]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blame_incremental - include/linux/sched.h
Linux 2.6.27-rc6
[mirror_ubuntu-zesty-kernel.git] / include / linux / sched.h
... / ...
CommitLineData
1#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
4/*
5 * cloning flags:
6 */
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 */
31
32/*
33 * Scheduling policies
34 */
35#define SCHED_NORMAL 0
36#define SCHED_FIFO 1
37#define SCHED_RR 2
38#define SCHED_BATCH 3
39/* SCHED_ISO: reserved but not implemented yet */
40#define SCHED_IDLE 5
41
42#ifdef __KERNEL__
43
44struct sched_param {
45 int sched_priority;
46};
47
48#include <asm/param.h> /* for HZ */
49
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>
62
63#include <asm/system.h>
64#include <asm/page.h>
65#include <asm/ptrace.h>
66#include <asm/cputime.h>
67
68#include <linux/smp.h>
69#include <linux/sem.h>
70#include <linux/signal.h>
71#include <linux/fs_struct.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>
81
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>
91
92#include <asm/processor.h>
93
94struct mem_cgroup;
95struct exec_domain;
96struct futex_pi_state;
97struct robust_list_head;
98struct bio;
99
100/*
101 * List of flags we want to share for kernel threads,
102 * if only because they are not used by them anyway.
103 */
104#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
105
106/*
107 * These are the constant used to fake the fixed-point load-average
108 * counting. Some notes:
109 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
110 * a load-average precision of 10 bits integer + 11 bits fractional
111 * - if you want to count load-averages more often, you need more
112 * precision, or rounding will get you. With 2-second counting freq,
113 * the EXP_n values would be 1981, 2034 and 2043 if still using only
114 * 11 bit fractions.
115 */
116extern unsigned long avenrun[]; /* Load averages */
117
118#define FSHIFT 11 /* nr of bits of precision */
119#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
120#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
121#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
122#define EXP_5 2014 /* 1/exp(5sec/5min) */
123#define EXP_15 2037 /* 1/exp(5sec/15min) */
124
125#define CALC_LOAD(load,exp,n) \
126 load *= exp; \
127 load += n*(FIXED_1-exp); \
128 load >>= FSHIFT;
129
130extern unsigned long total_forks;
131extern int nr_threads;
132DECLARE_PER_CPU(unsigned long, process_counts);
133extern int nr_processes(void);
134extern unsigned long nr_running(void);
135extern unsigned long nr_uninterruptible(void);
136extern unsigned long nr_active(void);
137extern unsigned long nr_iowait(void);
138
139struct seq_file;
140struct cfs_rq;
141struct task_group;
142#ifdef CONFIG_SCHED_DEBUG
143extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
144extern void proc_sched_set_task(struct task_struct *p);
145extern void
146print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
147#else
148static inline void
149proc_sched_show_task(struct task_struct *p, struct seq_file *m)
150{
151}
152static inline void proc_sched_set_task(struct task_struct *p)
153{
154}
155static inline void
156print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
157{
158}
159#endif
160
161extern unsigned long long time_sync_thresh;
162
163/*
164 * Task state bitmask. NOTE! These bits are also
165 * encoded in fs/proc/array.c: get_task_state().
166 *
167 * We have two separate sets of flags: task->state
168 * is about runnability, while task->exit_state are
169 * about the task exiting. Confusing, but this way
170 * modifying one set can't modify the other one by
171 * mistake.
172 */
173#define TASK_RUNNING 0
174#define TASK_INTERRUPTIBLE 1
175#define TASK_UNINTERRUPTIBLE 2
176#define __TASK_STOPPED 4
177#define __TASK_TRACED 8
178/* in tsk->exit_state */
179#define EXIT_ZOMBIE 16
180#define EXIT_DEAD 32
181/* in tsk->state again */
182#define TASK_DEAD 64
183#define TASK_WAKEKILL 128
184
185/* Convenience macros for the sake of set_task_state */
186#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
187#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
188#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
189
190/* Convenience macros for the sake of wake_up */
191#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
192#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
193
194/* get_task_state() */
195#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
196 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
197 __TASK_TRACED)
198
199#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
200#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
201#define task_is_stopped_or_traced(task) \
202 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
203#define task_contributes_to_load(task) \
204 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
205
206#define __set_task_state(tsk, state_value) \
207 do { (tsk)->state = (state_value); } while (0)
208#define set_task_state(tsk, state_value) \
209 set_mb((tsk)->state, (state_value))
210
211/*
212 * set_current_state() includes a barrier so that the write of current->state
213 * is correctly serialised wrt the caller's subsequent test of whether to
214 * actually sleep:
215 *
216 * set_current_state(TASK_UNINTERRUPTIBLE);
217 * if (do_i_need_to_sleep())
218 * schedule();
219 *
220 * If the caller does not need such serialisation then use __set_current_state()
221 */
222#define __set_current_state(state_value) \
223 do { current->state = (state_value); } while (0)
224#define set_current_state(state_value) \
225 set_mb(current->state, (state_value))
226
227/* Task command name length */
228#define TASK_COMM_LEN 16
229
230#include <linux/spinlock.h>
231
232/*
233 * This serializes "schedule()" and also protects
234 * the run-queue from deletions/modifications (but
235 * _adding_ to the beginning of the run-queue has
236 * a separate lock).
237 */
238extern rwlock_t tasklist_lock;
239extern spinlock_t mmlist_lock;
240
241struct task_struct;
242
243extern void sched_init(void);
244extern void sched_init_smp(void);
245extern asmlinkage void schedule_tail(struct task_struct *prev);
246extern void init_idle(struct task_struct *idle, int cpu);
247extern void init_idle_bootup_task(struct task_struct *idle);
248
249extern int runqueue_is_locked(void);
250
251extern cpumask_t nohz_cpu_mask;
252#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
253extern int select_nohz_load_balancer(int cpu);
254#else
255static inline int select_nohz_load_balancer(int cpu)
256{
257 return 0;
258}
259#endif
260
261extern unsigned long rt_needs_cpu(int cpu);
262
263/*
264 * Only dump TASK_* tasks. (0 for all tasks)
265 */
266extern void show_state_filter(unsigned long state_filter);
267
268static inline void show_state(void)
269{
270 show_state_filter(0);
271}
272
273extern void show_regs(struct pt_regs *);
274
275/*
276 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
277 * task), SP is the stack pointer of the first frame that should be shown in the back
278 * trace (or NULL if the entire call-chain of the task should be shown).
