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/path.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;
103 * List of flags we want to share for kernel threads,
104 * if only because they are not used by them anyway.
106 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
109 * These are the constant used to fake the fixed-point load-average
110 * counting. Some notes:
111 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
112 * a load-average precision of 10 bits integer + 11 bits fractional
113 * - if you want to count load-averages more often, you need more
114 * precision, or rounding will get you. With 2-second counting freq,
115 * the EXP_n values would be 1981, 2034 and 2043 if still using only
118 extern unsigned long avenrun[]; /* Load averages */
120 #define FSHIFT 11 /* nr of bits of precision */
121 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
122 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
123 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
124 #define EXP_5 2014 /* 1/exp(5sec/5min) */
125 #define EXP_15 2037 /* 1/exp(5sec/15min) */
127 #define CALC_LOAD(load,exp,n) \
129 load += n*(FIXED_1-exp); \
132 extern unsigned long total_forks;
133 extern int nr_threads;
134 DECLARE_PER_CPU(unsigned long, process_counts);
135 extern int nr_processes(void);
136 extern unsigned long nr_running(void);
137 extern unsigned long nr_uninterruptible(void);
138 extern unsigned long nr_active(void);
139 extern unsigned long nr_iowait(void);
144 #ifdef CONFIG_SCHED_DEBUG
145 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
146 extern void proc_sched_set_task(struct task_struct *p);
148 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
151 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
154 static inline void proc_sched_set_task(struct task_struct *p)
158 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
163 extern unsigned long long time_sync_thresh;
166 * Task state bitmask. NOTE! These bits are also
167 * encoded in fs/proc/array.c: get_task_state().
169 * We have two separate sets of flags: task->state
170 * is about runnability, while task->exit_state are
171 * about the task exiting. Confusing, but this way
172 * modifying one set can't modify the other one by
175 #define TASK_RUNNING 0
176 #define TASK_INTERRUPTIBLE 1
177 #define TASK_UNINTERRUPTIBLE 2
178 #define __TASK_STOPPED 4
179 #define __TASK_TRACED 8
180 /* in tsk->exit_state */
181 #define EXIT_ZOMBIE 16
183 /* in tsk->state again */
185 #define TASK_WAKEKILL 128
187 /* Convenience macros for the sake of set_task_state */
188 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
189 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
190 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
192 /* Convenience macros for the sake of wake_up */
193 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
194 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
196 /* get_task_state() */
197 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
198 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
201 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
202 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
203 #define task_is_stopped_or_traced(task) \
204 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
205 #define task_contributes_to_load(task) \
206 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
208 #define __set_task_state(tsk, state_value) \
209 do { (tsk)->state = (state_value); } while (0)
210 #define set_task_state(tsk, state_value) \
211 set_mb((tsk)->state, (state_value))
214 * set_current_state() includes a barrier so that the write of current->state
215 * is correctly serialised wrt the caller's subsequent test of whether to
218 * set_current_state(TASK_UNINTERRUPTIBLE);
219 * if (do_i_need_to_sleep())
222 * If the caller does not need such serialisation then use __set_current_state()
224 #define __set_current_state(state_value) \
225 do { current->state = (state_value); } while (0)
226 #define set_current_state(state_value) \
227 set_mb(current->state, (state_value))
229 /* Task command name length */
230 #define TASK_COMM_LEN 16
232 #include <linux/spinlock.h>
235 * This serializes "schedule()" and also protects
236 * the run-queue from deletions/modifications (but
237 * _adding_ to the beginning of the run-queue has
240 extern rwlock_t tasklist_lock;
241 extern spinlock_t mmlist_lock;
245 extern void sched_init(void);
246 extern void sched_init_smp(void);
247 extern asmlinkage void schedule_tail(struct task_struct *prev);
248 extern void init_idle(struct task_struct *idle, int cpu);
249 extern void init_idle_bootup_task(struct task_struct *idle);
251 extern int runqueue_is_locked(void);
252 extern void task_rq_unlock_wait(struct task_struct *p);
254 extern cpumask_var_t nohz_cpu_mask;
255 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
256 extern int select_nohz_load_balancer(int cpu);
258 static inline int select_nohz_load_balancer(int cpu)
265 * Only dump TASK_* tasks. (0 for all tasks)
267 extern void show_state_filter(unsigned long state_filter);
269 static inline void show_state(void)
271 show_state_filter(0);
274 extern void show_regs(struct pt_regs *);
277 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
278 * task), SP is the stack pointer of the first frame that should be shown in the back
279 * trace (or NULL if the entire call-chain of the task should be shown).
281 extern void show_stack(struct task_struct *task, unsigned long *sp);
283 void io_schedule(void);
284 long io_schedule_timeout(long timeout);
286 extern void cpu_init (void);
287 extern void trap_init(void);
288 extern void update_process_times(int user);
289 extern void scheduler_tick(void);
291 extern void sched_show_task(struct task_struct *p);
293 #ifdef CONFIG_DETECT_SOFTLOCKUP
294 extern void softlockup_tick(void);
295 extern void touch_softlockup_watchdog(void);
296 extern void touch_all_softlockup_watchdogs(void);
297 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
298 struct file *filp, void __user *buffer,
299 size_t *lenp, loff_t *ppos);
300 extern unsigned int softlockup_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 softlockup_thresh;
306 static inline void softlockup_tick(void)
309 static inline void spawn_softlockup_task(void)
312 static inline void touch_softlockup_watchdog(void)
315 static inline void touch_all_softlockup_watchdogs(void)
321 /* Attach to any functions which should be ignored in wchan output. */
322 #define __sched __attribute__((__section__(".sched.text")))
324 /* Linker adds these: start and end of __sched functions */
325 extern char __sched_text_start[], __sched_text_end[];
327 /* Is this address in the __sched functions? */
328 extern int in_sched_functions(unsigned long addr);
330 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
331 extern signed long schedule_timeout(signed long timeout);
332 extern signed long schedule_timeout_interruptible(signed long timeout);
333 extern signed long schedule_timeout_killable(signed long timeout);
334 extern signed long schedule_timeout_uninterruptible(signed long timeout);
335 asmlinkage void __schedule(void);
336 asmlinkage void schedule(void);
337 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
340 struct user_namespace;
342 /* Maximum number of active map areas.. This is a random (large) number */
343 #define DEFAULT_MAX_MAP_COUNT 65536
345 extern int sysctl_max_map_count;
347 #include <linux/aio.h>
350 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
351 unsigned long, unsigned long);
353 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
354 unsigned long len, unsigned long pgoff,
355 unsigned long flags);
356 extern void arch_unmap_area(struct mm_struct *, unsigned long);
357 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
359 #if USE_SPLIT_PTLOCKS
361 * The mm counters are not protected by its page_table_lock,
362 * so must be incremented atomically.
364 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
365 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
366 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
367 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
368 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
370 #else /* !USE_SPLIT_PTLOCKS */
372 * The mm counters are protected by its page_table_lock,
373 * so can be incremented directly.
