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/perf_counter.h>
75 #include <linux/pid.h>
76 #include <linux/percpu.h>
77 #include <linux/topology.h>
78 #include <linux/proportions.h>
79 #include <linux/seccomp.h>
80 #include <linux/rcupdate.h>
81 #include <linux/rtmutex.h>
83 #include <linux/time.h>
84 #include <linux/param.h>
85 #include <linux/resource.h>
86 #include <linux/timer.h>
87 #include <linux/hrtimer.h>
88 #include <linux/task_io_accounting.h>
89 #include <linux/kobject.h>
90 #include <linux/latencytop.h>
91 #include <linux/cred.h>
93 #include <asm/processor.h>
97 struct futex_pi_state;
98 struct robust_list_head;
104 * List of flags we want to share for kernel threads,
105 * if only because they are not used by them anyway.
107 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
110 * These are the constant used to fake the fixed-point load-average
111 * counting. Some notes:
112 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
113 * a load-average precision of 10 bits integer + 11 bits fractional
114 * - if you want to count load-averages more often, you need more
115 * precision, or rounding will get you. With 2-second counting freq,
116 * the EXP_n values would be 1981, 2034 and 2043 if still using only
119 extern unsigned long avenrun[]; /* Load averages */
121 #define FSHIFT 11 /* nr of bits of precision */
122 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
123 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
124 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
125 #define EXP_5 2014 /* 1/exp(5sec/5min) */
126 #define EXP_15 2037 /* 1/exp(5sec/15min) */
128 #define CALC_LOAD(load,exp,n) \
130 load += n*(FIXED_1-exp); \
133 extern unsigned long total_forks;
134 extern int nr_threads;
135 DECLARE_PER_CPU(unsigned long, process_counts);
136 extern int nr_processes(void);
137 extern unsigned long nr_running(void);
138 extern unsigned long nr_uninterruptible(void);
139 extern unsigned long nr_active(void);
140 extern unsigned long nr_iowait(void);
141 extern u64 cpu_nr_migrations(int cpu);
143 extern unsigned long get_parent_ip(unsigned long addr);
148 #ifdef CONFIG_SCHED_DEBUG
149 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
150 extern void proc_sched_set_task(struct task_struct *p);
152 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
155 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
158 static inline void proc_sched_set_task(struct task_struct *p)
162 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
167 extern unsigned long long time_sync_thresh;
170 * Task state bitmask. NOTE! These bits are also
171 * encoded in fs/proc/array.c: get_task_state().
173 * We have two separate sets of flags: task->state
174 * is about runnability, while task->exit_state are
175 * about the task exiting. Confusing, but this way
176 * modifying one set can't modify the other one by
179 #define TASK_RUNNING 0
180 #define TASK_INTERRUPTIBLE 1
181 #define TASK_UNINTERRUPTIBLE 2
182 #define __TASK_STOPPED 4
183 #define __TASK_TRACED 8
184 /* in tsk->exit_state */
185 #define EXIT_ZOMBIE 16
187 /* in tsk->state again */
189 #define TASK_WAKEKILL 128
191 /* Convenience macros for the sake of set_task_state */
192 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
193 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
194 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
196 /* Convenience macros for the sake of wake_up */
197 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
198 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
200 /* get_task_state() */
201 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
202 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
205 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
206 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
207 #define task_is_stopped_or_traced(task) \
208 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
209 #define task_contributes_to_load(task) \
210 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
212 #define __set_task_state(tsk, state_value) \
213 do { (tsk)->state = (state_value); } while (0)
214 #define set_task_state(tsk, state_value) \
215 set_mb((tsk)->state, (state_value))
218 * set_current_state() includes a barrier so that the write of current->state
219 * is correctly serialised wrt the caller's subsequent test of whether to
222 * set_current_state(TASK_UNINTERRUPTIBLE);
223 * if (do_i_need_to_sleep())
226 * If the caller does not need such serialisation then use __set_current_state()
228 #define __set_current_state(state_value) \
229 do { current->state = (state_value); } while (0)
230 #define set_current_state(state_value) \
231 set_mb(current->state, (state_value))
233 /* Task command name length */
234 #define TASK_COMM_LEN 16
236 #include <linux/spinlock.h>
239 * This serializes "schedule()" and also protects
240 * the run-queue from deletions/modifications (but
241 * _adding_ to the beginning of the run-queue has
244 extern rwlock_t tasklist_lock;
245 extern spinlock_t mmlist_lock;
249 extern void sched_init(void);
250 extern void sched_init_smp(void);
251 extern asmlinkage void schedule_tail(struct task_struct *prev);
252 extern void init_idle(struct task_struct *idle, int cpu);
253 extern void init_idle_bootup_task(struct task_struct *idle);
255 extern int runqueue_is_locked(void);
256 extern void task_rq_unlock_wait(struct task_struct *p);
258 extern cpumask_var_t nohz_cpu_mask;
259 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
260 extern int select_nohz_load_balancer(int cpu);
262 static inline int select_nohz_load_balancer(int cpu)
269 * Only dump TASK_* tasks. (0 for all tasks)
271 extern void show_state_filter(unsigned long state_filter);
273 static inline void show_state(void)
275 show_state_filter(0);
278 extern void show_regs(struct pt_regs *);
281 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
282 * task), SP is the stack pointer of the first frame that should be shown in the back
283 * trace (or NULL if the entire call-chain of the task should be shown).
285 extern void show_stack(struct task_struct *task, unsigned long *sp);
287 void io_schedule(void);
288 long io_schedule_timeout(long timeout);
290 extern void cpu_init (void);
291 extern void trap_init(void);
292 extern void update_process_times(int user);
293 extern void scheduler_tick(void);
295 extern void sched_show_task(struct task_struct *p);
297 #ifdef CONFIG_DETECT_SOFTLOCKUP
298 extern void softlockup_tick(void);
299 extern void touch_softlockup_watchdog(void);
300 extern void touch_all_softlockup_watchdogs(void);
301 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
302 struct file *filp, void __user *buffer,
303 size_t *lenp, loff_t *ppos);
304 extern unsigned int softlockup_panic;
305 extern unsigned long sysctl_hung_task_check_count;
306 extern unsigned long sysctl_hung_task_timeout_secs;
307 extern unsigned long sysctl_hung_task_warnings;
308 extern int softlockup_thresh;
310 static inline void softlockup_tick(void)
313 static inline void spawn_softlockup_task(void)
316 static inline void touch_softlockup_watchdog(void)
319 static inline void touch_all_softlockup_watchdogs(void)
325 /* Attach to any functions which should be ignored in wchan output. */
326 #define __sched __attribute__((__section__(".sched.text")))
328 /* Linker adds these: start and end of __sched functions */
329 extern char __sched_text_start[], __sched_text_end[];
331 /* Is this address in the __sched functions? */
332 extern int in_sched_functions(unsigned long addr);
334 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
335 extern signed long schedule_timeout(signed long timeout);
336 extern signed long schedule_timeout_interruptible(signed long timeout);
337 extern signed long schedule_timeout_killable(signed long timeout);
338 extern signed long schedule_timeout_uninterruptible(signed long timeout);
339 asmlinkage void __schedule(void);
340 asmlinkage void schedule(void);
341 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
344 struct user_namespace;
346 /* Maximum number of active map areas.. This is a random (large) number */
347 #define DEFAULT_MAX_MAP_COUNT 65536
349 extern int sysctl_max_map_count;
351 #include <linux/aio.h>
354 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
355 unsigned long, unsigned long);
357 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
358 unsigned long len, unsigned long pgoff,
359 unsigned long flags);
360 extern void arch_unmap_area(struct mm_struct *, unsigned long);
361 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
363 #if USE_SPLIT_PTLOCKS
365 * The mm counters are not protected by its page_table_lock,
366 * so must be incremented atomically.
