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/rculist.h>
81 #include <linux/rtmutex.h>
83 #include <linux/time.h>
84 #include <linux/param.h>
85 #include <linux/resource.h>
86 #include <linux/timer.h>
87 #include <linux/hrtimer.h>
88 #include <linux/task_io_accounting.h>
89 #include <linux/kobject.h>
90 #include <linux/latencytop.h>
91 #include <linux/cred.h>
93 #include <asm/processor.h>
97 struct futex_pi_state;
98 struct robust_list_head;
104 * List of flags we want to share for kernel threads,
105 * if only because they are not used by them anyway.
107 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
110 * These are the constant used to fake the fixed-point load-average
111 * counting. Some notes:
112 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
113 * a load-average precision of 10 bits integer + 11 bits fractional
114 * - if you want to count load-averages more often, you need more
115 * precision, or rounding will get you. With 2-second counting freq,
116 * the EXP_n values would be 1981, 2034 and 2043 if still using only
119 extern unsigned long avenrun[]; /* Load averages */
120 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
122 #define FSHIFT 11 /* nr of bits of precision */
123 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
124 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
125 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
126 #define EXP_5 2014 /* 1/exp(5sec/5min) */
127 #define EXP_15 2037 /* 1/exp(5sec/15min) */
129 #define CALC_LOAD(load,exp,n) \
131 load += n*(FIXED_1-exp); \
134 extern unsigned long total_forks;
135 extern int nr_threads;
136 DECLARE_PER_CPU(unsigned long, process_counts);
137 extern int nr_processes(void);
138 extern unsigned long nr_running(void);
139 extern unsigned long nr_uninterruptible(void);
140 extern unsigned long nr_iowait(void);
141 extern void calc_global_load(void);
143 extern unsigned long get_parent_ip(unsigned long addr);
148 #ifdef CONFIG_SCHED_DEBUG
149 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
150 extern void proc_sched_set_task(struct task_struct *p);
152 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
155 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
158 static inline void proc_sched_set_task(struct task_struct *p)
162 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
167 extern unsigned long long time_sync_thresh;
170 * Task state bitmask. NOTE! These bits are also
171 * encoded in fs/proc/array.c: get_task_state().
173 * We have two separate sets of flags: task->state
174 * is about runnability, while task->exit_state are
175 * about the task exiting. Confusing, but this way
176 * modifying one set can't modify the other one by
179 #define TASK_RUNNING 0
180 #define TASK_INTERRUPTIBLE 1
181 #define TASK_UNINTERRUPTIBLE 2
182 #define __TASK_STOPPED 4
183 #define __TASK_TRACED 8
184 /* in tsk->exit_state */
185 #define EXIT_ZOMBIE 16
187 /* in tsk->state again */
189 #define TASK_WAKEKILL 128
191 /* Convenience macros for the sake of set_task_state */
192 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
193 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
194 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
196 /* Convenience macros for the sake of wake_up */
197 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
198 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
200 /* get_task_state() */
201 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
202 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
205 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
206 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
207 #define task_is_stopped_or_traced(task) \
208 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
209 #define task_contributes_to_load(task) \
210 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
211 (task->flags & PF_FROZEN) == 0)
213 #define __set_task_state(tsk, state_value) \
214 do { (tsk)->state = (state_value); } while (0)
215 #define set_task_state(tsk, state_value) \
216 set_mb((tsk)->state, (state_value))
219 * set_current_state() includes a barrier so that the write of current->state
220 * is correctly serialised wrt the caller's subsequent test of whether to
223 * set_current_state(TASK_UNINTERRUPTIBLE);
224 * if (do_i_need_to_sleep())
227 * If the caller does not need such serialisation then use __set_current_state()
229 #define __set_current_state(state_value) \
230 do { current->state = (state_value); } while (0)
231 #define set_current_state(state_value) \
232 set_mb(current->state, (state_value))
234 /* Task command name length */
235 #define TASK_COMM_LEN 16
237 #include <linux/spinlock.h>
240 * This serializes "schedule()" and also protects
241 * the run-queue from deletions/modifications (but
242 * _adding_ to the beginning of the run-queue has
245 extern rwlock_t tasklist_lock;
246 extern spinlock_t mmlist_lock;
250 extern void sched_init(void);
251 extern void sched_init_smp(void);
252 extern asmlinkage void schedule_tail(struct task_struct *prev);
253 extern void init_idle(struct task_struct *idle, int cpu);
254 extern void init_idle_bootup_task(struct task_struct *idle);
256 extern int runqueue_is_locked(void);
257 extern void task_rq_unlock_wait(struct task_struct *p);
259 extern cpumask_var_t nohz_cpu_mask;
260 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
261 extern int select_nohz_load_balancer(int cpu);
263 static inline int select_nohz_load_balancer(int cpu)
270 * Only dump TASK_* tasks. (0 for all tasks)
272 extern void show_state_filter(unsigned long state_filter);
274 static inline void show_state(void)
276 show_state_filter(0);
279 extern void show_regs(struct pt_regs *);
282 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
283 * task), SP is the stack pointer of the first frame that should be shown in the back
284 * trace (or NULL if the entire call-chain of the task should be shown).
286 extern void show_stack(struct task_struct *task, unsigned long *sp);
288 void io_schedule(void);
289 long io_schedule_timeout(long timeout);
291 extern void cpu_init (void);
292 extern void trap_init(void);
293 extern void update_process_times(int user);
294 extern void scheduler_tick(void);
296 extern void sched_show_task(struct task_struct *p);
298 #ifdef CONFIG_DETECT_SOFTLOCKUP
299 extern void softlockup_tick(void);
300 extern void touch_softlockup_watchdog(void);
301 extern void touch_all_softlockup_watchdogs(void);
302 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
303 struct file *filp, void __user *buffer,
304 size_t *lenp, loff_t *ppos);
305 extern unsigned int softlockup_panic;
306 extern int softlockup_thresh;
308 static inline void softlockup_tick(void)
311 static inline void touch_softlockup_watchdog(void)
314 static inline void touch_all_softlockup_watchdogs(void)
319 #ifdef CONFIG_DETECT_HUNG_TASK
320 extern unsigned int sysctl_hung_task_panic;
321 extern unsigned long sysctl_hung_task_check_count;
322 extern unsigned long sysctl_hung_task_timeout_secs;
323 extern unsigned long sysctl_hung_task_warnings;
324 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
325 struct file *filp, void __user *buffer,
326 size_t *lenp, loff_t *ppos);
329 /* Attach to any functions which should be ignored in wchan output. */
330 #define __sched __attribute__((__section__(".sched.text")))
332 /* Linker adds these: start and end of __sched functions */
333 extern char __sched_text_start[], __sched_text_end[];
335 /* Is this address in the __sched functions? */
336 extern int in_sched_functions(unsigned long addr);
338 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
339 extern signed long schedule_timeout(signed long timeout);
340 extern signed long schedule_timeout_interruptible(signed long timeout);
341 extern signed long schedule_timeout_killable(signed long timeout);
342 extern signed long schedule_timeout_uninterruptible(signed long timeout);
343 asmlinkage void __schedule(void);
344 asmlinkage void schedule(void);
345 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
348 struct user_namespace;
350 /* Maximum number of active map areas.. This is a random (large) number */
351 #define DEFAULT_MAX_MAP_COUNT 65536
353 extern int sysctl_max_map_count;
355 #include <linux/aio.h>
358 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
359 unsigned long, unsigned long);
361 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
362 unsigned long len, unsigned long pgoff,
363 unsigned long flags);
364 extern void arch_unmap_area(struct mm_struct *, unsigned long);
365 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
367 #if USE_SPLIT_PTLOCKS
369 * The mm counters are not protected by its page_table_lock,
370 * so must be incremented atomically.