279 */
280extern void show_stack(struct task_struct *task, unsigned long *sp);
281
282void io_schedule(void);
283long io_schedule_timeout(long timeout);
284
285extern void cpu_init (void);
286extern void trap_init(void);
287extern void account_process_tick(struct task_struct *task, int user);
288extern void update_process_times(int user);
289extern void scheduler_tick(void);
290extern void hrtick_resched(void);
291
292extern void sched_show_task(struct task_struct *p);
293
294#ifdef CONFIG_DETECT_SOFTLOCKUP
295extern void softlockup_tick(void);
296extern void touch_softlockup_watchdog(void);
297extern void touch_all_softlockup_watchdogs(void);
298extern unsigned int softlockup_panic;
299extern unsigned long sysctl_hung_task_check_count;
300extern unsigned long sysctl_hung_task_timeout_secs;
301extern unsigned long sysctl_hung_task_warnings;
302extern int softlockup_thresh;
303#else
304static inline void softlockup_tick(void)
305{
306}
307static inline void spawn_softlockup_task(void)
308{
309}
310static inline void touch_softlockup_watchdog(void)
311{
312}
313static inline void touch_all_softlockup_watchdogs(void)
314{
315}
316#endif
317
318
319/* Attach to any functions which should be ignored in wchan output. */
320#define __sched __attribute__((__section__(".sched.text")))
321
322/* Linker adds these: start and end of __sched functions */
323extern char __sched_text_start[], __sched_text_end[];
324
325/* Is this address in the __sched functions? */
326extern int in_sched_functions(unsigned long addr);
327
328#define MAX_SCHEDULE_TIMEOUT LONG_MAX
329extern signed long schedule_timeout(signed long timeout);
330extern signed long schedule_timeout_interruptible(signed long timeout);
331extern signed long schedule_timeout_killable(signed long timeout);
332extern signed long schedule_timeout_uninterruptible(signed long timeout);
333asmlinkage void schedule(void);
334
335struct nsproxy;
336struct user_namespace;
337
338/* Maximum number of active map areas.. This is a random (large) number */
339#define DEFAULT_MAX_MAP_COUNT 65536
340
341extern int sysctl_max_map_count;
342
343#include <linux/aio.h>
344
345extern unsigned long
346arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
347 unsigned long, unsigned long);
348extern unsigned long
349arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
350 unsigned long len, unsigned long pgoff,
351 unsigned long flags);
352extern void arch_unmap_area(struct mm_struct *, unsigned long);
353extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
354
355#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
356/*
357 * The mm counters are not protected by its page_table_lock,
358 * so must be incremented atomically.
359 */
360#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
361#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
362#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
363#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
364#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
365
366#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
367/*
368 * The mm counters are protected by its page_table_lock,
369 * so can be incremented directly.
370 */
371#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
372#define get_mm_counter(mm, member) ((mm)->_##member)
373#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
374#define inc_mm_counter(mm, member) (mm)->_##member++
375#define dec_mm_counter(mm, member) (mm)->_##member--
376
377#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
378
379#define get_mm_rss(mm) \
380 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
381#define update_hiwater_rss(mm) do { \
382 unsigned long _rss = get_mm_rss(mm); \
383 if ((mm)->hiwater_rss < _rss) \
384 (mm)->hiwater_rss = _rss; \
385} while (0)
386#define update_hiwater_vm(mm) do { \
387 if ((mm)->hiwater_vm < (mm)->total_vm) \
388 (mm)->hiwater_vm = (mm)->total_vm; \
389} while (0)
390
391extern void set_dumpable(struct mm_struct *mm, int value);
392extern int get_dumpable(struct mm_struct *mm);
393
394/* mm flags */
395/* dumpable bits */
396#define MMF_DUMPABLE 0 /* core dump is permitted */
397#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
398#define MMF_DUMPABLE_BITS 2
399
400/* coredump filter bits */
401#define MMF_DUMP_ANON_PRIVATE 2
402#define MMF_DUMP_ANON_SHARED 3
403#define MMF_DUMP_MAPPED_PRIVATE 4
404#define MMF_DUMP_MAPPED_SHARED 5
405#define MMF_DUMP_ELF_HEADERS 6
406#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
407#define MMF_DUMP_FILTER_BITS 5
408#define MMF_DUMP_FILTER_MASK \
409 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
410#define MMF_DUMP_FILTER_DEFAULT \
411 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
412
413struct sighand_struct {
414 atomic_t count;
415 struct k_sigaction action[_NSIG];
416 spinlock_t siglock;
417 wait_queue_head_t signalfd_wqh;
418};
419
420struct pacct_struct {
421 int ac_flag;
422 long ac_exitcode;
423 unsigned long ac_mem;
424 cputime_t ac_utime, ac_stime;
425 unsigned long ac_minflt, ac_majflt;
426};
427
428/*
429 * NOTE! "signal_struct" does not have it's own
430 * locking, because a shared signal_struct always
431 * implies a shared sighand_struct, so locking
432 * sighand_struct is always a proper superset of
433 * the locking of signal_struct.
434 */
435struct signal_struct {
436 atomic_t count;
437 atomic_t live;
438
439 wait_queue_head_t wait_chldexit; /* for wait4() */
440
441 /* current thread group signal load-balancing target: */
442 struct task_struct *curr_target;
443
444 /* shared signal handling: */
445 struct sigpending shared_pending;
446
447 /* thread group exit support */
448 int group_exit_code;
449 /* overloaded:
450 * - notify group_exit_task when ->count is equal to notify_count
451 * - everyone except group_exit_task is stopped during signal delivery
452 * of fatal signals, group_exit_task processes the signal.
453 */
454 struct task_struct *group_exit_task;
455 int notify_count;
456
457 /* thread group stop support, overloads group_exit_code too */
458 int group_stop_count;
459 unsigned int flags; /* see SIGNAL_* flags below */
460
461 /* POSIX.1b Interval Timers */
462 struct list_head posix_timers;
463
464 /* ITIMER_REAL timer for the process */
465 struct hrtimer real_timer;
466 struct pid *leader_pid;
467 ktime_t it_real_incr;
468
469 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
470 cputime_t it_prof_expires, it_virt_expires;
471 cputime_t it_prof_incr, it_virt_incr;
472
473 /* job control IDs */
474
475 /*
476 * pgrp and session fields are deprecated.
477 * use the task_session_Xnr and task_pgrp_Xnr routines below
478 */
479
480 union {
481 pid_t pgrp __deprecated;
482 pid_t __pgrp;
483 };
484
485 struct pid *tty_old_pgrp;
486
487 union {
488 pid_t session __deprecated;
489 pid_t __session;
490 };
491
492 /* boolean value for session group leader */
493 int leader;
494
495 struct tty_struct *tty; /* NULL if no tty */
496
497 /*
498 * Cumulative resource counters for dead threads in the group,
499 * and for reaped dead child processes forked by this group.
500 * Live threads maintain their own counters and add to these
501 * in __exit_signal, except for the group leader.
502 */
503 cputime_t utime, stime, cutime, cstime;
504 cputime_t gtime;
505 cputime_t cgtime;
506 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
507 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
508 unsigned long inblock, oublock, cinblock, coublock;
509 struct task_io_accounting ioac;
510
511 /*
512 * Cumulative ns of scheduled CPU time for dead threads in the
513 * group, not including a zombie group leader. (This only differs
514 * from jiffies_to_ns(utime + stime) if sched_clock uses something
515 * other than jiffies.)
516 */
517 unsigned long long sum_sched_runtime;
518
519 /*
520 * We don't bother to synchronize most readers of this at all,
521 * because there is no reader checking a limit that actually needs
522 * to get both rlim_cur and rlim_max atomically, and either one
523 * alone is a single word that can safely be read normally.
524 * getrlimit/setrlimit use task_lock(current->group_leader) to
525 * protect this instead of the siglock, because they really
526 * have no need to disable irqs.