368 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
369 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
370 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
371 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
372 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
374 #else /* !USE_SPLIT_PTLOCKS */
376 * The mm counters are protected by its page_table_lock,
377 * so can be incremented directly.
379 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
380 #define get_mm_counter(mm, member) ((mm)->_##member)
381 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
382 #define inc_mm_counter(mm, member) (mm)->_##member++
383 #define dec_mm_counter(mm, member) (mm)->_##member--
385 #endif /* !USE_SPLIT_PTLOCKS */
387 #define get_mm_rss(mm) \
388 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
389 #define update_hiwater_rss(mm) do { \
390 unsigned long _rss = get_mm_rss(mm); \
391 if ((mm)->hiwater_rss < _rss) \
392 (mm)->hiwater_rss = _rss; \
394 #define update_hiwater_vm(mm) do { \
395 if ((mm)->hiwater_vm < (mm)->total_vm) \
396 (mm)->hiwater_vm = (mm)->total_vm; \
399 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
401 return max(mm->hiwater_rss, get_mm_rss(mm));
404 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
406 return max(mm->hiwater_vm, mm->total_vm);
409 extern void set_dumpable(struct mm_struct *mm, int value);
410 extern int get_dumpable(struct mm_struct *mm);
414 #define MMF_DUMPABLE 0 /* core dump is permitted */
415 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
416 #define MMF_DUMPABLE_BITS 2
418 /* coredump filter bits */
419 #define MMF_DUMP_ANON_PRIVATE 2
420 #define MMF_DUMP_ANON_SHARED 3
421 #define MMF_DUMP_MAPPED_PRIVATE 4
422 #define MMF_DUMP_MAPPED_SHARED 5
423 #define MMF_DUMP_ELF_HEADERS 6
424 #define MMF_DUMP_HUGETLB_PRIVATE 7
425 #define MMF_DUMP_HUGETLB_SHARED 8
426 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
427 #define MMF_DUMP_FILTER_BITS 7
428 #define MMF_DUMP_FILTER_MASK \
429 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
430 #define MMF_DUMP_FILTER_DEFAULT \
431 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
432 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
434 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
435 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
437 # define MMF_DUMP_MASK_DEFAULT_ELF 0
440 struct sighand_struct {
442 struct k_sigaction action[_NSIG];
444 wait_queue_head_t signalfd_wqh;
447 struct pacct_struct {
450 unsigned long ac_mem;
451 cputime_t ac_utime, ac_stime;
452 unsigned long ac_minflt, ac_majflt;
456 * struct task_cputime - collected CPU time counts
457 * @utime: time spent in user mode, in &cputime_t units
458 * @stime: time spent in kernel mode, in &cputime_t units
459 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
461 * This structure groups together three kinds of CPU time that are
462 * tracked for threads and thread groups. Most things considering
463 * CPU time want to group these counts together and treat all three
464 * of them in parallel.
466 struct task_cputime {
469 unsigned long long sum_exec_runtime;
471 /* Alternate field names when used to cache expirations. */
472 #define prof_exp stime
473 #define virt_exp utime
474 #define sched_exp sum_exec_runtime
476 #define INIT_CPUTIME \
477 (struct task_cputime) { \
478 .utime = cputime_zero, \
479 .stime = cputime_zero, \
480 .sum_exec_runtime = 0, \
484 * struct thread_group_cputimer - thread group interval timer counts
485 * @cputime: thread group interval timers.
486 * @running: non-zero when there are timers running and
487 * @cputime receives updates.
488 * @lock: lock for fields in this struct.
490 * This structure contains the version of task_cputime, above, that is
491 * used for thread group CPU timer calculations.
493 struct thread_group_cputimer {
494 struct task_cputime cputime;
500 * NOTE! "signal_struct" does not have it's own
501 * locking, because a shared signal_struct always
502 * implies a shared sighand_struct, so locking
503 * sighand_struct is always a proper superset of
504 * the locking of signal_struct.
506 struct signal_struct {
510 wait_queue_head_t wait_chldexit; /* for wait4() */
512 /* current thread group signal load-balancing target: */
513 struct task_struct *curr_target;
515 /* shared signal handling: */
516 struct sigpending shared_pending;
518 /* thread group exit support */
521 * - notify group_exit_task when ->count is equal to notify_count
522 * - everyone except group_exit_task is stopped during signal delivery
523 * of fatal signals, group_exit_task processes the signal.
526 struct task_struct *group_exit_task;
528 /* thread group stop support, overloads group_exit_code too */
529 int group_stop_count;
530 unsigned int flags; /* see SIGNAL_* flags below */
532 /* POSIX.1b Interval Timers */
533 struct list_head posix_timers;
535 /* ITIMER_REAL timer for the process */
536 struct hrtimer real_timer;
537 struct pid *leader_pid;
538 ktime_t it_real_incr;
540 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
541 cputime_t it_prof_expires, it_virt_expires;
542 cputime_t it_prof_incr, it_virt_incr;
545 * Thread group totals for process CPU timers.
546 * See thread_group_cputimer(), et al, for details.
548 struct thread_group_cputimer cputimer;
550 /* Earliest-expiration cache. */
551 struct task_cputime cputime_expires;
553 struct list_head cpu_timers[3];
555 struct pid *tty_old_pgrp;
557 /* boolean value for session group leader */
560 struct tty_struct *tty; /* NULL if no tty */
563 * Cumulative resource counters for dead threads in the group,
564 * and for reaped dead child processes forked by this group.
565 * Live threads maintain their own counters and add to these
566 * in __exit_signal, except for the group leader.
568 cputime_t utime, stime, cutime, cstime;
571 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
572 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
573 unsigned long inblock, oublock, cinblock, coublock;
574 struct task_io_accounting ioac;
577 * Cumulative ns of schedule CPU time fo dead threads in the
578 * group, not including a zombie group leader, (This only differs
579 * from jiffies_to_ns(utime + stime) if sched_clock uses something
580 * other than jiffies.)
582 unsigned long long sum_sched_runtime;
585 * We don't bother to synchronize most readers of this at all,
586 * because there is no reader checking a limit that actually needs
587 * to get both rlim_cur and rlim_max atomically, and either one
588 * alone is a single word that can safely be read normally.
589 * getrlimit/setrlimit use task_lock(current->group_leader) to
590 * protect this instead of the siglock, because they really
591 * have no need to disable irqs.
593 struct rlimit rlim[RLIM_NLIMITS];
595 #ifdef CONFIG_BSD_PROCESS_ACCT
596 struct pacct_struct pacct; /* per-process accounting information */
598 #ifdef CONFIG_TASKSTATS
599 struct taskstats *stats;
603 struct tty_audit_buf *tty_audit_buf;
607 /* Context switch must be unlocked if interrupts are to be enabled */
608 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
609 # define __ARCH_WANT_UNLOCKED_CTXSW
613 * Bits in flags field of signal_struct.
