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 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
397 return max(mm->hiwater_rss, get_mm_rss(mm));
400 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
402 return max(mm->hiwater_vm, mm->total_vm);
405 extern void set_dumpable(struct mm_struct *mm, int value);
406 extern int get_dumpable(struct mm_struct *mm);
410 #define MMF_DUMPABLE 0 /* core dump is permitted */
411 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
412 #define MMF_DUMPABLE_BITS 2
414 /* coredump filter bits */
415 #define MMF_DUMP_ANON_PRIVATE 2
416 #define MMF_DUMP_ANON_SHARED 3
417 #define MMF_DUMP_MAPPED_PRIVATE 4
418 #define MMF_DUMP_MAPPED_SHARED 5
419 #define MMF_DUMP_ELF_HEADERS 6
420 #define MMF_DUMP_HUGETLB_PRIVATE 7
421 #define MMF_DUMP_HUGETLB_SHARED 8
422 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
423 #define MMF_DUMP_FILTER_BITS 7
424 #define MMF_DUMP_FILTER_MASK \
425 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
426 #define MMF_DUMP_FILTER_DEFAULT \
427 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
428 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
430 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
431 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
433 # define MMF_DUMP_MASK_DEFAULT_ELF 0
436 struct sighand_struct {
438 struct k_sigaction action[_NSIG];
440 wait_queue_head_t signalfd_wqh;
443 struct pacct_struct {
446 unsigned long ac_mem;
447 cputime_t ac_utime, ac_stime;
448 unsigned long ac_minflt, ac_majflt;
452 * struct task_cputime - collected CPU time counts
453 * @utime: time spent in user mode, in &cputime_t units
454 * @stime: time spent in kernel mode, in &cputime_t units
455 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
457 * This structure groups together three kinds of CPU time that are
458 * tracked for threads and thread groups. Most things considering
459 * CPU time want to group these counts together and treat all three
460 * of them in parallel.
462 struct task_cputime {
465 unsigned long long sum_exec_runtime;
467 /* Alternate field names when used to cache expirations. */
468 #define prof_exp stime
469 #define virt_exp utime
470 #define sched_exp sum_exec_runtime
472 #define INIT_CPUTIME \
473 (struct task_cputime) { \
474 .utime = cputime_zero, \
475 .stime = cputime_zero, \
476 .sum_exec_runtime = 0, \
480 * struct thread_group_cputimer - thread group interval timer counts
481 * @cputime: thread group interval timers.
482 * @running: non-zero when there are timers running and
483 * @cputime receives updates.
484 * @lock: lock for fields in this struct.
486 * This structure contains the version of task_cputime, above, that is
487 * used for thread group CPU timer calculations.
489 struct thread_group_cputimer {
490 struct task_cputime cputime;
496 * NOTE! "signal_struct" does not have it's own
497 * locking, because a shared signal_struct always
498 * implies a shared sighand_struct, so locking
499 * sighand_struct is always a proper superset of
500 * the locking of signal_struct.
502 struct signal_struct {
506 wait_queue_head_t wait_chldexit; /* for wait4() */
508 /* current thread group signal load-balancing target: */
509 struct task_struct *curr_target;
511 /* shared signal handling: */
512 struct sigpending shared_pending;
514 /* thread group exit support */
517 * - notify group_exit_task when ->count is equal to notify_count
518 * - everyone except group_exit_task is stopped during signal delivery
519 * of fatal signals, group_exit_task processes the signal.
522 struct task_struct *group_exit_task;
524 /* thread group stop support, overloads group_exit_code too */
525 int group_stop_count;
526 unsigned int flags; /* see SIGNAL_* flags below */
528 /* POSIX.1b Interval Timers */
529 struct list_head posix_timers;
531 /* ITIMER_REAL timer for the process */
532 struct hrtimer real_timer;
533 struct pid *leader_pid;
534 ktime_t it_real_incr;
536 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
537 cputime_t it_prof_expires, it_virt_expires;
538 cputime_t it_prof_incr, it_virt_incr;
541 * Thread group totals for process CPU timers.
542 * See thread_group_cputimer(), et al, for details.
544 struct thread_group_cputimer cputimer;
546 /* Earliest-expiration cache. */
547 struct task_cputime cputime_expires;
549 struct list_head cpu_timers[3];
551 struct pid *tty_old_pgrp;
553 /* boolean value for session group leader */
556 struct tty_struct *tty; /* NULL if no tty */
559 * Cumulative resource counters for dead threads in the group,
560 * and for reaped dead child processes forked by this group.
561 * Live threads maintain their own counters and add to these
562 * in __exit_signal, except for the group leader.
564 cputime_t utime, stime, cutime, cstime;
567 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
568 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
569 unsigned long inblock, oublock, cinblock, coublock;
570 struct task_io_accounting ioac;
573 * Cumulative ns of schedule CPU time fo dead threads in the
574 * group, not including a zombie group leader, (This only differs
575 * from jiffies_to_ns(utime + stime) if sched_clock uses something
576 * other than jiffies.)
578 unsigned long long sum_sched_runtime;
581 * We don't bother to synchronize most readers of this at all,
582 * because there is no reader checking a limit that actually needs
583 * to get both rlim_cur and rlim_max atomically, and either one
584 * alone is a single word that can safely be read normally.