375 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
376 #define get_mm_counter(mm, member) ((mm)->_##member)
377 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
378 #define inc_mm_counter(mm, member) (mm)->_##member++
379 #define dec_mm_counter(mm, member) (mm)->_##member--
381 #endif /* !USE_SPLIT_PTLOCKS */
383 #define get_mm_rss(mm) \
384 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
385 #define update_hiwater_rss(mm) do { \
386 unsigned long _rss = get_mm_rss(mm); \
387 if ((mm)->hiwater_rss < _rss) \
388 (mm)->hiwater_rss = _rss; \
390 #define update_hiwater_vm(mm) do { \
391 if ((mm)->hiwater_vm < (mm)->total_vm) \
392 (mm)->hiwater_vm = (mm)->total_vm; \
395 #define get_mm_hiwater_rss(mm) max((mm)->hiwater_rss, get_mm_rss(mm))
396 #define get_mm_hiwater_vm(mm) max((mm)->hiwater_vm, (mm)->total_vm)
398 extern void set_dumpable(struct mm_struct *mm, int value);
399 extern int get_dumpable(struct mm_struct *mm);
403 #define MMF_DUMPABLE 0 /* core dump is permitted */
404 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
405 #define MMF_DUMPABLE_BITS 2
407 /* coredump filter bits */
408 #define MMF_DUMP_ANON_PRIVATE 2
409 #define MMF_DUMP_ANON_SHARED 3
410 #define MMF_DUMP_MAPPED_PRIVATE 4
411 #define MMF_DUMP_MAPPED_SHARED 5
412 #define MMF_DUMP_ELF_HEADERS 6
413 #define MMF_DUMP_HUGETLB_PRIVATE 7
414 #define MMF_DUMP_HUGETLB_SHARED 8
415 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
416 #define MMF_DUMP_FILTER_BITS 7
417 #define MMF_DUMP_FILTER_MASK \
418 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
419 #define MMF_DUMP_FILTER_DEFAULT \
420 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
421 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
423 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
424 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
426 # define MMF_DUMP_MASK_DEFAULT_ELF 0
429 struct sighand_struct {
431 struct k_sigaction action[_NSIG];
433 wait_queue_head_t signalfd_wqh;
436 struct pacct_struct {
439 unsigned long ac_mem;
440 cputime_t ac_utime, ac_stime;
441 unsigned long ac_minflt, ac_majflt;
445 * struct task_cputime - collected CPU time counts
446 * @utime: time spent in user mode, in &cputime_t units
447 * @stime: time spent in kernel mode, in &cputime_t units
448 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
450 * This structure groups together three kinds of CPU time that are
451 * tracked for threads and thread groups. Most things considering
452 * CPU time want to group these counts together and treat all three
453 * of them in parallel.
455 struct task_cputime {
458 unsigned long long sum_exec_runtime;
460 /* Alternate field names when used to cache expirations. */
461 #define prof_exp stime
462 #define virt_exp utime
463 #define sched_exp sum_exec_runtime
465 #define INIT_CPUTIME \
466 (struct task_cputime) { \
467 .utime = cputime_zero, \
468 .stime = cputime_zero, \
469 .sum_exec_runtime = 0, \
473 * struct thread_group_cputimer - thread group interval timer counts
474 * @cputime: thread group interval timers.
475 * @running: non-zero when there are timers running and
476 * @cputime receives updates.
477 * @lock: lock for fields in this struct.
479 * This structure contains the version of task_cputime, above, that is
480 * used for thread group CPU timer calculations.
482 struct thread_group_cputimer {
483 struct task_cputime cputime;
489 * NOTE! "signal_struct" does not have it's own
490 * locking, because a shared signal_struct always
491 * implies a shared sighand_struct, so locking
492 * sighand_struct is always a proper superset of
493 * the locking of signal_struct.
495 struct signal_struct {
499 wait_queue_head_t wait_chldexit; /* for wait4() */
501 /* current thread group signal load-balancing target: */
502 struct task_struct *curr_target;
504 /* shared signal handling: */
505 struct sigpending shared_pending;
507 /* thread group exit support */
510 * - notify group_exit_task when ->count is equal to notify_count
511 * - everyone except group_exit_task is stopped during signal delivery
512 * of fatal signals, group_exit_task processes the signal.
515 struct task_struct *group_exit_task;
517 /* thread group stop support, overloads group_exit_code too */
518 int group_stop_count;
519 unsigned int flags; /* see SIGNAL_* flags below */
521 /* POSIX.1b Interval Timers */
522 struct list_head posix_timers;
524 /* ITIMER_REAL timer for the process */
525 struct hrtimer real_timer;
526 struct pid *leader_pid;
527 ktime_t it_real_incr;
529 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
530 cputime_t it_prof_expires, it_virt_expires;
531 cputime_t it_prof_incr, it_virt_incr;
534 * Thread group totals for process CPU timers.
535 * See thread_group_cputimer(), et al, for details.
537 struct thread_group_cputimer cputimer;
539 /* Earliest-expiration cache. */
540 struct task_cputime cputime_expires;
542 struct list_head cpu_timers[3];
544 /* job control IDs */
547 * pgrp and session fields are deprecated.
548 * use the task_session_Xnr and task_pgrp_Xnr routines below
552 pid_t pgrp __deprecated;
556 struct pid *tty_old_pgrp;
559 pid_t session __deprecated;
563 /* boolean value for session group leader */
566 struct tty_struct *tty; /* NULL if no tty */
569 * Cumulative resource counters for dead threads in the group,
570 * and for reaped dead child processes forked by this group.
571 * Live threads maintain their own counters and add to these
572 * in __exit_signal, except for the group leader.
574 cputime_t utime, stime, cutime, cstime;
577 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
578 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
579 unsigned long inblock, oublock, cinblock, coublock;
580 struct task_io_accounting ioac;
583 * Cumulative ns of schedule CPU time fo dead threads in the
584 * group, not including a zombie group leader, (This only differs
585 * from jiffies_to_ns(utime + stime) if sched_clock uses something
586 * other than jiffies.)
588 unsigned long long sum_sched_runtime;
591 * We don't bother to synchronize most readers of this at all,
592 * because there is no reader checking a limit that actually needs
593 * to get both rlim_cur and rlim_max atomically, and either one
594 * alone is a single word that can safely be read normally.
595 * getrlimit/setrlimit use task_lock(current->group_leader) to
596 * protect this instead of the siglock, because they really
597 * have no need to disable irqs.
599 struct rlimit rlim[RLIM_NLIMITS];
601 #ifdef CONFIG_BSD_PROCESS_ACCT
602 struct pacct_struct pacct; /* per-process accounting information */
604 #ifdef CONFIG_TASKSTATS
605 struct taskstats *stats;
609 struct tty_audit_buf *tty_audit_buf;
613 /* Context switch must be unlocked if interrupts are to be enabled */
614 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
615 # define __ARCH_WANT_UNLOCKED_CTXSW
619 * Bits in flags field of signal_struct.
621 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
622 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
623 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
624 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
626 * Pending notifications to parent.