615 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
616 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
617 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
618 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
620 * Pending notifications to parent.
622 #define SIGNAL_CLD_STOPPED 0x00000010
623 #define SIGNAL_CLD_CONTINUED 0x00000020
624 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
626 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
628 /* If true, all threads except ->group_exit_task have pending SIGKILL */
629 static inline int signal_group_exit(const struct signal_struct *sig)
631 return (sig->flags & SIGNAL_GROUP_EXIT) ||
632 (sig->group_exit_task != NULL);
636 * Some day this will be a full-fledged user tracking system..
639 atomic_t __count; /* reference count */
640 atomic_t processes; /* How many processes does this user have? */
641 atomic_t files; /* How many open files does this user have? */
642 atomic_t sigpending; /* How many pending signals does this user have? */
643 #ifdef CONFIG_INOTIFY_USER
644 atomic_t inotify_watches; /* How many inotify watches does this user have? */
645 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
648 atomic_t epoll_watches; /* The number of file descriptors currently watched */
650 #ifdef CONFIG_POSIX_MQUEUE
651 /* protected by mq_lock */
652 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
654 unsigned long locked_shm; /* How many pages of mlocked shm ? */
657 struct key *uid_keyring; /* UID specific keyring */
658 struct key *session_keyring; /* UID's default session keyring */
661 /* Hash table maintenance information */
662 struct hlist_node uidhash_node;
664 struct user_namespace *user_ns;
666 #ifdef CONFIG_USER_SCHED
667 struct task_group *tg;
670 struct work_struct work;
675 extern int uids_sysfs_init(void);
677 extern struct user_struct *find_user(uid_t);
679 extern struct user_struct root_user;
680 #define INIT_USER (&root_user)
683 struct backing_dev_info;
684 struct reclaim_state;
686 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
688 /* cumulative counters */
689 unsigned long pcount; /* # of times run on this cpu */
690 unsigned long long run_delay; /* time spent waiting on a runqueue */
693 unsigned long long last_arrival,/* when we last ran on a cpu */
694 last_queued; /* when we were last queued to run */
695 #ifdef CONFIG_SCHEDSTATS
697 unsigned int bkl_count;
700 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
702 #ifdef CONFIG_TASK_DELAY_ACCT
703 struct task_delay_info {
705 unsigned int flags; /* Private per-task flags */
707 /* For each stat XXX, add following, aligned appropriately
709 * struct timespec XXX_start, XXX_end;
713 * Atomicity of updates to XXX_delay, XXX_count protected by
714 * single lock above (split into XXX_lock if contention is an issue).
718 * XXX_count is incremented on every XXX operation, the delay
719 * associated with the operation is added to XXX_delay.
720 * XXX_delay contains the accumulated delay time in nanoseconds.
722 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
723 u64 blkio_delay; /* wait for sync block io completion */
724 u64 swapin_delay; /* wait for swapin block io completion */
725 u32 blkio_count; /* total count of the number of sync block */
726 /* io operations performed */
727 u32 swapin_count; /* total count of the number of swapin block */
728 /* io operations performed */
730 struct timespec freepages_start, freepages_end;
731 u64 freepages_delay; /* wait for memory reclaim */
732 u32 freepages_count; /* total count of memory reclaim */
734 #endif /* CONFIG_TASK_DELAY_ACCT */
736 static inline int sched_info_on(void)
738 #ifdef CONFIG_SCHEDSTATS
740 #elif defined(CONFIG_TASK_DELAY_ACCT)
741 extern int delayacct_on;
756 * sched-domains (multiprocessor balancing) declarations:
760 * Increase resolution of nice-level calculations:
762 #define SCHED_LOAD_SHIFT 10
763 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
765 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
768 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
769 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
770 #define SD_BALANCE_EXEC 4 /* Balance on exec */
771 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
772 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
773 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
774 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
775 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
776 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
777 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
778 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
779 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
781 enum powersavings_balance_level {
782 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
783 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
784 * first for long running threads
786 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
787 * cpu package for power savings
789 MAX_POWERSAVINGS_BALANCE_LEVELS
792 extern int sched_mc_power_savings, sched_smt_power_savings;
794 static inline int sd_balance_for_mc_power(void)
796 if (sched_smt_power_savings)
797 return SD_POWERSAVINGS_BALANCE;
802 static inline int sd_balance_for_package_power(void)
804 if (sched_mc_power_savings | sched_smt_power_savings)
805 return SD_POWERSAVINGS_BALANCE;
811 * Optimise SD flags for power savings:
812 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
813 * Keep default SD flags if sched_{smt,mc}_power_saving=0
816 static inline int sd_power_saving_flags(void)
818 if (sched_mc_power_savings | sched_smt_power_savings)
819 return SD_BALANCE_NEWIDLE;
825 struct sched_group *next; /* Must be a circular list */
828 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
829 * single CPU. This is read only (except for setup, hotplug CPU).
830 * Note : Never change cpu_power without recompute its reciprocal
832 unsigned int __cpu_power;
834 * reciprocal value of cpu_power to avoid expensive divides
835 * (see include/linux/reciprocal_div.h)
837 u32 reciprocal_cpu_power;
839 unsigned long cpumask[];
842 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
844 return to_cpumask(sg->cpumask);
847 enum sched_domain_level {
857 struct sched_domain_attr {
858 int relax_domain_level;
861 #define SD_ATTR_INIT (struct sched_domain_attr) { \
862 .relax_domain_level = -1, \
865 struct sched_domain {
866 /* These fields must be setup */
867 struct sched_domain *parent; /* top domain must be null terminated */
868 struct sched_domain *child; /* bottom domain must be null terminated */
869 struct sched_group *groups; /* the balancing groups of the domain */
870 unsigned long min_interval; /* Minimum balance interval ms */
871 unsigned long max_interval; /* Maximum balance interval ms */
872 unsigned int busy_factor; /* less balancing by factor if busy */
873 unsigned int imbalance_pct; /* No balance until over watermark */
874 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
875 unsigned int busy_idx;
876 unsigned int idle_idx;
877 unsigned int newidle_idx;
878 unsigned int wake_idx;
879 unsigned int forkexec_idx;
880 int flags; /* See SD_* */
881 enum sched_domain_level level;
883 /* Runtime fields. */
884 unsigned long last_balance; /* init to jiffies. units in jiffies */
885 unsigned int balance_interval; /* initialise to 1. units in ms. */
886 unsigned int nr_balance_failed; /* initialise to 0 */
890 #ifdef CONFIG_SCHEDSTATS
891 /* load_balance() stats */
892 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
893 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
894 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