585 * getrlimit/setrlimit use task_lock(current->group_leader) to
586 * protect this instead of the siglock, because they really
587 * have no need to disable irqs.
589 struct rlimit rlim[RLIM_NLIMITS];
591 #ifdef CONFIG_BSD_PROCESS_ACCT
592 struct pacct_struct pacct; /* per-process accounting information */
594 #ifdef CONFIG_TASKSTATS
595 struct taskstats *stats;
599 struct tty_audit_buf *tty_audit_buf;
603 /* Context switch must be unlocked if interrupts are to be enabled */
604 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
605 # define __ARCH_WANT_UNLOCKED_CTXSW
609 * Bits in flags field of signal_struct.
611 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
612 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
613 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
614 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
616 * Pending notifications to parent.
618 #define SIGNAL_CLD_STOPPED 0x00000010
619 #define SIGNAL_CLD_CONTINUED 0x00000020
620 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
622 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
624 /* If true, all threads except ->group_exit_task have pending SIGKILL */
625 static inline int signal_group_exit(const struct signal_struct *sig)
627 return (sig->flags & SIGNAL_GROUP_EXIT) ||
628 (sig->group_exit_task != NULL);
632 * Some day this will be a full-fledged user tracking system..
635 atomic_t __count; /* reference count */
636 atomic_t processes; /* How many processes does this user have? */
637 atomic_t files; /* How many open files does this user have? */
638 atomic_t sigpending; /* How many pending signals does this user have? */
639 #ifdef CONFIG_INOTIFY_USER
640 atomic_t inotify_watches; /* How many inotify watches does this user have? */
641 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
644 atomic_t epoll_watches; /* The number of file descriptors currently watched */
646 #ifdef CONFIG_POSIX_MQUEUE
647 /* protected by mq_lock */
648 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
650 unsigned long locked_shm; /* How many pages of mlocked shm ? */
653 struct key *uid_keyring; /* UID specific keyring */
654 struct key *session_keyring; /* UID's default session keyring */
657 /* Hash table maintenance information */
658 struct hlist_node uidhash_node;
660 struct user_namespace *user_ns;
662 #ifdef CONFIG_USER_SCHED
663 struct task_group *tg;
666 struct work_struct work;
671 extern int uids_sysfs_init(void);
673 extern struct user_struct *find_user(uid_t);
675 extern struct user_struct root_user;
676 #define INIT_USER (&root_user)
679 struct backing_dev_info;
680 struct reclaim_state;
682 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
684 /* cumulative counters */
685 unsigned long pcount; /* # of times run on this cpu */
686 unsigned long long run_delay; /* time spent waiting on a runqueue */
689 unsigned long long last_arrival,/* when we last ran on a cpu */
690 last_queued; /* when we were last queued to run */
691 #ifdef CONFIG_SCHEDSTATS
693 unsigned int bkl_count;
696 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
698 #ifdef CONFIG_TASK_DELAY_ACCT
699 struct task_delay_info {
701 unsigned int flags; /* Private per-task flags */
703 /* For each stat XXX, add following, aligned appropriately
705 * struct timespec XXX_start, XXX_end;
709 * Atomicity of updates to XXX_delay, XXX_count protected by
710 * single lock above (split into XXX_lock if contention is an issue).
714 * XXX_count is incremented on every XXX operation, the delay
715 * associated with the operation is added to XXX_delay.
716 * XXX_delay contains the accumulated delay time in nanoseconds.
718 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
719 u64 blkio_delay; /* wait for sync block io completion */
720 u64 swapin_delay; /* wait for swapin block io completion */
721 u32 blkio_count; /* total count of the number of sync block */
722 /* io operations performed */
723 u32 swapin_count; /* total count of the number of swapin block */
724 /* io operations performed */
726 struct timespec freepages_start, freepages_end;
727 u64 freepages_delay; /* wait for memory reclaim */
728 u32 freepages_count; /* total count of memory reclaim */
730 #endif /* CONFIG_TASK_DELAY_ACCT */
732 static inline int sched_info_on(void)
734 #ifdef CONFIG_SCHEDSTATS
736 #elif defined(CONFIG_TASK_DELAY_ACCT)
737 extern int delayacct_on;
752 * sched-domains (multiprocessor balancing) declarations:
756 * Increase resolution of nice-level calculations:
758 #define SCHED_LOAD_SHIFT 10
759 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
761 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
764 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
765 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
766 #define SD_BALANCE_EXEC 4 /* Balance on exec */
767 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
768 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
769 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
770 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
771 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
772 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
773 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
774 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
775 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
777 enum powersavings_balance_level {
778 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
779 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
780 * first for long running threads
782 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
783 * cpu package for power savings
785 MAX_POWERSAVINGS_BALANCE_LEVELS
788 extern int sched_mc_power_savings, sched_smt_power_savings;
790 static inline int sd_balance_for_mc_power(void)
792 if (sched_smt_power_savings)
793 return SD_POWERSAVINGS_BALANCE;
798 static inline int sd_balance_for_package_power(void)
800 if (sched_mc_power_savings | sched_smt_power_savings)
801 return SD_POWERSAVINGS_BALANCE;
807 * Optimise SD flags for power savings:
808 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
809 * Keep default SD flags if sched_{smt,mc}_power_saving=0
812 static inline int sd_power_saving_flags(void)
814 if (sched_mc_power_savings | sched_smt_power_savings)
815 return SD_BALANCE_NEWIDLE;
821 struct sched_group *next; /* Must be a circular list */
824 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
825 * single CPU. This is read only (except for setup, hotplug CPU).
