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 */
119 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
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_iowait(void);
140 extern void calc_global_load(void);
142 extern unsigned long get_parent_ip(unsigned long addr);
147 #ifdef CONFIG_SCHED_DEBUG
148 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
149 extern void proc_sched_set_task(struct task_struct *p);
151 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
154 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
157 static inline void proc_sched_set_task(struct task_struct *p)
161 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
166 extern unsigned long long time_sync_thresh;
169 * Task state bitmask. NOTE! These bits are also
170 * encoded in fs/proc/array.c: get_task_state().
172 * We have two separate sets of flags: task->state
173 * is about runnability, while task->exit_state are
174 * about the task exiting. Confusing, but this way
175 * modifying one set can't modify the other one by
178 #define TASK_RUNNING 0
179 #define TASK_INTERRUPTIBLE 1
180 #define TASK_UNINTERRUPTIBLE 2
181 #define __TASK_STOPPED 4
182 #define __TASK_TRACED 8
183 /* in tsk->exit_state */
184 #define EXIT_ZOMBIE 16
186 /* in tsk->state again */
188 #define TASK_WAKEKILL 128
190 /* Convenience macros for the sake of set_task_state */
191 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
192 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
193 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
195 /* Convenience macros for the sake of wake_up */
196 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
197 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
199 /* get_task_state() */
200 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
201 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
204 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
205 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
206 #define task_is_stopped_or_traced(task) \
207 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
208 #define task_contributes_to_load(task) \
209 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
210 (task->flags & PF_FROZEN) == 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 int softlockup_thresh;
307 static inline void softlockup_tick(void)
310 static inline void touch_softlockup_watchdog(void)
313 static inline void touch_all_softlockup_watchdogs(void)
318 #ifdef CONFIG_DETECT_HUNG_TASK
319 extern unsigned int sysctl_hung_task_panic;
320 extern unsigned long sysctl_hung_task_check_count;
321 extern unsigned long sysctl_hung_task_timeout_secs;
322 extern unsigned long sysctl_hung_task_warnings;
323 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
324 struct file *filp, void __user *buffer,
325 size_t *lenp, loff_t *ppos);
328 /* Attach to any functions which should be ignored in wchan output. */
329 #define __sched __attribute__((__section__(".sched.text")))
331 /* Linker adds these: start and end of __sched functions */
332 extern char __sched_text_start[], __sched_text_end[];
334 /* Is this address in the __sched functions? */
335 extern int in_sched_functions(unsigned long addr);
337 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
338 extern signed long schedule_timeout(signed long timeout);
339 extern signed long schedule_timeout_interruptible(signed long timeout);
340 extern signed long schedule_timeout_killable(signed long timeout);
341 extern signed long schedule_timeout_uninterruptible(signed long timeout);
342 asmlinkage void __schedule(void);
343 asmlinkage void schedule(void);
344 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
347 struct user_namespace;
349 /* Maximum number of active map areas.. This is a random (large) number */
350 #define DEFAULT_MAX_MAP_COUNT 65536
352 extern int sysctl_max_map_count;
354 #include <linux/aio.h>
357 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
358 unsigned long, unsigned long);
360 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
361 unsigned long len, unsigned long pgoff,
362 unsigned long flags);
363 extern void arch_unmap_area(struct mm_struct *, unsigned long);
364 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
366 #if USE_SPLIT_PTLOCKS
368 * The mm counters are not protected by its page_table_lock,
369 * so must be incremented atomically.
371 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
372 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
373 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
374 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
375 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
377 #else /* !USE_SPLIT_PTLOCKS */
379 * The mm counters are protected by its page_table_lock,
380 * so can be incremented directly.
382 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
383 #define get_mm_counter(mm, member) ((mm)->_##member)
384 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
385 #define inc_mm_counter(mm, member) (mm)->_##member++
386 #define dec_mm_counter(mm, member) (mm)->_##member--
388 #endif /* !USE_SPLIT_PTLOCKS */
390 #define get_mm_rss(mm) \
391 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
392 #define update_hiwater_rss(mm) do { \
393 unsigned long _rss = get_mm_rss(mm); \
394 if ((mm)->hiwater_rss < _rss) \
395 (mm)->hiwater_rss = _rss; \
397 #define update_hiwater_vm(mm) do { \
398 if ((mm)->hiwater_vm < (mm)->total_vm) \
399 (mm)->hiwater_vm = (mm)->total_vm; \
402 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
404 return max(mm->hiwater_rss, get_mm_rss(mm));
407 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
409 return max(mm->hiwater_vm, mm->total_vm);
412 extern void set_dumpable(struct mm_struct *mm, int value);
413 extern int get_dumpable(struct mm_struct *mm);
417 #define MMF_DUMPABLE 0 /* core dump is permitted */
418 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
419 #define MMF_DUMPABLE_BITS 2
421 /* coredump filter bits */
422 #define MMF_DUMP_ANON_PRIVATE 2
423 #define MMF_DUMP_ANON_SHARED 3
424 #define MMF_DUMP_MAPPED_PRIVATE 4
425 #define MMF_DUMP_MAPPED_SHARED 5
426 #define MMF_DUMP_ELF_HEADERS 6
427 #define MMF_DUMP_HUGETLB_PRIVATE 7
428 #define MMF_DUMP_HUGETLB_SHARED 8
429 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
430 #define MMF_DUMP_FILTER_BITS 7
431 #define MMF_DUMP_FILTER_MASK \
432 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
433 #define MMF_DUMP_FILTER_DEFAULT \
434 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
435 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
437 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
438 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
440 # define MMF_DUMP_MASK_DEFAULT_ELF 0
443 struct sighand_struct {
445 struct k_sigaction action[_NSIG];
447 wait_queue_head_t signalfd_wqh;
450 struct pacct_struct {
453 unsigned long ac_mem;
454 cputime_t ac_utime, ac_stime;
455 unsigned long ac_minflt, ac_majflt;
459 * struct task_cputime - collected CPU time counts
460 * @utime: time spent in user mode, in &cputime_t units
461 * @stime: time spent in kernel mode, in &cputime_t units
462 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
464 * This structure groups together three kinds of CPU time that are
465 * tracked for threads and thread groups. Most things considering
466 * CPU time want to group these counts together and treat all three
467 * of them in parallel.
469 struct task_cputime {
472 unsigned long long sum_exec_runtime;
474 /* Alternate field names when used to cache expirations. */
475 #define prof_exp stime
476 #define virt_exp utime
477 #define sched_exp sum_exec_runtime
479 #define INIT_CPUTIME \
480 (struct task_cputime) { \
481 .utime = cputime_zero, \
482 .stime = cputime_zero, \
483 .sum_exec_runtime = 0, \
487 * struct thread_group_cputimer - thread group interval timer counts
488 * @cputime: thread group interval timers.