372 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
373 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
374 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
375 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
376 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
378 #else /* !USE_SPLIT_PTLOCKS */
380 * The mm counters are protected by its page_table_lock,
381 * so can be incremented directly.
383 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
384 #define get_mm_counter(mm, member) ((mm)->_##member)
385 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
386 #define inc_mm_counter(mm, member) (mm)->_##member++
387 #define dec_mm_counter(mm, member) (mm)->_##member--
389 #endif /* !USE_SPLIT_PTLOCKS */
391 #define get_mm_rss(mm) \
392 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
393 #define update_hiwater_rss(mm) do { \
394 unsigned long _rss = get_mm_rss(mm); \
395 if ((mm)->hiwater_rss < _rss) \
396 (mm)->hiwater_rss = _rss; \
398 #define update_hiwater_vm(mm) do { \
399 if ((mm)->hiwater_vm < (mm)->total_vm) \
400 (mm)->hiwater_vm = (mm)->total_vm; \
403 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
405 return max(mm->hiwater_rss, get_mm_rss(mm));
408 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
410 return max(mm->hiwater_vm, mm->total_vm);
413 extern void set_dumpable(struct mm_struct *mm, int value);
414 extern int get_dumpable(struct mm_struct *mm);
418 #define MMF_DUMPABLE 0 /* core dump is permitted */
419 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
420 #define MMF_DUMPABLE_BITS 2
422 /* coredump filter bits */
423 #define MMF_DUMP_ANON_PRIVATE 2
424 #define MMF_DUMP_ANON_SHARED 3
425 #define MMF_DUMP_MAPPED_PRIVATE 4
426 #define MMF_DUMP_MAPPED_SHARED 5
427 #define MMF_DUMP_ELF_HEADERS 6
428 #define MMF_DUMP_HUGETLB_PRIVATE 7
429 #define MMF_DUMP_HUGETLB_SHARED 8
430 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
431 #define MMF_DUMP_FILTER_BITS 7
432 #define MMF_DUMP_FILTER_MASK \
433 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
434 #define MMF_DUMP_FILTER_DEFAULT \
435 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
436 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
438 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
439 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
441 # define MMF_DUMP_MASK_DEFAULT_ELF 0
444 struct sighand_struct {
446 struct k_sigaction action[_NSIG];
448 wait_queue_head_t signalfd_wqh;
451 struct pacct_struct {
454 unsigned long ac_mem;
455 cputime_t ac_utime, ac_stime;
456 unsigned long ac_minflt, ac_majflt;
460 * struct task_cputime - collected CPU time counts
461 * @utime: time spent in user mode, in &cputime_t units
462 * @stime: time spent in kernel mode, in &cputime_t units
463 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
465 * This structure groups together three kinds of CPU time that are
466 * tracked for threads and thread groups. Most things considering
467 * CPU time want to group these counts together and treat all three
468 * of them in parallel.
470 struct task_cputime {
473 unsigned long long sum_exec_runtime;
475 /* Alternate field names when used to cache expirations. */
476 #define prof_exp stime
477 #define virt_exp utime
478 #define sched_exp sum_exec_runtime
480 #define INIT_CPUTIME \
481 (struct task_cputime) { \
482 .utime = cputime_zero, \
483 .stime = cputime_zero, \
484 .sum_exec_runtime = 0, \
488 * struct thread_group_cputimer - thread group interval timer counts
489 * @cputime: thread group interval timers.
490 * @running: non-zero when there are timers running and
491 * @cputime receives updates.
492 * @lock: lock for fields in this struct.
494 * This structure contains the version of task_cputime, above, that is
495 * used for thread group CPU timer calculations.
497 struct thread_group_cputimer {
498 struct task_cputime cputime;
504 * NOTE! "signal_struct" does not have it's own
505 * locking, because a shared signal_struct always
506 * implies a shared sighand_struct, so locking
507 * sighand_struct is always a proper superset of
508 * the locking of signal_struct.
510 struct signal_struct {
514 wait_queue_head_t wait_chldexit; /* for wait4() */
516 /* current thread group signal load-balancing target: */
517 struct task_struct *curr_target;
519 /* shared signal handling: */
520 struct sigpending shared_pending;
522 /* thread group exit support */
525 * - notify group_exit_task when ->count is equal to notify_count
526 * - everyone except group_exit_task is stopped during signal delivery
527 * of fatal signals, group_exit_task processes the signal.
530 struct task_struct *group_exit_task;
532 /* thread group stop support, overloads group_exit_code too */
533 int group_stop_count;
534 unsigned int flags; /* see SIGNAL_* flags below */
536 /* POSIX.1b Interval Timers */
537 struct list_head posix_timers;
539 /* ITIMER_REAL timer for the process */
540 struct hrtimer real_timer;
541 struct pid *leader_pid;
542 ktime_t it_real_incr;
544 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
545 cputime_t it_prof_expires, it_virt_expires;
546 cputime_t it_prof_incr, it_virt_incr;
549 * Thread group totals for process CPU timers.
550 * See thread_group_cputimer(), et al, for details.
552 struct thread_group_cputimer cputimer;
554 /* Earliest-expiration cache. */
555 struct task_cputime cputime_expires;
557 struct list_head cpu_timers[3];
559 struct pid *tty_old_pgrp;
561 /* boolean value for session group leader */
564 struct tty_struct *tty; /* NULL if no tty */
567 * Cumulative resource counters for dead threads in the group,
568 * and for reaped dead child processes forked by this group.
569 * Live threads maintain their own counters and add to these
570 * in __exit_signal, except for the group leader.
572 cputime_t utime, stime, cutime, cstime;
575 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
576 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
577 unsigned long inblock, oublock, cinblock, coublock;
578 struct task_io_accounting ioac;
581 * Cumulative ns of schedule CPU time fo dead threads in the
582 * group, not including a zombie group leader, (This only differs
583 * from jiffies_to_ns(utime + stime) if sched_clock uses something
584 * other than jiffies.)
586 unsigned long long sum_sched_runtime;
589 * We don't bother to synchronize most readers of this at all,
590 * because there is no reader checking a limit that actually needs
591 * to get both rlim_cur and rlim_max atomically, and either one
592 * alone is a single word that can safely be read normally.
593 * getrlimit/setrlimit use task_lock(current->group_leader) to
594 * protect this instead of the siglock, because they really
595 * have no need to disable irqs.
597 struct rlimit rlim[RLIM_NLIMITS];
599 #ifdef CONFIG_BSD_PROCESS_ACCT
600 struct pacct_struct pacct; /* per-process accounting information */
602 #ifdef CONFIG_TASKSTATS
603 struct taskstats *stats;
607 struct tty_audit_buf *tty_audit_buf;
611 /* Context switch must be unlocked if interrupts are to be enabled */
612 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
613 # define __ARCH_WANT_UNLOCKED_CTXSW
617 * Bits in flags field of signal_struct.