826 * Note : Never change cpu_power without recompute its reciprocal
828 unsigned int __cpu_power;
830 * reciprocal value of cpu_power to avoid expensive divides
831 * (see include/linux/reciprocal_div.h)
833 u32 reciprocal_cpu_power;
835 unsigned long cpumask[];
838 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
840 return to_cpumask(sg->cpumask);
843 enum sched_domain_level {
853 struct sched_domain_attr {
854 int relax_domain_level;
857 #define SD_ATTR_INIT (struct sched_domain_attr) { \
858 .relax_domain_level = -1, \
861 struct sched_domain {
862 /* These fields must be setup */
863 struct sched_domain *parent; /* top domain must be null terminated */
864 struct sched_domain *child; /* bottom domain must be null terminated */
865 struct sched_group *groups; /* the balancing groups of the domain */
866 unsigned long min_interval; /* Minimum balance interval ms */
867 unsigned long max_interval; /* Maximum balance interval ms */
868 unsigned int busy_factor; /* less balancing by factor if busy */
869 unsigned int imbalance_pct; /* No balance until over watermark */
870 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
871 unsigned int busy_idx;
872 unsigned int idle_idx;
873 unsigned int newidle_idx;
874 unsigned int wake_idx;
875 unsigned int forkexec_idx;
876 int flags; /* See SD_* */
877 enum sched_domain_level level;
879 /* Runtime fields. */
880 unsigned long last_balance; /* init to jiffies. units in jiffies */
881 unsigned int balance_interval; /* initialise to 1. units in ms. */
882 unsigned int nr_balance_failed; /* initialise to 0 */
886 #ifdef CONFIG_SCHEDSTATS
887 /* load_balance() stats */
888 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
889 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
890 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
891 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
892 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
893 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
894 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
895 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
897 /* Active load balancing */
898 unsigned int alb_count;
899 unsigned int alb_failed;
900 unsigned int alb_pushed;
902 /* SD_BALANCE_EXEC stats */
903 unsigned int sbe_count;
904 unsigned int sbe_balanced;
905 unsigned int sbe_pushed;
907 /* SD_BALANCE_FORK stats */
908 unsigned int sbf_count;
909 unsigned int sbf_balanced;
910 unsigned int sbf_pushed;
912 /* try_to_wake_up() stats */
913 unsigned int ttwu_wake_remote;
914 unsigned int ttwu_move_affine;
915 unsigned int ttwu_move_balance;
917 #ifdef CONFIG_SCHED_DEBUG
921 /* span of all CPUs in this domain */
922 unsigned long span[];
925 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
927 return to_cpumask(sd->span);
930 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
931 struct sched_domain_attr *dattr_new);
933 /* Test a flag in parent sched domain */
934 static inline int test_sd_parent(struct sched_domain *sd, int flag)
936 if (sd->parent && (sd->parent->flags & flag))
942 #else /* CONFIG_SMP */
944 struct sched_domain_attr;
947 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
948 struct sched_domain_attr *dattr_new)
951 #endif /* !CONFIG_SMP */
953 struct io_context; /* See blkdev.h */
956 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
957 extern void prefetch_stack(struct task_struct *t);
959 static inline void prefetch_stack(struct task_struct *t) { }
962 struct audit_context; /* See audit.c */
964 struct pipe_inode_info;
965 struct uts_namespace;
971 const struct sched_class *next;
973 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
974 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
975 void (*yield_task) (struct rq *rq);
977 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
979 struct task_struct * (*pick_next_task) (struct rq *rq);
980 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
983 int (*select_task_rq)(struct task_struct *p, int sync);
985 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
986 struct rq *busiest, unsigned long max_load_move,
987 struct sched_domain *sd, enum cpu_idle_type idle,
988 int *all_pinned, int *this_best_prio);
990 int (*move_one_task) (struct rq *this_rq, int this_cpu,
991 struct rq *busiest, struct sched_domain *sd,
992 enum cpu_idle_type idle);
993 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
994 int (*needs_post_schedule) (struct rq *this_rq);
995 void (*post_schedule) (struct rq *this_rq);
996 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
998 void (*set_cpus_allowed)(struct task_struct *p,
999 const struct cpumask *newmask);
1001 void (*rq_online)(struct rq *rq);
1002 void (*rq_offline)(struct rq *rq);
1005 void (*set_curr_task) (struct rq *rq);