489 * @running: non-zero when there are timers running and
490 * @cputime receives updates.
491 * @lock: lock for fields in this struct.
493 * This structure contains the version of task_cputime, above, that is
494 * used for thread group CPU timer calculations.
496 struct thread_group_cputimer {
497 struct task_cputime cputime;
503 * NOTE! "signal_struct" does not have it's own
504 * locking, because a shared signal_struct always
505 * implies a shared sighand_struct, so locking
506 * sighand_struct is always a proper superset of
507 * the locking of signal_struct.
509 struct signal_struct {
513 wait_queue_head_t wait_chldexit; /* for wait4() */
515 /* current thread group signal load-balancing target: */
516 struct task_struct *curr_target;
518 /* shared signal handling: */
519 struct sigpending shared_pending;
521 /* thread group exit support */
524 * - notify group_exit_task when ->count is equal to notify_count
525 * - everyone except group_exit_task is stopped during signal delivery
526 * of fatal signals, group_exit_task processes the signal.
529 struct task_struct *group_exit_task;
531 /* thread group stop support, overloads group_exit_code too */
532 int group_stop_count;
533 unsigned int flags; /* see SIGNAL_* flags below */
535 /* POSIX.1b Interval Timers */
536 struct list_head posix_timers;
538 /* ITIMER_REAL timer for the process */
539 struct hrtimer real_timer;
540 struct pid *leader_pid;
541 ktime_t it_real_incr;
543 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
544 cputime_t it_prof_expires, it_virt_expires;
545 cputime_t it_prof_incr, it_virt_incr;
548 * Thread group totals for process CPU timers.
549 * See thread_group_cputimer(), et al, for details.
551 struct thread_group_cputimer cputimer;
553 /* Earliest-expiration cache. */
554 struct task_cputime cputime_expires;
556 struct list_head cpu_timers[3];
558 struct pid *tty_old_pgrp;
560 /* boolean value for session group leader */
563 struct tty_struct *tty; /* NULL if no tty */
566 * Cumulative resource counters for dead threads in the group,
567 * and for reaped dead child processes forked by this group.
568 * Live threads maintain their own counters and add to these
569 * in __exit_signal, except for the group leader.
571 cputime_t utime, stime, cutime, cstime;
574 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
575 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
576 unsigned long inblock, oublock, cinblock, coublock;
577 struct task_io_accounting ioac;
580 * Cumulative ns of schedule CPU time fo dead threads in the
581 * group, not including a zombie group leader, (This only differs
582 * from jiffies_to_ns(utime + stime) if sched_clock uses something
583 * other than jiffies.)
585 unsigned long long sum_sched_runtime;
588 * We don't bother to synchronize most readers of this at all,
589 * because there is no reader checking a limit that actually needs
590 * to get both rlim_cur and rlim_max atomically, and either one
591 * alone is a single word that can safely be read normally.
592 * getrlimit/setrlimit use task_lock(current->group_leader) to
593 * protect this instead of the siglock, because they really
594 * have no need to disable irqs.
596 struct rlimit rlim[RLIM_NLIMITS];
598 #ifdef CONFIG_BSD_PROCESS_ACCT
599 struct pacct_struct pacct; /* per-process accounting information */
601 #ifdef CONFIG_TASKSTATS
602 struct taskstats *stats;
606 struct tty_audit_buf *tty_audit_buf;
610 /* Context switch must be unlocked if interrupts are to be enabled */
611 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
612 # define __ARCH_WANT_UNLOCKED_CTXSW
616 * Bits in flags field of signal_struct.
618 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
619 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
620 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
621 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
623 * Pending notifications to parent.
625 #define SIGNAL_CLD_STOPPED 0x00000010
626 #define SIGNAL_CLD_CONTINUED 0x00000020
627 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
629 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
631 /* If true, all threads except ->group_exit_task have pending SIGKILL */
632 static inline int signal_group_exit(const struct signal_struct *sig)
634 return (sig->flags & SIGNAL_GROUP_EXIT) ||
635 (sig->group_exit_task != NULL);
639 * Some day this will be a full-fledged user tracking system..
642 atomic_t __count; /* reference count */
643 atomic_t processes; /* How many processes does this user have? */
644 atomic_t files; /* How many open files does this user have? */
645 atomic_t sigpending; /* How many pending signals does this user have? */
646 #ifdef CONFIG_INOTIFY_USER
647 atomic_t inotify_watches; /* How many inotify watches does this user have? */
648 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
651 atomic_t epoll_watches; /* The number of file descriptors currently watched */
653 #ifdef CONFIG_POSIX_MQUEUE
654 /* protected by mq_lock */
655 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
657 unsigned long locked_shm; /* How many pages of mlocked shm ? */
660 struct key *uid_keyring; /* UID specific keyring */
661 struct key *session_keyring; /* UID's default session keyring */
664 /* Hash table maintenance information */
665 struct hlist_node uidhash_node;
667 struct user_namespace *user_ns;
669 #ifdef CONFIG_USER_SCHED
670 struct task_group *tg;
673 struct work_struct work;
678 extern int uids_sysfs_init(void);
680 extern struct user_struct *find_user(uid_t);
682 extern struct user_struct root_user;
683 #define INIT_USER (&root_user)
686 struct backing_dev_info;
687 struct reclaim_state;
689 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
691 /* cumulative counters */
692 unsigned long pcount; /* # of times run on this cpu */
693 unsigned long long run_delay; /* time spent waiting on a runqueue */
696 unsigned long long last_arrival,/* when we last ran on a cpu */
697 last_queued; /* when we were last queued to run */
698 #ifdef CONFIG_SCHEDSTATS
700 unsigned int bkl_count;
703 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
705 #ifdef CONFIG_TASK_DELAY_ACCT
706 struct task_delay_info {
708 unsigned int flags; /* Private per-task flags */
710 /* For each stat XXX, add following, aligned appropriately
712 * struct timespec XXX_start, XXX_end;
716 * Atomicity of updates to XXX_delay, XXX_count protected by
717 * single lock above (split into XXX_lock if contention is an issue).
721 * XXX_count is incremented on every XXX operation, the delay
722 * associated with the operation is added to XXX_delay.
723 * XXX_delay contains the accumulated delay time in nanoseconds.
