_message;
1363 siginfo_t *last_siginfo; /* For ptrace use. */
1364 struct task_io_accounting ioac;
1365 #if defined(CONFIG_TASK_XACCT)
1366 u64 acct_rss_mem1; /* accumulated rss usage */
1367 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1368 cputime_t acct_timexpd; /* stime + utime since last update */
1370 #ifdef CONFIG_CPUSETS
1371 nodemask_t mems_allowed;
1372 int cpuset_mems_generation;
1373 int cpuset_mem_spread_rotor;
1375 #ifdef CONFIG_CGROUPS
1376 /* Control Group info protected by css_set_lock */
1377 struct css_set *cgroups;
1378 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1379 struct list_head cg_list;
1382 struct robust_list_head __user *robust_list;
1383 #ifdef CONFIG_COMPAT
1384 struct compat_robust_list_head __user *compat_robust_list;
1386 struct list_head pi_state_list;
1387 struct futex_pi_state *pi_state_cache;
1390 struct mempolicy *mempolicy;
1393 atomic_t fs_excl; /* holding fs exclusive resources */
1394 struct rcu_head rcu;
1397 * cache last used pipe for splice
1399 struct pipe_inode_info *splice_pipe;
1400 #ifdef CONFIG_TASK_DELAY_ACCT
1401 struct task_delay_info *delays;
1403 #ifdef CONFIG_FAULT_INJECTION
1406 struct prop_local_single dirties;
1407 #ifdef CONFIG_LATENCYTOP
1408 int latency_record_count;
1409 struct latency_record latency_record[LT_SAVECOUNT];
1412 * time slack values; these are used to round up poll() and
1413 * select() etc timeout values. These are in nanoseconds.
1415 unsigned long timer_slack_ns;
1416 unsigned long default_timer_slack_ns;
1418 struct list_head *scm_work_list;
1419 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1420 /* Index of current stored adress in ret_stack */
1422 /* Stack of return addresses for return function tracing */
1423 struct ftrace_ret_stack *ret_stack;
1425 * Number of functions that haven't been traced
1426 * because of depth overrun.
1428 atomic_t trace_overrun;
1429 /* Pause for the tracing */
1430 atomic_t tracing_graph_pause;
1432 #ifdef CONFIG_TRACING
1433 /* state flags for use by tracers */
1434 unsigned long trace;
1438 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1439 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1442 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1443 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1444 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1445 * values are inverted: lower p->prio value means higher priority.
1447 * The MAX_USER_RT_PRIO value allows the actual maximum
1448 * RT priority to be separate from the value exported to
1449 * user-space. This allows kernel threads to set their
1450 * priority to a value higher than any user task. Note:
1451 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1454 #define MAX_USER_RT_PRIO 100
1455 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1457 #define MAX_PRIO (MAX_RT_PRIO + 40)
1458 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1460 static inline int rt_prio(int prio)
1462 if (unlikely(prio < MAX_RT_PRIO))
1467 static inline int rt_task(struct task_struct *p)
1469 return rt_prio(p->prio);
1472 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1474 tsk->signal->__session = session;
1477 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1479 tsk->signal->__pgrp = pgrp;
1482 static inline struct pid *task_pid(struct task_struct *task)
1484 return task->pids[PIDTYPE_PID].pid;
1487 static inline struct pid *task_tgid(struct task_struct *task)
1489 return task->group_leader->pids[PIDTYPE_PID].pid;
1493 * Without tasklist or rcu lock it is not safe to dereference
1494 * the result of task_pgrp/task_session even if task == current,
1495 * we can race with another thread doing sys_setsid/sys_setpgid.