527 */
528 struct rlimit rlim[RLIM_NLIMITS];
529
530 struct list_head cpu_timers[3];
531
532 /* keep the process-shared keyrings here so that they do the right
533 * thing in threads created with CLONE_THREAD */
534#ifdef CONFIG_KEYS
535 struct key *session_keyring; /* keyring inherited over fork */
536 struct key *process_keyring; /* keyring private to this process */
537#endif
538#ifdef CONFIG_BSD_PROCESS_ACCT
539 struct pacct_struct pacct; /* per-process accounting information */
540#endif
541#ifdef CONFIG_TASKSTATS
542 struct taskstats *stats;
543#endif
544#ifdef CONFIG_AUDIT
545 unsigned audit_tty;
546 struct tty_audit_buf *tty_audit_buf;
547#endif
548};
549
550/* Context switch must be unlocked if interrupts are to be enabled */
551#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
552# define __ARCH_WANT_UNLOCKED_CTXSW
553#endif
554
555/*
556 * Bits in flags field of signal_struct.
557 */
558#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
559#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
560#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
561#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
562/*
563 * Pending notifications to parent.
564 */
565#define SIGNAL_CLD_STOPPED 0x00000010
566#define SIGNAL_CLD_CONTINUED 0x00000020
567#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
568
569#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
570
571/* If true, all threads except ->group_exit_task have pending SIGKILL */
572static inline int signal_group_exit(const struct signal_struct *sig)
573{
574 return (sig->flags & SIGNAL_GROUP_EXIT) ||
575 (sig->group_exit_task != NULL);
576}
577
578/*
579 * Some day this will be a full-fledged user tracking system..
580 */
581struct user_struct {
582 atomic_t __count; /* reference count */
583 atomic_t processes; /* How many processes does this user have? */
584 atomic_t files; /* How many open files does this user have? */
585 atomic_t sigpending; /* How many pending signals does this user have? */
586#ifdef CONFIG_INOTIFY_USER
587 atomic_t inotify_watches; /* How many inotify watches does this user have? */
588 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
589#endif
590#ifdef CONFIG_POSIX_MQUEUE
591 /* protected by mq_lock */
592 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
593#endif
594 unsigned long locked_shm; /* How many pages of mlocked shm ? */
595
596#ifdef CONFIG_KEYS
597 struct key *uid_keyring; /* UID specific keyring */
598 struct key *session_keyring; /* UID's default session keyring */
599#endif
600
601 /* Hash table maintenance information */
602 struct hlist_node uidhash_node;
603 uid_t uid;
604
605#ifdef CONFIG_USER_SCHED
606 struct task_group *tg;
607#ifdef CONFIG_SYSFS
608 struct kobject kobj;
609 struct work_struct work;
610#endif
611#endif
612};
613
614extern int uids_sysfs_init(void);
615
616extern struct user_struct *find_user(uid_t);
617
618extern struct user_struct root_user;
619#define INIT_USER (&root_user)
620
621struct backing_dev_info;
622struct reclaim_state;
623
624#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
625struct sched_info {
626 /* cumulative counters */
627 unsigned long pcount; /* # of times run on this cpu */
628 unsigned long long cpu_time, /* time spent on the cpu */
629 run_delay; /* time spent waiting on a runqueue */
630
631 /* timestamps */
632 unsigned long long last_arrival,/* when we last ran on a cpu */
633 last_queued; /* when we were last queued to run */
634#ifdef CONFIG_SCHEDSTATS
635 /* BKL stats */
636 unsigned int bkl_count;
637#endif
638};
639#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
640
641#ifdef CONFIG_SCHEDSTATS
642extern const struct file_operations proc_schedstat_operations;
643#endif /* CONFIG_SCHEDSTATS */
644
645#ifdef CONFIG_TASK_DELAY_ACCT
646struct task_delay_info {
647 spinlock_t lock;
648 unsigned int flags; /* Private per-task flags */
649
650 /* For each stat XXX, add following, aligned appropriately
651 *
652 * struct timespec XXX_start, XXX_end;
653 * u64 XXX_delay;
654 * u32 XXX_count;
655 *
656 * Atomicity of updates to XXX_delay, XXX_count protected by
657 * single lock above (split into XXX_lock if contention is an issue).
658 */
659
660 /*
661 * XXX_count is incremented on every XXX operation, the delay
662 * associated with the operation is added to XXX_delay.
663 * XXX_delay contains the accumulated delay time in nanoseconds.
664 */
665 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
666 u64 blkio_delay; /* wait for sync block io completion */
667 u64 swapin_delay; /* wait for swapin block io completion */
668 u32 blkio_count; /* total count of the number of sync block */
669 /* io operations performed */
670 u32 swapin_count; /* total count of the number of swapin block */
671 /* io operations performed */
672
673 struct timespec freepages_start, freepages_end;
674 u64 freepages_delay; /* wait for memory reclaim */
675 u32 freepages_count; /* total count of memory reclaim */
676};
677#endif /* CONFIG_TASK_DELAY_ACCT */
678
679static inline int sched_info_on(void)
680{
681#ifdef CONFIG_SCHEDSTATS
682 return 1;
683#elif defined(CONFIG_TASK_DELAY_ACCT)
684 extern int delayacct_on;
685 return delayacct_on;
686#else
687 return 0;
688#endif
689}
690
691enum cpu_idle_type {
692 CPU_IDLE,
693 CPU_NOT_IDLE,
694 CPU_NEWLY_IDLE,
695 CPU_MAX_IDLE_TYPES
696};
697
698/*
699 * sched-domains (multiprocessor balancing) declarations:
700 */
701
702/*
703 * Increase resolution of nice-level calculations:
704 */
705#define SCHED_LOAD_SHIFT 10
706#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
707
708#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
709
710#ifdef CONFIG_SMP
711#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
712#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
713#define SD_BALANCE_EXEC 4 /* Balance on exec */
714#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
715#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
716#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
717#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
718#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
719#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
720#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
721#define SD_SERIALIZE 1024 /* Only a single load balancing instance */
722#define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
723
724#define BALANCE_FOR_MC_POWER \
725 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
726
727#define BALANCE_FOR_PKG_POWER \
728 ((sched_mc_power_savings || sched_smt_power_savings) ? \
729 SD_POWERSAVINGS_BALANCE : 0)
730
731#define test_sd_parent(sd, flag) ((sd->parent && \
732 (sd->parent->flags & flag)) ? 1 : 0)
733
734
735struct sched_group {
736 struct sched_group *next; /* Must be a circular list */
737 cpumask_t cpumask;
738
739 /*
740 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
741 * single CPU. This is read only (except for setup, hotplug CPU).