1006 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1007 void (*task_new) (struct rq *rq, struct task_struct *p);
1009 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1011 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1013 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1014 int oldprio, int running);
1016 #ifdef CONFIG_FAIR_GROUP_SCHED
1017 void (*moved_group) (struct task_struct *p);
1021 struct load_weight {
1022 unsigned long weight, inv_weight;
1026 * CFS stats for a schedulable entity (task, task-group etc)
1028 * Current field usage histogram:
1035 struct sched_entity {
1036 struct load_weight load; /* for load-balancing */
1037 struct rb_node run_node;
1038 struct list_head group_node;
1042 u64 sum_exec_runtime;
1044 u64 prev_sum_exec_runtime;
1053 #ifdef CONFIG_SCHEDSTATS
1061 s64 sum_sleep_runtime;
1068 u64 nr_migrations_cold;
1069 u64 nr_failed_migrations_affine;
1070 u64 nr_failed_migrations_running;
1071 u64 nr_failed_migrations_hot;
1072 u64 nr_forced_migrations;
1073 u64 nr_forced2_migrations;
1076 u64 nr_wakeups_sync;
1077 u64 nr_wakeups_migrate;
1078 u64 nr_wakeups_local;
1079 u64 nr_wakeups_remote;
1080 u64 nr_wakeups_affine;
1081 u64 nr_wakeups_affine_attempts;
1082 u64 nr_wakeups_passive;
1083 u64 nr_wakeups_idle;
1086 #ifdef CONFIG_FAIR_GROUP_SCHED
1087 struct sched_entity *parent;
1088 /* rq on which this entity is (to be) queued: */
1089 struct cfs_rq *cfs_rq;
1090 /* rq "owned" by this entity/group: */
1091 struct cfs_rq *my_q;
1095 struct sched_rt_entity {
1096 struct list_head run_list;
1097 unsigned long timeout;
1098 unsigned int time_slice;
1099 int nr_cpus_allowed;
1101 struct sched_rt_entity *back;
1102 #ifdef CONFIG_RT_GROUP_SCHED
1103 struct sched_rt_entity *parent;
1104 /* rq on which this entity is (to be) queued: */
1105 struct rt_rq *rt_rq;
1106 /* rq "owned" by this entity/group: */
1111 struct task_struct {
1112 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1115 unsigned int flags; /* per process flags, defined below */
1116 unsigned int ptrace;
1118 int lock_depth; /* BKL lock depth */
1121 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1126 int prio, static_prio, normal_prio;
1127 unsigned int rt_priority;
1128 const struct sched_class *sched_class;
1129 struct sched_entity se;
1130 struct sched_rt_entity rt;
1132 #ifdef CONFIG_PREEMPT_NOTIFIERS
1133 /* list of struct preempt_notifier: */
1134 struct hlist_head preempt_notifiers;
1138 * fpu_counter contains the number of consecutive context switches
1139 * that the FPU is used. If this is over a threshold, the lazy fpu
1140 * saving becomes unlazy to save the trap. This is an unsigned char
1141 * so that after 256 times the counter wraps and the behavior turns
1142 * lazy again; this to deal with bursty apps that only use FPU for
1145 unsigned char fpu_counter;
1146 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1147 #ifdef CONFIG_BLK_DEV_IO_TRACE
1148 unsigned int btrace_seq;
1151 unsigned int policy;
1152 cpumask_t cpus_allowed;
1154 #ifdef CONFIG_PREEMPT_RCU
1155 int rcu_read_lock_nesting;
1156 int rcu_flipctr_idx;
1157 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1159 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1160 struct sched_info sched_info;
1163 struct list_head tasks;
1164 struct plist_node pushable_tasks;
1166 struct mm_struct *mm, *active_mm;
1169 struct linux_binfmt *binfmt;
1171 int exit_code, exit_signal;
1172 int pdeath_signal; /* The signal sent when the parent dies */
1174 unsigned int personality;
1175 unsigned did_exec:1;
1176 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1181 /* Canary value for the -fstack-protector gcc feature */
1182 unsigned long stack_canary;
1185 * pointers to (original) parent process, youngest child, younger sibling,
1186 * older sibling, respectively. (p->father can be replaced with
1187 * p->real_parent->pid)
1189 struct task_struct *real_parent; /* real parent process */
1190 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1192 * children/sibling forms the list of my natural children
1194 struct list_head children; /* list of my children */
1195 struct list_head sibling; /* linkage in my parent's children list */
1196 struct task_struct *group_leader; /* threadgroup leader */
1199 * ptraced is the list of tasks this task is using ptrace on.
1200 * This includes both natural children and PTRACE_ATTACH targets.
1201 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1203 struct list_head ptraced;
1204 struct list_head ptrace_entry;
1206 #ifdef CONFIG_X86_PTRACE_BTS
1208 * This is the tracer handle for the ptrace BTS extension.
1209 * This field actually belongs to the ptracer task.
1211 struct bts_tracer *bts;
1213 * The buffer to hold the BTS data.