628 #define SIGNAL_CLD_STOPPED 0x00000010
629 #define SIGNAL_CLD_CONTINUED 0x00000020
630 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
632 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
634 /* If true, all threads except ->group_exit_task have pending SIGKILL */
635 static inline int signal_group_exit(const struct signal_struct *sig)
637 return (sig->flags & SIGNAL_GROUP_EXIT) ||
638 (sig->group_exit_task != NULL);
642 * Some day this will be a full-fledged user tracking system..
645 atomic_t __count; /* reference count */
646 atomic_t processes; /* How many processes does this user have? */
647 atomic_t files; /* How many open files does this user have? */
648 atomic_t sigpending; /* How many pending signals does this user have? */
649 #ifdef CONFIG_INOTIFY_USER
650 atomic_t inotify_watches; /* How many inotify watches does this user have? */
651 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
654 atomic_t epoll_watches; /* The number of file descriptors currently watched */
656 #ifdef CONFIG_POSIX_MQUEUE
657 /* protected by mq_lock */
658 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
660 unsigned long locked_shm; /* How many pages of mlocked shm ? */
663 struct key *uid_keyring; /* UID specific keyring */
664 struct key *session_keyring; /* UID's default session keyring */
667 /* Hash table maintenance information */
668 struct hlist_node uidhash_node;
670 struct user_namespace *user_ns;
672 #ifdef CONFIG_USER_SCHED
673 struct task_group *tg;
676 struct work_struct work;
681 extern int uids_sysfs_init(void);
683 extern struct user_struct *find_user(uid_t);
685 extern struct user_struct root_user;
686 #define INIT_USER (&root_user)
689 struct backing_dev_info;
690 struct reclaim_state;
692 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
694 /* cumulative counters */
695 unsigned long pcount; /* # of times run on this cpu */
696 unsigned long long run_delay; /* time spent waiting on a runqueue */
699 unsigned long long last_arrival,/* when we last ran on a cpu */
700 last_queued; /* when we were last queued to run */
701 #ifdef CONFIG_SCHEDSTATS
703 unsigned int bkl_count;
706 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
708 #ifdef CONFIG_TASK_DELAY_ACCT
709 struct task_delay_info {
711 unsigned int flags; /* Private per-task flags */
713 /* For each stat XXX, add following, aligned appropriately
715 * struct timespec XXX_start, XXX_end;
719 * Atomicity of updates to XXX_delay, XXX_count protected by
720 * single lock above (split into XXX_lock if contention is an issue).
724 * XXX_count is incremented on every XXX operation, the delay
725 * associated with the operation is added to XXX_delay.
726 * XXX_delay contains the accumulated delay time in nanoseconds.
728 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
729 u64 blkio_delay; /* wait for sync block io completion */
730 u64 swapin_delay; /* wait for swapin block io completion */
731 u32 blkio_count; /* total count of the number of sync block */
732 /* io operations performed */
733 u32 swapin_count; /* total count of the number of swapin block */
734 /* io operations performed */
736 struct timespec freepages_start, freepages_end;
737 u64 freepages_delay; /* wait for memory reclaim */
738 u32 freepages_count; /* total count of memory reclaim */
740 #endif /* CONFIG_TASK_DELAY_ACCT */
742 static inline int sched_info_on(void)
744 #ifdef CONFIG_SCHEDSTATS
746 #elif defined(CONFIG_TASK_DELAY_ACCT)
747 extern int delayacct_on;
762 * sched-domains (multiprocessor balancing) declarations:
766 * Increase resolution of nice-level calculations:
768 #define SCHED_LOAD_SHIFT 10
769 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
771 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
774 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
775 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
776 #define SD_BALANCE_EXEC 4 /* Balance on exec */
777 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
778 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
779 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
780 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
781 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
782 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
783 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
784 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
785 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
787 enum powersavings_balance_level {
788 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
789 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
790 * first for long running threads
792 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
793 * cpu package for power savings
795 MAX_POWERSAVINGS_BALANCE_LEVELS
798 extern int sched_mc_power_savings, sched_smt_power_savings;
800 static inline int sd_balance_for_mc_power(void)
802 if (sched_smt_power_savings)
803 return SD_POWERSAVINGS_BALANCE;
808 static inline int sd_balance_for_package_power(void)
810 if (sched_mc_power_savings | sched_smt_power_savings)
811 return SD_POWERSAVINGS_BALANCE;
817 * Optimise SD flags for power savings:
818 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
819 * Keep default SD flags if sched_{smt,mc}_power_saving=0
822 static inline int sd_power_saving_flags(void)
824 if (sched_mc_power_savings | sched_smt_power_savings)
825 return SD_BALANCE_NEWIDLE;
831 struct sched_group *next; /* Must be a circular list */
834 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
835 * single CPU. This is read only (except for setup, hotplug CPU).
836 * Note : Never change cpu_power without recompute its reciprocal
838 unsigned int __cpu_power;
840 * reciprocal value of cpu_power to avoid expensive divides
841 * (see include/linux/reciprocal_div.h)
843 u32 reciprocal_cpu_power;
845 unsigned long cpumask[];
848 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
850 return to_cpumask(sg->cpumask);
853 enum sched_domain_level {
863 struct sched_domain_attr {
864 int relax_domain_level;
867 #define SD_ATTR_INIT (struct sched_domain_attr) { \
868 .relax_domain_level = -1, \
871 struct sched_domain {
872 /* These fields must be setup */
873 struct sched_domain *parent; /* top domain must be null terminated */
874 struct sched_domain *child; /* bottom domain must be null terminated */
875 struct sched_group *groups; /* the balancing groups of the domain */
876 unsigned long min_interval; /* Minimum balance interval ms */
877 unsigned long max_interval; /* Maximum balance interval ms */
878 unsigned int busy_factor; /* less balancing by factor if busy */
879 unsigned int imbalance_pct; /* No balance until over watermark */
880 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
881 unsigned int busy_idx;
882 unsigned int idle_idx;
883 unsigned int newidle_idx;
884 unsigned int wake_idx;
885 unsigned int forkexec_idx;
886 int flags; /* See SD_* */
887 enum sched_domain_level level;
889 /* Runtime fields. */
890 unsigned long last_balance; /* init to jiffies. units in jiffies */
891 unsigned int balance_interval; /* initialise to 1. units in ms. */
892 unsigned int nr_balance_failed; /* initialise to 0 */
896 #ifdef CONFIG_SCHEDSTATS
897 /* load_balance() stats */
898 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
899 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
900 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
901 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
902 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
903 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