1497 static inline struct pid *task_pgrp(struct task_struct *task)
1499 return task->group_leader->pids[PIDTYPE_PGID].pid;
1502 static inline struct pid *task_session(struct task_struct *task)
1504 return task->group_leader->pids[PIDTYPE_SID].pid;
1507 struct pid_namespace;
1510 * the helpers to get the task's different pids as they are seen
1511 * from various namespaces
1513 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1514 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1516 * task_xid_nr_ns() : id seen from the ns specified;
1518 * set_task_vxid() : assigns a virtual id to a task;
1520 * see also pid_nr() etc in include/linux/pid.h
1522 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1523 struct pid_namespace *ns);
1525 static inline pid_t task_pid_nr(struct task_struct *tsk)
1530 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1531 struct pid_namespace *ns)
1533 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1536 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1538 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1542 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1547 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1549 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1551 return pid_vnr(task_tgid(tsk));
1555 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1557 return tsk->signal->__pgrp;
1560 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1561 struct pid_namespace *ns)
1563 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1566 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1568 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1572 static inline pid_t task_session_nr(struct task_struct *tsk)
1574 return tsk->signal->__session;
1577 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1578 struct pid_namespace *ns)
1580 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1583 static inline pid_t task_session_vnr(struct task_struct *tsk)
1585 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1589 * pid_alive - check that a task structure is not stale
1590 * @p: Task structure to be checked.
1592 * Test if a process is not yet dead (at most zombie state)
1593 * If pid_alive fails, then pointers within the task structure
1594 * can be stale and must not be dereferenced.
1596 static inline int pid_alive(struct task_struct *p)
1598 return p->pids[PIDTYPE_PID].pid != NULL;
1602 * is_global_init - check if a task structure is init
1603 * @tsk: Task structure to be checked.
1605 * Check if a task structure is the first user space task the kernel created.
1607 static inline int is_global_init(struct task_struct *tsk)
1609 return tsk->pid == 1;
1613 * is_container_init:
1614 * check whether in the task is init in its own pid namespace.
1616 extern int is_container_init(struct task_struct *tsk);
1618 extern struct pid *cad_pid;
1620 extern void free_task(struct task_struct *tsk);
1621 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1623 extern void __put_task_struct(struct task_struct *t);
1625 static inline void put_task_struct(struct task_struct *t)
1627 if (atomic_dec_and_test(&t->usage))
1628 __put_task_struct(t);
1631 extern cputime_t task_utime(struct task_struct *p);
1632 extern cputime_t task_stime(struct task_struct *p);
1633 extern cputime_t task_gtime(struct task_struct *p);
1638 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1639 /* Not implemented yet, only for 486*/
1640 #define PF_STARTING 0x00000002 /* being created */
1641 #define PF_EXITING 0x00000004 /* getting shut down */
1642 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1643 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1644 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1645 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1646 #define PF_DUMPCORE 0x00000200 /* dumped core */
1647 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1648 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1649 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1650 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1651 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1652 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1653 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1654 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1655 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1656 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1657 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1658 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1659 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1660 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1661 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1662 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1663 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1664 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1665 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1666 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1669 * Only the _current_ task can read/write to tsk->flags, but other
1670 * tasks can access tsk->flags in readonly mode for example
1671 * with tsk_used_math (like during threaded core dumping).
1672 * There is however an exception to this rule during ptrace
1673 * or during fork: the ptracer task is allowed to write to the
1674 * child->flags of its traced child (same goes for fork, the parent
1675 * can write to the child->flags), because we're guaranteed the
1676 * child is not running and in turn not changing child->flags
1677 * at the same time the parent does it.