1217 #endif /* CONFIG_X86_PTRACE_BTS */
1219 /* PID/PID hash table linkage. */
1220 struct pid_link pids[PIDTYPE_MAX];
1221 struct list_head thread_group;
1223 struct completion *vfork_done; /* for vfork() */
1224 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1225 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1227 cputime_t utime, stime, utimescaled, stimescaled;
1229 cputime_t prev_utime, prev_stime;
1230 unsigned long nvcsw, nivcsw; /* context switch counts */
1231 struct timespec start_time; /* monotonic time */
1232 struct timespec real_start_time; /* boot based time */
1233 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1234 unsigned long min_flt, maj_flt;
1236 struct task_cputime cputime_expires;
1237 struct list_head cpu_timers[3];
1239 /* process credentials */
1240 const struct cred *real_cred; /* objective and real subjective task
1241 * credentials (COW) */
1242 const struct cred *cred; /* effective (overridable) subjective task
1243 * credentials (COW) */
1244 struct mutex cred_exec_mutex; /* execve vs ptrace cred calculation mutex */
1246 char comm[TASK_COMM_LEN]; /* executable name excluding path
1247 - access with [gs]et_task_comm (which lock
1248 it with task_lock())
1249 - initialized normally by flush_old_exec */
1250 /* file system info */
1251 int link_count, total_link_count;
1252 #ifdef CONFIG_SYSVIPC
1254 struct sysv_sem sysvsem;
1256 #ifdef CONFIG_DETECT_SOFTLOCKUP
1257 /* hung task detection */
1258 unsigned long last_switch_timestamp;
1259 unsigned long last_switch_count;
1261 /* CPU-specific state of this task */
1262 struct thread_struct thread;
1263 /* filesystem information */
1264 struct fs_struct *fs;
1265 /* open file information */
1266 struct files_struct *files;
1268 struct nsproxy *nsproxy;
1269 /* signal handlers */
1270 struct signal_struct *signal;
1271 struct sighand_struct *sighand;
1273 sigset_t blocked, real_blocked;
1274 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1275 struct sigpending pending;
1277 unsigned long sas_ss_sp;
1279 int (*notifier)(void *priv);
1280 void *notifier_data;
1281 sigset_t *notifier_mask;
1282 struct audit_context *audit_context;
1283 #ifdef CONFIG_AUDITSYSCALL
1285 unsigned int sessionid;
1289 /* Thread group tracking */
1292 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1293 spinlock_t alloc_lock;
1295 /* Protection of the PI data structures: */
1298 #ifdef CONFIG_RT_MUTEXES
1299 /* PI waiters blocked on a rt_mutex held by this task */
1300 struct plist_head pi_waiters;
1301 /* Deadlock detection and priority inheritance handling */
1302 struct rt_mutex_waiter *pi_blocked_on;
1305 #ifdef CONFIG_DEBUG_MUTEXES
1306 /* mutex deadlock detection */
1307 struct mutex_waiter *blocked_on;
1309 #ifdef CONFIG_TRACE_IRQFLAGS
1310 unsigned int irq_events;
1311 int hardirqs_enabled;
1312 unsigned long hardirq_enable_ip;
1313 unsigned int hardirq_enable_event;
1314 unsigned long hardirq_disable_ip;
1315 unsigned int hardirq_disable_event;
1316 int softirqs_enabled;
1317 unsigned long softirq_disable_ip;
1318 unsigned int softirq_disable_event;
1319 unsigned long softirq_enable_ip;
1320 unsigned int softirq_enable_event;
1321 int hardirq_context;
1322 int softirq_context;
1324 #ifdef CONFIG_LOCKDEP
1325 # define MAX_LOCK_DEPTH 48UL
1328 unsigned int lockdep_recursion;
1329 struct held_lock held_locks[MAX_LOCK_DEPTH];
1330 gfp_t lockdep_reclaim_gfp;
1333 /* journalling filesystem info */
1336 /* stacked block device info */
1337 struct bio *bio_list, **bio_tail;
1340 struct reclaim_state *reclaim_state;
1342 struct backing_dev_info *backing_dev_info;
1344 struct io_context *io_context;
1346 unsigned long ptrace_message;
1347 siginfo_t *last_siginfo; /* For ptrace use. */
1348 struct task_io_accounting ioac;
1349 #if defined(CONFIG_TASK_XACCT)
1350 u64 acct_rss_mem1; /* accumulated rss usage */
1351 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1352 cputime_t acct_timexpd; /* stime + utime since last update */
1354 #ifdef CONFIG_CPUSETS
1355 nodemask_t mems_allowed;
1356 int cpuset_mems_generation;
1357 int cpuset_mem_spread_rotor;
1359 #ifdef CONFIG_CGROUPS
1360 /* Control Group info protected by css_set_lock */
1361 struct css_set *cgroups;
1362 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1363 struct list_head cg_list;
1366 struct robust_list_head __user *robust_list;
1367 #ifdef CONFIG_COMPAT
1368 struct compat_robust_list_head __user *compat_robust_list;
1370 struct list_head pi_state_list;
1371 struct futex_pi_state *pi_state_cache;
1374 struct mempolicy *mempolicy;
1377 atomic_t fs_excl; /* holding fs exclusive resources */
1378 struct rcu_head rcu;
1381 * cache last used pipe for splice
1383 struct pipe_inode_info *splice_pipe;
1384 #ifdef CONFIG_TASK_DELAY_ACCT
1385 struct task_delay_info *delays;
1387 #ifdef CONFIG_FAULT_INJECTION
1390 struct prop_local_single dirties;
1391 #ifdef CONFIG_LATENCYTOP
1392 int latency_record_count;
1393 struct latency_record latency_record[LT_SAVECOUNT];
1396 * time slack values; these are used to round up poll() and
1397 * select() etc timeout values. These are in nanoseconds.
1399 unsigned long timer_slack_ns;
1400 unsigned long default_timer_slack_ns;
1402 struct list_head *scm_work_list;
1403 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1404 /* Index of current stored adress in ret_stack */
1406 /* Stack of return addresses for return function tracing */
1407 struct ftrace_ret_stack *ret_stack;
1409 * Number of functions that haven't been traced
1410 * because of depth overrun.