904 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
905 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
907 /* Active load balancing */
908 unsigned int alb_count;
909 unsigned int alb_failed;
910 unsigned int alb_pushed;
912 /* SD_BALANCE_EXEC stats */
913 unsigned int sbe_count;
914 unsigned int sbe_balanced;
915 unsigned int sbe_pushed;
917 /* SD_BALANCE_FORK stats */
918 unsigned int sbf_count;
919 unsigned int sbf_balanced;
920 unsigned int sbf_pushed;
922 /* try_to_wake_up() stats */
923 unsigned int ttwu_wake_remote;
924 unsigned int ttwu_move_affine;
925 unsigned int ttwu_move_balance;
927 #ifdef CONFIG_SCHED_DEBUG
931 /* span of all CPUs in this domain */
932 unsigned long span[];
935 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
937 return to_cpumask(sd->span);
940 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
941 struct sched_domain_attr *dattr_new);
943 /* Test a flag in parent sched domain */
944 static inline int test_sd_parent(struct sched_domain *sd, int flag)
946 if (sd->parent && (sd->parent->flags & flag))
952 #else /* CONFIG_SMP */
954 struct sched_domain_attr;
957 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
958 struct sched_domain_attr *dattr_new)
961 #endif /* !CONFIG_SMP */
963 struct io_context; /* See blkdev.h */
966 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
967 extern void prefetch_stack(struct task_struct *t);
969 static inline void prefetch_stack(struct task_struct *t) { }
972 struct audit_context; /* See audit.c */
974 struct pipe_inode_info;
975 struct uts_namespace;
981 const struct sched_class *next;
983 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
984 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
985 void (*yield_task) (struct rq *rq);
987 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
989 struct task_struct * (*pick_next_task) (struct rq *rq);
990 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
993 int (*select_task_rq)(struct task_struct *p, int sync);
995 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
996 struct rq *busiest, unsigned long max_load_move,
997 struct sched_domain *sd, enum cpu_idle_type idle,
998 int *all_pinned, int *this_best_prio);
1000 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1001 struct rq *busiest, struct sched_domain *sd,
1002 enum cpu_idle_type idle);
1003 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1004 int (*needs_post_schedule) (struct rq *this_rq);
1005 void (*post_schedule) (struct rq *this_rq);
1006 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1008 void (*set_cpus_allowed)(struct task_struct *p,
1009 const struct cpumask *newmask);
1011 void (*rq_online)(struct rq *rq);
1012 void (*rq_offline)(struct rq *rq);
1015 void (*set_curr_task) (struct rq *rq);
1016 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1017 void (*task_new) (struct rq *rq, struct task_struct *p);
1019 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1021 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1023 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1024 int oldprio, int running);
1026 #ifdef CONFIG_FAIR_GROUP_SCHED
1027 void (*moved_group) (struct task_struct *p);
1031 struct load_weight {
1032 unsigned long weight, inv_weight;
1036 * CFS stats for a schedulable entity (task, task-group etc)
1038 * Current field usage histogram:
1045 struct sched_entity {
1046 struct load_weight load; /* for load-balancing */
1047 struct rb_node run_node;
1048 struct list_head group_node;
1052 u64 sum_exec_runtime;
1054 u64 prev_sum_exec_runtime;
1063 #ifdef CONFIG_SCHEDSTATS
1071 s64 sum_sleep_runtime;
1078 u64 nr_migrations_cold;
1079 u64 nr_failed_migrations_affine;
1080 u64 nr_failed_migrations_running;
1081 u64 nr_failed_migrations_hot;
1082 u64 nr_forced_migrations;
1083 u64 nr_forced2_migrations;
1086 u64 nr_wakeups_sync;
1087 u64 nr_wakeups_migrate;
1088 u64 nr_wakeups_local;
1089 u64 nr_wakeups_remote;
1090 u64 nr_wakeups_affine;
1091 u64 nr_wakeups_affine_attempts;
1092 u64 nr_wakeups_passive;
1093 u64 nr_wakeups_idle;
1096 #ifdef CONFIG_FAIR_GROUP_SCHED
1097 struct sched_entity *parent;
1098 /* rq on which this entity is (to be) queued: */
1099 struct cfs_rq *cfs_rq;
1100 /* rq "owned" by this entity/group: */
1101 struct cfs_rq *my_q;
1105 struct sched_rt_entity {
1106 struct list_head run_list;
1107 unsigned long timeout;
1108 unsigned int time_slice;
1109 int nr_cpus_allowed;
1111 struct sched_rt_entity *back;
1112 #ifdef CONFIG_RT_GROUP_SCHED
1113 struct sched_rt_entity *parent;
1114 /* rq on which this entity is (to be) queued: */
1115 struct rt_rq *rt_rq;
1116 /* rq "owned" by this entity/group: */
1121 struct task_struct {
1122 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1125 unsigned int flags; /* per process flags, defined below */
1126 unsigned int ptrace;
1128 int lock_depth; /* BKL lock depth */
1131 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1136 int prio, static_prio, normal_prio;
1137 unsigned int rt_priority;
1138 const struct sched_class *sched_class;
1139 struct sched_entity se;
1140 struct sched_rt_entity rt;
1142 #ifdef CONFIG_PREEMPT_NOTIFIERS
1143 /* list of struct preempt_notifier: */
1144 struct hlist_head preempt_notifiers;
1148 * fpu_counter contains the number of consecutive context switches
1149 * that the FPU is used. If this is over a threshold, the lazy fpu
1150 * saving becomes unlazy to save the trap. This is an unsigned char
1151 * so that after 256 times the counter wraps and the behavior turns
1152 * lazy again; this to deal with bursty apps that only use FPU for
1155 unsigned char fpu_counter;
1156 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1157 #ifdef CONFIG_BLK_DEV_IO_TRACE
1158 unsigned int btrace_seq;
1161 unsigned int policy;
1162 cpumask_t cpus_allowed;
1164 #ifdef CONFIG_PREEMPT_RCU
1165 int rcu_read_lock_nesting;
1166 int rcu_flipctr_idx;
1167 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1169 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1170 struct sched_info sched_info;
1173 struct list_head tasks;
1174 struct plist_node pushable_tasks;
1176 struct mm_struct *mm, *active_mm;
1179 struct linux_binfmt *binfmt;
1181 int exit_code, exit_signal;
1182 int pdeath_signal; /* The signal sent when the parent dies */
1184 unsigned int personality;
1185 unsigned did_exec:1;
1186 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1191 /* Canary value for the -fstack-protector gcc feature */
1192 unsigned long stack_canary;
1195 * pointers to (original) parent process, youngest child, younger sibling,
1196 * older sibling, respectively. (p->father can be replaced with
1197 * p->real_parent->pid)
1199 struct task_struct *real_parent; /* real parent process */
1200 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1202 * children/sibling forms the list of my natural children
1204 struct list_head children; /* list of my children */
1205 struct list_head sibling; /* linkage in my parent's children list */
1206 struct task_struct *group_leader; /* threadgroup leader */
1209 * ptraced is the list of tasks this task is using ptrace on.
1210 * This includes both natural children and PTRACE_ATTACH targets.
1211 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1213 struct list_head ptraced;
1214 struct list_head ptrace_entry;
1216 #ifdef CONFIG_X86_PTRACE_BTS
1218 * This is the tracer handle for the ptrace BTS extension.
1219 * This field actually belongs to the ptracer task.
1221 struct bts_tracer *bts;
1223 * The buffer to hold the BTS data.