725 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
726 u64 blkio_delay; /* wait for sync block io completion */
727 u64 swapin_delay; /* wait for swapin block io completion */
728 u32 blkio_count; /* total count of the number of sync block */
729 /* io operations performed */
730 u32 swapin_count; /* total count of the number of swapin block */
731 /* io operations performed */
733 struct timespec freepages_start, freepages_end;
734 u64 freepages_delay; /* wait for memory reclaim */
735 u32 freepages_count; /* total count of memory reclaim */
737 #endif /* CONFIG_TASK_DELAY_ACCT */
739 static inline int sched_info_on(void)
741 #ifdef CONFIG_SCHEDSTATS
743 #elif defined(CONFIG_TASK_DELAY_ACCT)
744 extern int delayacct_on;
759 * sched-domains (multiprocessor balancing) declarations:
763 * Increase resolution of nice-level calculations:
765 #define SCHED_LOAD_SHIFT 10
766 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
768 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
771 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
772 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
773 #define SD_BALANCE_EXEC 4 /* Balance on exec */
774 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
775 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
776 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
777 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
778 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
779 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
780 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
781 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
782 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
784 enum powersavings_balance_level {
785 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
786 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
787 * first for long running threads
789 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
790 * cpu package for power savings
792 MAX_POWERSAVINGS_BALANCE_LEVELS
795 extern int sched_mc_power_savings, sched_smt_power_savings;
797 static inline int sd_balance_for_mc_power(void)
799 if (sched_smt_power_savings)
800 return SD_POWERSAVINGS_BALANCE;
805 static inline int sd_balance_for_package_power(void)
807 if (sched_mc_power_savings | sched_smt_power_savings)
808 return SD_POWERSAVINGS_BALANCE;
814 * Optimise SD flags for power savings:
815 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
816 * Keep default SD flags if sched_{smt,mc}_power_saving=0
819 static inline int sd_power_saving_flags(void)
821 if (sched_mc_power_savings | sched_smt_power_savings)
822 return SD_BALANCE_NEWIDLE;
828 struct sched_group *next; /* Must be a circular list */
831 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
832 * single CPU. This is read only (except for setup, hotplug CPU).
833 * Note : Never change cpu_power without recompute its reciprocal
835 unsigned int __cpu_power;
837 * reciprocal value of cpu_power to avoid expensive divides
838 * (see include/linux/reciprocal_div.h)
840 u32 reciprocal_cpu_power;
843 * The CPUs this group covers.
845 * NOTE: this field is variable length. (Allocated dynamically
846 * by attaching extra space to the end of the structure,
847 * depending on how many CPUs the kernel has booted up with)
849 * It is also be embedded into static data structures at build
850 * time. (See 'struct static_sched_group' in kernel/sched.c)
852 unsigned long cpumask[0];
855 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
857 return to_cpumask(sg->cpumask);
860 enum sched_domain_level {
870 struct sched_domain_attr {
871 int relax_domain_level;
874 #define SD_ATTR_INIT (struct sched_domain_attr) { \
875 .relax_domain_level = -1, \
878 struct sched_domain {
879 /* These fields must be setup */
880 struct sched_domain *parent; /* top domain must be null terminated */
881 struct sched_domain *child; /* bottom domain must be null terminated */
882 struct sched_group *groups; /* the balancing groups of the domain */
883 unsigned long min_interval; /* Minimum balance interval ms */
884 unsigned long max_interval; /* Maximum balance interval ms */
885 unsigned int busy_factor; /* less balancing by factor if busy */
886 unsigned int imbalance_pct; /* No balance until over watermark */
887 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
888 unsigned int busy_idx;
889 unsigned int idle_idx;
890 unsigned int newidle_idx;
891 unsigned int wake_idx;
892 unsigned int forkexec_idx;
893 int flags; /* See SD_* */
894 enum sched_domain_level level;
896 /* Runtime fields. */
897 unsigned long last_balance; /* init to jiffies. units in jiffies */
898 unsigned int balance_interval; /* initialise to 1. units in ms. */
899 unsigned int nr_balance_failed; /* initialise to 0 */
903 #ifdef CONFIG_SCHEDSTATS
904 /* load_balance() stats */
905 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
906 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
907 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
908 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
909 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
910 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
911 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
912 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
914 /* Active load balancing */
915 unsigned int alb_count;
916 unsigned int alb_failed;
917 unsigned int alb_pushed;
919 /* SD_BALANCE_EXEC stats */
920 unsigned int sbe_count;
921 unsigned int sbe_balanced;
922 unsigned int sbe_pushed;
924 /* SD_BALANCE_FORK stats */
925 unsigned int sbf_count;
926 unsigned int sbf_balanced;
927 unsigned int sbf_pushed;
929 /* try_to_wake_up() stats */
930 unsigned int ttwu_wake_remote;
931 unsigned int ttwu_move_affine;
932 unsigned int ttwu_move_balance;
934 #ifdef CONFIG_SCHED_DEBUG
939 * Span of all CPUs in this domain.
941 * NOTE: this field is variable length. (Allocated dynamically
942 * by attaching extra space to the end of the structure,
943 * depending on how many CPUs the kernel has booted up with)
945 * It is also be embedded into static data structures at build
946 * time. (See 'struct static_sched_domain' in kernel/sched.c)
948 unsigned long span[0];
951 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
953 return to_cpumask(sd->span);
956 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
957 struct sched_domain_attr *dattr_new);
959 /* Test a flag in parent sched domain */
960 static inline int test_sd_parent(struct sched_domain *sd, int flag)
962 if (sd->parent && (sd->parent->flags & flag))
968 #else /* CONFIG_SMP */
970 struct sched_domain_attr;
973 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
974 struct sched_domain_attr *dattr_new)
977 #endif /* !CONFIG_SMP */
979 struct io_context; /* See blkdev.h */
982 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
983 extern void prefetch_stack(struct task_struct *t);
985 static inline void prefetch_stack(struct task_struct *t) { }
988 struct audit_context; /* See audit.c */
990 struct pipe_inode_info;
991 struct uts_namespace;
997 const struct sched_class *next;
999 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
1000 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1001 void (*yield_task) (struct rq *rq);
1003 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
1005 struct task_struct * (*pick_next_task) (struct rq *rq);
1006 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1009 int (*select_task_rq)(struct task_struct *p, int sync);