895 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
896 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
897 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
898 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
899 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
901 /* Active load balancing */
902 unsigned int alb_count;
903 unsigned int alb_failed;
904 unsigned int alb_pushed;
906 /* SD_BALANCE_EXEC stats */
907 unsigned int sbe_count;
908 unsigned int sbe_balanced;
909 unsigned int sbe_pushed;
911 /* SD_BALANCE_FORK stats */
912 unsigned int sbf_count;
913 unsigned int sbf_balanced;
914 unsigned int sbf_pushed;
916 /* try_to_wake_up() stats */
917 unsigned int ttwu_wake_remote;
918 unsigned int ttwu_move_affine;
919 unsigned int ttwu_move_balance;
921 #ifdef CONFIG_SCHED_DEBUG
925 /* span of all CPUs in this domain */
926 unsigned long span[];
929 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
931 return to_cpumask(sd->span);
934 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
935 struct sched_domain_attr *dattr_new);
937 /* Test a flag in parent sched domain */
938 static inline int test_sd_parent(struct sched_domain *sd, int flag)
940 if (sd->parent && (sd->parent->flags & flag))
946 #else /* CONFIG_SMP */
948 struct sched_domain_attr;
951 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
952 struct sched_domain_attr *dattr_new)
955 #endif /* !CONFIG_SMP */
957 struct io_context; /* See blkdev.h */
960 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
961 extern void prefetch_stack(struct task_struct *t);
963 static inline void prefetch_stack(struct task_struct *t) { }
966 struct audit_context; /* See audit.c */
968 struct pipe_inode_info;
969 struct uts_namespace;
975 const struct sched_class *next;
977 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
978 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
979 void (*yield_task) (struct rq *rq);
981 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
983 struct task_struct * (*pick_next_task) (struct rq *rq);
984 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
987 int (*select_task_rq)(struct task_struct *p, int sync);
989 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
990 struct rq *busiest, unsigned long max_load_move,
991 struct sched_domain *sd, enum cpu_idle_type idle,
992 int *all_pinned, int *this_best_prio);
994 int (*move_one_task) (struct rq *this_rq, int this_cpu,
995 struct rq *busiest, struct sched_domain *sd,
996 enum cpu_idle_type idle);
997 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
998 int (*needs_post_schedule) (struct rq *this_rq);
999 void (*post_schedule) (struct rq *this_rq);
1000 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1002 void (*set_cpus_allowed)(struct task_struct *p,
1003 const struct cpumask *newmask);
1005 void (*rq_online)(struct rq *rq);
1006 void (*rq_offline)(struct rq *rq);
1009 void (*set_curr_task) (struct rq *rq);
1010 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1011 void (*task_new) (struct rq *rq, struct task_struct *p);
1013 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1015 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1017 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1018 int oldprio, int running);
1020 #ifdef CONFIG_FAIR_GROUP_SCHED
1021 void (*moved_group) (struct task_struct *p);
1025 struct load_weight {
1026 unsigned long weight, inv_weight;
1030 * CFS stats for a schedulable entity (task, task-group etc)
1032 * Current field usage histogram:
1039 struct sched_entity {
1040 struct load_weight load; /* for load-balancing */
1041 struct rb_node run_node;
1042 struct list_head group_node;
1046 u64 sum_exec_runtime;
1048 u64 prev_sum_exec_runtime;
1058 #ifdef CONFIG_SCHEDSTATS
1066 s64 sum_sleep_runtime;
1073 u64 nr_migrations_cold;
1074 u64 nr_failed_migrations_affine;
1075 u64 nr_failed_migrations_running;
1076 u64 nr_failed_migrations_hot;
1077 u64 nr_forced_migrations;
1078 u64 nr_forced2_migrations;
1081 u64 nr_wakeups_sync;
1082 u64 nr_wakeups_migrate;
1083 u64 nr_wakeups_local;
1084 u64 nr_wakeups_remote;
1085 u64 nr_wakeups_affine;
1086 u64 nr_wakeups_affine_attempts;
1087 u64 nr_wakeups_passive;
1088 u64 nr_wakeups_idle;
1091 #ifdef CONFIG_FAIR_GROUP_SCHED
1092 struct sched_entity *parent;
1093 /* rq on which this entity is (to be) queued: */
1094 struct cfs_rq *cfs_rq;
1095 /* rq "owned" by this entity/group: */
1096 struct cfs_rq *my_q;
1100 struct sched_rt_entity {
1101 struct list_head run_list;
1102 unsigned long timeout;
1103 unsigned int time_slice;
1104 int nr_cpus_allowed;
1106 struct sched_rt_entity *back;
1107 #ifdef CONFIG_RT_GROUP_SCHED
1108 struct sched_rt_entity *parent;
1109 /* rq on which this entity is (to be) queued: */
1110 struct rt_rq *rt_rq;
1111 /* rq "owned" by this entity/group: */
1116 struct task_struct {
1117 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1120 unsigned int flags; /* per process flags, defined below */
1121 unsigned int ptrace;
1123 int lock_depth; /* BKL lock depth */
1126 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1131 int prio, static_prio, normal_prio;
1132 unsigned int rt_priority;
1133 const struct sched_class *sched_class;
1134 struct sched_entity se;
1135 struct sched_rt_entity rt;
1137 #ifdef CONFIG_PREEMPT_NOTIFIERS
1138 /* list of struct preempt_notifier: */
1139 struct hlist_head preempt_notifiers;
1143 * fpu_counter contains the number of consecutive context switches
1144 * that the FPU is used. If this is over a threshold, the lazy fpu
1145 * saving becomes unlazy to save the trap. This is an unsigned char
1146 * so that after 256 times the counter wraps and the behavior turns
1147 * lazy again; this to deal with bursty apps that only use FPU for
1150 unsigned char fpu_counter;
1151 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1152 #ifdef CONFIG_BLK_DEV_IO_TRACE
1153 unsigned int btrace_seq;
1156 unsigned int policy;
1157 cpumask_t cpus_allowed;
1159 #ifdef CONFIG_PREEMPT_RCU
1160 int rcu_read_lock_nesting;
1161 int rcu_flipctr_idx;
1162 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1164 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1165 struct sched_info sched_info;
1168 struct list_head tasks;
1169 struct plist_node pushable_tasks;
1171 struct mm_struct *mm, *active_mm;
1174 struct linux_binfmt *binfmt;
1176 int exit_code, exit_signal;
1177 int pdeath_signal; /* The signal sent when the parent dies */
1179 unsigned int personality;
1180 unsigned did_exec:1;
1181 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1186 /* Canary value for the -fstack-protector gcc feature */
1187 unsigned long stack_canary;
1190 * pointers to (original) parent process, youngest child, younger sibling,
1191 * older sibling, respectively. (p->father can be replaced with
1192 * p->real_parent->pid)
1194 struct task_struct *real_parent; /* real parent process */
1195 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1197 * children/sibling forms the list of my natural children
1199 struct list_head children; /* list of my children */
1200 struct list_head sibling; /* linkage in my parent's children list */
1201 struct task_struct *group_leader; /* threadgroup leader */
1204 * ptraced is the list of tasks this task is using ptrace on.
1205 * This includes both natural children and PTRACE_ATTACH targets.
1206 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1208 struct list_head ptraced;
1209 struct list_head ptrace_entry;
1211 #ifdef CONFIG_X86_PTRACE_BTS
1213 * This is the tracer handle for the ptrace BTS extension.
1214 * This field actually belongs to the ptracer task.
1216 struct bts_tracer *bts;
1218 * The buffer to hold the BTS data.