619 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
620 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
621 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
622 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
624 * Pending notifications to parent.
626 #define SIGNAL_CLD_STOPPED 0x00000010
627 #define SIGNAL_CLD_CONTINUED 0x00000020
628 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
630 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
632 /* If true, all threads except ->group_exit_task have pending SIGKILL */
633 static inline int signal_group_exit(const struct signal_struct *sig)
635 return (sig->flags & SIGNAL_GROUP_EXIT) ||
636 (sig->group_exit_task != NULL);
640 * Some day this will be a full-fledged user tracking system..
643 atomic_t __count; /* reference count */
644 atomic_t processes; /* How many processes does this user have? */
645 atomic_t files; /* How many open files does this user have? */
646 atomic_t sigpending; /* How many pending signals does this user have? */
647 #ifdef CONFIG_INOTIFY_USER
648 atomic_t inotify_watches; /* How many inotify watches does this user have? */
649 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
652 atomic_t epoll_watches; /* The number of file descriptors currently watched */
654 #ifdef CONFIG_POSIX_MQUEUE
655 /* protected by mq_lock */
656 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
658 unsigned long locked_shm; /* How many pages of mlocked shm ? */
661 struct key *uid_keyring; /* UID specific keyring */
662 struct key *session_keyring; /* UID's default session keyring */
665 /* Hash table maintenance information */
666 struct hlist_node uidhash_node;
668 struct user_namespace *user_ns;
670 #ifdef CONFIG_USER_SCHED
671 struct task_group *tg;
674 struct work_struct work;
679 extern int uids_sysfs_init(void);
681 extern struct user_struct *find_user(uid_t);
683 extern struct user_struct root_user;
684 #define INIT_USER (&root_user)
687 struct backing_dev_info;
688 struct reclaim_state;
690 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
692 /* cumulative counters */
693 unsigned long pcount; /* # of times run on this cpu */
694 unsigned long long run_delay; /* time spent waiting on a runqueue */
697 unsigned long long last_arrival,/* when we last ran on a cpu */
698 last_queued; /* when we were last queued to run */
699 #ifdef CONFIG_SCHEDSTATS
701 unsigned int bkl_count;
704 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
706 #ifdef CONFIG_TASK_DELAY_ACCT
707 struct task_delay_info {
709 unsigned int flags; /* Private per-task flags */
711 /* For each stat XXX, add following, aligned appropriately
713 * struct timespec XXX_start, XXX_end;
717 * Atomicity of updates to XXX_delay, XXX_count protected by
718 * single lock above (split into XXX_lock if contention is an issue).
722 * XXX_count is incremented on every XXX operation, the delay
723 * associated with the operation is added to XXX_delay.
724 * XXX_delay contains the accumulated delay time in nanoseconds.
726 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
727 u64 blkio_delay; /* wait for sync block io completion */
728 u64 swapin_delay; /* wait for swapin block io completion */
729 u32 blkio_count; /* total count of the number of sync block */
730 /* io operations performed */
731 u32 swapin_count; /* total count of the number of swapin block */
732 /* io operations performed */
734 struct timespec freepages_start, freepages_end;
735 u64 freepages_delay; /* wait for memory reclaim */
736 u32 freepages_count; /* total count of memory reclaim */
738 #endif /* CONFIG_TASK_DELAY_ACCT */
740 static inline int sched_info_on(void)
742 #ifdef CONFIG_SCHEDSTATS
744 #elif defined(CONFIG_TASK_DELAY_ACCT)
745 extern int delayacct_on;
760 * sched-domains (multiprocessor balancing) declarations:
764 * Increase resolution of nice-level calculations:
766 #define SCHED_LOAD_SHIFT 10
767 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
769 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
772 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
773 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
774 #define SD_BALANCE_EXEC 4 /* Balance on exec */
775 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
776 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
777 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
778 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
779 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
780 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
781 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
782 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
783 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
785 enum powersavings_balance_level {
786 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
787 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
788 * first for long running threads
790 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
791 * cpu package for power savings
793 MAX_POWERSAVINGS_BALANCE_LEVELS
796 extern int sched_mc_power_savings, sched_smt_power_savings;
798 static inline int sd_balance_for_mc_power(void)
800 if (sched_smt_power_savings)
801 return SD_POWERSAVINGS_BALANCE;
806 static inline int sd_balance_for_package_power(void)
808 if (sched_mc_power_savings | sched_smt_power_savings)
809 return SD_POWERSAVINGS_BALANCE;
815 * Optimise SD flags for power savings:
816 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
817 * Keep default SD flags if sched_{smt,mc}_power_saving=0
820 static inline int sd_power_saving_flags(void)
822 if (sched_mc_power_savings | sched_smt_power_savings)
823 return SD_BALANCE_NEWIDLE;
829 struct sched_group *next; /* Must be a circular list */
832 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
833 * single CPU. This is read only (except for setup, hotplug CPU).
834 * Note : Never change cpu_power without recompute its reciprocal
836 unsigned int __cpu_power;
838 * reciprocal value of cpu_power to avoid expensive divides
839 * (see include/linux/reciprocal_div.h)
841 u32 reciprocal_cpu_power;
844 * The CPUs this group covers.
846 * NOTE: this field is variable length. (Allocated dynamically
847 * by attaching extra space to the end of the structure,
848 * depending on how many CPUs the kernel has booted up with)
850 * It is also be embedded into static data structures at build
851 * time. (See 'struct static_sched_group' in kernel/sched.c)
853 unsigned long cpumask[0];
856 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
858 return to_cpumask(sg->cpumask);
861 enum sched_domain_level {
871 struct sched_domain_attr {
872 int relax_domain_level;
875 #define SD_ATTR_INIT (struct sched_domain_attr) { \
876 .relax_domain_level = -1, \
879 struct sched_domain {
880 /* These fields must be setup */
881 struct sched_domain *parent; /* top domain must be null terminated */
882 struct sched_domain *child; /* bottom domain must be null terminated */
883 struct sched_group *groups; /* the balancing groups of the domain */
884 unsigned long min_interval; /* Minimum balance interval ms */
885 unsigned long max_interval; /* Maximum balance interval ms */
886 unsigned int busy_factor; /* less balancing by factor if busy */
887 unsigned int imbalance_pct; /* No balance until over watermark */
888 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
889 unsigned int busy_idx;
890 unsigned int idle_idx;
891 unsigned int newidle_idx;
892 unsigned int wake_idx;
893 unsigned int forkexec_idx;
894 int flags; /* See SD_* */
895 enum sched_domain_level level;
897 /* Runtime fields. */
898 unsigned long last_balance; /* init to jiffies. units in jiffies */
899 unsigned int balance_interval; /* initialise to 1. units in ms. */
900 unsigned int nr_balance_failed; /* initialise to 0 */
904 #ifdef CONFIG_SCHEDSTATS
905 /* load_balance() stats */
906 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
907 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
908 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
909 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
910 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
911 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
912 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
913 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
915 /* Active load balancing */
916 unsigned int alb_count;
917 unsigned int alb_failed;
918 unsigned int alb_pushed;
920 /* SD_BALANCE_EXEC stats */
921 unsigned int sbe_count;
922 unsigned int sbe_balanced;
923 unsigned int sbe_pushed;
925 /* SD_BALANCE_FORK stats */
926 unsigned int sbf_count;
927 unsigned int sbf_balanced;
928 unsigned int sbf_pushed;
930 /* try_to_wake_up() stats */
931 unsigned int ttwu_wake_remote;
932 unsigned int ttwu_move_affine;
933 unsigned int ttwu_move_balance;
935 #ifdef CONFIG_SCHED_DEBUG
940 * Span of all CPUs in this domain.