1011 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1012 struct rq *busiest, unsigned long max_load_move,
1013 struct sched_domain *sd, enum cpu_idle_type idle,
1014 int *all_pinned, int *this_best_prio);
1016 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1017 struct rq *busiest, struct sched_domain *sd,
1018 enum cpu_idle_type idle);
1019 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1020 int (*needs_post_schedule) (struct rq *this_rq);
1021 void (*post_schedule) (struct rq *this_rq);
1022 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1024 void (*set_cpus_allowed)(struct task_struct *p,
1025 const struct cpumask *newmask);
1027 void (*rq_online)(struct rq *rq);
1028 void (*rq_offline)(struct rq *rq);
1031 void (*set_curr_task) (struct rq *rq);
1032 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1033 void (*task_new) (struct rq *rq, struct task_struct *p);
1035 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1037 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1039 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1040 int oldprio, int running);
1042 #ifdef CONFIG_FAIR_GROUP_SCHED
1043 void (*moved_group) (struct task_struct *p);
1047 struct load_weight {
1048 unsigned long weight, inv_weight;
1052 * CFS stats for a schedulable entity (task, task-group etc)
1054 * Current field usage histogram:
1061 struct sched_entity {
1062 struct load_weight load; /* for load-balancing */
1063 struct rb_node run_node;
1064 struct list_head group_node;
1068 u64 sum_exec_runtime;
1070 u64 prev_sum_exec_runtime;
1079 #ifdef CONFIG_SCHEDSTATS
1087 s64 sum_sleep_runtime;
1094 u64 nr_migrations_cold;
1095 u64 nr_failed_migrations_affine;
1096 u64 nr_failed_migrations_running;
1097 u64 nr_failed_migrations_hot;
1098 u64 nr_forced_migrations;
1099 u64 nr_forced2_migrations;
1102 u64 nr_wakeups_sync;
1103 u64 nr_wakeups_migrate;
1104 u64 nr_wakeups_local;
1105 u64 nr_wakeups_remote;
1106 u64 nr_wakeups_affine;
1107 u64 nr_wakeups_affine_attempts;
1108 u64 nr_wakeups_passive;
1109 u64 nr_wakeups_idle;
1112 #ifdef CONFIG_FAIR_GROUP_SCHED
1113 struct sched_entity *parent;
1114 /* rq on which this entity is (to be) queued: */
1115 struct cfs_rq *cfs_rq;
1116 /* rq "owned" by this entity/group: */
1117 struct cfs_rq *my_q;
1121 struct sched_rt_entity {
1122 struct list_head run_list;
1123 unsigned long timeout;
1124 unsigned int time_slice;
1125 int nr_cpus_allowed;
1127 struct sched_rt_entity *back;
1128 #ifdef CONFIG_RT_GROUP_SCHED
1129 struct sched_rt_entity *parent;
1130 /* rq on which this entity is (to be) queued: */
1131 struct rt_rq *rt_rq;
1132 /* rq "owned" by this entity/group: */
1137 struct task_struct {
1138 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1141 unsigned int flags; /* per process flags, defined below */
1142 unsigned int ptrace;
1144 int lock_depth; /* BKL lock depth */
1147 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1152 int prio, static_prio, normal_prio;
1153 unsigned int rt_priority;
1154 const struct sched_class *sched_class;
1155 struct sched_entity se;
1156 struct sched_rt_entity rt;
1158 #ifdef CONFIG_PREEMPT_NOTIFIERS
1159 /* list of struct preempt_notifier: */
1160 struct hlist_head preempt_notifiers;
1164 * fpu_counter contains the number of consecutive context switches
1165 * that the FPU is used. If this is over a threshold, the lazy fpu
1166 * saving becomes unlazy to save the trap. This is an unsigned char
1167 * so that after 256 times the counter wraps and the behavior turns
1168 * lazy again; this to deal with bursty apps that only use FPU for
1171 unsigned char fpu_counter;
1172 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1173 #ifdef CONFIG_BLK_DEV_IO_TRACE
1174 unsigned int btrace_seq;
1177 unsigned int policy;
1178 cpumask_t cpus_allowed;
1180 #ifdef CONFIG_PREEMPT_RCU
1181 int rcu_read_lock_nesting;
1182 int rcu_flipctr_idx;
1183 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1185 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1186 struct sched_info sched_info;
1189 struct list_head tasks;
1190 struct plist_node pushable_tasks;
1192 struct mm_struct *mm, *active_mm;
1195 struct linux_binfmt *binfmt;
1197 int exit_code, exit_signal;
1198 int pdeath_signal; /* The signal sent when the parent dies */
1200 unsigned int personality;
1201 unsigned did_exec:1;
1202 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1207 /* Canary value for the -fstack-protector gcc feature */
1208 unsigned long stack_canary;
1211 * pointers to (original) parent process, youngest child, younger sibling,
1212 * older sibling, respectively. (p->father can be replaced with
1213 * p->real_parent->pid)
1215 struct task_struct *real_parent; /* real parent process */
1216 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1218 * children/sibling forms the list of my natural children
1220 struct list_head children; /* list of my children */
1221 struct list_head sibling; /* linkage in my parent's children list */
1222 struct task_struct *group_leader; /* threadgroup leader */
1225 * ptraced is the list of tasks this task is using ptrace on.
1226 * This includes both natural children and PTRACE_ATTACH targets.
1227 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1229 struct list_head ptraced;
1230 struct list_head ptrace_entry;
1232 #ifdef CONFIG_X86_PTRACE_BTS
1234 * This is the tracer handle for the ptrace BTS extension.
1235 * This field actually belongs to the ptracer task.
1237 struct bts_tracer *bts;
1239 * The buffer to hold the BTS data.
1243 #endif /* CONFIG_X86_PTRACE_BTS */
1245 /* PID/PID hash table linkage. */
1246 struct pid_link pids[PIDTYPE_MAX];
1247 struct list_head thread_group;
1249 struct completion *vfork_done; /* for vfork() */
1250 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1251 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1253 cputime_t utime, stime, utimescaled, stimescaled;
1255 cputime_t prev_utime, prev_stime;
1256 unsigned long nvcsw, nivcsw; /* context switch counts */
1257 struct timespec start_time; /* monotonic time */
1258 struct timespec real_start_time; /* boot based time */
1259 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1260 unsigned long min_flt, maj_flt;
1262 struct task_cputime cputime_expires;
1263 struct list_head cpu_timers[3];
1265 /* process credentials */
1266 const struct cred *real_cred; /* objective and real subjective task
1267 * credentials (COW) */
1268 const struct cred *cred; /* effective (overridable) subjective task
1269 * credentials (COW) */
1270 struct mutex cred_exec_mutex; /* execve vs ptrace cred calculation mutex */
1272 char comm[TASK_COMM_LEN]; /* executable name excluding path
1273 - access with [gs]et_task_comm (which lock
1274 it with task_lock())
1275 - initialized normally by flush_old_exec */
1276 /* file system info */
1277 int link_count, total_link_count;
1278 #ifdef CONFIG_SYSVIPC
1280 struct sysv_sem sysvsem;
1282 #ifdef CONFIG_DETECT_HUNG_TASK
1283 /* hung task detection */
1284 unsigned long last_switch_count;
1286 /* CPU-specific state of this task */
1287 struct thread_struct thread;
1288 /* filesystem information */
1289 struct fs_struct *fs;
1290 /* open file information */
1291 struct files_struct *files;
1293 struct nsproxy *nsproxy;
1294 /* signal handlers */
1295 struct signal_struct *signal;
1296 struct sighand_struct *sighand;
1298 sigset_t blocked, real_blocked;