1227 #endif /* CONFIG_X86_PTRACE_BTS */
1229 /* PID/PID hash table linkage. */
1230 struct pid_link pids[PIDTYPE_MAX];
1231 struct list_head thread_group;
1233 struct completion *vfork_done; /* for vfork() */
1234 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1235 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1237 cputime_t utime, stime, utimescaled, stimescaled;
1239 cputime_t prev_utime, prev_stime;
1240 unsigned long nvcsw, nivcsw; /* context switch counts */
1241 struct timespec start_time; /* monotonic time */
1242 struct timespec real_start_time; /* boot based time */
1243 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1244 unsigned long min_flt, maj_flt;
1246 struct task_cputime cputime_expires;
1247 struct list_head cpu_timers[3];
1249 /* process credentials */
1250 const struct cred *real_cred; /* objective and real subjective task
1251 * credentials (COW) */
1252 const struct cred *cred; /* effective (overridable) subjective task
1253 * credentials (COW) */
1254 struct mutex cred_exec_mutex; /* execve vs ptrace cred calculation mutex */
1256 char comm[TASK_COMM_LEN]; /* executable name excluding path
1257 - access with [gs]et_task_comm (which lock
1258 it with task_lock())
1259 - initialized normally by flush_old_exec */
1260 /* file system info */
1261 int link_count, total_link_count;
1262 #ifdef CONFIG_SYSVIPC
1264 struct sysv_sem sysvsem;
1266 #ifdef CONFIG_DETECT_SOFTLOCKUP
1267 /* hung task detection */
1268 unsigned long last_switch_timestamp;
1269 unsigned long last_switch_count;
1271 /* CPU-specific state of this task */
1272 struct thread_struct thread;
1273 /* filesystem information */
1274 struct fs_struct *fs;
1275 /* open file information */
1276 struct files_struct *files;
1278 struct nsproxy *nsproxy;
1279 /* signal handlers */
1280 struct signal_struct *signal;
1281 struct sighand_struct *sighand;
1283 sigset_t blocked, real_blocked;
1284 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1285 struct sigpending pending;
1287 unsigned long sas_ss_sp;
1289 int (*notifier)(void *priv);
1290 void *notifier_data;
1291 sigset_t *notifier_mask;
1292 struct audit_context *audit_context;
1293 #ifdef CONFIG_AUDITSYSCALL
1295 unsigned int sessionid;
1299 /* Thread group tracking */
1302 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1303 spinlock_t alloc_lock;
1305 /* Protection of the PI data structures: */
1308 #ifdef CONFIG_RT_MUTEXES
1309 /* PI waiters blocked on a rt_mutex held by this task */
1310 struct plist_head pi_waiters;
1311 /* Deadlock detection and priority inheritance handling */
1312 struct rt_mutex_waiter *pi_blocked_on;
1315 #ifdef CONFIG_DEBUG_MUTEXES
1316 /* mutex deadlock detection */
1317 struct mutex_waiter *blocked_on;
1319 #ifdef CONFIG_TRACE_IRQFLAGS
1320 unsigned int irq_events;
1321 int hardirqs_enabled;
1322 unsigned long hardirq_enable_ip;
1323 unsigned int hardirq_enable_event;
1324 unsigned long hardirq_disable_ip;
1325 unsigned int hardirq_disable_event;
1326 int softirqs_enabled;
1327 unsigned long softirq_disable_ip;
1328 unsigned int softirq_disable_event;
1329 unsigned long softirq_enable_ip;
1330 unsigned int softirq_enable_event;
1331 int hardirq_context;
1332 int softirq_context;
1334 #ifdef CONFIG_LOCKDEP
1335 # define MAX_LOCK_DEPTH 48UL
1338 unsigned int lockdep_recursion;
1339 struct held_lock held_locks[MAX_LOCK_DEPTH];
1340 gfp_t lockdep_reclaim_gfp;
1343 /* journalling filesystem info */
1346 /* stacked block device info */
1347 struct bio *bio_list, **bio_tail;
1350 struct reclaim_state *reclaim_state;
1352 struct backing_dev_info *backing_dev_info;
1354 struct io_context *io_context;
1356 unsigned long ptrace_message;
1357 siginfo_t *last_siginfo; /* For ptrace use. */
1358 struct task_io_accounting ioac;
1359 #if defined(CONFIG_TASK_XACCT)
1360 u64 acct_rss_mem1; /* accumulated rss usage */
1361 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1362 cputime_t acct_timexpd; /* stime + utime since last update */
1364 #ifdef CONFIG_CPUSETS
1365 nodemask_t mems_allowed;
1366 int cpuset_mems_generation;
1367 int cpuset_mem_spread_rotor;
1369 #ifdef CONFIG_CGROUPS
1370 /* Control Group info protected by css_set_lock */
1371 struct css_set *cgroups;
1372 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1373 struct list_head cg_list;
1376 struct robust_list_head __user *robust_list;
1377 #ifdef CONFIG_COMPAT
1378 struct compat_robust_list_head __user *compat_robust_list;
1380 struct list_head pi_state_list;
1381 struct futex_pi_state *pi_state_cache;
1384 struct mempolicy *mempolicy;
1387 atomic_t fs_excl; /* holding fs exclusive resources */
1388 struct rcu_head rcu;
1391 * cache last used pipe for splice
1393 struct pipe_inode_info *splice_pipe;
1394 #ifdef CONFIG_TASK_DELAY_ACCT
1395 struct task_delay_info *delays;
1397 #ifdef CONFIG_FAULT_INJECTION
1400 struct prop_local_single dirties;
1401 #ifdef CONFIG_LATENCYTOP
1402 int latency_record_count;
1403 struct latency_record latency_record[LT_SAVECOUNT];
1406 * time slack values; these are used to round up poll() and
1407 * select() etc timeout values. These are in nanoseconds.
1409 unsigned long timer_slack_ns;
1410 unsigned long default_timer_slack_ns;
1412 struct list_head *scm_work_list;
1413 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1414 /* Index of current stored adress in ret_stack */
1416 /* Stack of return addresses for return function tracing */
1417 struct ftrace_ret_stack *ret_stack;
1419 * Number of functions that haven't been traced
1420 * because of depth overrun.
1422 atomic_t trace_overrun;
1423 /* Pause for the tracing */
1424 atomic_t tracing_graph_pause;
1426 #ifdef CONFIG_TRACING
1427 /* state flags for use by tracers */
1428 unsigned long trace;
1432 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1433 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1436 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1437 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1438 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1439 * values are inverted: lower p->prio value means higher priority.