942 * NOTE: this field is variable length. (Allocated dynamically
943 * by attaching extra space to the end of the structure,
944 * depending on how many CPUs the kernel has booted up with)
946 * It is also be embedded into static data structures at build
947 * time. (See 'struct static_sched_domain' in kernel/sched.c)
949 unsigned long span[0];
952 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
954 return to_cpumask(sd->span);
957 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
958 struct sched_domain_attr *dattr_new);
960 /* Test a flag in parent sched domain */
961 static inline int test_sd_parent(struct sched_domain *sd, int flag)
963 if (sd->parent && (sd->parent->flags & flag))
969 #else /* CONFIG_SMP */
971 struct sched_domain_attr;
974 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
975 struct sched_domain_attr *dattr_new)
978 #endif /* !CONFIG_SMP */
980 struct io_context; /* See blkdev.h */
983 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
984 extern void prefetch_stack(struct task_struct *t);
986 static inline void prefetch_stack(struct task_struct *t) { }
989 struct audit_context; /* See audit.c */
991 struct pipe_inode_info;
992 struct uts_namespace;
998 const struct sched_class *next;
1000 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
1001 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1002 void (*yield_task) (struct rq *rq);
1004 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
1006 struct task_struct * (*pick_next_task) (struct rq *rq);
1007 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1010 int (*select_task_rq)(struct task_struct *p, int sync);
1012 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1013 struct rq *busiest, unsigned long max_load_move,
1014 struct sched_domain *sd, enum cpu_idle_type idle,
1015 int *all_pinned, int *this_best_prio);
1017 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1018 struct rq *busiest, struct sched_domain *sd,
1019 enum cpu_idle_type idle);
1020 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1021 int (*needs_post_schedule) (struct rq *this_rq);
1022 void (*post_schedule) (struct rq *this_rq);
1023 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1025 void (*set_cpus_allowed)(struct task_struct *p,
1026 const struct cpumask *newmask);
1028 void (*rq_online)(struct rq *rq);
1029 void (*rq_offline)(struct rq *rq);
1032 void (*set_curr_task) (struct rq *rq);
1033 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1034 void (*task_new) (struct rq *rq, struct task_struct *p);
1036 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1038 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1040 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1041 int oldprio, int running);
1043 #ifdef CONFIG_FAIR_GROUP_SCHED
1044 void (*moved_group) (struct task_struct *p);
1048 struct load_weight {
1049 unsigned long weight, inv_weight;
1053 * CFS stats for a schedulable entity (task, task-group etc)
1055 * Current field usage histogram:
1062 struct sched_entity {
1063 struct load_weight load; /* for load-balancing */
1064 struct rb_node run_node;
1065 struct list_head group_node;
1069 u64 sum_exec_runtime;
1071 u64 prev_sum_exec_runtime;
1080 #ifdef CONFIG_SCHEDSTATS
1088 s64 sum_sleep_runtime;
1095 u64 nr_migrations_cold;
1096 u64 nr_failed_migrations_affine;
1097 u64 nr_failed_migrations_running;
1098 u64 nr_failed_migrations_hot;
1099 u64 nr_forced_migrations;
1100 u64 nr_forced2_migrations;
1103 u64 nr_wakeups_sync;
1104 u64 nr_wakeups_migrate;
1105 u64 nr_wakeups_local;
1106 u64 nr_wakeups_remote;
1107 u64 nr_wakeups_affine;
1108 u64 nr_wakeups_affine_attempts;
1109 u64 nr_wakeups_passive;
1110 u64 nr_wakeups_idle;
1113 #ifdef CONFIG_FAIR_GROUP_SCHED
1114 struct sched_entity *parent;
1115 /* rq on which this entity is (to be) queued: */
1116 struct cfs_rq *cfs_rq;
1117 /* rq "owned" by this entity/group: */
1118 struct cfs_rq *my_q;
1122 struct sched_rt_entity {
1123 struct list_head run_list;
1124 unsigned long timeout;
1125 unsigned int time_slice;
1126 int nr_cpus_allowed;
1128 struct sched_rt_entity *back;
1129 #ifdef CONFIG_RT_GROUP_SCHED
1130 struct sched_rt_entity *parent;
1131 /* rq on which this entity is (to be) queued: */
1132 struct rt_rq *rt_rq;
1133 /* rq "owned" by this entity/group: */
1138 struct task_struct {
1139 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1142 unsigned int flags; /* per process flags, defined below */
1143 unsigned int ptrace;
1145 int lock_depth; /* BKL lock depth */
1148 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1153 int prio, static_prio, normal_prio;
1154 unsigned int rt_priority;
1155 const struct sched_class *sched_class;
1156 struct sched_entity se;
1157 struct sched_rt_entity rt;
1159 #ifdef CONFIG_PREEMPT_NOTIFIERS
1160 /* list of struct preempt_notifier: */
1161 struct hlist_head preempt_notifiers;
1165 * fpu_counter contains the number of consecutive context switches
1166 * that the FPU is used. If this is over a threshold, the lazy fpu
1167 * saving becomes unlazy to save the trap. This is an unsigned char
1168 * so that after 256 times the counter wraps and the behavior turns
1169 * lazy again; this to deal with bursty apps that only use FPU for
1172 unsigned char fpu_counter;
1173 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1174 #ifdef CONFIG_BLK_DEV_IO_TRACE
1175 unsigned int btrace_seq;
1178 unsigned int policy;
1179 cpumask_t cpus_allowed;
1181 #ifdef CONFIG_PREEMPT_RCU
1182 int rcu_read_lock_nesting;
1183 int rcu_flipctr_idx;
1184 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1186 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1187 struct sched_info sched_info;
1190 struct list_head tasks;
1191 struct plist_node pushable_tasks;
1193 struct mm_struct *mm, *active_mm;
1196 struct linux_binfmt *binfmt;
1198 int exit_code, exit_signal;
1199 int pdeath_signal; /* The signal sent when the parent dies */
1201 unsigned int personality;
1202 unsigned did_exec:1;
1203 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1208 /* Canary value for the -fstack-protector gcc feature */
1209 unsigned long stack_canary;
1212 * pointers to (original) parent process, youngest child, younger sibling,
1213 * older sibling, respectively. (p->father can be replaced with
1214 * p->real_parent->pid)
1216 struct task_struct *real_parent; /* real parent process */
1217 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1219 * children/sibling forms the list of my natural children
1221 struct list_head children; /* list of my children */
1222 struct list_head sibling; /* linkage in my parent's children list */
1223 struct task_struct *group_leader; /* threadgroup leader */
1226 * ptraced is the list of tasks this task is using ptrace on.
1227 * This includes both natural children and PTRACE_ATTACH targets.
1228 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1230 struct list_head ptraced;
1231 struct list_head ptrace_entry;
1234 * This is the tracer handle for the ptrace BTS extension.
1235 * This field actually belongs to the ptracer task.