1412 atomic_t trace_overrun;
1413 /* Pause for the tracing */
1414 atomic_t tracing_graph_pause;
1416 #ifdef CONFIG_TRACING
1417 /* state flags for use by tracers */
1418 unsigned long trace;
1422 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1423 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1426 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1427 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1428 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1429 * values are inverted: lower p->prio value means higher priority.
1431 * The MAX_USER_RT_PRIO value allows the actual maximum
1432 * RT priority to be separate from the value exported to
1433 * user-space. This allows kernel threads to set their
1434 * priority to a value higher than any user task. Note:
1435 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1438 #define MAX_USER_RT_PRIO 100
1439 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1441 #define MAX_PRIO (MAX_RT_PRIO + 40)
1442 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1444 static inline int rt_prio(int prio)
1446 if (unlikely(prio < MAX_RT_PRIO))
1451 static inline int rt_task(struct task_struct *p)
1453 return rt_prio(p->prio);
1456 static inline struct pid *task_pid(struct task_struct *task)
1458 return task->pids[PIDTYPE_PID].pid;
1461 static inline struct pid *task_tgid(struct task_struct *task)
1463 return task->group_leader->pids[PIDTYPE_PID].pid;
1467 * Without tasklist or rcu lock it is not safe to dereference
1468 * the result of task_pgrp/task_session even if task == current,
1469 * we can race with another thread doing sys_setsid/sys_setpgid.
1471 static inline struct pid *task_pgrp(struct task_struct *task)
1473 return task->group_leader->pids[PIDTYPE_PGID].pid;
1476 static inline struct pid *task_session(struct task_struct *task)
1478 return task->group_leader->pids[PIDTYPE_SID].pid;
1481 struct pid_namespace;
1484 * the helpers to get the task's different pids as they are seen
1485 * from various namespaces
1487 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1488 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1490 * task_xid_nr_ns() : id seen from the ns specified;
1492 * set_task_vxid() : assigns a virtual id to a task;
1494 * see also pid_nr() etc in include/linux/pid.h
1496 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1497 struct pid_namespace *ns);
1499 static inline pid_t task_pid_nr(struct task_struct *tsk)
1504 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1505 struct pid_namespace *ns)
1507 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1510 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1512 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1516 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1521 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1523 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1525 return pid_vnr(task_tgid(tsk));
1529 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1530 struct pid_namespace *ns)
1532 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1535 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1537 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1541 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1542 struct pid_namespace *ns)
1544 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1547 static inline pid_t task_session_vnr(struct task_struct *tsk)
1549 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1552 /* obsolete, do not use */
1553 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1555 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1559 * pid_alive - check that a task structure is not stale
1560 * @p: Task structure to be checked.
1562 * Test if a process is not yet dead (at most zombie state)
1563 * If pid_alive fails, then pointers within the task structure
1564 * can be stale and must not be dereferenced.
1566 static inline int pid_alive(struct task_struct *p)
1568 return p->pids[PIDTYPE_PID].pid != NULL;
1572 * is_global_init - check if a task structure is init
1573 * @tsk: Task structure to be checked.
1575 * Check if a task structure is the first user space task the kernel created.
1577 static inline int is_global_init(struct task_struct *tsk)
1579 return tsk->pid == 1;
1583 * is_container_init:
1584 * check whether in the task is init in its own pid namespace.
1586 extern int is_container_init(struct task_struct *tsk);
1588 extern struct pid *cad_pid;
1590 extern void free_task(struct task_struct *tsk);
1591 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1593 extern void __put_task_struct(struct task_struct *t);
1595 static inline void put_task_struct(struct task_struct *t)
1597 if (atomic_dec_and_test(&t->usage))
1598 __put_task_struct(t);
1601 extern cputime_t task_utime(struct task_struct *p);
1602 extern cputime_t task_stime(struct task_struct *p);
1603 extern cputime_t task_gtime(struct task_struct *p);
1608 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1609 /* Not implemented yet, only for 486*/
1610 #define PF_STARTING 0x00000002 /* being created */
1611 #define PF_EXITING 0x00000004 /* getting shut down */
1612 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1613 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1614 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1615 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1616 #define PF_DUMPCORE 0x00000200 /* dumped core */
1617 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1618 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1619 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1620 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1621 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1622 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1623 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1624 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1625 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1626 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1627 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1628 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1629 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1630 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1631 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1632 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1633 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1634 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1635 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1636 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1639 * Only the _current_ task can read/write to tsk->flags, but other
1640 * tasks can access tsk->flags in readonly mode for example
1641 * with tsk_used_math (like during threaded core dumping).
1642 * There is however an exception to this rule during ptrace
1643 * or during fork: the ptracer task is allowed to write to the
1644 * child->flags of its traced child (same goes for fork, the parent
1645 * can write to the child->flags), because we're guaranteed the
1646 * child is not running and in turn not changing child->flags
1647 * at the same time the parent does it.