1299 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1300 struct sigpending pending;
1302 unsigned long sas_ss_sp;
1304 int (*notifier)(void *priv);
1305 void *notifier_data;
1306 sigset_t *notifier_mask;
1307 struct audit_context *audit_context;
1308 #ifdef CONFIG_AUDITSYSCALL
1310 unsigned int sessionid;
1314 /* Thread group tracking */
1317 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1318 spinlock_t alloc_lock;
1320 #ifdef CONFIG_GENERIC_HARDIRQS
1321 /* IRQ handler threads */
1322 struct irqaction *irqaction;
1325 /* Protection of the PI data structures: */
1328 #ifdef CONFIG_RT_MUTEXES
1329 /* PI waiters blocked on a rt_mutex held by this task */
1330 struct plist_head pi_waiters;
1331 /* Deadlock detection and priority inheritance handling */
1332 struct rt_mutex_waiter *pi_blocked_on;
1335 #ifdef CONFIG_DEBUG_MUTEXES
1336 /* mutex deadlock detection */
1337 struct mutex_waiter *blocked_on;
1339 #ifdef CONFIG_TRACE_IRQFLAGS
1340 unsigned int irq_events;
1341 int hardirqs_enabled;
1342 unsigned long hardirq_enable_ip;
1343 unsigned int hardirq_enable_event;
1344 unsigned long hardirq_disable_ip;
1345 unsigned int hardirq_disable_event;
1346 int softirqs_enabled;
1347 unsigned long softirq_disable_ip;
1348 unsigned int softirq_disable_event;
1349 unsigned long softirq_enable_ip;
1350 unsigned int softirq_enable_event;
1351 int hardirq_context;
1352 int softirq_context;
1354 #ifdef CONFIG_LOCKDEP
1355 # define MAX_LOCK_DEPTH 48UL
1358 unsigned int lockdep_recursion;
1359 struct held_lock held_locks[MAX_LOCK_DEPTH];
1360 gfp_t lockdep_reclaim_gfp;
1363 /* journalling filesystem info */
1366 /* stacked block device info */
1367 struct bio *bio_list, **bio_tail;
1370 struct reclaim_state *reclaim_state;
1372 struct backing_dev_info *backing_dev_info;
1374 struct io_context *io_context;
1376 unsigned long ptrace_message;
1377 siginfo_t *last_siginfo; /* For ptrace use. */
1378 struct task_io_accounting ioac;
1379 #if defined(CONFIG_TASK_XACCT)
1380 u64 acct_rss_mem1; /* accumulated rss usage */
1381 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1382 cputime_t acct_timexpd; /* stime + utime since last update */
1384 #ifdef CONFIG_CPUSETS
1385 nodemask_t mems_allowed;
1386 int cpuset_mems_generation;
1387 int cpuset_mem_spread_rotor;
1389 #ifdef CONFIG_CGROUPS
1390 /* Control Group info protected by css_set_lock */
1391 struct css_set *cgroups;
1392 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1393 struct list_head cg_list;
1396 struct robust_list_head __user *robust_list;
1397 #ifdef CONFIG_COMPAT
1398 struct compat_robust_list_head __user *compat_robust_list;
1400 struct list_head pi_state_list;
1401 struct futex_pi_state *pi_state_cache;
1404 struct mempolicy *mempolicy;
1407 atomic_t fs_excl; /* holding fs exclusive resources */
1408 struct rcu_head rcu;
1411 * cache last used pipe for splice
1413 struct pipe_inode_info *splice_pipe;
1414 #ifdef CONFIG_TASK_DELAY_ACCT
1415 struct task_delay_info *delays;
1417 #ifdef CONFIG_FAULT_INJECTION
1420 struct prop_local_single dirties;
1421 #ifdef CONFIG_LATENCYTOP
1422 int latency_record_count;
1423 struct latency_record latency_record[LT_SAVECOUNT];
1426 * time slack values; these are used to round up poll() and
1427 * select() etc timeout values. These are in nanoseconds.
1429 unsigned long timer_slack_ns;
1430 unsigned long default_timer_slack_ns;
1432 struct list_head *scm_work_list;
1433 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1434 /* Index of current stored adress in ret_stack */
1436 /* Stack of return addresses for return function tracing */
1437 struct ftrace_ret_stack *ret_stack;
1438 /* time stamp for last schedule */
1439 unsigned long long ftrace_timestamp;
1441 * Number of functions that haven't been traced
1442 * because of depth overrun.
1444 atomic_t trace_overrun;
1445 /* Pause for the tracing */
1446 atomic_t tracing_graph_pause;
1448 #ifdef CONFIG_TRACING
1449 /* state flags for use by tracers */
1450 unsigned long trace;
1454 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1455 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1458 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1459 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1460 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1461 * values are inverted: lower p->prio value means higher priority.
1463 * The MAX_USER_RT_PRIO value allows the actual maximum
1464 * RT priority to be separate from the value exported to
1465 * user-space. This allows kernel threads to set their
1466 * priority to a value higher than any user task. Note:
1467 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1470 #define MAX_USER_RT_PRIO 100
1471 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1473 #define MAX_PRIO (MAX_RT_PRIO + 40)
1474 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1476 static inline int rt_prio(int prio)
1478 if (unlikely(prio < MAX_RT_PRIO))
1483 static inline int rt_task(struct task_struct *p)
1485 return rt_prio(p->prio);
1488 static inline struct pid *task_pid(struct task_struct *task)
1490 return task->pids[PIDTYPE_PID].pid;
1493 static inline struct pid *task_tgid(struct task_struct *task)
1495 return task->group_leader->pids[PIDTYPE_PID].pid;
1499 * Without tasklist or rcu lock it is not safe to dereference
1500 * the result of task_pgrp/task_session even if task == current,
1501 * we can race with another thread doing sys_setsid/sys_setpgid.
1503 static inline struct pid *task_pgrp(struct task_struct *task)
1505 return task->group_leader->pids[PIDTYPE_PGID].pid;
1508 static inline struct pid *task_session(struct task_struct *task)
1510 return task->group_leader->pids[PIDTYPE_SID].pid;
1513 struct pid_namespace;
1516 * the helpers to get the task's different pids as they are seen
1517 * from various namespaces
1519 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1520 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1522 * task_xid_nr_ns() : id seen from the ns specified;
1524 * set_task_vxid() : assigns a virtual id to a task;
1526 * see also pid_nr() etc in include/linux/pid.h
1528 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1529 struct pid_namespace *ns);
1531 static inline pid_t task_pid_nr(struct task_struct *tsk)
1536 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1537 struct pid_namespace *ns)
1539 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1542 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1544 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1548 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1553 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1555 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1557 return pid_vnr(task_tgid(tsk));
1561 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1562 struct pid_namespace *ns)
1564 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1567 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1569 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1573 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1574 struct pid_namespace *ns)
1576 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1579 static inline pid_t task_session_vnr(struct task_struct *tsk)
1581 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1584 /* obsolete, do not use */
1585 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1587 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1591 * pid_alive - check that a task structure is not stale
1592 * @p: Task structure to be checked.