742 * Note : Never change cpu_power without recompute its reciprocal
743 */
744 unsigned int __cpu_power;
745 /*
746 * reciprocal value of cpu_power to avoid expensive divides
747 * (see include/linux/reciprocal_div.h)
748 */
749 u32 reciprocal_cpu_power;
750};
751
752enum sched_domain_level {
753 SD_LV_NONE = 0,
754 SD_LV_SIBLING,
755 SD_LV_MC,
756 SD_LV_CPU,
757 SD_LV_NODE,
758 SD_LV_ALLNODES,
759 SD_LV_MAX
760};
761
762struct sched_domain_attr {
763 int relax_domain_level;
764};
765
766#define SD_ATTR_INIT (struct sched_domain_attr) { \
767 .relax_domain_level = -1, \
768}
769
770struct sched_domain {
771 /* These fields must be setup */
772 struct sched_domain *parent; /* top domain must be null terminated */
773 struct sched_domain *child; /* bottom domain must be null terminated */
774 struct sched_group *groups; /* the balancing groups of the domain */
775 cpumask_t span; /* span of all CPUs in this domain */
776 unsigned long min_interval; /* Minimum balance interval ms */
777 unsigned long max_interval; /* Maximum balance interval ms */
778 unsigned int busy_factor; /* less balancing by factor if busy */
779 unsigned int imbalance_pct; /* No balance until over watermark */
780 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
781 unsigned int busy_idx;
782 unsigned int idle_idx;
783 unsigned int newidle_idx;
784 unsigned int wake_idx;
785 unsigned int forkexec_idx;
786 int flags; /* See SD_* */
787 enum sched_domain_level level;
788
789 /* Runtime fields. */
790 unsigned long last_balance; /* init to jiffies. units in jiffies */
791 unsigned int balance_interval; /* initialise to 1. units in ms. */
792 unsigned int nr_balance_failed; /* initialise to 0 */
793
794 u64 last_update;
795
796#ifdef CONFIG_SCHEDSTATS
797 /* load_balance() stats */
798 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
799 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
800 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
801 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
802 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
803 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
804 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
805 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
806
807 /* Active load balancing */
808 unsigned int alb_count;
809 unsigned int alb_failed;
810 unsigned int alb_pushed;
811
812 /* SD_BALANCE_EXEC stats */
813 unsigned int sbe_count;
814 unsigned int sbe_balanced;
815 unsigned int sbe_pushed;
816
817 /* SD_BALANCE_FORK stats */
818 unsigned int sbf_count;
819 unsigned int sbf_balanced;
820 unsigned int sbf_pushed;
821
822 /* try_to_wake_up() stats */
823 unsigned int ttwu_wake_remote;
824 unsigned int ttwu_move_affine;
825 unsigned int ttwu_move_balance;
826#endif
827};
828
829extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new,
830 struct sched_domain_attr *dattr_new);
831extern int arch_reinit_sched_domains(void);
832
833#else /* CONFIG_SMP */
834
835struct sched_domain_attr;
836
837static inline void
838partition_sched_domains(int ndoms_new, cpumask_t *doms_new,
839 struct sched_domain_attr *dattr_new)
840{
841}
842#endif /* !CONFIG_SMP */
843
844struct io_context; /* See blkdev.h */
845#define NGROUPS_SMALL 32
846#define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
847struct group_info {
848 int ngroups;
849 atomic_t usage;
850 gid_t small_block[NGROUPS_SMALL];
851 int nblocks;
852 gid_t *blocks[0];
853};
854
855/*
856 * get_group_info() must be called with the owning task locked (via task_lock())
857 * when task != current. The reason being that the vast majority of callers are
858 * looking at current->group_info, which can not be changed except by the
859 * current task. Changing current->group_info requires the task lock, too.
860 */
861#define get_group_info(group_info) do { \
862 atomic_inc(&(group_info)->usage); \
863} while (0)
864
865#define put_group_info(group_info) do { \
866 if (atomic_dec_and_test(&(group_info)->usage)) \
867 groups_free(group_info); \
868} while (0)
869
870extern struct group_info *groups_alloc(int gidsetsize);
871extern void groups_free(struct group_info *group_info);
872extern int set_current_groups(struct group_info *group_info);
873extern int groups_search(struct group_info *group_info, gid_t grp);
874/* access the groups "array" with this macro */
875#define GROUP_AT(gi, i) \
876 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
877
878#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
879extern void prefetch_stack(struct task_struct *t);
880#else
881static inline void prefetch_stack(struct task_struct *t) { }
882#endif
883
884struct audit_context; /* See audit.c */
885struct mempolicy;
886struct pipe_inode_info;
887struct uts_namespace;
888
889struct rq;
890struct sched_domain;
891
892struct sched_class {
893 const struct sched_class *next;
894
895 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
896 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
897 void (*yield_task) (struct rq *rq);
898 int (*select_task_rq)(struct task_struct *p, int sync);
899
900 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
901
902 struct task_struct * (*pick_next_task) (struct rq *rq);
903 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
904
905#ifdef CONFIG_SMP
906 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
907 struct rq *busiest, unsigned long max_load_move,
908 struct sched_domain *sd, enum cpu_idle_type idle,
909 int *all_pinned, int *this_best_prio);
910
911 int (*move_one_task) (struct rq *this_rq, int this_cpu,
912 struct rq *busiest, struct sched_domain *sd,
913 enum cpu_idle_type idle);
914 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
915 void (*post_schedule) (struct rq *this_rq);
916 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
917#endif
918
919 void (*set_curr_task) (struct rq *rq);
920 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
921 void (*task_new) (struct rq *rq, struct task_struct *p);
922 void (*set_cpus_allowed)(struct task_struct *p,
923 const cpumask_t *newmask);
924
925 void (*rq_online)(struct rq *rq);
926 void (*rq_offline)(struct rq *rq);
927
928 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
929 int running);
930 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
931 int running);
932 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
933 int oldprio, int running);
934
935#ifdef CONFIG_FAIR_GROUP_SCHED
936 void (*moved_group) (struct task_struct *p);
937#endif
938};
939
940struct load_weight {
941 unsigned long weight, inv_weight;
942};
943
944/*
945 * CFS stats for a schedulable entity (task, task-group etc)
946 *
947 * Current field usage histogram:
948 *
949 * 4 se->block_start
950 * 4 se->run_node
951 * 4 se->sleep_start
952 * 6 se->load.weight
953 */
954struct sched_entity {
955 struct load_weight load; /* for load-balancing */
956 struct rb_node run_node;
957 struct list_head group_node;
958 unsigned int on_rq;