1237 struct bts_context *bts;
1239 /* PID/PID hash table linkage. */
1240 struct pid_link pids[PIDTYPE_MAX];
1241 struct list_head thread_group;
1243 struct completion *vfork_done; /* for vfork() */
1244 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1245 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1247 cputime_t utime, stime, utimescaled, stimescaled;
1249 cputime_t prev_utime, prev_stime;
1250 unsigned long nvcsw, nivcsw; /* context switch counts */
1251 struct timespec start_time; /* monotonic time */
1252 struct timespec real_start_time; /* boot based time */
1253 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1254 unsigned long min_flt, maj_flt;
1256 struct task_cputime cputime_expires;
1257 struct list_head cpu_timers[3];
1259 /* process credentials */
1260 const struct cred *real_cred; /* objective and real subjective task
1261 * credentials (COW) */
1262 const struct cred *cred; /* effective (overridable) subjective task
1263 * credentials (COW) */
1264 struct mutex cred_guard_mutex; /* guard against foreign influences on
1265 * credential calculations
1266 * (notably. ptrace) */
1268 char comm[TASK_COMM_LEN]; /* executable name excluding path
1269 - access with [gs]et_task_comm (which lock
1270 it with task_lock())
1271 - initialized normally by flush_old_exec */
1272 /* file system info */
1273 int link_count, total_link_count;
1274 #ifdef CONFIG_SYSVIPC
1276 struct sysv_sem sysvsem;
1278 #ifdef CONFIG_DETECT_HUNG_TASK
1279 /* hung task detection */
1280 unsigned long last_switch_count;
1282 /* CPU-specific state of this task */
1283 struct thread_struct thread;
1284 /* filesystem information */
1285 struct fs_struct *fs;
1286 /* open file information */
1287 struct files_struct *files;
1289 struct nsproxy *nsproxy;
1290 /* signal handlers */
1291 struct signal_struct *signal;
1292 struct sighand_struct *sighand;
1294 sigset_t blocked, real_blocked;
1295 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1296 struct sigpending pending;
1298 unsigned long sas_ss_sp;
1300 int (*notifier)(void *priv);
1301 void *notifier_data;
1302 sigset_t *notifier_mask;
1303 struct audit_context *audit_context;
1304 #ifdef CONFIG_AUDITSYSCALL
1306 unsigned int sessionid;
1310 /* Thread group tracking */
1313 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1314 spinlock_t alloc_lock;
1316 #ifdef CONFIG_GENERIC_HARDIRQS
1317 /* IRQ handler threads */
1318 struct irqaction *irqaction;
1321 /* Protection of the PI data structures: */
1324 #ifdef CONFIG_RT_MUTEXES
1325 /* PI waiters blocked on a rt_mutex held by this task */
1326 struct plist_head pi_waiters;
1327 /* Deadlock detection and priority inheritance handling */
1328 struct rt_mutex_waiter *pi_blocked_on;
1331 #ifdef CONFIG_DEBUG_MUTEXES
1332 /* mutex deadlock detection */
1333 struct mutex_waiter *blocked_on;
1335 #ifdef CONFIG_TRACE_IRQFLAGS
1336 unsigned int irq_events;
1337 int hardirqs_enabled;
1338 unsigned long hardirq_enable_ip;
1339 unsigned int hardirq_enable_event;
1340 unsigned long hardirq_disable_ip;
1341 unsigned int hardirq_disable_event;
1342 int softirqs_enabled;
1343 unsigned long softirq_disable_ip;
1344 unsigned int softirq_disable_event;
1345 unsigned long softirq_enable_ip;
1346 unsigned int softirq_enable_event;
1347 int hardirq_context;
1348 int softirq_context;
1350 #ifdef CONFIG_LOCKDEP
1351 # define MAX_LOCK_DEPTH 48UL
1354 unsigned int lockdep_recursion;
1355 struct held_lock held_locks[MAX_LOCK_DEPTH];
1356 gfp_t lockdep_reclaim_gfp;
1359 /* journalling filesystem info */
1362 /* stacked block device info */
1363 struct bio *bio_list, **bio_tail;
1366 struct reclaim_state *reclaim_state;
1368 struct backing_dev_info *backing_dev_info;
1370 struct io_context *io_context;
1372 unsigned long ptrace_message;
1373 siginfo_t *last_siginfo; /* For ptrace use. */
1374 struct task_io_accounting ioac;
1375 #if defined(CONFIG_TASK_XACCT)
1376 u64 acct_rss_mem1; /* accumulated rss usage */
1377 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1378 cputime_t acct_timexpd; /* stime + utime since last update */
1380 #ifdef CONFIG_CPUSETS
1381 nodemask_t mems_allowed;
1382 int cpuset_mems_generation;
1383 int cpuset_mem_spread_rotor;
1385 #ifdef CONFIG_CGROUPS
1386 /* Control Group info protected by css_set_lock */
1387 struct css_set *cgroups;
1388 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1389 struct list_head cg_list;
1392 struct robust_list_head __user *robust_list;
1393 #ifdef CONFIG_COMPAT
1394 struct compat_robust_list_head __user *compat_robust_list;
1396 struct list_head pi_state_list;
1397 struct futex_pi_state *pi_state_cache;
1400 struct mempolicy *mempolicy;
1403 atomic_t fs_excl; /* holding fs exclusive resources */
1404 struct rcu_head rcu;
1407 * cache last used pipe for splice
1409 struct pipe_inode_info *splice_pipe;
1410 #ifdef CONFIG_TASK_DELAY_ACCT
1411 struct task_delay_info *delays;
1413 #ifdef CONFIG_FAULT_INJECTION
1416 struct prop_local_single dirties;
1417 #ifdef CONFIG_LATENCYTOP
1418 int latency_record_count;
1419 struct latency_record latency_record[LT_SAVECOUNT];
1422 * time slack values; these are used to round up poll() and
1423 * select() etc timeout values. These are in nanoseconds.
1425 unsigned long timer_slack_ns;
1426 unsigned long default_timer_slack_ns;
1428 struct list_head *scm_work_list;
1429 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1430 /* Index of current stored adress in ret_stack */
1432 /* Stack of return addresses for return function tracing */
1433 struct ftrace_ret_stack *ret_stack;
1434 /* time stamp for last schedule */
1435 unsigned long long ftrace_timestamp;
1437 * Number of functions that haven't been traced
1438 * because of depth overrun.
1440 atomic_t trace_overrun;
1441 /* Pause for the tracing */
1442 atomic_t tracing_graph_pause;
1444 #ifdef CONFIG_TRACING
1445 /* state flags for use by tracers */
1446 unsigned long trace;
1447 /* bitmask of trace recursion */
1448 unsigned long trace_recursion;
1449 #endif /* CONFIG_TRACING */
1452 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1453 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1456 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1457 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1458 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1459 * values are inverted: lower p->prio value means higher priority.
1461 * The MAX_USER_RT_PRIO value allows the actual maximum
1462 * RT priority to be separate from the value exported to
1463 * user-space. This allows kernel threads to set their
1464 * priority to a value higher than any user task. Note:
1465 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1468 #define MAX_USER_RT_PRIO 100
1469 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1471 #define MAX_PRIO (MAX_RT_PRIO + 40)
1472 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1474 static inline int rt_prio(int prio)
1476 if (unlikely(prio < MAX_RT_PRIO))
1481 static inline int rt_task(struct task_struct *p)
1483 return rt_prio(p->prio);
1486 static inline struct pid *task_pid(struct task_struct *task)
1488 return task->pids[PIDTYPE_PID].pid;
1491 static inline struct pid *task_tgid(struct task_struct *task)
1493 return task->group_leader->pids[PIDTYPE_PID].pid;
1497 * Without tasklist or rcu lock it is not safe to dereference
1498 * the result of task_pgrp/task_session even if task == current,
1499 * we can race with another thread doing sys_setsid/sys_setpgid.