1441 * The MAX_USER_RT_PRIO value allows the actual maximum
1442 * RT priority to be separate from the value exported to
1443 * user-space. This allows kernel threads to set their
1444 * priority to a value higher than any user task. Note:
1445 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1448 #define MAX_USER_RT_PRIO 100
1449 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1451 #define MAX_PRIO (MAX_RT_PRIO + 40)
1452 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1454 static inline int rt_prio(int prio)
1456 if (unlikely(prio < MAX_RT_PRIO))
1461 static inline int rt_task(struct task_struct *p)
1463 return rt_prio(p->prio);
1466 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1468 tsk->signal->__session = session;
1471 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1473 tsk->signal->__pgrp = pgrp;
1476 static inline struct pid *task_pid(struct task_struct *task)
1478 return task->pids[PIDTYPE_PID].pid;
1481 static inline struct pid *task_tgid(struct task_struct *task)
1483 return task->group_leader->pids[PIDTYPE_PID].pid;
1486 static inline struct pid *task_pgrp(struct task_struct *task)
1488 return task->group_leader->pids[PIDTYPE_PGID].pid;
1491 static inline struct pid *task_session(struct task_struct *task)
1493 return task->group_leader->pids[PIDTYPE_SID].pid;
1496 struct pid_namespace;
1499 * the helpers to get the task's different pids as they are seen
1500 * from various namespaces
1502 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1503 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1505 * task_xid_nr_ns() : id seen from the ns specified;
1507 * set_task_vxid() : assigns a virtual id to a task;
1509 * see also pid_nr() etc in include/linux/pid.h
1512 static inline pid_t task_pid_nr(struct task_struct *tsk)
1517 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1519 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1521 return pid_vnr(task_pid(tsk));
1525 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1530 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1532 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1534 return pid_vnr(task_tgid(tsk));
1538 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1540 return tsk->signal->__pgrp;
1543 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1545 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1547 return pid_vnr(task_pgrp(tsk));
1551 static inline pid_t task_session_nr(struct task_struct *tsk)
1553 return tsk->signal->__session;
1556 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1558 static inline pid_t task_session_vnr(struct task_struct *tsk)
1560 return pid_vnr(task_session(tsk));
1565 * pid_alive - check that a task structure is not stale
1566 * @p: Task structure to be checked.
1568 * Test if a process is not yet dead (at most zombie state)
1569 * If pid_alive fails, then pointers within the task structure
1570 * can be stale and must not be dereferenced.
1572 static inline int pid_alive(struct task_struct *p)
1574 return p->pids[PIDTYPE_PID].pid != NULL;
1578 * is_global_init - check if a task structure is init
1579 * @tsk: Task structure to be checked.
1581 * Check if a task structure is the first user space task the kernel created.
1583 static inline int is_global_init(struct task_struct *tsk)
1585 return tsk->pid == 1;
1589 * is_container_init:
1590 * check whether in the task is init in its own pid namespace.
1592 extern int is_container_init(struct task_struct *tsk);
1594 extern struct pid *cad_pid;
1596 extern void free_task(struct task_struct *tsk);
1597 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1599 extern void __put_task_struct(struct task_struct *t);
1601 static inline void put_task_struct(struct task_struct *t)
1603 if (atomic_dec_and_test(&t->usage))
1604 __put_task_struct(t);
1607 extern cputime_t task_utime(struct task_struct *p);
1608 extern cputime_t task_stime(struct task_struct *p);
1609 extern cputime_t task_gtime(struct task_struct *p);
1614 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1615 /* Not implemented yet, only for 486*/
1616 #define PF_STARTING 0x00000002 /* being created */
1617 #define PF_EXITING 0x00000004 /* getting shut down */
1618 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1619 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1620 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1621 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1622 #define PF_DUMPCORE 0x00000200 /* dumped core */
1623 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1624 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1625 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1626 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1627 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1628 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1629 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1630 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1631 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1632 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1633 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1634 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1635 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1636 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1637 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1638 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1639 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1640 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1641 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1642 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1645 * Only the _current_ task can read/write to tsk->flags, but other
1646 * tasks can access tsk->flags in readonly mode for example
1647 * with tsk_used_math (like during threaded core dumping).
1648 * There is however an exception to this rule during ptrace
1649 * or during fork: the ptracer task is allowed to write to the
1650 * child->flags of its traced child (same goes for fork, the parent
1651 * can write to the child->flags), because we're guaranteed the
1652 * child is not running and in turn not changing child->flags
1653 * at the same time the parent does it.
1655 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1656 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1657 #define clear_used_math() clear_stopped_child_used_math(current)
1658 #define set_used_math() set_stopped_child_used_math(current)
1659 #define conditional_stopped_child_used_math(condition, child) \
1660 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1661 #define conditional_used_math(condition) \
1662 conditional_stopped_child_used_math(condition, current)
1663 #define copy_to_stopped_child_used_math(child) \
1664 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1665 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1666 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1667 #define used_math() tsk_used_math(current)
1670 extern int set_cpus_allowed_ptr(struct task_struct *p,
1671 const struct cpumask *new_mask);
1673 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1674 const struct cpumask *new_mask)
1676 if (!cpumask_test_cpu(0, new_mask))
1681 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1683 return set_cpus_allowed_ptr(p, &new_mask);
1687 * Architectures can set this to 1 if they have specified
1688 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1689 * but then during bootup it turns out that sched_clock()
1690 * is reliable after all:
1692 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1693 extern int sched_clock_stable;
1696 extern unsigned long long sched_clock(void);
1698 extern void sched_clock_init(void);
1699 extern u64 sched_clock_cpu(int cpu);
1701 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1702 static inline void sched_clock_tick(void)
1706 static inline void sched_clock_idle_sleep_event(void)
1710 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1714 extern void sched_clock_tick(void);
1715 extern void sched_clock_idle_sleep_event(void);
1716 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1720 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1721 * clock constructed from sched_clock():
1723 extern unsigned long long cpu_clock(int cpu);
1725 extern unsigned long long
1726 task_sched_runtime(struct task_struct *task);
1727 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1729 /* sched_exec is called by processes performing an exec */
1731 extern void sched_exec(void);
1733 #define sched_exec() {}
1736 extern void sched_clock_idle_sleep_event(void);
1737 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1739 #ifdef CONFIG_HOTPLUG_CPU
1740 extern void idle_task_exit(void);
1742 static inline void idle_task_exit(void) {}
1745 extern void sched_idle_next(void);
1747 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1748 extern void wake_up_idle_cpu(int cpu);
1750 static inline void wake_up_idle_cpu(int cpu) { }
1753 extern unsigned int sysctl_sched_latency;
1754 extern unsigned int sysctl_sched_min_granularity;
1755 extern unsigned int sysctl_sched_wakeup_granularity;
1756 extern unsigned int sysctl_sched_shares_ratelimit;
1757 extern unsigned int sysctl_sched_shares_thresh;
1758 #ifdef CONFIG_SCHED_DEBUG
1759 extern unsigned int sysctl_sched_child_runs_first;
1760 extern unsigned int sysctl_sched_features;
1761 extern unsigned int sysctl_sched_migration_cost;
1762 extern unsigned int sysctl_sched_nr_migrate;
1764 int sched_nr_latency_handler(struct ctl_table *table, int write,
1765 struct file *file, void __user *buffer, size_t *length,
1768 extern unsigned int sysctl_sched_rt_period;
1769 extern int sysctl_sched_rt_runtime;
1771 int sched_rt_handler(struct ctl_table *table, int write,
1772 struct file *filp, void __user *buffer, size_t *lenp,
1775 extern unsigned int sysctl_sched_compat_yield;
1777 #ifdef CONFIG_RT_MUTEXES
1778 extern int rt_mutex_getprio(struct task_struct *p);
1779 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1780 extern void rt_mutex_adjust_pi(struct task_struct *p);
1782 static inline int rt_mutex_getprio(struct task_struct *p)
1784 return p->normal_prio;
1786 # define rt_mutex_adjust_pi(p) do { } while (0)
1789 extern void set_user_nice(struct task_struct *p, long nice);
1790 extern int task_prio(const struct task_struct *p);
1791 extern int task_nice(const struct task_struct *p);
1792 extern int can_nice(const struct task_struct *p, const int nice);
1793 extern int task_curr(const struct task_struct *p);
1794 extern int idle_cpu(int cpu);
1795 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1796 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1797 struct sched_param *);
1798 extern struct task_struct *idle_task(int cpu);
1799 extern struct task_struct *curr_task(int cpu);
1800 extern void set_curr_task(int cpu, struct task_struct *p);
1805 * The default (Linux) execution domain.