1649 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1650 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1651 #define clear_used_math() clear_stopped_child_used_math(current)
1652 #define set_used_math() set_stopped_child_used_math(current)
1653 #define conditional_stopped_child_used_math(condition, child) \
1654 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1655 #define conditional_used_math(condition) \
1656 conditional_stopped_child_used_math(condition, current)
1657 #define copy_to_stopped_child_used_math(child) \
1658 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1659 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1660 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1661 #define used_math() tsk_used_math(current)
1664 extern int set_cpus_allowed_ptr(struct task_struct *p,
1665 const struct cpumask *new_mask);
1667 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1668 const struct cpumask *new_mask)
1670 if (!cpumask_test_cpu(0, new_mask))
1675 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1677 return set_cpus_allowed_ptr(p, &new_mask);
1681 * Architectures can set this to 1 if they have specified
1682 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1683 * but then during bootup it turns out that sched_clock()
1684 * is reliable after all:
1686 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1687 extern int sched_clock_stable;
1690 extern unsigned long long sched_clock(void);
1692 extern void sched_clock_init(void);
1693 extern u64 sched_clock_cpu(int cpu);
1695 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1696 static inline void sched_clock_tick(void)
1700 static inline void sched_clock_idle_sleep_event(void)
1704 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1708 extern void sched_clock_tick(void);
1709 extern void sched_clock_idle_sleep_event(void);
1710 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1714 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1715 * clock constructed from sched_clock():
1717 extern unsigned long long cpu_clock(int cpu);
1719 extern unsigned long long
1720 task_sched_runtime(struct task_struct *task);
1721 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1723 /* sched_exec is called by processes performing an exec */
1725 extern void sched_exec(void);
1727 #define sched_exec() {}
1730 extern void sched_clock_idle_sleep_event(void);
1731 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1733 #ifdef CONFIG_HOTPLUG_CPU
1734 extern void idle_task_exit(void);
1736 static inline void idle_task_exit(void) {}
1739 extern void sched_idle_next(void);
1741 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1742 extern void wake_up_idle_cpu(int cpu);
1744 static inline void wake_up_idle_cpu(int cpu) { }
1747 extern unsigned int sysctl_sched_latency;
1748 extern unsigned int sysctl_sched_min_granularity;
1749 extern unsigned int sysctl_sched_wakeup_granularity;
1750 extern unsigned int sysctl_sched_shares_ratelimit;
1751 extern unsigned int sysctl_sched_shares_thresh;
1752 #ifdef CONFIG_SCHED_DEBUG
1753 extern unsigned int sysctl_sched_child_runs_first;
1754 extern unsigned int sysctl_sched_features;
1755 extern unsigned int sysctl_sched_migration_cost;
1756 extern unsigned int sysctl_sched_nr_migrate;
1758 int sched_nr_latency_handler(struct ctl_table *table, int write,
1759 struct file *file, void __user *buffer, size_t *length,
1762 extern unsigned int sysctl_sched_rt_period;
1763 extern int sysctl_sched_rt_runtime;
1765 int sched_rt_handler(struct ctl_table *table, int write,
1766 struct file *filp, void __user *buffer, size_t *lenp,
1769 extern unsigned int sysctl_sched_compat_yield;
1771 #ifdef CONFIG_RT_MUTEXES
1772 extern int rt_mutex_getprio(struct task_struct *p);
1773 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1774 extern void rt_mutex_adjust_pi(struct task_struct *p);
1776 static inline int rt_mutex_getprio(struct task_struct *p)
1778 return p->normal_prio;
1780 # define rt_mutex_adjust_pi(p) do { } while (0)
1783 extern void set_user_nice(struct task_struct *p, long nice);
1784 extern int task_prio(const struct task_struct *p);
1785 extern int task_nice(const struct task_struct *p);
1786 extern int can_nice(const struct task_struct *p, const int nice);
1787 extern int task_curr(const struct task_struct *p);
1788 extern int idle_cpu(int cpu);
1789 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1790 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1791 struct sched_param *);
1792 extern struct task_struct *idle_task(int cpu);
1793 extern struct task_struct *curr_task(int cpu);
1794 extern void set_curr_task(int cpu, struct task_struct *p);
1799 * The default (Linux) execution domain.
1801 extern struct exec_domain default_exec_domain;
1803 union thread_union {
1804 struct thread_info thread_info;
1805 unsigned long stack[THREAD_SIZE/sizeof(long)];
1808 #ifndef __HAVE_ARCH_KSTACK_END
1809 static inline int kstack_end(void *addr)
1811 /* Reliable end of stack detection:
1812 * Some APM bios versions misalign the stack
1814 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1818 extern union thread_union init_thread_union;
1819 extern struct task_struct init_task;
1821 extern struct mm_struct init_mm;
1823 extern struct pid_namespace init_pid_ns;
1826 * find a task by one of its numerical ids
1828 * find_task_by_pid_type_ns():
1829 * it is the most generic call - it finds a task by all id,
1830 * type and namespace specified
1831 * find_task_by_pid_ns():
1832 * finds a task by its pid in the specified namespace
1833 * find_task_by_vpid():
1834 * finds a task by its virtual pid
1836 * see also find_vpid() etc in include/linux/pid.h
1839 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1840 struct pid_namespace *ns);
1842 extern struct task_struct *find_task_by_vpid(pid_t nr);
1843 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1844 struct pid_namespace *ns);
1846 extern void __set_special_pids(struct pid *pid);
1848 /* per-UID process charging. */
1849 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1850 static inline struct user_struct *get_uid(struct user_struct *u)
1852 atomic_inc(&u->__count);
1855 extern void free_uid(struct user_struct *);
1856 extern void release_uids(struct user_namespace *ns);
1858 #include <asm/current.h>
1860 extern void do_timer(unsigned long ticks);
1862 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1863 extern int wake_up_process(struct task_struct *tsk);