1222 #endif /* CONFIG_X86_PTRACE_BTS */
1224 /* PID/PID hash table linkage. */
1225 struct pid_link pids[PIDTYPE_MAX];
1226 struct list_head thread_group;
1228 struct completion *vfork_done; /* for vfork() */
1229 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1230 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1232 cputime_t utime, stime, utimescaled, stimescaled;
1234 cputime_t prev_utime, prev_stime;
1235 unsigned long nvcsw, nivcsw; /* context switch counts */
1236 struct timespec start_time; /* monotonic time */
1237 struct timespec real_start_time; /* boot based time */
1238 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1239 unsigned long min_flt, maj_flt;
1241 struct task_cputime cputime_expires;
1242 struct list_head cpu_timers[3];
1244 /* process credentials */
1245 const struct cred *real_cred; /* objective and real subjective task
1246 * credentials (COW) */
1247 const struct cred *cred; /* effective (overridable) subjective task
1248 * credentials (COW) */
1249 struct mutex cred_exec_mutex; /* execve vs ptrace cred calculation mutex */
1251 char comm[TASK_COMM_LEN]; /* executable name excluding path
1252 - access with [gs]et_task_comm (which lock
1253 it with task_lock())
1254 - initialized normally by flush_old_exec */
1255 /* file system info */
1256 int link_count, total_link_count;
1257 #ifdef CONFIG_SYSVIPC
1259 struct sysv_sem sysvsem;
1261 #ifdef CONFIG_DETECT_SOFTLOCKUP
1262 /* hung task detection */
1263 unsigned long last_switch_timestamp;
1264 unsigned long last_switch_count;
1266 /* CPU-specific state of this task */
1267 struct thread_struct thread;
1268 /* filesystem information */
1269 struct fs_struct *fs;
1270 /* open file information */
1271 struct files_struct *files;
1273 struct nsproxy *nsproxy;
1274 /* signal handlers */
1275 struct signal_struct *signal;
1276 struct sighand_struct *sighand;
1278 sigset_t blocked, real_blocked;
1279 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1280 struct sigpending pending;
1282 unsigned long sas_ss_sp;
1284 int (*notifier)(void *priv);
1285 void *notifier_data;
1286 sigset_t *notifier_mask;
1287 struct audit_context *audit_context;
1288 #ifdef CONFIG_AUDITSYSCALL
1290 unsigned int sessionid;
1294 /* Thread group tracking */
1297 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1298 spinlock_t alloc_lock;
1300 /* Protection of the PI data structures: */
1303 #ifdef CONFIG_RT_MUTEXES
1304 /* PI waiters blocked on a rt_mutex held by this task */
1305 struct plist_head pi_waiters;
1306 /* Deadlock detection and priority inheritance handling */
1307 struct rt_mutex_waiter *pi_blocked_on;
1310 #ifdef CONFIG_DEBUG_MUTEXES
1311 /* mutex deadlock detection */
1312 struct mutex_waiter *blocked_on;
1314 #ifdef CONFIG_TRACE_IRQFLAGS
1315 unsigned int irq_events;
1316 int hardirqs_enabled;
1317 unsigned long hardirq_enable_ip;
1318 unsigned int hardirq_enable_event;
1319 unsigned long hardirq_disable_ip;
1320 unsigned int hardirq_disable_event;
1321 int softirqs_enabled;
1322 unsigned long softirq_disable_ip;
1323 unsigned int softirq_disable_event;
1324 unsigned long softirq_enable_ip;
1325 unsigned int softirq_enable_event;
1326 int hardirq_context;
1327 int softirq_context;
1329 #ifdef CONFIG_LOCKDEP
1330 # define MAX_LOCK_DEPTH 48UL
1333 unsigned int lockdep_recursion;
1334 struct held_lock held_locks[MAX_LOCK_DEPTH];
1335 gfp_t lockdep_reclaim_gfp;
1338 /* journalling filesystem info */
1341 /* stacked block device info */
1342 struct bio *bio_list, **bio_tail;
1345 struct reclaim_state *reclaim_state;
1347 struct backing_dev_info *backing_dev_info;
1349 struct io_context *io_context;
1351 unsigned long ptrace_message;
1352 siginfo_t *last_siginfo; /* For ptrace use. */
1353 struct task_io_accounting ioac;
1354 #if defined(CONFIG_TASK_XACCT)
1355 u64 acct_rss_mem1; /* accumulated rss usage */
1356 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1357 cputime_t acct_timexpd; /* stime + utime since last update */
1359 #ifdef CONFIG_CPUSETS
1360 nodemask_t mems_allowed;
1361 int cpuset_mems_generation;
1362 int cpuset_mem_spread_rotor;
1364 #ifdef CONFIG_CGROUPS
1365 /* Control Group info protected by css_set_lock */
1366 struct css_set *cgroups;
1367 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1368 struct list_head cg_list;
1371 struct robust_list_head __user *robust_list;
1372 #ifdef CONFIG_COMPAT
1373 struct compat_robust_list_head __user *compat_robust_list;
1375 struct list_head pi_state_list;
1376 struct futex_pi_state *pi_state_cache;
1378 struct perf_counter_context perf_counter_ctx;
1380 struct mempolicy *mempolicy;
1383 atomic_t fs_excl; /* holding fs exclusive resources */
1384 struct rcu_head rcu;
1387 * cache last used pipe for splice
1389 struct pipe_inode_info *splice_pipe;
1390 #ifdef CONFIG_TASK_DELAY_ACCT
1391 struct task_delay_info *delays;
1393 #ifdef CONFIG_FAULT_INJECTION
1396 struct prop_local_single dirties;
1397 #ifdef CONFIG_LATENCYTOP
1398 int latency_record_count;
1399 struct latency_record latency_record[LT_SAVECOUNT];
1402 * time slack values; these are used to round up poll() and
1403 * select() etc timeout values. These are in nanoseconds.
1405 unsigned long timer_slack_ns;
1406 unsigned long default_timer_slack_ns;
1408 struct list_head *scm_work_list;
1409 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1410 /* Index of current stored adress in ret_stack */
1412 /* Stack of return addresses for return function tracing */
1413 struct ftrace_ret_stack *ret_stack;
1414 /* time stamp for last schedule */
1415 unsigned long long ftrace_timestamp;
1417 * Number of functions that haven't been traced
1418 * because of depth overrun.
1420 atomic_t trace_overrun;
1421 /* Pause for the tracing */
1422 atomic_t tracing_graph_pause;
1424 #ifdef CONFIG_TRACING
1425 /* state flags for use by tracers */
1426 unsigned long trace;
1430 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1431 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1434 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1435 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1436 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1437 * values are inverted: lower p->prio value means higher priority.
1439 * The MAX_USER_RT_PRIO value allows the actual maximum
1440 * RT priority to be separate from the value exported to
1441 * user-space. This allows kernel threads to set their
1442 * priority to a value higher than any user task. Note:
1443 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1446 #define MAX_USER_RT_PRIO 100
1447 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1449 #define MAX_PRIO (MAX_RT_PRIO + 40)
1450 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1452 static inline int rt_prio(int prio)
1454 if (unlikely(prio < MAX_RT_PRIO))
1459 static inline int rt_task(struct task_struct *p)
1461 return rt_prio(p->prio);
1464 static inline struct pid *task_pid(struct task_struct *task)
1466 return task->pids[PIDTYPE_PID].pid;
1469 static inline struct pid *task_tgid(struct task_struct *task)
1471 return task->group_leader->pids[PIDTYPE_PID].pid;
1475 * Without tasklist or rcu lock it is not safe to dereference
1476 * the result of task_pgrp/task_session even if task == current,
1477 * we can race with another thread doing sys_setsid/sys_setpgid.