1594 * Test if a process is not yet dead (at most zombie state)
1595 * If pid_alive fails, then pointers within the task structure
1596 * can be stale and must not be dereferenced.
1598 static inline int pid_alive(struct task_struct *p)
1600 return p->pids[PIDTYPE_PID].pid != NULL;
1604 * is_global_init - check if a task structure is init
1605 * @tsk: Task structure to be checked.
1607 * Check if a task structure is the first user space task the kernel created.
1609 static inline int is_global_init(struct task_struct *tsk)
1611 return tsk->pid == 1;
1615 * is_container_init:
1616 * check whether in the task is init in its own pid namespace.
1618 extern int is_container_init(struct task_struct *tsk);
1620 extern struct pid *cad_pid;
1622 extern void free_task(struct task_struct *tsk);
1623 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1625 extern void __put_task_struct(struct task_struct *t);
1627 static inline void put_task_struct(struct task_struct *t)
1629 if (atomic_dec_and_test(&t->usage))
1630 __put_task_struct(t);
1633 extern cputime_t task_utime(struct task_struct *p);
1634 extern cputime_t task_stime(struct task_struct *p);
1635 extern cputime_t task_gtime(struct task_struct *p);
1640 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1641 /* Not implemented yet, only for 486*/
1642 #define PF_STARTING 0x00000002 /* being created */
1643 #define PF_EXITING 0x00000004 /* getting shut down */
1644 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1645 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1646 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1647 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1648 #define PF_DUMPCORE 0x00000200 /* dumped core */
1649 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1650 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1651 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1652 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1653 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1654 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1655 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1656 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1657 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1658 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1659 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1660 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1661 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1662 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1663 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1664 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1665 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1666 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1667 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1668 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1671 * Only the _current_ task can read/write to tsk->flags, but other
1672 * tasks can access tsk->flags in readonly mode for example
1673 * with tsk_used_math (like during threaded core dumping).
1674 * There is however an exception to this rule during ptrace
1675 * or during fork: the ptracer task is allowed to write to the
1676 * child->flags of its traced child (same goes for fork, the parent
1677 * can write to the child->flags), because we're guaranteed the
1678 * child is not running and in turn not changing child->flags
1679 * at the same time the parent does it.
1681 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1682 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1683 #define clear_used_math() clear_stopped_child_used_math(current)
1684 #define set_used_math() set_stopped_child_used_math(current)
1685 #define conditional_stopped_child_used_math(condition, child) \
1686 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1687 #define conditional_used_math(condition) \
1688 conditional_stopped_child_used_math(condition, current)
1689 #define copy_to_stopped_child_used_math(child) \
1690 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1691 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1692 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1693 #define used_math() tsk_used_math(current)
1696 extern int set_cpus_allowed_ptr(struct task_struct *p,
1697 const struct cpumask *new_mask);
1699 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1700 const struct cpumask *new_mask)
1702 if (!cpumask_test_cpu(0, new_mask))
1707 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1709 return set_cpus_allowed_ptr(p, &new_mask);
1713 * Architectures can set this to 1 if they have specified
1714 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1715 * but then during bootup it turns out that sched_clock()
1716 * is reliable after all:
1718 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1719 extern int sched_clock_stable;
1722 extern unsigned long long sched_clock(void);
1724 extern void sched_clock_init(void);
1725 extern u64 sched_clock_cpu(int cpu);
1727 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1728 static inline void sched_clock_tick(void)
1732 static inline void sched_clock_idle_sleep_event(void)
1736 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1740 extern void sched_clock_tick(void);
1741 extern void sched_clock_idle_sleep_event(void);
1742 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1746 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1747 * clock constructed from sched_clock():
1749 extern unsigned long long cpu_clock(int cpu);
1751 extern unsigned long long
1752 task_sched_runtime(struct task_struct *task);
1753 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1755 /* sched_exec is called by processes performing an exec */
1757 extern void sched_exec(void);
1759 #define sched_exec() {}
1762 extern void sched_clock_idle_sleep_event(void);
1763 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1765 #ifdef CONFIG_HOTPLUG_CPU
1766 extern void idle_task_exit(void);
1768 static inline void idle_task_exit(void) {}
1771 extern void sched_idle_next(void);
1773 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1774 extern void wake_up_idle_cpu(int cpu);
1776 static inline void wake_up_idle_cpu(int cpu) { }
1779 extern unsigned int sysctl_sched_latency;
1780 extern unsigned int sysctl_sched_min_granularity;
1781 extern unsigned int sysctl_sched_wakeup_granularity;
1782 extern unsigned int sysctl_sched_shares_ratelimit;
1783 extern unsigned int sysctl_sched_shares_thresh;
1784 #ifdef CONFIG_SCHED_DEBUG
1785 extern unsigned int sysctl_sched_child_runs_first;
1786 extern unsigned int sysctl_sched_features;
1787 extern unsigned int sysctl_sched_migration_cost;
1788 extern unsigned int sysctl_sched_nr_migrate;
1790 int sched_nr_latency_handler(struct ctl_table *table, int write,
1791 struct file *file, void __user *buffer, size_t *length,
1794 extern unsigned int sysctl_sched_rt_period;
1795 extern int sysctl_sched_rt_runtime;
1797 int sched_rt_handler(struct ctl_table *table, int write,
1798 struct file *filp, void __user *buffer, size_t *lenp,
1801 extern unsigned int sysctl_sched_compat_yield;
1803 #ifdef CONFIG_RT_MUTEXES
1804 extern int rt_mutex_getprio(struct task_struct *p);
1805 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1806 extern void rt_mutex_adjust_pi(struct task_struct *p);
1808 static inline int rt_mutex_getprio(struct task_struct *p)
1810 return p->normal_prio;
1812 # define rt_mutex_adjust_pi(p) do { } while (0)
1815 extern void set_user_nice(struct task_struct *p, long nice);
1816 extern int task_prio(const struct task_struct *p);
1817 extern int task_nice(const struct task_struct *p);
1818 extern int can_nice(const struct task_struct *p, const int nice);
1819 extern int task_curr(const struct task_struct *p);
1820 extern int idle_cpu(int cpu);
1821 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1822 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1823 struct sched_param *);
1824 extern struct task_struct *idle_task(int cpu);
1825 extern struct task_struct *curr_task(int cpu);
1826 extern void set_curr_task(int cpu, struct task_struct *p);
1831 * The default (Linux) execution domain.