1679 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1680 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1681 #define clear_used_math() clear_stopped_child_used_math(current)
1682 #define set_used_math() set_stopped_child_used_math(current)
1683 #define conditional_stopped_child_used_math(condition, child) \
1684 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1685 #define conditional_used_math(condition) \
1686 conditional_stopped_child_used_math(condition, current)
1687 #define copy_to_stopped_child_used_math(child) \
1688 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1689 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1690 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1691 #define used_math() tsk_used_math(current)
1694 extern int set_cpus_allowed_ptr(struct task_struct *p,
1695 const struct cpumask *new_mask);
1697 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1698 const struct cpumask *new_mask)
1700 if (!cpumask_test_cpu(0, new_mask))
1705 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1707 return set_cpus_allowed_ptr(p, &new_mask);
1711 * Architectures can set this to 1 if they have specified
1712 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1713 * but then during bootup it turns out that sched_clock()
1714 * is reliable after all:
1716 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1717 extern int sched_clock_stable;
1720 extern unsigned long long sched_clock(void);
1722 extern void sched_clock_init(void);
1723 extern u64 sched_clock_cpu(int cpu);
1725 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1726 static inline void sched_clock_tick(void)
1730 static inline void sched_clock_idle_sleep_event(void)
1734 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1738 extern void sched_clock_tick(void);
1739 extern void sched_clock_idle_sleep_event(void);
1740 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1744 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1745 * clock constructed from sched_clock():
1747 extern unsigned long long cpu_clock(int cpu);
1749 extern unsigned long long
1750 task_sched_runtime(struct task_struct *task);
1751 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1753 /* sched_exec is called by processes performing an exec */
1755 extern void sched_exec(void);
1757 #define sched_exec() {}
1760 extern void sched_clock_idle_sleep_event(void);
1761 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1763 #ifdef CONFIG_HOTPLUG_CPU
1764 extern void idle_task_exit(void);
1766 static inline void idle_task_exit(void) {}
1769 extern void sched_idle_next(void);
1771 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1772 extern void wake_up_idle_cpu(int cpu);
1774 static inline void wake_up_idle_cpu(int cpu) { }
1777 extern unsigned int sysctl_sched_latency;
1778 extern unsigned int sysctl_sched_min_granularity;
1779 extern unsigned int sysctl_sched_wakeup_granularity;
1780 extern unsigned int sysctl_sched_shares_ratelimit;
1781 extern unsigned int sysctl_sched_shares_thresh;
1782 #ifdef CONFIG_SCHED_DEBUG
1783 extern unsigned int sysctl_sched_child_runs_first;
1784 extern unsigned int sysctl_sched_features;
1785 extern unsigned int sysctl_sched_migration_cost;
1786 extern unsigned int sysctl_sched_nr_migrate;
1788 int sched_nr_latency_handler(struct ctl_table *table, int write,
1789 struct file *file, void __user *buffer, size_t *length,
1792 extern unsigned int sysctl_sched_rt_period;
1793 extern int sysctl_sched_rt_runtime;
1795 int sched_rt_handler(struct ctl_table *table, int write,
1796 struct file *filp, void __user *buffer, size_t *lenp,
1799 extern unsigned int sysctl_sched_compat_yield;
1801 #ifdef CONFIG_RT_MUTEXES
1802 extern int rt_mutex_getprio(struct task_struct *p);
1803 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1804 extern void rt_mutex_adjust_pi(struct task_struct *p);
1806 static inline int rt_mutex_getprio(struct task_struct *p)
1808 return p->normal_prio;
1810 # define rt_mutex_adjust_pi(p) do { } while (0)
1813 extern void set_user_nice(struct task_struct *p, long nice);
1814 extern int task_prio(const struct task_struct *p);
1815 extern int task_nice(const struct task_struct *p);
1816 extern int can_nice(const struct task_struct *p, const int nice);
1817 extern int task_curr(const struct task_struct *p);
1818 extern int idle_cpu(int cpu);
1819 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1820 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1821 struct sched_param *);
1822 extern struct task_struct *idle_task(int cpu);
1823 extern struct task_struct *curr_task(int cpu);
1824 extern void set_curr_task(int cpu, struct task_struct *p);
1829 * The default (Linux) execution domain.