1807 extern struct exec_domain default_exec_domain;
1809 union thread_union {
1810 struct thread_info thread_info;
1811 unsigned long stack[THREAD_SIZE/sizeof(long)];
1814 #ifndef __HAVE_ARCH_KSTACK_END
1815 static inline int kstack_end(void *addr)
1817 /* Reliable end of stack detection:
1818 * Some APM bios versions misalign the stack
1820 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1824 extern union thread_union init_thread_union;
1825 extern struct task_struct init_task;
1827 extern struct mm_struct init_mm;
1829 extern struct pid_namespace init_pid_ns;
1832 * find a task by one of its numerical ids
1834 * find_task_by_pid_type_ns():
1835 * it is the most generic call - it finds a task by all id,
1836 * type and namespace specified
1837 * find_task_by_pid_ns():
1838 * finds a task by its pid in the specified namespace
1839 * find_task_by_vpid():
1840 * finds a task by its virtual pid
1842 * see also find_vpid() etc in include/linux/pid.h
1845 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1846 struct pid_namespace *ns);
1848 extern struct task_struct *find_task_by_vpid(pid_t nr);
1849 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1850 struct pid_namespace *ns);
1852 extern void __set_special_pids(struct pid *pid);
1854 /* per-UID process charging. */
1855 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1856 static inline struct user_struct *get_uid(struct user_struct *u)
1858 atomic_inc(&u->__count);
1861 extern void free_uid(struct user_struct *);
1862 extern void release_uids(struct user_namespace *ns);
1864 #include <asm/current.h>
1866 extern void do_timer(unsigned long ticks);
1868 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1869 extern int wake_up_process(struct task_struct *tsk);
1870 extern void wake_up_new_task(struct task_struct *tsk,
1871 unsigned long clone_flags);
1873 extern void kick_process(struct task_struct *tsk);
1875 static inline void kick_process(struct task_struct *tsk) { }
1877 extern void sched_fork(struct task_struct *p, int clone_flags);
1878 extern void sched_dead(struct task_struct *p);
1880 extern void proc_caches_init(void);
1881 extern void flush_signals(struct task_struct *);
1882 extern void ignore_signals(struct task_struct *);
1883 extern void flush_signal_handlers(struct task_struct *, int force_default);
1884 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1886 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1888 unsigned long flags;
1891 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1892 ret = dequeue_signal(tsk, mask, info);
1893 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1898 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1900 extern void unblock_all_signals(void);
1901 extern void release_task(struct task_struct * p);
1902 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1903 extern int force_sigsegv(int, struct task_struct *);
1904 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1905 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1906 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1907 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1908 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1909 extern int kill_pid(struct pid *pid, int sig, int priv);
1910 extern int kill_proc_info(int, struct siginfo *, pid_t);
1911 extern int do_notify_parent(struct task_struct *, int);
1912 extern void force_sig(int, struct task_struct *);
1913 extern void force_sig_specific(int, struct task_struct *);
1914 extern int send_sig(int, struct task_struct *, int);
1915 extern void zap_other_threads(struct task_struct *p);
1916 extern struct sigqueue *sigqueue_alloc(void);
1917 extern void sigqueue_free(struct sigqueue *);
1918 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1919 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1920 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1922 static inline int kill_cad_pid(int sig, int priv)
1924 return kill_pid(cad_pid, sig, priv);
1927 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1928 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1929 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1930 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1932 static inline int is_si_special(const struct siginfo *info)
1934 return info <= SEND_SIG_FORCED;
1937 /* True if we are on the alternate signal stack. */
1939 static inline int on_sig_stack(unsigned long sp)
1941 return (sp - current->sas_ss_sp < current->sas_ss_size);
1944 static inline int sas_ss_flags(unsigned long sp)
1946 return (current->sas_ss_size == 0 ? SS_DISABLE
1947 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1951 * Routines for handling mm_structs
1953 extern struct mm_struct * mm_alloc(void);
1955 /* mmdrop drops the mm and the page tables */
1956 extern void __mmdrop(struct mm_struct *);
1957 static inline void mmdrop(struct mm_struct * mm)
1959 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1963 /* mmput gets rid of the mappings and all user-space */
1964 extern void mmput(struct mm_struct *);
1965 /* Grab a reference to a task's mm, if it is not already going away */
1966 extern struct mm_struct *get_task_mm(struct task_struct *task);
1967 /* Remove the current tasks stale references to the old mm_struct */
1968 extern void mm_release(struct task_struct *, struct mm_struct *);
1969 /* Allocate a new mm structure and copy contents from tsk->mm */
1970 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1972 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1973 extern void flush_thread(void);
1974 extern void exit_thread(void);
1976 extern void exit_files(struct task_struct *);
1977 extern void __cleanup_signal(struct signal_struct *);
1978 extern void __cleanup_sighand(struct sighand_struct *);
1980 extern void exit_itimers(struct signal_struct *);
1981 extern void flush_itimer_signals(void);
1983 extern NORET_TYPE void do_group_exit(int);
1985 extern void daemonize(const char *, ...);
1986 extern int allow_signal(int);
1987 extern int disallow_signal(int);
1989 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1990 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1991 struct task_struct *fork_idle(int);
1993 extern void set_task_comm(struct task_struct *tsk, char *from);
1994 extern char *get_task_comm(char *to, struct task_struct *tsk);
1997 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1999 static inline unsigned long wait_task_inactive(struct task_struct *p,
2006 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2008 #define for_each_process(p) \
2009 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2011 extern bool is_single_threaded(struct task_struct *);
2014 * Careful: do_each_thread/while_each_thread is a double loop so
2015 * 'break' will not work as expected - use goto instead.
2017 #define do_each_thread(g, t) \
2018 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2020 #define while_each_thread(g, t) \
2021 while ((t = next_thread(t)) != g)
2023 /* de_thread depends on thread_group_leader not being a pid based check */
2024 #define thread_group_leader(p) (p == p->group_leader)
2026 /* Do to the insanities of de_thread it is possible for a process
2027 * to have the pid of the thread group leader without actually being
2028 * the thread group leader. For iteration through the pids in proc
2029 * all we care about is that we have a task with the appropriate
2030 * pid, we don't actually care if we have the right task.