959
960 u64 exec_start;
961 u64 sum_exec_runtime;
962 u64 vruntime;
963 u64 prev_sum_exec_runtime;
964
965 u64 last_wakeup;
966 u64 avg_overlap;
967
968#ifdef CONFIG_SCHEDSTATS
969 u64 wait_start;
970 u64 wait_max;
971 u64 wait_count;
972 u64 wait_sum;
973
974 u64 sleep_start;
975 u64 sleep_max;
976 s64 sum_sleep_runtime;
977
978 u64 block_start;
979 u64 block_max;
980 u64 exec_max;
981 u64 slice_max;
982
983 u64 nr_migrations;
984 u64 nr_migrations_cold;
985 u64 nr_failed_migrations_affine;
986 u64 nr_failed_migrations_running;
987 u64 nr_failed_migrations_hot;
988 u64 nr_forced_migrations;
989 u64 nr_forced2_migrations;
990
991 u64 nr_wakeups;
992 u64 nr_wakeups_sync;
993 u64 nr_wakeups_migrate;
994 u64 nr_wakeups_local;
995 u64 nr_wakeups_remote;
996 u64 nr_wakeups_affine;
997 u64 nr_wakeups_affine_attempts;
998 u64 nr_wakeups_passive;
999 u64 nr_wakeups_idle;
1000#endif
1001
1002#ifdef CONFIG_FAIR_GROUP_SCHED
1003 struct sched_entity *parent;
1004 /* rq on which this entity is (to be) queued: */
1005 struct cfs_rq *cfs_rq;
1006 /* rq "owned" by this entity/group: */
1007 struct cfs_rq *my_q;
1008#endif
1009};
1010
1011struct sched_rt_entity {
1012 struct list_head run_list;
1013 unsigned int time_slice;
1014 unsigned long timeout;
1015 int nr_cpus_allowed;
1016
1017 struct sched_rt_entity *back;
1018#ifdef CONFIG_RT_GROUP_SCHED
1019 struct sched_rt_entity *parent;
1020 /* rq on which this entity is (to be) queued: */
1021 struct rt_rq *rt_rq;
1022 /* rq "owned" by this entity/group: */
1023 struct rt_rq *my_q;
1024#endif
1025};
1026
1027struct task_struct {
1028 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1029 void *stack;
1030 atomic_t usage;
1031 unsigned int flags; /* per process flags, defined below */
1032 unsigned int ptrace;
1033
1034 int lock_depth; /* BKL lock depth */
1035
1036#ifdef CONFIG_SMP
1037#ifdef __ARCH_WANT_UNLOCKED_CTXSW
1038 int oncpu;
1039#endif
1040#endif
1041
1042 int prio, static_prio, normal_prio;
1043 unsigned int rt_priority;
1044 const struct sched_class *sched_class;
1045 struct sched_entity se;
1046 struct sched_rt_entity rt;
1047
1048#ifdef CONFIG_PREEMPT_NOTIFIERS
1049 /* list of struct preempt_notifier: */
1050 struct hlist_head preempt_notifiers;
1051#endif
1052
1053 /*
1054 * fpu_counter contains the number of consecutive context switches
1055 * that the FPU is used. If this is over a threshold, the lazy fpu
1056 * saving becomes unlazy to save the trap. This is an unsigned char
1057 * so that after 256 times the counter wraps and the behavior turns
1058 * lazy again; this to deal with bursty apps that only use FPU for
1059 * a short time
1060 */
1061 unsigned char fpu_counter;
1062 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1063#ifdef CONFIG_BLK_DEV_IO_TRACE
1064 unsigned int btrace_seq;
1065#endif
1066
1067 unsigned int policy;
1068 cpumask_t cpus_allowed;
1069
1070#ifdef CONFIG_PREEMPT_RCU
1071 int rcu_read_lock_nesting;
1072 int rcu_flipctr_idx;
1073#endif /* #ifdef CONFIG_PREEMPT_RCU */
1074
1075#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1076 struct sched_info sched_info;
1077#endif
1078
1079 struct list_head tasks;
1080
1081 struct mm_struct *mm, *active_mm;
1082
1083/* task state */
1084 struct linux_binfmt *binfmt;
1085 int exit_state;
1086 int exit_code, exit_signal;
1087 int pdeath_signal; /* The signal sent when the parent dies */
1088 /* ??? */
1089 unsigned int personality;
1090 unsigned did_exec:1;
1091 pid_t pid;
1092 pid_t tgid;
1093
1094#ifdef CONFIG_CC_STACKPROTECTOR
1095 /* Canary value for the -fstack-protector gcc feature */
1096 unsigned long stack_canary;
1097#endif
1098 /*
1099 * pointers to (original) parent process, youngest child, younger sibling,
1100 * older sibling, respectively. (p->father can be replaced with
1101 * p->real_parent->pid)
1102 */
1103 struct task_struct *real_parent; /* real parent process */
1104 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1105 /*
1106 * children/sibling forms the list of my natural children
1107 */
1108 struct list_head children; /* list of my children */
1109 struct list_head sibling; /* linkage in my parent's children list */
1110 struct task_struct *group_leader; /* threadgroup leader */
1111
1112 /*
1113 * ptraced is the list of tasks this task is using ptrace on.
1114 * This includes both natural children and PTRACE_ATTACH targets.
1115 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1116 */
1117 struct list_head ptraced;
1118 struct list_head ptrace_entry;
1119
1120 /* PID/PID hash table linkage. */
1121 struct pid_link pids[PIDTYPE_MAX];
1122 struct list_head thread_group;
1123
1124 struct completion *vfork_done; /* for vfork() */
1125 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1126 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1127
1128 cputime_t utime, stime, utimescaled, stimescaled;
1129 cputime_t gtime;
1130 cputime_t prev_utime, prev_stime;
1131 unsigned long nvcsw, nivcsw; /* context switch counts */
1132 struct timespec start_time; /* monotonic time */
1133 struct timespec real_start_time; /* boot based time */
1134/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1135 unsigned long min_flt, maj_flt;
1136
1137 cputime_t it_prof_expires, it_virt_expires;
1138 unsigned long long it_sched_expires;
1139 struct list_head cpu_timers[3];
1140
1141/* process credentials */
1142 uid_t uid,euid,suid,fsuid;
1143 gid_t gid,egid,sgid,fsgid;
1144 struct group_info *group_info;
1145 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset;
1146 struct user_struct *user;
1147 unsigned securebits;
1148#ifdef CONFIG_KEYS
1149 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1150 struct key *request_key_auth; /* assumed request_key authority */
1151 struct key *thread_keyring; /* keyring private to this thread */
1152#endif
1153 char comm[TASK_COMM_LEN]; /* executable name excluding path
1154 - access with [gs]et_task_comm (which lock
1155 it with task_lock())
1156 - initialized normally by flush_old_exec */
1157/* file system info */
1158 int link_count, total_link_count;
1159#ifdef CONFIG_SYSVIPC
1160/* ipc stuff */
1161 struct sysv_sem sysvsem;
1162#endif
1163#ifdef CONFIG_DETECT_SOFTLOCKUP
1164/* hung task detection */
1165 unsigned long last_switch_timestamp;
1166 unsigned long last_switch_count;
1167#endif
1168/* CPU-specific state of this task */
1169 struct thread_struct thread;
1170/* filesystem information */
1171 struct fs_struct *fs;
1172/* open file information */
1173 struct files_struct *files;
1174/* namespaces */
1175 struct nsproxy *nsproxy;
1176/* signal handlers */
1177 struct signal_struct *signal;
1178 struct sighand_struct *sighand;
1179
1180 sigset_t blocked, real_blocked;
1181 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1182 struct sigpending pending;
1183
1184 unsigned long sas_ss_sp;
1185 size_t sas_ss_size;
1186 int (*notifier)(void *priv);
1187 void *notifier_data;
1188 sigset_t *notifier_mask;