1501 static inline struct pid *task_pgrp(struct task_struct *task)
1503 return task->group_leader->pids[PIDTYPE_PGID].pid;
1506 static inline struct pid *task_session(struct task_struct *task)
1508 return task->group_leader->pids[PIDTYPE_SID].pid;
1511 struct pid_namespace;
1514 * the helpers to get the task's different pids as they are seen
1515 * from various namespaces
1517 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1518 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1520 * task_xid_nr_ns() : id seen from the ns specified;
1522 * set_task_vxid() : assigns a virtual id to a task;
1524 * see also pid_nr() etc in include/linux/pid.h
1526 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1527 struct pid_namespace *ns);
1529 static inline pid_t task_pid_nr(struct task_struct *tsk)
1534 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1535 struct pid_namespace *ns)
1537 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1540 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1542 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1546 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1551 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1553 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1555 return pid_vnr(task_tgid(tsk));
1559 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1560 struct pid_namespace *ns)
1562 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1565 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1567 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1571 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1572 struct pid_namespace *ns)
1574 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1577 static inline pid_t task_session_vnr(struct task_struct *tsk)
1579 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1582 /* obsolete, do not use */
1583 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1585 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1589 * pid_alive - check that a task structure is not stale
1590 * @p: Task structure to be checked.
1592 * Test if a process is not yet dead (at most zombie state)
1593 * If pid_alive fails, then pointers within the task structure
1594 * can be stale and must not be dereferenced.
1596 static inline int pid_alive(struct task_struct *p)
1598 return p->pids[PIDTYPE_PID].pid != NULL;
1602 * is_global_init - check if a task structure is init
1603 * @tsk: Task structure to be checked.
1605 * Check if a task structure is the first user space task the kernel created.
1607 static inline int is_global_init(struct task_struct *tsk)
1609 return tsk->pid == 1;
1613 * is_container_init:
1614 * check whether in the task is init in its own pid namespace.
1616 extern int is_container_init(struct task_struct *tsk);
1618 extern struct pid *cad_pid;
1620 extern void free_task(struct task_struct *tsk);
1621 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1623 extern void __put_task_struct(struct task_struct *t);
1625 static inline void put_task_struct(struct task_struct *t)
1627 if (atomic_dec_and_test(&t->usage))
1628 __put_task_struct(t);
1631 extern cputime_t task_utime(struct task_struct *p);
1632 extern cputime_t task_stime(struct task_struct *p);
1633 extern cputime_t task_gtime(struct task_struct *p);
1638 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1639 /* Not implemented yet, only for 486*/
1640 #define PF_STARTING 0x00000002 /* being created */
1641 #define PF_EXITING 0x00000004 /* getting shut down */
1642 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1643 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1644 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1645 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1646 #define PF_DUMPCORE 0x00000200 /* dumped core */
1647 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1648 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1649 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1650 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1651 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1652 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1653 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1654 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1655 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1656 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1657 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1658 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1659 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1660 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1661 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1662 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1663 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1664 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1665 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1666 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1669 * Only the _current_ task can read/write to tsk->flags, but other
1670 * tasks can access tsk->flags in readonly mode for example
1671 * with tsk_used_math (like during threaded core dumping).
1672 * There is however an exception to this rule during ptrace
1673 * or during fork: the ptracer task is allowed to write to the
1674 * child->flags of its traced child (same goes for fork, the parent
1675 * can write to the child->flags), because we're guaranteed the
1676 * child is not running and in turn not changing child->flags
1677 * at the same time the parent does it.
1679 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1680 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1681 #define clear_used_math() clear_stopped_child_used_math(current)
1682 #define set_used_math() set_stopped_child_used_math(current)
1683 #define conditional_stopped_child_used_math(condition, child) \
1684 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1685 #define conditional_used_math(condition) \
1686 conditional_stopped_child_used_math(condition, current)
1687 #define copy_to_stopped_child_used_math(child) \
1688 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1689 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1690 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1691 #define used_math() tsk_used_math(current)
1694 extern int set_cpus_allowed_ptr(struct task_struct *p,
1695 const struct cpumask *new_mask);
1697 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1698 const struct cpumask *new_mask)
1700 if (!cpumask_test_cpu(0, new_mask))
1705 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1707 return set_cpus_allowed_ptr(p, &new_mask);
1711 * Architectures can set this to 1 if they have specified
1712 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1713 * but then during bootup it turns out that sched_clock()
1714 * is reliable after all:
1716 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1717 extern int sched_clock_stable;
1720 extern unsigned long long sched_clock(void);
1722 extern void sched_clock_init(void);
1723 extern u64 sched_clock_cpu(int cpu);
1725 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1726 static inline void sched_clock_tick(void)
1730 static inline void sched_clock_idle_sleep_event(void)
1734 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1738 extern void sched_clock_tick(void);
1739 extern void sched_clock_idle_sleep_event(void);
1740 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1744 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1745 * clock constructed from sched_clock():
1747 extern unsigned long long cpu_clock(int cpu);
1749 extern unsigned long long
1750 task_sched_runtime(struct task_struct *task);
1751 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1753 /* sched_exec is called by processes performing an exec */
1755 extern void sched_exec(void);
1757 #define sched_exec() {}
1760 extern void sched_clock_idle_sleep_event(void);
1761 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1763 #ifdef CONFIG_HOTPLUG_CPU
1764 extern void idle_task_exit(void);
1766 static inline void idle_task_exit(void) {}
1769 extern void sched_idle_next(void);
1771 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1772 extern void wake_up_idle_cpu(int cpu);
1774 static inline void wake_up_idle_cpu(int cpu) { }
1777 extern unsigned int sysctl_sched_latency;
1778 extern unsigned int sysctl_sched_min_granularity;
1779 extern unsigned int sysctl_sched_wakeup_granularity;
1780 extern unsigned int sysctl_sched_shares_ratelimit;
1781 extern unsigned int sysctl_sched_shares_thresh;
1782 #ifdef CONFIG_SCHED_DEBUG
1783 extern unsigned int sysctl_sched_child_runs_first;
1784 extern unsigned int sysctl_sched_features;
1785 extern unsigned int sysctl_sched_migration_cost;
1786 extern unsigned int sysctl_sched_nr_migrate;
1788 int sched_nr_latency_handler(struct ctl_table *table, int write,
1789 struct file *file, void __user *buffer, size_t *length,
1792 extern unsigned int sysctl_sched_rt_period;
1793 extern int sysctl_sched_rt_runtime;
1795 int sched_rt_handler(struct ctl_table *table, int write,
1796 struct file *filp, void __user *buffer, size_t *lenp,
1799 extern unsigned int sysctl_sched_compat_yield;
1801 #ifdef CONFIG_RT_MUTEXES
1802 extern int rt_mutex_getprio(struct task_struct *p);
1803 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1804 extern void rt_mutex_adjust_pi(struct task_struct *p);
1806 static inline int rt_mutex_getprio(struct task_struct *p)
1808 return p->normal_prio;
1810 # define rt_mutex_adjust_pi(p) do { } while (0)
1813 extern void set_user_nice(struct task_struct *p, long nice);
1814 extern int task_prio(const struct task_struct *p);
1815 extern int task_nice(const struct task_struct *p);
1816 extern int can_nice(const struct task_struct *p, const int nice);
1817 extern int task_curr(const struct task_struct *p);
1818 extern int idle_cpu(int cpu);
1819 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1820 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1821 struct sched_param *);
1822 extern struct task_struct *idle_task(int cpu);
1823 extern struct task_struct *curr_task(int cpu);
1824 extern void set_curr_task(int cpu, struct task_struct *p);
1829 * The default (Linux) execution domain.