1864 extern void wake_up_new_task(struct task_struct *tsk,
1865 unsigned long clone_flags);
1867 extern void kick_process(struct task_struct *tsk);
1869 static inline void kick_process(struct task_struct *tsk) { }
1871 extern void sched_fork(struct task_struct *p, int clone_flags);
1872 extern void sched_dead(struct task_struct *p);
1874 extern void proc_caches_init(void);
1875 extern void flush_signals(struct task_struct *);
1876 extern void ignore_signals(struct task_struct *);
1877 extern void flush_signal_handlers(struct task_struct *, int force_default);
1878 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1880 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1882 unsigned long flags;
1885 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1886 ret = dequeue_signal(tsk, mask, info);
1887 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1892 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1894 extern void unblock_all_signals(void);
1895 extern void release_task(struct task_struct * p);
1896 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1897 extern int force_sigsegv(int, struct task_struct *);
1898 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1899 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1900 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1901 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1902 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1903 extern int kill_pid(struct pid *pid, int sig, int priv);
1904 extern int kill_proc_info(int, struct siginfo *, pid_t);
1905 extern int do_notify_parent(struct task_struct *, int);
1906 extern void force_sig(int, struct task_struct *);
1907 extern void force_sig_specific(int, struct task_struct *);
1908 extern int send_sig(int, struct task_struct *, int);
1909 extern void zap_other_threads(struct task_struct *p);
1910 extern struct sigqueue *sigqueue_alloc(void);
1911 extern void sigqueue_free(struct sigqueue *);
1912 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1913 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1914 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1916 static inline int kill_cad_pid(int sig, int priv)
1918 return kill_pid(cad_pid, sig, priv);
1921 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1922 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1923 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1924 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1926 static inline int is_si_special(const struct siginfo *info)
1928 return info <= SEND_SIG_FORCED;
1931 /* True if we are on the alternate signal stack. */
1933 static inline int on_sig_stack(unsigned long sp)
1935 return (sp - current->sas_ss_sp < current->sas_ss_size);
1938 static inline int sas_ss_flags(unsigned long sp)
1940 return (current->sas_ss_size == 0 ? SS_DISABLE
1941 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1945 * Routines for handling mm_structs
1947 extern struct mm_struct * mm_alloc(void);
1949 /* mmdrop drops the mm and the page tables */
1950 extern void __mmdrop(struct mm_struct *);
1951 static inline void mmdrop(struct mm_struct * mm)
1953 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1957 /* mmput gets rid of the mappings and all user-space */
1958 extern void mmput(struct mm_struct *);
1959 /* Grab a reference to a task's mm, if it is not already going away */
1960 extern struct mm_struct *get_task_mm(struct task_struct *task);
1961 /* Remove the current tasks stale references to the old mm_struct */
1962 extern void mm_release(struct task_struct *, struct mm_struct *);
1963 /* Allocate a new mm structure and copy contents from tsk->mm */
1964 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1966 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1967 struct task_struct *, struct pt_regs *);
1968 extern void flush_thread(void);
1969 extern void exit_thread(void);
1971 extern void exit_files(struct task_struct *);
1972 extern void __cleanup_signal(struct signal_struct *);
1973 extern void __cleanup_sighand(struct sighand_struct *);
1975 extern void exit_itimers(struct signal_struct *);
1976 extern void flush_itimer_signals(void);
1978 extern NORET_TYPE void do_group_exit(int);
1980 extern void daemonize(const char *, ...);
1981 extern int allow_signal(int);
1982 extern int disallow_signal(int);
1984 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1985 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1986 struct task_struct *fork_idle(int);
1988 extern void set_task_comm(struct task_struct *tsk, char *from);
1989 extern char *get_task_comm(char *to, struct task_struct *tsk);
1992 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1994 static inline unsigned long wait_task_inactive(struct task_struct *p,
2001 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2003 #define for_each_process(p) \
2004 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2006 extern bool is_single_threaded(struct task_struct *);
2009 * Careful: do_each_thread/while_each_thread is a double loop so
2010 * 'break' will not work as expected - use goto instead.
2012 #define do_each_thread(g, t) \
2013 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2015 #define while_each_thread(g, t) \
2016 while ((t = next_thread(t)) != g)
2018 /* de_thread depends on thread_group_leader not being a pid based check */
2019 #define thread_group_leader(p) (p == p->group_leader)
2021 /* Do to the insanities of de_thread it is possible for a process
2022 * to have the pid of the thread group leader without actually being
2023 * the thread group leader. For iteration through the pids in proc
2024 * all we care about is that we have a task with the appropriate
2025 * pid, we don't actually care if we have the right task.
2027 static inline int has_group_leader_pid(struct task_struct *p)
2029 return p->pid == p->tgid;
2033 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2035 return p1->tgid == p2->tgid;
2038 static inline struct task_struct *next_thread(const struct task_struct *p)
2040 return list_entry(rcu_dereference(p->thread_group.next),
2041 struct task_struct, thread_group);
2044 static inline int thread_group_empty(struct task_struct *p)
2046 return list_empty(&p->thread_group);
2049 #define delay_group_leader(p) \
2050 (thread_group_leader(p) && !thread_group_empty(p))
2052 static inline int task_detached(struct task_struct *p)
2054 return p->exit_signal == -1;
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__ */