1831 extern struct exec_domain default_exec_domain;
1833 union thread_union {
1834 struct thread_info thread_info;
1835 unsigned long stack[THREAD_SIZE/sizeof(long)];
1838 #ifndef __HAVE_ARCH_KSTACK_END
1839 static inline int kstack_end(void *addr)
1841 /* Reliable end of stack detection:
1842 * Some APM bios versions misalign the stack
1844 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1848 extern union thread_union init_thread_union;
1849 extern struct task_struct init_task;
1851 extern struct mm_struct init_mm;
1853 extern struct pid_namespace init_pid_ns;
1856 * find a task by one of its numerical ids
1858 * find_task_by_pid_type_ns():
1859 * it is the most generic call - it finds a task by all id,
1860 * type and namespace specified
1861 * find_task_by_pid_ns():
1862 * finds a task by its pid in the specified namespace
1863 * find_task_by_vpid():
1864 * finds a task by its virtual pid
1866 * see also find_vpid() etc in include/linux/pid.h
1869 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1870 struct pid_namespace *ns);
1872 extern struct task_struct *find_task_by_vpid(pid_t nr);
1873 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1874 struct pid_namespace *ns);
1876 extern void __set_special_pids(struct pid *pid);
1878 /* per-UID process charging. */
1879 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1880 static inline struct user_struct *get_uid(struct user_struct *u)
1882 atomic_inc(&u->__count);
1885 extern void free_uid(struct user_struct *);
1886 extern void release_uids(struct user_namespace *ns);
1888 #include <asm/current.h>
1890 extern void do_timer(unsigned long ticks);
1892 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1893 extern int wake_up_process(struct task_struct *tsk);
1894 extern void wake_up_new_task(struct task_struct *tsk,
1895 unsigned long clone_flags);
1897 extern void kick_process(struct task_struct *tsk);
1899 static inline void kick_process(struct task_struct *tsk) { }
1901 extern void sched_fork(struct task_struct *p, int clone_flags);
1902 extern void sched_dead(struct task_struct *p);
1904 extern void proc_caches_init(void);
1905 extern void flush_signals(struct task_struct *);
1906 extern void ignore_signals(struct task_struct *);
1907 extern void flush_signal_handlers(struct task_struct *, int force_default);
1908 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1910 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1912 unsigned long flags;
1915 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1916 ret = dequeue_signal(tsk, mask, info);
1917 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1922 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1924 extern void unblock_all_signals(void);
1925 extern void release_task(struct task_struct * p);
1926 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1927 extern int force_sigsegv(int, struct task_struct *);
1928 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1929 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1930 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1931 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1932 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1933 extern int kill_pid(struct pid *pid, int sig, int priv);
1934 extern int kill_proc_info(int, struct siginfo *, pid_t);
1935 extern int do_notify_parent(struct task_struct *, int);
1936 extern void force_sig(int, struct task_struct *);
1937 extern void force_sig_specific(int, struct task_struct *);
1938 extern int send_sig(int, struct task_struct *, int);
1939 extern void zap_other_threads(struct task_struct *p);
1940 extern struct sigqueue *sigqueue_alloc(void);
1941 extern void sigqueue_free(struct sigqueue *);
1942 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1943 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1944 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1946 static inline int kill_cad_pid(int sig, int priv)
1948 return kill_pid(cad_pid, sig, priv);
1951 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1952 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1953 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1954 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1956 static inline int is_si_special(const struct siginfo *info)
1958 return info <= SEND_SIG_FORCED;
1961 /* True if we are on the alternate signal stack. */
1963 static inline int on_sig_stack(unsigned long sp)
1965 return (sp - current->sas_ss_sp < current->sas_ss_size);
1968 static inline int sas_ss_flags(unsigned long sp)
1970 return (current->sas_ss_size == 0 ? SS_DISABLE
1971 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1975 * Routines for handling mm_structs
1977 extern struct mm_struct * mm_alloc(void);
1979 /* mmdrop drops the mm and the page tables */
1980 extern void __mmdrop(struct mm_struct *);
1981 static inline void mmdrop(struct mm_struct * mm)
1983 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1987 /* mmput gets rid of the mappings and all user-space */
1988 extern void mmput(struct mm_struct *);
1989 /* Grab a reference to a task's mm, if it is not already going away */
1990 extern struct mm_struct *get_task_mm(struct task_struct *task);
1991 /* Remove the current tasks stale references to the old mm_struct */
1992 extern void mm_release(struct task_struct *, struct mm_struct *);
1993 /* Allocate a new mm structure and copy contents from tsk->mm */
1994 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1996 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1997 struct task_struct *, struct pt_regs *);
1998 extern void flush_thread(void);
1999 extern void exit_thread(void);
2001 extern void exit_files(struct task_struct *);
2002 extern void __cleanup_signal(struct signal_struct *);
2003 extern void __cleanup_sighand(struct sighand_struct *);
2005 extern void exit_itimers(struct signal_struct *);
2006 extern void flush_itimer_signals(void);
2008 extern NORET_TYPE void do_group_exit(int);
2010 extern void daemonize(const char *, ...);
2011 extern int allow_signal(int);
2012 extern int disallow_signal(int);
2014 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2015 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2016 struct task_struct *fork_idle(int);
2018 extern void set_task_comm(struct task_struct *tsk, char *from);
2019 extern char *get_task_comm(char *to, struct task_struct *tsk);
2022 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2024 static inline unsigned long wait_task_inactive(struct task_struct *p,
2031 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2033 #define for_each_process(p) \
2034 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2036 extern bool is_single_threaded(struct task_struct *);
2039 * Careful: do_each_thread/while_each_thread is a double loop so
2040 * 'break' will not work as expected - use goto instead.