1189#ifdef CONFIG_SECURITY
1190 void *security;
1191#endif
1192 struct audit_context *audit_context;
1193#ifdef CONFIG_AUDITSYSCALL
1194 uid_t loginuid;
1195 unsigned int sessionid;
1196#endif
1197 seccomp_t seccomp;
1198
1199/* Thread group tracking */
1200 u32 parent_exec_id;
1201 u32 self_exec_id;
1202/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1203 spinlock_t alloc_lock;
1204
1205 /* Protection of the PI data structures: */
1206 spinlock_t pi_lock;
1207
1208#ifdef CONFIG_RT_MUTEXES
1209 /* PI waiters blocked on a rt_mutex held by this task */
1210 struct plist_head pi_waiters;
1211 /* Deadlock detection and priority inheritance handling */
1212 struct rt_mutex_waiter *pi_blocked_on;
1213#endif
1214
1215#ifdef CONFIG_DEBUG_MUTEXES
1216 /* mutex deadlock detection */
1217 struct mutex_waiter *blocked_on;
1218#endif
1219#ifdef CONFIG_TRACE_IRQFLAGS
1220 unsigned int irq_events;
1221 int hardirqs_enabled;
1222 unsigned long hardirq_enable_ip;
1223 unsigned int hardirq_enable_event;
1224 unsigned long hardirq_disable_ip;
1225 unsigned int hardirq_disable_event;
1226 int softirqs_enabled;
1227 unsigned long softirq_disable_ip;
1228 unsigned int softirq_disable_event;
1229 unsigned long softirq_enable_ip;
1230 unsigned int softirq_enable_event;
1231 int hardirq_context;
1232 int softirq_context;
1233#endif
1234#ifdef CONFIG_LOCKDEP
1235# define MAX_LOCK_DEPTH 48UL
1236 u64 curr_chain_key;
1237 int lockdep_depth;
1238 unsigned int lockdep_recursion;
1239 struct held_lock held_locks[MAX_LOCK_DEPTH];
1240#endif
1241
1242/* journalling filesystem info */
1243 void *journal_info;
1244
1245/* stacked block device info */
1246 struct bio *bio_list, **bio_tail;
1247
1248/* VM state */
1249 struct reclaim_state *reclaim_state;
1250
1251 struct backing_dev_info *backing_dev_info;
1252
1253 struct io_context *io_context;
1254
1255 unsigned long ptrace_message;
1256 siginfo_t *last_siginfo; /* For ptrace use. */
1257 struct task_io_accounting ioac;
1258#if defined(CONFIG_TASK_XACCT)
1259 u64 acct_rss_mem1; /* accumulated rss usage */
1260 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1261 cputime_t acct_timexpd; /* stime + utime since last update */
1262#endif
1263#ifdef CONFIG_CPUSETS
1264 nodemask_t mems_allowed;
1265 int cpuset_mems_generation;
1266 int cpuset_mem_spread_rotor;
1267#endif
1268#ifdef CONFIG_CGROUPS
1269 /* Control Group info protected by css_set_lock */
1270 struct css_set *cgroups;
1271 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1272 struct list_head cg_list;
1273#endif
1274#ifdef CONFIG_FUTEX
1275 struct robust_list_head __user *robust_list;
1276#ifdef CONFIG_COMPAT
1277 struct compat_robust_list_head __user *compat_robust_list;
1278#endif
1279 struct list_head pi_state_list;
1280 struct futex_pi_state *pi_state_cache;
1281#endif
1282#ifdef CONFIG_NUMA
1283 struct mempolicy *mempolicy;
1284 short il_next;
1285#endif
1286 atomic_t fs_excl; /* holding fs exclusive resources */
1287 struct rcu_head rcu;
1288
1289 /*
1290 * cache last used pipe for splice
1291 */
1292 struct pipe_inode_info *splice_pipe;
1293#ifdef CONFIG_TASK_DELAY_ACCT
1294 struct task_delay_info *delays;
1295#endif
1296#ifdef CONFIG_FAULT_INJECTION
1297 int make_it_fail;
1298#endif
1299 struct prop_local_single dirties;
1300#ifdef CONFIG_LATENCYTOP
1301 int latency_record_count;
1302 struct latency_record latency_record[LT_SAVECOUNT];
1303#endif
1304};
1305
1306/*
1307 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1308 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1309 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1310 * values are inverted: lower p->prio value means higher priority.
1311 *
1312 * The MAX_USER_RT_PRIO value allows the actual maximum
1313 * RT priority to be separate from the value exported to
1314 * user-space. This allows kernel threads to set their
1315 * priority to a value higher than any user task. Note:
1316 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1317 */
1318
1319#define MAX_USER_RT_PRIO 100
1320#define MAX_RT_PRIO MAX_USER_RT_PRIO
1321
1322#define MAX_PRIO (MAX_RT_PRIO + 40)
1323#define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1324
1325static inline int rt_prio(int prio)
1326{
1327 if (unlikely(prio < MAX_RT_PRIO))
1328 return 1;
1329 return 0;
1330}
1331
1332static inline int rt_task(struct task_struct *p)
1333{
1334 return rt_prio(p->prio);
1335}
1336
1337static inline void set_task_session(struct task_struct *tsk, pid_t session)
1338{
1339 tsk->signal->__session = session;
1340}
1341
1342static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1343{
1344 tsk->signal->__pgrp = pgrp;
1345}
1346
1347static inline struct pid *task_pid(struct task_struct *task)
1348{
1349 return task->pids[PIDTYPE_PID].pid;
1350}
1351
1352static inline struct pid *task_tgid(struct task_struct *task)
1353{
1354 return task->group_leader->pids[PIDTYPE_PID].pid;
1355}
1356
1357static inline struct pid *task_pgrp(struct task_struct *task)
1358{
1359 return task->group_leader->pids[PIDTYPE_PGID].pid;
1360}
1361
1362static inline struct pid *task_session(struct task_struct *task)
1363{
1364 return task->group_leader->pids[PIDTYPE_SID].pid;
1365}
1366
1367struct pid_namespace;
1368
1369/*
1370 * the helpers to get the task's different pids as they are seen
1371 * from various namespaces
1372 *
1373 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1374 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1375 * current.
1376 * task_xid_nr_ns() : id seen from the ns specified;
1377 *
1378 * set_task_vxid() : assigns a virtual id to a task;
1379 *
1380 * see also pid_nr() etc in include/linux/pid.h
1381 */
1382
1383static inline pid_t task_pid_nr(struct task_struct *tsk)
1384{
1385 return tsk->pid;
1386}
1387
1388pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1389
1390static inline pid_t task_pid_vnr(struct task_struct *tsk)
1391{
1392 return pid_vnr(task_pid(tsk));
1393}
1394
1395
1396static inline pid_t task_tgid_nr(struct task_struct *tsk)
1397{
1398 return tsk->tgid;
1399}
1400
1401pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1402
1403static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1404{
1405 return pid_vnr(task_tgid(tsk));
1406}
1407
1408
1409static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1410{
1411 return tsk->signal->__pgrp;
1412}
1413
1414pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1415
1416static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1417{
1418 return pid_vnr(task_pgrp(tsk));
1419}
1420
1421
1422static inline pid_t task_session_nr(struct task_struct *tsk)
1423{
1424 return tsk->signal->__session;
1425}
1426
1427pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1428
1429static inline pid_t task_session_vnr(struct task_struct *tsk)
1430{
1431 return pid_vnr(task_session(tsk));
1432}
1433
1434
1435/**
1436 * pid_alive - check that a task structure is not stale
1437 * @p: Task structure to be checked.