1831 extern struct exec_domain default_exec_domain;
1833 union thread_union {
1834 struct thread_info thread_info;
1835 unsigned long stack[THREAD_SIZE/sizeof(long)];
1838 #ifndef __HAVE_ARCH_KSTACK_END
1839 static inline int kstack_end(void *addr)
1841 /* Reliable end of stack detection:
1842 * Some APM bios versions misalign the stack
1844 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1848 extern union thread_union init_thread_union;
1849 extern struct task_struct init_task;
1851 extern struct mm_struct init_mm;
1853 extern struct pid_namespace init_pid_ns;
1856 * find a task by one of its numerical ids
1858 * find_task_by_pid_type_ns():
1859 * it is the most generic call - it finds a task by all id,
1860 * type and namespace specified
1861 * find_task_by_pid_ns():
1862 * finds a task by its pid in the specified namespace
1863 * find_task_by_vpid():
1864 * finds a task by its virtual pid
1866 * see also find_vpid() etc in include/linux/pid.h
1869 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1870 struct pid_namespace *ns);
1872 extern struct task_struct *find_task_by_vpid(pid_t nr);
1873 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1874 struct pid_namespace *ns);
1876 extern void __set_special_pids(struct pid *pid);
1878 /* per-UID process charging. */
1879 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1880 static inline struct user_struct *get_uid(struct user_struct *u)
1882 atomic_inc(&u->__count);
1885 extern void free_uid(struct user_struct *);
1886 extern void release_uids(struct user_namespace *ns);
1888 #include <asm/current.h>
1890 extern void do_timer(unsigned long ticks);
1892 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1893 extern int wake_up_process(struct task_struct *tsk);
1894 extern void wake_up_new_task(struct task_struct *tsk,
1895 unsigned long clone_flags);
1897 extern void kick_process(struct task_struct *tsk);
1899 static inline void kick_process(struct task_struct *tsk) { }
1901 extern void sched_fork(struct task_struct *p, int clone_flags);
1902 extern void sched_dead(struct task_struct *p);
1904 extern void proc_caches_init(void);
1905 extern void flush_signals(struct task_struct *);
1906 extern void __flush_signals(struct task_struct *);
1907 extern void ignore_signals(struct task_struct *);
1908 extern void flush_signal_handlers(struct task_struct *, int force_default);
1909 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1911 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1913 unsigned long flags;
1916 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1917 ret = dequeue_signal(tsk, mask, info);
1918 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1923 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1925 extern void unblock_all_signals(void);
1926 extern void release_task(struct task_struct * p);
1927 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1928 extern int force_sigsegv(int, struct task_struct *);
1929 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1930 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1931 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1932 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1933 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1934 extern int kill_pid(struct pid *pid, int sig, int priv);
1935 extern int kill_proc_info(int, struct siginfo *, pid_t);
1936 extern int do_notify_parent(struct task_struct *, int);
1937 extern void force_sig(int, struct task_struct *);
1938 extern void force_sig_specific(int, struct task_struct *);
1939 extern int send_sig(int, struct task_struct *, int);
1940 extern void zap_other_threads(struct task_struct *p);
1941 extern struct sigqueue *sigqueue_alloc(void);
1942 extern void sigqueue_free(struct sigqueue *);
1943 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1944 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1945 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1947 static inline int kill_cad_pid(int sig, int priv)
1949 return kill_pid(cad_pid, sig, priv);
1952 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1953 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1954 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1955 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1957 static inline int is_si_special(const struct siginfo *info)
1959 return info <= SEND_SIG_FORCED;
1962 /* True if we are on the alternate signal stack. */
1964 static inline int on_sig_stack(unsigned long sp)
1966 return (sp - current->sas_ss_sp < current->sas_ss_size);
1969 static inline int sas_ss_flags(unsigned long sp)
1971 return (current->sas_ss_size == 0 ? SS_DISABLE
1972 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1976 * Routines for handling mm_structs
1978 extern struct mm_struct * mm_alloc(void);
1980 /* mmdrop drops the mm and the page tables */
1981 extern void __mmdrop(struct mm_struct *);
1982 static inline void mmdrop(struct mm_struct * mm)
1984 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1988 /* mmput gets rid of the mappings and all user-space */
1989 extern void mmput(struct mm_struct *);
1990 /* Grab a reference to a task's mm, if it is not already going away */
1991 extern struct mm_struct *get_task_mm(struct task_struct *task);
1992 /* Remove the current tasks stale references to the old mm_struct */
1993 extern void mm_release(struct task_struct *, struct mm_struct *);
1994 /* Allocate a new mm structure and copy contents from tsk->mm */
1995 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1997 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1998 struct task_struct *, struct pt_regs *);
1999 extern void flush_thread(void);
2000 extern void exit_thread(void);
2002 extern void exit_files(struct task_struct *);
2003 extern void __cleanup_signal(struct signal_struct *);
2004 extern void __cleanup_sighand(struct sighand_struct *);
2006 extern void exit_itimers(struct signal_struct *);
2007 extern void flush_itimer_signals(void);
2009 extern NORET_TYPE void do_group_exit(int);
2011 extern void daemonize(const char *, ...);
2012 extern int allow_signal(int);
2013 extern int disallow_signal(int);
2015 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2016 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2017 struct task_struct *fork_idle(int);
2019 extern void set_task_comm(struct task_struct *tsk, char *from);
2020 extern char *get_task_comm(char *to, struct task_struct *tsk);
2023 extern void wait_task_context_switch(struct task_struct *p);
2024 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2026 static inline void wait_task_context_switch(struct task_struct *p) {}
2027 static inline unsigned long wait_task_inactive(struct task_struct *p,
2034 #define next_task(p) \
2035 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2037 #define for_each_process(p) \
2038 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2040 extern bool is_single_threaded(struct task_struct *);
2043 * Careful: do_each_thread/while_each_thread is a double loop so
2044 * 'break' will not work as expected - use goto instead.
2046 #define do_each_thread(g, t) \
2047 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2049 #define while_each_thread(g, t) \
2050 while ((t = next_thread(t)) != g)
2052 /* de_thread depends on thread_group_leader not being a pid based check */
2053 #define thread_group_leader(p) (p == p->group_leader)
2055 /* Do to the insanities of de_thread it is possible for a process
2056 * to have the pid of the thread group leader without actually being
2057 * the thread group leader. For iteration through the pids in proc
2058 * all we care about is that we have a task with the appropriate
2059 * pid, we don't actually care if we have the right task.