2032 static inline int has_group_leader_pid(struct task_struct *p)
2034 return p->pid == p->tgid;
2038 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2040 return p1->tgid == p2->tgid;
2043 static inline struct task_struct *next_thread(const struct task_struct *p)
2045 return list_entry(rcu_dereference(p->thread_group.next),
2046 struct task_struct, thread_group);
2049 static inline int thread_group_empty(struct task_struct *p)
2051 return list_empty(&p->thread_group);
2054 #define delay_group_leader(p) \
2055 (thread_group_leader(p) && !thread_group_empty(p))
2058 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2059 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2060 * pins the final release of task.io_context. Also protects ->cpuset and
2061 * ->cgroup.subsys[].
2063 * Nests both inside and outside of read_lock(&tasklist_lock).
2064 * It must not be nested with write_lock_irq(&tasklist_lock),
2065 * neither inside nor outside.
2067 static inline void task_lock(struct task_struct *p)
2069 spin_lock(&p->alloc_lock);
2072 static inline void task_unlock(struct task_struct *p)
2074 spin_unlock(&p->alloc_lock);
2077 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2078 unsigned long *flags);
2080 static inline void unlock_task_sighand(struct task_struct *tsk,
2081 unsigned long *flags)
2083 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2086 #ifndef __HAVE_THREAD_FUNCTIONS
2088 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2089 #define task_stack_page(task) ((task)->stack)
2091 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2093 *task_thread_info(p) = *task_thread_info(org);
2094 task_thread_info(p)->task = p;
2097 static inline unsigned long *end_of_stack(struct task_struct *p)
2099 return (unsigned long *)(task_thread_info(p) + 1);
2104 static inline int object_is_on_stack(void *obj)
2106 void *stack = task_stack_page(current);
2108 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2111 extern void thread_info_cache_init(void);
2113 #ifdef CONFIG_DEBUG_STACK_USAGE
2114 static inline unsigned long stack_not_used(struct task_struct *p)
2116 unsigned long *n = end_of_stack(p);
2118 do { /* Skip over canary */
2122 return (unsigned long)n - (unsigned long)end_of_stack(p);
2126 /* set thread flags in other task's structures
2127 * - see asm/thread_info.h for TIF_xxxx flags available
2129 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2131 set_ti_thread_flag(task_thread_info(tsk), flag);
2134 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2136 clear_ti_thread_flag(task_thread_info(tsk), flag);
2139 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2141 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2144 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2146 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2149 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2151 return test_ti_thread_flag(task_thread_info(tsk), flag);
2154 static inline void set_tsk_need_resched(struct task_struct *tsk)
2156 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2159 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2161 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2164 static inline int test_tsk_need_resched(struct task_struct *tsk)
2166 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2169 static inline int signal_pending(struct task_struct *p)
2171 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2174 extern int __fatal_signal_pending(struct task_struct *p);
2176 static inline int fatal_signal_pending(struct task_struct *p)
2178 return signal_pending(p) && __fatal_signal_pending(p);
2181 static inline int signal_pending_state(long state, struct task_struct *p)
2183 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2185 if (!signal_pending(p))
2188 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2191 static inline int need_resched(void)
2193 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2197 * cond_resched() and cond_resched_lock(): latency reduction via
2198 * explicit rescheduling in places that are safe. The return
2199 * value indicates whether a reschedule was done in fact.
2200 * cond_resched_lock() will drop the spinlock before scheduling,
2201 * cond_resched_softirq() will enable bhs before scheduling.
2203 extern int _cond_resched(void);
2204 #ifdef CONFIG_PREEMPT_BKL
2205 static inline int cond_resched(void)
2210 static inline int cond_resched(void)
2212 return _cond_resched();
2215 extern int cond_resched_lock(spinlock_t * lock);
2216 extern int cond_resched_softirq(void);
2217 static inline int cond_resched_bkl(void)
2219 return _cond_resched();
2223 * Does a critical section need to be broken due to another
2224 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2225 * but a general need for low latency)
2227 static inline int spin_needbreak(spinlock_t *lock)
2229 #ifdef CONFIG_PREEMPT
2230 return spin_is_contended(lock);
2237 * Thread group CPU time accounting.
2239 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2240 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2242 static inline void thread_group_cputime_init(struct signal_struct *sig)
2244 sig->cputimer.cputime = INIT_CPUTIME;
2245 spin_lock_init(&sig->cputimer.lock);
2246 sig->cputimer.running = 0;
2249 static inline void thread_group_cputime_free(struct signal_struct *sig)
2254 * Reevaluate whether the task has signals pending delivery.
2255 * Wake the task if so.
2256 * This is required every time the blocked sigset_t changes.
2257 * callers must hold sighand->siglock.
2259 extern void recalc_sigpending_and_wake(struct task_struct *t);
2260 extern void recalc_sigpending(void);
2262 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2265 * Wrappers for p->thread_info->cpu access. No-op on UP.
2269 static inline unsigned int task_cpu(const struct task_struct *p)
2271 return task_thread_info(p)->cpu;
2274 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2278 static inline unsigned int task_cpu(const struct task_struct *p)
2283 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2287 #endif /* CONFIG_SMP */
2289 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2291 #ifdef CONFIG_TRACING
2293 __trace_special(void *__tr, void *__data,
2294 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2297 __trace_special(void *__tr, void *__data,
2298 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2303 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2304 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2306 extern void normalize_rt_tasks(void);
2308 #ifdef CONFIG_GROUP_SCHED
2310 extern struct task_group init_task_group;
2311 #ifdef CONFIG_USER_SCHED
2312 extern struct task_group root_task_group;
2313 extern void set_tg_uid(struct user_struct *user);
2316 extern struct task_group *sched_create_group(struct task_group *parent);
2317 extern void sched_destroy_group(struct task_group *tg);
2318 extern void sched_move_task(struct task_struct *tsk);
2319 #ifdef CONFIG_FAIR_GROUP_SCHED
2320 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2321 extern unsigned long sched_group_shares(struct task_group *tg);
2323 #ifdef CONFIG_RT_GROUP_SCHED
2324 extern int sched_group_set_rt_runtime(struct task_group *tg,
2325 long rt_runtime_us);
2326 extern long sched_group_rt_runtime(struct task_group *tg);
2327 extern int sched_group_set_rt_period(struct task_group *tg,
2329 extern long sched_group_rt_period(struct task_group *tg);
2330 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2334 extern int task_can_switch_user(struct user_struct *up,
2335 struct task_struct *tsk);
2337 #ifdef CONFIG_TASK_XACCT
2338 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2340 tsk->ioac.rchar += amt;
2343 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2345 tsk->ioac.wchar += amt;
2348 static inline void inc_syscr(struct task_struct *tsk)
2353 static inline void inc_syscw(struct task_struct *tsk)
2358 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2362 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2366 static inline void inc_syscr(struct task_struct *tsk)
2370 static inline void inc_syscw(struct task_struct *tsk)
2375 #ifndef TASK_SIZE_OF
2376 #define TASK_SIZE_OF(tsk) TASK_SIZE
2379 #ifdef CONFIG_MM_OWNER
2380 extern void mm_update_next_owner(struct mm_struct *mm);
2381 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2383 static inline void mm_update_next_owner(struct mm_struct *mm)
2387 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2390 #endif /* CONFIG_MM_OWNER */
2392 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2394 #endif /* __KERNEL__ */