1438 *
1439 * Test if a process is not yet dead (at most zombie state)
1440 * If pid_alive fails, then pointers within the task structure
1441 * can be stale and must not be dereferenced.
1442 */
1443static inline int pid_alive(struct task_struct *p)
1444{
1445 return p->pids[PIDTYPE_PID].pid != NULL;
1446}
1447
1448/**
1449 * is_global_init - check if a task structure is init
1450 * @tsk: Task structure to be checked.
1451 *
1452 * Check if a task structure is the first user space task the kernel created.
1453 */
1454static inline int is_global_init(struct task_struct *tsk)
1455{
1456 return tsk->pid == 1;
1457}
1458
1459/*
1460 * is_container_init:
1461 * check whether in the task is init in its own pid namespace.
1462 */
1463extern int is_container_init(struct task_struct *tsk);
1464
1465extern struct pid *cad_pid;
1466
1467extern void free_task(struct task_struct *tsk);
1468#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1469
1470extern void __put_task_struct(struct task_struct *t);
1471
1472static inline void put_task_struct(struct task_struct *t)
1473{
1474 if (atomic_dec_and_test(&t->usage))
1475 __put_task_struct(t);
1476}
1477
1478extern cputime_t task_utime(struct task_struct *p);
1479extern cputime_t task_stime(struct task_struct *p);
1480extern cputime_t task_gtime(struct task_struct *p);
1481
1482/*
1483 * Per process flags
1484 */
1485#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1486 /* Not implemented yet, only for 486*/
1487#define PF_STARTING 0x00000002 /* being created */
1488#define PF_EXITING 0x00000004 /* getting shut down */
1489#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1490#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1491#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1492#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1493#define PF_DUMPCORE 0x00000200 /* dumped core */
1494#define PF_SIGNALED 0x00000400 /* killed by a signal */
1495#define PF_MEMALLOC 0x00000800 /* Allocating memory */
1496#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1497#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1498#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1499#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1500#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1501#define PF_KSWAPD 0x00040000 /* I am kswapd */
1502#define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1503#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1504#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1505#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1506#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1507#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1508#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1509#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1510#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1511#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1512#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1513#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1514
1515/*
1516 * Only the _current_ task can read/write to tsk->flags, but other
1517 * tasks can access tsk->flags in readonly mode for example
1518 * with tsk_used_math (like during threaded core dumping).
1519 * There is however an exception to this rule during ptrace
1520 * or during fork: the ptracer task is allowed to write to the
1521 * child->flags of its traced child (same goes for fork, the parent
1522 * can write to the child->flags), because we're guaranteed the
1523 * child is not running and in turn not changing child->flags
1524 * at the same time the parent does it.
1525 */
1526#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1527#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1528#define clear_used_math() clear_stopped_child_used_math(current)
1529#define set_used_math() set_stopped_child_used_math(current)
1530#define conditional_stopped_child_used_math(condition, child) \
1531 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1532#define conditional_used_math(condition) \
1533 conditional_stopped_child_used_math(condition, current)
1534#define copy_to_stopped_child_used_math(child) \
1535 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1536/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1537#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1538#define used_math() tsk_used_math(current)
1539
1540#ifdef CONFIG_SMP
1541extern int set_cpus_allowed_ptr(struct task_struct *p,
1542 const cpumask_t *new_mask);
1543#else
1544static inline int set_cpus_allowed_ptr(struct task_struct *p,
1545 const cpumask_t *new_mask)
1546{
1547 if (!cpu_isset(0, *new_mask))
1548 return -EINVAL;
1549 return 0;
1550}
1551#endif
1552static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1553{
1554 return set_cpus_allowed_ptr(p, &new_mask);
1555}
1556
1557extern unsigned long long sched_clock(void);
1558
1559extern void sched_clock_init(void);
1560extern u64 sched_clock_cpu(int cpu);
1561
1562#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1563static inline void sched_clock_tick(void)
1564{
1565}
1566
1567static inline void sched_clock_idle_sleep_event(void)
1568{
1569}
1570
1571static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1572{
1573}
1574#else
1575extern void sched_clock_tick(void);
1576extern void sched_clock_idle_sleep_event(void);
1577extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1578#endif
1579
1580/*
1581 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1582 * clock constructed from sched_clock():
1583 */
1584extern unsigned long long cpu_clock(int cpu);
1585
1586extern unsigned long long
1587task_sched_runtime(struct task_struct *task);
1588
1589/* sched_exec is called by processes performing an exec */
1590#ifdef CONFIG_SMP
1591extern void sched_exec(void);
1592#else
1593#define sched_exec() {}
1594#endif
1595
1596extern void sched_clock_idle_sleep_event(void);
1597extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1598
1599#ifdef CONFIG_HOTPLUG_CPU
1600extern void idle_task_exit(void);
1601#else
1602static inline void idle_task_exit(void) {}
1603#endif
1604
1605extern void sched_idle_next(void);
1606
1607#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1608extern void wake_up_idle_cpu(int cpu);
1609#else
1610static inline void wake_up_idle_cpu(int cpu) { }
1611#endif
1612
1613#ifdef CONFIG_SCHED_DEBUG
1614extern unsigned int sysctl_sched_latency;
1615extern unsigned int sysctl_sched_min_granularity;
1616extern unsigned int sysctl_sched_wakeup_granularity;
1617extern unsigned int sysctl_sched_child_runs_first;
1618extern unsigned int sysctl_sched_features;
1619extern unsigned int sysctl_sched_migration_cost;
1620extern unsigned int sysctl_sched_nr_migrate;
1621extern unsigned int sysctl_sched_shares_ratelimit;
1622
1623int sched_nr_latency_handler(struct ctl_table *table, int write,
1624 struct file *file, void __user *buffer, size_t *length,
1625 loff_t *ppos);
1626#endif
1627extern unsigned int sysctl_sched_rt_period;
1628extern int sysctl_sched_rt_runtime;
1629
1630int sched_rt_handler(struct ctl_table *table, int write,
1631 struct file *filp, void __user *buffer, size_t *lenp,
1632 loff_t *ppos);
1633
1634extern unsigned int sysctl_sched_compat_yield;
1635
1636#ifdef CONFIG_RT_MUTEXES
1637extern int rt_mutex_getprio(struct task_struct *p);
1638extern void rt_mutex_setprio(struct task_struct *p, int prio);
1639extern void rt_mutex_adjust_pi(struct task_struct *p);
1640#else
1641static inline int rt_mutex_getprio(struct task_struct *p)
1642{
1643 return p->normal_prio;
1644}
1645# define rt_mutex_adjust_pi(p) do { } while (0)
1646#endif
1647
1648extern void set_user_nice(struct task_struct *p, long nice);
1649extern int task_prio(const struct task_struct *p);
1650extern int task_nice(const struct task_struct *p);
1651extern int can_nice(const struct task_struct *p, const int nice);
1652extern int task_curr(const struct task_struct *p);
1653extern int idle_cpu(int cpu);
1654extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1655extern int sched_setscheduler_nocheck(struct task_struct *, int,
1656 struct sched_param *);
1657extern struct task_struct *idle_task(int cpu);
1658extern struct task_struct *curr_task(int cpu);
1659extern void set_curr_task(int cpu, struct task_struct *p);
1660
1661void yield(void);
1662
1663/*
1664 * The default (Linux) execution domain.