1833 extern struct exec_domain default_exec_domain;
1835 union thread_union {
1836 struct thread_info thread_info;
1837 unsigned long stack[THREAD_SIZE/sizeof(long)];
1840 #ifndef __HAVE_ARCH_KSTACK_END
1841 static inline int kstack_end(void *addr)
1843 /* Reliable end of stack detection:
1844 * Some APM bios versions misalign the stack
1846 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1850 extern union thread_union init_thread_union;
1851 extern struct task_struct init_task;
1853 extern struct mm_struct init_mm;
1855 extern struct pid_namespace init_pid_ns;
1858 * find a task by one of its numerical ids
1860 * find_task_by_pid_type_ns():
1861 * it is the most generic call - it finds a task by all id,
1862 * type and namespace specified
1863 * find_task_by_pid_ns():
1864 * finds a task by its pid in the specified namespace
1865 * find_task_by_vpid():
1866 * finds a task by its virtual pid
1868 * see also find_vpid() etc in include/linux/pid.h
1871 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1872 struct pid_namespace *ns);
1874 extern struct task_struct *find_task_by_vpid(pid_t nr);
1875 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1876 struct pid_namespace *ns);
1878 extern void __set_special_pids(struct pid *pid);
1880 /* per-UID process charging. */
1881 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1882 static inline struct user_struct *get_uid(struct user_struct *u)
1884 atomic_inc(&u->__count);
1887 extern void free_uid(struct user_struct *);
1888 extern void release_uids(struct user_namespace *ns);
1890 #include <asm/current.h>
1892 extern void do_timer(unsigned long ticks);
1894 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1895 extern int wake_up_process(struct task_struct *tsk);
1896 extern void wake_up_new_task(struct task_struct *tsk,
1897 unsigned long clone_flags);
1899 extern void kick_process(struct task_struct *tsk);
1901 static inline void kick_process(struct task_struct *tsk) { }
1903 extern void sched_fork(struct task_struct *p, int clone_flags);
1904 extern void sched_dead(struct task_struct *p);
1906 extern void proc_caches_init(void);
1907 extern void flush_signals(struct task_struct *);
1908 extern void ignore_signals(struct task_struct *);
1909 extern void flush_signal_handlers(struct task_struct *, int force_default);
1910 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1912 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1914 unsigned long flags;
1917 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1918 ret = dequeue_signal(tsk, mask, info);
1919 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1924 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1926 extern void unblock_all_signals(void);
1927 extern void release_task(struct task_struct * p);
1928 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1929 extern int force_sigsegv(int, struct task_struct *);
1930 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1931 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1932 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1933 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1934 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1935 extern int kill_pid(struct pid *pid, int sig, int priv);
1936 extern int kill_proc_info(int, struct siginfo *, pid_t);
1937 extern int do_notify_parent(struct task_struct *, int);
1938 extern void force_sig(int, struct task_struct *);
1939 extern void force_sig_specific(int, struct task_struct *);
1940 extern int send_sig(int, struct task_struct *, int);
1941 extern void zap_other_threads(struct task_struct *p);
1942 extern struct sigqueue *sigqueue_alloc(void);
1943 extern void sigqueue_free(struct sigqueue *);
1944 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1945 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1946 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1948 static inline int kill_cad_pid(int sig, int priv)
1950 return kill_pid(cad_pid, sig, priv);
1953 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1954 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1955 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1956 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1958 static inline int is_si_special(const struct siginfo *info)
1960 return info <= SEND_SIG_FORCED;
1963 /* True if we are on the alternate signal stack. */
1965 static inline int on_sig_stack(unsigned long sp)
1967 return (sp - current->sas_ss_sp < current->sas_ss_size);
1970 static inline int sas_ss_flags(unsigned long sp)
1972 return (current->sas_ss_size == 0 ? SS_DISABLE
1973 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1977 * Routines for handling mm_structs
1979 extern struct mm_struct * mm_alloc(void);
1981 /* mmdrop drops the mm and the page tables */
1982 extern void __mmdrop(struct mm_struct *);
1983 static inline void mmdrop(struct mm_struct * mm)
1985 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1989 /* mmput gets rid of the mappings and all user-space */
1990 extern void mmput(struct mm_struct *);
1991 /* Grab a reference to a task's mm, if it is not already going away */
1992 extern struct mm_struct *get_task_mm(struct task_struct *task);
1993 /* Remove the current tasks stale references to the old mm_struct */
1994 extern void mm_release(struct task_struct *, struct mm_struct *);
1995 /* Allocate a new mm structure and copy contents from tsk->mm */
1996 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1998 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1999 struct task_struct *, struct pt_regs *);
2000 extern void flush_thread(void);
2001 extern void exit_thread(void);
2003 extern void exit_files(struct task_struct *);
2004 extern void __cleanup_signal(struct signal_struct *);
2005 extern void __cleanup_sighand(struct sighand_struct *);
2007 extern void exit_itimers(struct signal_struct *);
2008 extern void flush_itimer_signals(void);
2010 extern NORET_TYPE void do_group_exit(int);
2012 extern void daemonize(const char *, ...);
2013 extern int allow_signal(int);
2014 extern int disallow_signal(int);
2016 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2017 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2018 struct task_struct *fork_idle(int);
2020 extern void set_task_comm(struct task_struct *tsk, char *from);
2021 extern char *get_task_comm(char *to, struct task_struct *tsk);
2024 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2026 static inline unsigned long wait_task_inactive(struct task_struct *p,
2033 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2035 #define for_each_process(p) \
2036 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2038 extern bool is_single_threaded(struct task_struct *);
2041 * Careful: do_each_thread/while_each_thread is a double loop so
2042 * 'break' will not work as expected - use goto instead.
2044 #define do_each_thread(g, t) \
2045 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2047 #define while_each_thread(g, t) \
2048 while ((t = next_thread(t)) != g)
2050 /* de_thread depends on thread_group_leader not being a pid based check */
2051 #define thread_group_leader(p) (p == p->group_leader)
2053 /* Do to the insanities of de_thread it is possible for a process
2054 * to have the pid of the thread group leader without actually being
2055 * the thread group leader. For iteration through the pids in proc
2056 * all we care about is that we have a task with the appropriate
2057 * pid, we don't actually care if we have the right task.
2059 static inline int has_group_leader_pid(struct task_struct *p)
2061 return p->pid == p->tgid;
2065 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2067 return p1->tgid == p2->tgid;
2070 static inline struct task_struct *next_thread(const struct task_struct *p)
2072 return list_entry(rcu_dereference(p->thread_group.next),
2073 struct task_struct, thread_group);
2076 static inline int thread_group_empty(struct task_struct *p)
2078 return list_empty(&p->thread_group);
2081 #define delay_group_leader(p) \
2082 (thread_group_leader(p) && !thread_group_empty(p))
2084 static inline int task_detached(struct task_struct *p)
2086 return p->exit_signal == -1;
2090 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2091 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2092 * pins the final release of task.io_context. Also protects ->cpuset and
2093 * ->cgroup.subsys[].