1479 static inline struct pid *task_pgrp(struct task_struct *task)
1481 return task->group_leader->pids[PIDTYPE_PGID].pid;
1484 static inline struct pid *task_session(struct task_struct *task)
1486 return task->group_leader->pids[PIDTYPE_SID].pid;
1489 struct pid_namespace;
1492 * the helpers to get the task's different pids as they are seen
1493 * from various namespaces
1495 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1496 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1498 * task_xid_nr_ns() : id seen from the ns specified;
1500 * set_task_vxid() : assigns a virtual id to a task;
1502 * see also pid_nr() etc in include/linux/pid.h
1504 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1505 struct pid_namespace *ns);
1507 static inline pid_t task_pid_nr(struct task_struct *tsk)
1512 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1513 struct pid_namespace *ns)
1515 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1518 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1520 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1524 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1529 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1531 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1533 return pid_vnr(task_tgid(tsk));
1537 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1538 struct pid_namespace *ns)
1540 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1543 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1545 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1549 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1550 struct pid_namespace *ns)
1552 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1555 static inline pid_t task_session_vnr(struct task_struct *tsk)
1557 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1560 /* obsolete, do not use */
1561 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1563 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1567 * pid_alive - check that a task structure is not stale
1568 * @p: Task structure to be checked.
1570 * Test if a process is not yet dead (at most zombie state)
1571 * If pid_alive fails, then pointers within the task structure
1572 * can be stale and must not be dereferenced.
1574 static inline int pid_alive(struct task_struct *p)
1576 return p->pids[PIDTYPE_PID].pid != NULL;
1580 * is_global_init - check if a task structure is init
1581 * @tsk: Task structure to be checked.
1583 * Check if a task structure is the first user space task the kernel created.
1585 static inline int is_global_init(struct task_struct *tsk)
1587 return tsk->pid == 1;
1591 * is_container_init:
1592 * check whether in the task is init in its own pid namespace.
1594 extern int is_container_init(struct task_struct *tsk);
1596 extern struct pid *cad_pid;
1598 extern void free_task(struct task_struct *tsk);
1599 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1601 extern void __put_task_struct(struct task_struct *t);
1603 static inline void put_task_struct(struct task_struct *t)
1605 if (atomic_dec_and_test(&t->usage))
1606 __put_task_struct(t);
1609 extern cputime_t task_utime(struct task_struct *p);
1610 extern cputime_t task_stime(struct task_struct *p);
1611 extern cputime_t task_gtime(struct task_struct *p);
1616 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1617 /* Not implemented yet, only for 486*/
1618 #define PF_STARTING 0x00000002 /* being created */
1619 #define PF_EXITING 0x00000004 /* getting shut down */
1620 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1621 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1622 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1623 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1624 #define PF_DUMPCORE 0x00000200 /* dumped core */
1625 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1626 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1627 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1628 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1629 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1630 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1631 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1632 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1633 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1634 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1635 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1636 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1637 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1638 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1639 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1640 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1641 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1642 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1643 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1644 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1647 * Only the _current_ task can read/write to tsk->flags, but other
1648 * tasks can access tsk->flags in readonly mode for example
1649 * with tsk_used_math (like during threaded core dumping).
1650 * There is however an exception to this rule during ptrace
1651 * or during fork: the ptracer task is allowed to write to the
1652 * child->flags of its traced child (same goes for fork, the parent
1653 * can write to the child->flags), because we're guaranteed the
1654 * child is not running and in turn not changing child->flags
1655 * at the same time the parent does it.
1657 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1658 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1659 #define clear_used_math() clear_stopped_child_used_math(current)
1660 #define set_used_math() set_stopped_child_used_math(current)
1661 #define conditional_stopped_child_used_math(condition, child) \
1662 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1663 #define conditional_used_math(condition) \
1664 conditional_stopped_child_used_math(condition, current)
1665 #define copy_to_stopped_child_used_math(child) \
1666 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1667 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1668 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1669 #define used_math() tsk_used_math(current)
1672 extern int set_cpus_allowed_ptr(struct task_struct *p,
1673 const struct cpumask *new_mask);
1675 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1676 const struct cpumask *new_mask)
1678 if (!cpumask_test_cpu(0, new_mask))
1683 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1685 return set_cpus_allowed_ptr(p, &new_mask);
1689 * Architectures can set this to 1 if they have specified
1690 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1691 * but then during bootup it turns out that sched_clock()
1692 * is reliable after all:
1694 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1695 extern int sched_clock_stable;
1698 extern unsigned long long sched_clock(void);
1700 extern void sched_clock_init(void);
1701 extern u64 sched_clock_cpu(int cpu);
1703 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1704 static inline void sched_clock_tick(void)
1708 static inline void sched_clock_idle_sleep_event(void)
1712 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1716 extern void sched_clock_tick(void);
1717 extern void sched_clock_idle_sleep_event(void);
1718 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1722 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1723 * clock constructed from sched_clock():
1725 extern unsigned long long cpu_clock(int cpu);
1727 extern unsigned long long
1728 task_sched_runtime(struct task_struct *task);
1729 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1731 /* sched_exec is called by processes performing an exec */
1733 extern void sched_exec(void);
1735 #define sched_exec() {}
1738 extern void sched_clock_idle_sleep_event(void);
1739 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1741 #ifdef CONFIG_HOTPLUG_CPU
1742 extern void idle_task_exit(void);
1744 static inline void idle_task_exit(void) {}
1747 extern void sched_idle_next(void);
1749 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1750 extern void wake_up_idle_cpu(int cpu);
1752 static inline void wake_up_idle_cpu(int cpu) { }
1755 extern unsigned int sysctl_sched_latency;
1756 extern unsigned int sysctl_sched_min_granularity;
1757 extern unsigned int sysctl_sched_wakeup_granularity;
1758 extern unsigned int sysctl_sched_shares_ratelimit;
1759 extern unsigned int sysctl_sched_shares_thresh;
1760 #ifdef CONFIG_SCHED_DEBUG
1761 extern unsigned int sysctl_sched_child_runs_first;
1762 extern unsigned int sysctl_sched_features;
1763 extern unsigned int sysctl_sched_migration_cost;
1764 extern unsigned int sysctl_sched_nr_migrate;
1766 int sched_nr_latency_handler(struct ctl_table *table, int write,
1767 struct file *file, void __user *buffer, size_t *length,
1770 extern unsigned int sysctl_sched_rt_period;
1771 extern int sysctl_sched_rt_runtime;
1773 int sched_rt_handler(struct ctl_table *table, int write,
1774 struct file *filp, void __user *buffer, size_t *lenp,
1777 extern unsigned int sysctl_sched_compat_yield;
1779 #ifdef CONFIG_RT_MUTEXES
1780 extern int rt_mutex_getprio(struct task_struct *p);
1781 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1782 extern void rt_mutex_adjust_pi(struct task_struct *p);
1784 static inline int rt_mutex_getprio(struct task_struct *p)
1786 return p->normal_prio;
1788 # define rt_mutex_adjust_pi(p) do { } while (0)
1791 extern void set_user_nice(struct task_struct *p, long nice);
1792 extern int task_prio(const struct task_struct *p);
1793 extern int task_nice(const struct task_struct *p);
1794 extern int can_nice(const struct task_struct *p, const int nice);
1795 extern int task_curr(const struct task_struct *p);
1796 extern int idle_cpu(int cpu);
1797 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1798 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1799 struct sched_param *);
1800 extern struct task_struct *idle_task(int cpu);
1801 extern struct task_struct *curr_task(int cpu);
1802 extern void set_curr_task(int cpu, struct task_struct *p);
1807 * The default (Linux) execution domain.