2042 #define do_each_thread(g, t) \
2043 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2045 #define while_each_thread(g, t) \
2046 while ((t = next_thread(t)) != g)
2048 /* de_thread depends on thread_group_leader not being a pid based check */
2049 #define thread_group_leader(p) (p == p->group_leader)
2051 /* Do to the insanities of de_thread it is possible for a process
2052 * to have the pid of the thread group leader without actually being
2053 * the thread group leader. For iteration through the pids in proc
2054 * all we care about is that we have a task with the appropriate
2055 * pid, we don't actually care if we have the right task.
2057 static inline int has_group_leader_pid(struct task_struct *p)
2059 return p->pid == p->tgid;
2063 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2065 return p1->tgid == p2->tgid;
2068 static inline struct task_struct *next_thread(const struct task_struct *p)
2070 return list_entry(rcu_dereference(p->thread_group.next),
2071 struct task_struct, thread_group);
2074 static inline int thread_group_empty(struct task_struct *p)
2076 return list_empty(&p->thread_group);
2079 #define delay_group_leader(p) \
2080 (thread_group_leader(p) && !thread_group_empty(p))
2082 static inline int task_detached(struct task_struct *p)
2084 return p->exit_signal == -1;
2088 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2089 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2090 * pins the final release of task.io_context. Also protects ->cpuset and
2091 * ->cgroup.subsys[].
2093 * Nests both inside and outside of read_lock(&tasklist_lock).
2094 * It must not be nested with write_lock_irq(&tasklist_lock),
2095 * neither inside nor outside.
2097 static inline void task_lock(struct task_struct *p)
2099 spin_lock(&p->alloc_lock);
2102 static inline void task_unlock(struct task_struct *p)
2104 spin_unlock(&p->alloc_lock);
2107 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2108 unsigned long *flags);
2110 static inline void unlock_task_sighand(struct task_struct *tsk,
2111 unsigned long *flags)
2113 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2116 #ifndef __HAVE_THREAD_FUNCTIONS
2118 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2119 #define task_stack_page(task) ((task)->stack)
2121 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2123 *task_thread_info(p) = *task_thread_info(org);
2124 task_thread_info(p)->task = p;
2127 static inline unsigned long *end_of_stack(struct task_struct *p)
2129 return (unsigned long *)(task_thread_info(p) + 1);
2134 static inline int object_is_on_stack(void *obj)
2136 void *stack = task_stack_page(current);
2138 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2141 extern void thread_info_cache_init(void);
2143 #ifdef CONFIG_DEBUG_STACK_USAGE
2144 static inline unsigned long stack_not_used(struct task_struct *p)
2146 unsigned long *n = end_of_stack(p);
2148 do { /* Skip over canary */
2152 return (unsigned long)n - (unsigned long)end_of_stack(p);
2156 /* set thread flags in other task's structures
2157 * - see asm/thread_info.h for TIF_xxxx flags available
2159 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2161 set_ti_thread_flag(task_thread_info(tsk), flag);
2164 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2166 clear_ti_thread_flag(task_thread_info(tsk), flag);
2169 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2171 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2174 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2176 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2179 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2181 return test_ti_thread_flag(task_thread_info(tsk), flag);
2184 static inline void set_tsk_need_resched(struct task_struct *tsk)
2186 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2189 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2191 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2194 static inline int test_tsk_need_resched(struct task_struct *tsk)
2196 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2199 static inline int signal_pending(struct task_struct *p)
2201 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2204 extern int __fatal_signal_pending(struct task_struct *p);
2206 static inline int fatal_signal_pending(struct task_struct *p)
2208 return signal_pending(p) && __fatal_signal_pending(p);
2211 static inline int signal_pending_state(long state, struct task_struct *p)
2213 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2215 if (!signal_pending(p))
2218 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2221 static inline int need_resched(void)
2223 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2227 * cond_resched() and cond_resched_lock(): latency reduction via
2228 * explicit rescheduling in places that are safe. The return
2229 * value indicates whether a reschedule was done in fact.