2061 static inline int has_group_leader_pid(struct task_struct *p)
2063 return p->pid == p->tgid;
2067 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2069 return p1->tgid == p2->tgid;
2072 static inline struct task_struct *next_thread(const struct task_struct *p)
2074 return list_entry_rcu(p->thread_group.next,
2075 struct task_struct, thread_group);
2078 static inline int thread_group_empty(struct task_struct *p)
2080 return list_empty(&p->thread_group);
2083 #define delay_group_leader(p) \
2084 (thread_group_leader(p) && !thread_group_empty(p))
2086 static inline int task_detached(struct task_struct *p)
2088 return p->exit_signal == -1;
2092 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2093 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2094 * pins the final release of task.io_context. Also protects ->cpuset and
2095 * ->cgroup.subsys[].
2097 * Nests both inside and outside of read_lock(&tasklist_lock).
2098 * It must not be nested with write_lock_irq(&tasklist_lock),
2099 * neither inside nor outside.
2101 static inline void task_lock(struct task_struct *p)
2103 spin_lock(&p->alloc_lock);
2106 static inline void task_unlock(struct task_struct *p)
2108 spin_unlock(&p->alloc_lock);
2111 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2112 unsigned long *flags);
2114 static inline void unlock_task_sighand(struct task_struct *tsk,
2115 unsigned long *flags)
2117 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2120 #ifndef __HAVE_THREAD_FUNCTIONS
2122 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2123 #define task_stack_page(task) ((task)->stack)
2125 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2127 *task_thread_info(p) = *task_thread_info(org);
2128 task_thread_info(p)->task = p;
2131 static inline unsigned long *end_of_stack(struct task_struct *p)
2133 return (unsigned long *)(task_thread_info(p) + 1);
2138 static inline int object_is_on_stack(void *obj)
2140 void *stack = task_stack_page(current);
2142 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2145 extern void thread_info_cache_init(void);
2147 #ifdef CONFIG_DEBUG_STACK_USAGE
2148 static inline unsigned long stack_not_used(struct task_struct *p)
2150 unsigned long *n = end_of_stack(p);
2152 do { /* Skip over canary */
2156 return (unsigned long)n - (unsigned long)end_of_stack(p);
2160 /* set thread flags in other task's structures
2161 * - see asm/thread_info.h for TIF_xxxx flags available
2163 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2165 set_ti_thread_flag(task_thread_info(tsk), flag);
2168 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2170 clear_ti_thread_flag(task_thread_info(tsk), flag);
2173 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2175 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2178 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2180 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2183 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2185 return test_ti_thread_flag(task_thread_info(tsk), flag);
2188 static inline void set_tsk_need_resched(struct task_struct *tsk)
2190 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2193 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2195 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2198 static inline int test_tsk_need_resched(struct task_struct *tsk)
2200 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2203 static inline int signal_pending(struct task_struct *p)
2205 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2208 extern int __fatal_signal_pending(struct task_struct *p);
2210 static inline int fatal_signal_pending(struct task_struct *p)
2212 return signal_pending(p) && __fatal_signal_pending(p);
2215 static inline int signal_pending_state(long state, struct task_struct *p)
2217 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2219 if (!signal_pending(p))
2222 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2225 static inline int need_resched(void)
2227 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2231 * cond_resched() and cond_resched_lock(): latency reduction via
2232 * explicit rescheduling in places that are safe. The return
2233 * value indicates whether a reschedule was done in fact.
2234 * cond_resched_lock() will drop the spinlock before scheduling,
2235 * cond_resched_softirq() will enable bhs before scheduling.
2237 extern int _cond_resched(void);
2238 #ifdef CONFIG_PREEMPT_BKL
2239 static inline int cond_resched(void)
2244 static inline int cond_resched(void)
2246 return _cond_resched();
2249 extern int cond_resched_lock(spinlock_t * lock);
2250 extern int cond_resched_softirq(void);
2251 static inline int cond_resched_bkl(void)
2253 return _cond_resched();
2257 * Does a critical section need to be broken due to another
2258 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2259 * but a general need for low latency)
2261 static inline int spin_needbreak(spinlock_t *lock)
2263 #ifdef CONFIG_PREEMPT
2264 return spin_is_contended(lock);
2271 * Thread group CPU time accounting.
2273 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2274 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2276 static inline void thread_group_cputime_init(struct signal_struct *sig)
2278 sig->cputimer.cputime = INIT_CPUTIME;
2279 spin_lock_init(&sig->cputimer.lock);
2280 sig->cputimer.running = 0;
2283 static inline void thread_group_cputime_free(struct signal_struct *sig)
2288 * Reevaluate whether the task has signals pending delivery.
2289 * Wake the task if so.
2290 * This is required every time the blocked sigset_t changes.
2291 * callers must hold sighand->siglock.
2293 extern void recalc_sigpending_and_wake(struct task_struct *t);
2294 extern void recalc_sigpending(void);
2296 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2299 * Wrappers for p->thread_info->cpu access. No-op on UP.
2303 static inline unsigned int task_cpu(const struct task_struct *p)
2305 return task_thread_info(p)->cpu;
2308 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2312 static inline unsigned int task_cpu(const struct task_struct *p)
2317 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2321 #endif /* CONFIG_SMP */
2323 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2325 #ifdef CONFIG_TRACING
2327 __trace_special(void *__tr, void *__data,
2328 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2331 __trace_special(void *__tr, void *__data,
2332 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2337 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2338 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2340 extern void normalize_rt_tasks(void);
2342 #ifdef CONFIG_GROUP_SCHED
2344 extern struct task_group init_task_group;
2345 #ifdef CONFIG_USER_SCHED
2346 extern struct task_group root_task_group;
2347 extern void set_tg_uid(struct user_struct *user);
2350 extern struct task_group *sched_create_group(struct task_group *parent);
2351 extern void sched_destroy_group(struct task_group *tg);
2352 extern void sched_move_task(struct task_struct *tsk);
2353 #ifdef CONFIG_FAIR_GROUP_SCHED
2354 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2355 extern unsigned long sched_group_shares(struct task_group *tg);
2357 #ifdef CONFIG_RT_GROUP_SCHED
2358 extern int sched_group_set_rt_runtime(struct task_group *tg,
2359 long rt_runtime_us);
2360 extern long sched_group_rt_runtime(struct task_group *tg);
2361 extern int sched_group_set_rt_period(struct task_group *tg,
2363 extern long sched_group_rt_period(struct task_group *tg);
2364 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2368 extern int task_can_switch_user(struct user_struct *up,
2369 struct task_struct *tsk);
2371 #ifdef CONFIG_TASK_XACCT
2372 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2374 tsk->ioac.rchar += amt;
2377 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2379 tsk->ioac.wchar += amt;
2382 static inline void inc_syscr(struct task_struct *tsk)
2387 static inline void inc_syscw(struct task_struct *tsk)
2392 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2396 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2400 static inline void inc_syscr(struct task_struct *tsk)
2404 static inline void inc_syscw(struct task_struct *tsk)
2409 #ifndef TASK_SIZE_OF
2410 #define TASK_SIZE_OF(tsk) TASK_SIZE
2413 #ifdef CONFIG_MM_OWNER
2414 extern void mm_update_next_owner(struct mm_struct *mm);
2415 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2417 static inline void mm_update_next_owner(struct mm_struct *mm)
2421 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2424 #endif /* CONFIG_MM_OWNER */
2426 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2428 #endif /* __KERNEL__ */