1809 extern struct exec_domain default_exec_domain;
1811 union thread_union {
1812 struct thread_info thread_info;
1813 unsigned long stack[THREAD_SIZE/sizeof(long)];
1816 #ifndef __HAVE_ARCH_KSTACK_END
1817 static inline int kstack_end(void *addr)
1819 /* Reliable end of stack detection:
1820 * Some APM bios versions misalign the stack
1822 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1826 extern union thread_union init_thread_union;
1827 extern struct task_struct init_task;
1829 extern struct mm_struct init_mm;
1831 extern struct pid_namespace init_pid_ns;
1834 * find a task by one of its numerical ids
1836 * find_task_by_pid_type_ns():
1837 * it is the most generic call - it finds a task by all id,
1838 * type and namespace specified
1839 * find_task_by_pid_ns():
1840 * finds a task by its pid in the specified namespace
1841 * find_task_by_vpid():
1842 * finds a task by its virtual pid
1844 * see also find_vpid() etc in include/linux/pid.h
1847 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1848 struct pid_namespace *ns);
1850 extern struct task_struct *find_task_by_vpid(pid_t nr);
1851 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1852 struct pid_namespace *ns);
1854 extern void __set_special_pids(struct pid *pid);
1856 /* per-UID process charging. */
1857 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1858 static inline struct user_struct *get_uid(struct user_struct *u)
1860 atomic_inc(&u->__count);
1863 extern void free_uid(struct user_struct *);
1864 extern void release_uids(struct user_namespace *ns);
1866 #include <asm/current.h>
1868 extern void do_timer(unsigned long ticks);
1870 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1871 extern int wake_up_process(struct task_struct *tsk);
1872 extern void wake_up_new_task(struct task_struct *tsk,
1873 unsigned long clone_flags);
1875 extern void kick_process(struct task_struct *tsk);
1877 static inline void kick_process(struct task_struct *tsk) { }
1879 extern void sched_fork(struct task_struct *p, int clone_flags);
1880 extern void sched_dead(struct task_struct *p);
1882 extern void proc_caches_init(void);
1883 extern void flush_signals(struct task_struct *);
1884 extern void ignore_signals(struct task_struct *);
1885 extern void flush_signal_handlers(struct task_struct *, int force_default);
1886 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1888 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1890 unsigned long flags;
1893 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1894 ret = dequeue_signal(tsk, mask, info);
1895 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1900 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1902 extern void unblock_all_signals(void);
1903 extern void release_task(struct task_struct * p);
1904 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1905 extern int force_sigsegv(int, struct task_struct *);
1906 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1907 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1908 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1909 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1910 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1911 extern int kill_pid(struct pid *pid, int sig, int priv);
1912 extern int kill_proc_info(int, struct siginfo *, pid_t);
1913 extern int do_notify_parent(struct task_struct *, int);
1914 extern void force_sig(int, struct task_struct *);
1915 extern void force_sig_specific(int, struct task_struct *);
1916 extern int send_sig(int, struct task_struct *, int);
1917 extern void zap_other_threads(struct task_struct *p);
1918 extern struct sigqueue *sigqueue_alloc(void);
1919 extern void sigqueue_free(struct sigqueue *);
1920 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1921 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1922 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1924 static inline int kill_cad_pid(int sig, int priv)
1926 return kill_pid(cad_pid, sig, priv);
1929 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1930 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1931 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1932 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1934 static inline int is_si_special(const struct siginfo *info)
1936 return info <= SEND_SIG_FORCED;
1939 /* True if we are on the alternate signal stack. */
1941 static inline int on_sig_stack(unsigned long sp)
1943 return (sp - current->sas_ss_sp < current->sas_ss_size);
1946 static inline int sas_ss_flags(unsigned long sp)
1948 return (current->sas_ss_size == 0 ? SS_DISABLE
1949 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1953 * Routines for handling mm_structs
1955 extern struct mm_struct * mm_alloc(void);
1957 /* mmdrop drops the mm and the page tables */
1958 extern void __mmdrop(struct mm_struct *);
1959 static inline void mmdrop(struct mm_struct * mm)
1961 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1965 /* mmput gets rid of the mappings and all user-space */
1966 extern void mmput(struct mm_struct *);
1967 /* Grab a reference to a task's mm, if it is not already going away */
1968 extern struct mm_struct *get_task_mm(struct task_struct *task);
1969 /* Remove the current tasks stale references to the old mm_struct */
1970 extern void mm_release(struct task_struct *, struct mm_struct *);
1971 /* Allocate a new mm structure and copy contents from tsk->mm */
1972 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1974 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1975 struct task_struct *, struct pt_regs *);
1976 extern void flush_thread(void);
1977 extern void exit_thread(void);
1979 extern void exit_files(struct task_struct *);
1980 extern void __cleanup_signal(struct signal_struct *);
1981 extern void __cleanup_sighand(struct sighand_struct *);
1983 extern void exit_itimers(struct signal_struct *);
1984 extern void flush_itimer_signals(void);
1986 extern NORET_TYPE void do_group_exit(int);
1988 extern void daemonize(const char *, ...);
1989 extern int allow_signal(int);
1990 extern int disallow_signal(int);
1992 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1993 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1994 struct task_struct *fork_idle(int);
1996 extern void set_task_comm(struct task_struct *tsk, char *from);
1997 extern char *get_task_comm(char *to, struct task_struct *tsk);
2000 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2002 static inline unsigned long wait_task_inactive(struct task_struct *p,
2009 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2011 #define for_each_process(p) \
2012 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2014 extern bool is_single_threaded(struct task_struct *);
2017 * Careful: do_each_thread/while_each_thread is a double loop so
2018 * 'break' will not work as expected - use goto instead.
2020 #define do_each_thread(g, t) \
2021 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2023 #define while_each_thread(g, t) \
2024 while ((t = next_thread(t)) != g)
2026 /* de_thread depends on thread_group_leader not being a pid based check */
2027 #define thread_group_leader(p) (p == p->group_leader)
2029 /* Do to the insanities of de_thread it is possible for a process
2030 * to have the pid of the thread group leader without actually being
2031 * the thread group leader. For iteration through the pids in proc
2032 * all we care about is that we have a task with the appropriate
2033 * pid, we don't actually care if we have the right task.