2095 * Nests both inside and outside of read_lock(&tasklist_lock).
2096 * It must not be nested with write_lock_irq(&tasklist_lock),
2097 * neither inside nor outside.
2099 static inline void task_lock(struct task_struct *p)
2101 spin_lock(&p->alloc_lock);
2104 static inline void task_unlock(struct task_struct *p)
2106 spin_unlock(&p->alloc_lock);
2109 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2110 unsigned long *flags);
2112 static inline void unlock_task_sighand(struct task_struct *tsk,
2113 unsigned long *flags)
2115 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2118 #ifndef __HAVE_THREAD_FUNCTIONS
2120 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2121 #define task_stack_page(task) ((task)->stack)
2123 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2125 *task_thread_info(p) = *task_thread_info(org);
2126 task_thread_info(p)->task = p;
2129 static inline unsigned long *end_of_stack(struct task_struct *p)
2131 return (unsigned long *)(task_thread_info(p) + 1);
2136 static inline int object_is_on_stack(void *obj)
2138 void *stack = task_stack_page(current);
2140 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2143 extern void thread_info_cache_init(void);
2145 #ifdef CONFIG_DEBUG_STACK_USAGE
2146 static inline unsigned long stack_not_used(struct task_struct *p)
2148 unsigned long *n = end_of_stack(p);
2150 do { /* Skip over canary */
2154 return (unsigned long)n - (unsigned long)end_of_stack(p);
2158 /* set thread flags in other task's structures
2159 * - see asm/thread_info.h for TIF_xxxx flags available
2161 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2163 set_ti_thread_flag(task_thread_info(tsk), flag);
2166 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2168 clear_ti_thread_flag(task_thread_info(tsk), flag);
2171 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2173 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2176 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2178 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2181 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2183 return test_ti_thread_flag(task_thread_info(tsk), flag);
2186 static inline void set_tsk_need_resched(struct task_struct *tsk)
2188 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2191 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2193 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2196 static inline int test_tsk_need_resched(struct task_struct *tsk)
2198 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2201 static inline int signal_pending(struct task_struct *p)
2203 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2206 extern int __fatal_signal_pending(struct task_struct *p);
2208 static inline int fatal_signal_pending(struct task_struct *p)
2210 return signal_pending(p) && __fatal_signal_pending(p);
2213 static inline int signal_pending_state(long state, struct task_struct *p)
2215 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2217 if (!signal_pending(p))
2220 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2223 static inline int need_resched(void)
2225 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2229 * cond_resched() and cond_resched_lock(): latency reduction via
2230 * explicit rescheduling in places that are safe. The return
2231 * value indicates whether a reschedule was done in fact.
2232 * cond_resched_lock() will drop the spinlock before scheduling,
2233 * cond_resched_softirq() will enable bhs before scheduling.
2235 extern int _cond_resched(void);
2236 #ifdef CONFIG_PREEMPT_BKL
2237 static inline int cond_resched(void)
2242 static inline int cond_resched(void)
2244 return _cond_resched();
2247 extern int cond_resched_lock(spinlock_t * lock);
2248 extern int cond_resched_softirq(void);
2249 static inline int cond_resched_bkl(void)
2251 return _cond_resched();
2255 * Does a critical section need to be broken due to another
2256 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2257 * but a general need for low latency)
2259 static inline int spin_needbreak(spinlock_t *lock)
2261 #ifdef CONFIG_PREEMPT
2262 return spin_is_contended(lock);
2269 * Thread group CPU time accounting.
2271 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2272 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2274 static inline void thread_group_cputime_init(struct signal_struct *sig)
2276 sig->cputimer.cputime = INIT_CPUTIME;
2277 spin_lock_init(&sig->cputimer.lock);
2278 sig->cputimer.running = 0;
2281 static inline void thread_group_cputime_free(struct signal_struct *sig)
2286 * Reevaluate whether the task has signals pending delivery.
2287 * Wake the task if so.
2288 * This is required every time the blocked sigset_t changes.
2289 * callers must hold sighand->siglock.
2291 extern void recalc_sigpending_and_wake(struct task_struct *t);
2292 extern void recalc_sigpending(void);
2294 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2297 * Wrappers for p->thread_info->cpu access. No-op on UP.
2301 static inline unsigned int task_cpu(const struct task_struct *p)
2303 return task_thread_info(p)->cpu;
2306 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2310 static inline unsigned int task_cpu(const struct task_struct *p)
2315 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2319 #endif /* CONFIG_SMP */
2321 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2323 #ifdef CONFIG_TRACING
2325 __trace_special(void *__tr, void *__data,
2326 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2329 __trace_special(void *__tr, void *__data,
2330 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2335 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2336 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2338 extern void normalize_rt_tasks(void);
2340 #ifdef CONFIG_GROUP_SCHED
2342 extern struct task_group init_task_group;
2343 #ifdef CONFIG_USER_SCHED
2344 extern struct task_group root_task_group;
2345 extern void set_tg_uid(struct user_struct *user);
2348 extern struct task_group *sched_create_group(struct task_group *parent);
2349 extern void sched_destroy_group(struct task_group *tg);
2350 extern void sched_move_task(struct task_struct *tsk);
2351 #ifdef CONFIG_FAIR_GROUP_SCHED
2352 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2353 extern unsigned long sched_group_shares(struct task_group *tg);
2355 #ifdef CONFIG_RT_GROUP_SCHED
2356 extern int sched_group_set_rt_runtime(struct task_group *tg,
2357 long rt_runtime_us);
2358 extern long sched_group_rt_runtime(struct task_group *tg);
2359 extern int sched_group_set_rt_period(struct task_group *tg,
2361 extern long sched_group_rt_period(struct task_group *tg);
2362 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2366 extern int task_can_switch_user(struct user_struct *up,
2367 struct task_struct *tsk);
2369 #ifdef CONFIG_TASK_XACCT
2370 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2372 tsk->ioac.rchar += amt;
2375 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2377 tsk->ioac.wchar += amt;
2380 static inline void inc_syscr(struct task_struct *tsk)
2385 static inline void inc_syscw(struct task_struct *tsk)
2390 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2394 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2398 static inline void inc_syscr(struct task_struct *tsk)
2402 static inline void inc_syscw(struct task_struct *tsk)
2407 #ifndef TASK_SIZE_OF
2408 #define TASK_SIZE_OF(tsk) TASK_SIZE
2411 #ifdef CONFIG_MM_OWNER
2412 extern void mm_update_next_owner(struct mm_struct *mm);
2413 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2415 static inline void mm_update_next_owner(struct mm_struct *mm)
2419 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2422 #endif /* CONFIG_MM_OWNER */
2424 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2426 #endif /* __KERNEL__ */