2230 * cond_resched_lock() will drop the spinlock before scheduling,
2231 * cond_resched_softirq() will enable bhs before scheduling.
2233 extern int _cond_resched(void);
2234 #ifdef CONFIG_PREEMPT_BKL
2235 static inline int cond_resched(void)
2240 static inline int cond_resched(void)
2242 return _cond_resched();
2245 extern int cond_resched_lock(spinlock_t * lock);
2246 extern int cond_resched_softirq(void);
2247 static inline int cond_resched_bkl(void)
2249 return _cond_resched();
2253 * Does a critical section need to be broken due to another
2254 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2255 * but a general need for low latency)
2257 static inline int spin_needbreak(spinlock_t *lock)
2259 #ifdef CONFIG_PREEMPT
2260 return spin_is_contended(lock);
2267 * Thread group CPU time accounting.
2269 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2270 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2272 static inline void thread_group_cputime_init(struct signal_struct *sig)
2274 sig->cputimer.cputime = INIT_CPUTIME;
2275 spin_lock_init(&sig->cputimer.lock);
2276 sig->cputimer.running = 0;
2279 static inline void thread_group_cputime_free(struct signal_struct *sig)
2284 * Reevaluate whether the task has signals pending delivery.
2285 * Wake the task if so.
2286 * This is required every time the blocked sigset_t changes.
2287 * callers must hold sighand->siglock.
2289 extern void recalc_sigpending_and_wake(struct task_struct *t);
2290 extern void recalc_sigpending(void);
2292 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2295 * Wrappers for p->thread_info->cpu access. No-op on UP.
2299 static inline unsigned int task_cpu(const struct task_struct *p)
2301 return task_thread_info(p)->cpu;
2304 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2308 static inline unsigned int task_cpu(const struct task_struct *p)
2313 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2317 #endif /* CONFIG_SMP */
2319 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2321 #ifdef CONFIG_TRACING
2323 __trace_special(void *__tr, void *__data,
2324 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2327 __trace_special(void *__tr, void *__data,
2328 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2333 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2334 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2336 extern void normalize_rt_tasks(void);
2338 #ifdef CONFIG_GROUP_SCHED
2340 extern struct task_group init_task_group;
2341 #ifdef CONFIG_USER_SCHED
2342 extern struct task_group root_task_group;
2343 extern void set_tg_uid(struct user_struct *user);
2346 extern struct task_group *sched_create_group(struct task_group *parent);
2347 extern void sched_destroy_group(struct task_group *tg);
2348 extern void sched_move_task(struct task_struct *tsk);
2349 #ifdef CONFIG_FAIR_GROUP_SCHED
2350 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2351 extern unsigned long sched_group_shares(struct task_group *tg);
2353 #ifdef CONFIG_RT_GROUP_SCHED
2354 extern int sched_group_set_rt_runtime(struct task_group *tg,
2355 long rt_runtime_us);
2356 extern long sched_group_rt_runtime(struct task_group *tg);
2357 extern int sched_group_set_rt_period(struct task_group *tg,
2359 extern long sched_group_rt_period(struct task_group *tg);
2360 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2364 extern int task_can_switch_user(struct user_struct *up,
2365 struct task_struct *tsk);
2367 #ifdef CONFIG_TASK_XACCT
2368 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2370 tsk->ioac.rchar += amt;
2373 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2375 tsk->ioac.wchar += amt;
2378 static inline void inc_syscr(struct task_struct *tsk)
2383 static inline void inc_syscw(struct task_struct *tsk)
2388 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2392 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2396 static inline void inc_syscr(struct task_struct *tsk)
2400 static inline void inc_syscw(struct task_struct *tsk)
2405 #ifndef TASK_SIZE_OF
2406 #define TASK_SIZE_OF(tsk) TASK_SIZE
2409 #ifdef CONFIG_MM_OWNER
2410 extern void mm_update_next_owner(struct mm_struct *mm);
2411 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2413 static inline void mm_update_next_owner(struct mm_struct *mm)
2417 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2420 #endif /* CONFIG_MM_OWNER */
2422 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2424 #endif /* __KERNEL__ */