2035 static inline int has_group_leader_pid(struct task_struct *p)
2037 return p->pid == p->tgid;
2041 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2043 return p1->tgid == p2->tgid;
2046 static inline struct task_struct *next_thread(const struct task_struct *p)
2048 return list_entry(rcu_dereference(p->thread_group.next),
2049 struct task_struct, thread_group);
2052 static inline int thread_group_empty(struct task_struct *p)
2054 return list_empty(&p->thread_group);
2057 #define delay_group_leader(p) \
2058 (thread_group_leader(p) && !thread_group_empty(p))
2060 static inline int task_detached(struct task_struct *p)
2062 return p->exit_signal == -1;
2066 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2067 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2068 * pins the final release of task.io_context. Also protects ->cpuset and
2069 * ->cgroup.subsys[].
2071 * Nests both inside and outside of read_lock(&tasklist_lock).
2072 * It must not be nested with write_lock_irq(&tasklist_lock),
2073 * neither inside nor outside.
2075 static inline void task_lock(struct task_struct *p)
2077 spin_lock(&p->alloc_lock);
2080 static inline void task_unlock(struct task_struct *p)
2082 spin_unlock(&p->alloc_lock);
2085 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2086 unsigned long *flags);
2088 static inline void unlock_task_sighand(struct task_struct *tsk,
2089 unsigned long *flags)
2091 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2094 #ifndef __HAVE_THREAD_FUNCTIONS
2096 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2097 #define task_stack_page(task) ((task)->stack)
2099 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2101 *task_thread_info(p) = *task_thread_info(org);
2102 task_thread_info(p)->task = p;
2105 static inline unsigned long *end_of_stack(struct task_struct *p)
2107 return (unsigned long *)(task_thread_info(p) + 1);
2112 static inline int object_is_on_stack(void *obj)
2114 void *stack = task_stack_page(current);
2116 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2119 extern void thread_info_cache_init(void);
2121 #ifdef CONFIG_DEBUG_STACK_USAGE
2122 static inline unsigned long stack_not_used(struct task_struct *p)
2124 unsigned long *n = end_of_stack(p);
2126 do { /* Skip over canary */
2130 return (unsigned long)n - (unsigned long)end_of_stack(p);
2134 /* set thread flags in other task's structures
2135 * - see asm/thread_info.h for TIF_xxxx flags available
2137 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2139 set_ti_thread_flag(task_thread_info(tsk), flag);
2142 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2144 clear_ti_thread_flag(task_thread_info(tsk), flag);
2147 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2149 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2152 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2154 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2157 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2159 return test_ti_thread_flag(task_thread_info(tsk), flag);
2162 static inline void set_tsk_need_resched(struct task_struct *tsk)
2164 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2167 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2169 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2172 static inline int test_tsk_need_resched(struct task_struct *tsk)
2174 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2177 static inline int signal_pending(struct task_struct *p)
2179 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2182 extern int __fatal_signal_pending(struct task_struct *p);
2184 static inline int fatal_signal_pending(struct task_struct *p)
2186 return signal_pending(p) && __fatal_signal_pending(p);
2189 static inline int signal_pending_state(long state, struct task_struct *p)
2191 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2193 if (!signal_pending(p))
2196 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2199 static inline int need_resched(void)
2201 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2205 * cond_resched() and cond_resched_lock(): latency reduction via
2206 * explicit rescheduling in places that are safe. The return
2207 * value indicates whether a reschedule was done in fact.
2208 * cond_resched_lock() will drop the spinlock before scheduling,
2209 * cond_resched_softirq() will enable bhs before scheduling.
2211 extern int _cond_resched(void);
2212 #ifdef CONFIG_PREEMPT_BKL
2213 static inline int cond_resched(void)
2218 static inline int cond_resched(void)
2220 return _cond_resched();
2223 extern int cond_resched_lock(spinlock_t * lock);
2224 extern int cond_resched_softirq(void);
2225 static inline int cond_resched_bkl(void)
2227 return _cond_resched();
2231 * Does a critical section need to be broken due to another
2232 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2233 * but a general need for low latency)
2235 static inline int spin_needbreak(spinlock_t *lock)
2237 #ifdef CONFIG_PREEMPT
2238 return spin_is_contended(lock);
2245 * Thread group CPU time accounting.
2247 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2248 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2250 static inline void thread_group_cputime_init(struct signal_struct *sig)
2252 sig->cputimer.cputime = INIT_CPUTIME;
2253 spin_lock_init(&sig->cputimer.lock);
2254 sig->cputimer.running = 0;
2257 static inline void thread_group_cputime_free(struct signal_struct *sig)
2262 * Reevaluate whether the task has signals pending delivery.
2263 * Wake the task if so.
2264 * This is required every time the blocked sigset_t changes.
2265 * callers must hold sighand->siglock.
2267 extern void recalc_sigpending_and_wake(struct task_struct *t);
2268 extern void recalc_sigpending(void);
2270 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2273 * Wrappers for p->thread_info->cpu access. No-op on UP.
2277 static inline unsigned int task_cpu(const struct task_struct *p)
2279 return task_thread_info(p)->cpu;
2282 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2286 static inline unsigned int task_cpu(const struct task_struct *p)
2291 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2295 #endif /* CONFIG_SMP */
2297 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2299 #ifdef CONFIG_TRACING
2301 __trace_special(void *__tr, void *__data,
2302 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2305 __trace_special(void *__tr, void *__data,
2306 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2311 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2312 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2314 extern void normalize_rt_tasks(void);
2316 #ifdef CONFIG_GROUP_SCHED
2318 extern struct task_group init_task_group;
2319 #ifdef CONFIG_USER_SCHED
2320 extern struct task_group root_task_group;
2321 extern void set_tg_uid(struct user_struct *user);
2324 extern struct task_group *sched_create_group(struct task_group *parent);
2325 extern void sched_destroy_group(struct task_group *tg);
2326 extern void sched_move_task(struct task_struct *tsk);
2327 #ifdef CONFIG_FAIR_GROUP_SCHED
2328 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2329 extern unsigned long sched_group_shares(struct task_group *tg);
2331 #ifdef CONFIG_RT_GROUP_SCHED
2332 extern int sched_group_set_rt_runtime(struct task_group *tg,
2333 long rt_runtime_us);
2334 extern long sched_group_rt_runtime(struct task_group *tg);
2335 extern int sched_group_set_rt_period(struct task_group *tg,
2337 extern long sched_group_rt_period(struct task_group *tg);
2338 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2342 extern int task_can_switch_user(struct user_struct *up,
2343 struct task_struct *tsk);
2345 #ifdef CONFIG_TASK_XACCT
2346 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2348 tsk->ioac.rchar += amt;
2351 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2353 tsk->ioac.wchar += amt;
2356 static inline void inc_syscr(struct task_struct *tsk)
2361 static inline void inc_syscw(struct task_struct *tsk)
2366 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2370 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2374 static inline void inc_syscr(struct task_struct *tsk)
2378 static inline void inc_syscw(struct task_struct *tsk)
2383 #ifndef TASK_SIZE_OF
2384 #define TASK_SIZE_OF(tsk) TASK_SIZE
2388 * Call the function if the target task is executing on a CPU right now:
2390 extern void task_oncpu_function_call(struct task_struct *p,
2391 void (*func) (void *info), void *info);
2394 #ifdef CONFIG_MM_OWNER
2395 extern void mm_update_next_owner(struct mm_struct *mm);
2396 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2398 static inline void mm_update_next_owner(struct mm_struct *mm)
2402 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2405 #endif /* CONFIG_MM_OWNER */
2407 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2409 #endif /* __KERNEL__ */