1665 */
1666extern struct exec_domain default_exec_domain;
1667
1668union thread_union {
1669 struct thread_info thread_info;
1670 unsigned long stack[THREAD_SIZE/sizeof(long)];
1671};
1672
1673#ifndef __HAVE_ARCH_KSTACK_END
1674static inline int kstack_end(void *addr)
1675{
1676 /* Reliable end of stack detection:
1677 * Some APM bios versions misalign the stack
1678 */
1679 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1680}
1681#endif
1682
1683extern union thread_union init_thread_union;
1684extern struct task_struct init_task;
1685
1686extern struct mm_struct init_mm;
1687
1688extern struct pid_namespace init_pid_ns;
1689
1690/*
1691 * find a task by one of its numerical ids
1692 *
1693 * find_task_by_pid_type_ns():
1694 * it is the most generic call - it finds a task by all id,
1695 * type and namespace specified
1696 * find_task_by_pid_ns():
1697 * finds a task by its pid in the specified namespace
1698 * find_task_by_vpid():
1699 * finds a task by its virtual pid
1700 *
1701 * see also find_vpid() etc in include/linux/pid.h
1702 */
1703
1704extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1705 struct pid_namespace *ns);
1706
1707extern struct task_struct *find_task_by_vpid(pid_t nr);
1708extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1709 struct pid_namespace *ns);
1710
1711extern void __set_special_pids(struct pid *pid);
1712
1713/* per-UID process charging. */
1714extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1715static inline struct user_struct *get_uid(struct user_struct *u)
1716{
1717 atomic_inc(&u->__count);
1718 return u;
1719}
1720extern void free_uid(struct user_struct *);
1721extern void switch_uid(struct user_struct *);
1722extern void release_uids(struct user_namespace *ns);
1723
1724#include <asm/current.h>
1725
1726extern void do_timer(unsigned long ticks);
1727
1728extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1729extern int wake_up_process(struct task_struct *tsk);
1730extern void wake_up_new_task(struct task_struct *tsk,
1731 unsigned long clone_flags);
1732#ifdef CONFIG_SMP
1733 extern void kick_process(struct task_struct *tsk);
1734#else
1735 static inline void kick_process(struct task_struct *tsk) { }
1736#endif
1737extern void sched_fork(struct task_struct *p, int clone_flags);
1738extern void sched_dead(struct task_struct *p);
1739
1740extern int in_group_p(gid_t);
1741extern int in_egroup_p(gid_t);
1742
1743extern void proc_caches_init(void);
1744extern void flush_signals(struct task_struct *);
1745extern void ignore_signals(struct task_struct *);
1746extern void flush_signal_handlers(struct task_struct *, int force_default);
1747extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1748
1749static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1750{
1751 unsigned long flags;
1752 int ret;
1753
1754 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1755 ret = dequeue_signal(tsk, mask, info);
1756 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1757
1758 return ret;
1759}
1760
1761extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1762 sigset_t *mask);
1763extern void unblock_all_signals(void);
1764extern void release_task(struct task_struct * p);
1765extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1766extern int force_sigsegv(int, struct task_struct *);
1767extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1768extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1769extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1770extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1771extern int kill_pgrp(struct pid *pid, int sig, int priv);
1772extern int kill_pid(struct pid *pid, int sig, int priv);
1773extern int kill_proc_info(int, struct siginfo *, pid_t);
1774extern int do_notify_parent(struct task_struct *, int);
1775extern void force_sig(int, struct task_struct *);
1776extern void force_sig_specific(int, struct task_struct *);
1777extern int send_sig(int, struct task_struct *, int);
1778extern void zap_other_threads(struct task_struct *p);
1779extern struct sigqueue *sigqueue_alloc(void);
1780extern void sigqueue_free(struct sigqueue *);
1781extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1782extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1783extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1784
1785static inline int kill_cad_pid(int sig, int priv)
1786{
1787 return kill_pid(cad_pid, sig, priv);
1788}
1789
1790/* These can be the second arg to send_sig_info/send_group_sig_info. */
1791#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1792#define SEND_SIG_PRIV ((struct siginfo *) 1)
1793#define SEND_SIG_FORCED ((struct siginfo *) 2)
1794
1795static inline int is_si_special(const struct siginfo *info)
1796{
1797 return info <= SEND_SIG_FORCED;
1798}
1799
1800/* True if we are on the alternate signal stack. */
1801
1802static inline int on_sig_stack(unsigned long sp)
1803{
1804 return (sp - current->sas_ss_sp < current->sas_ss_size);
1805}
1806
1807static inline int sas_ss_flags(unsigned long sp)
1808{
1809 return (current->sas_ss_size == 0 ? SS_DISABLE
1810 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1811}
1812
1813/*
1814 * Routines for handling mm_structs
1815 */
1816extern struct mm_struct * mm_alloc(void);
1817
1818/* mmdrop drops the mm and the page tables */
1819extern void __mmdrop(struct mm_struct *);
1820static inline void mmdrop(struct mm_struct * mm)
1821{
1822 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1823 __mmdrop(mm);
1824}
1825
1826/* mmput gets rid of the mappings and all user-space */
1827extern void mmput(struct mm_struct *);
1828/* Grab a reference to a task's mm, if it is not already going away */
1829extern struct mm_struct *get_task_mm(struct task_struct *task);
1830/* Remove the current tasks stale references to the old mm_struct */
1831extern void mm_release(struct task_struct *, struct mm_struct *);
1832/* Allocate a new mm structure and copy contents from tsk->mm */
1833extern struct mm_struct *dup_mm(struct task_struct *tsk);
1834
1835extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1836extern void flush_thread(void);
1837extern void exit_thread(void);
1838
1839extern void exit_files(struct task_struct *);
1840extern void __cleanup_signal(struct signal_struct *);
1841extern void __cleanup_sighand(struct sighand_struct *);
1842
1843extern void exit_itimers(struct signal_struct *);
1844extern void flush_itimer_signals(void);
1845
1846extern NORET_TYPE void do_group_exit(int);
1847
1848extern void daemonize(const char *, ...);
1849extern int allow_signal(int);
1850extern int disallow_signal(int);
1851
1852extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1853extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1854struct task_struct *fork_idle(int);
1855
1856extern void set_task_comm(struct task_struct *tsk, char *from);
1857extern char *get_task_comm(char *to, struct task_struct *tsk);
1858
1859#ifdef CONFIG_SMP
1860extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1861#else
1862static inline unsigned long wait_task_inactive(struct task_struct *p,
1863 long match_state)
1864{
1865 return 1;
1866}
1867#endif
1868
1869#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1870
1871#define for_each_process(p) \
1872 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1873
1874/*
1875 * Careful: do_each_thread/while_each_thread is a double loop so
1876 * 'break' will not work as expected - use goto instead.
1877 */
1878#define do_each_thread(g, t) \
1879 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1880
1881#define while_each_thread(g, t) \
1882 while ((t = next_thread(t)) != g)
1883
1884/* de_thread depends on thread_group_leader not being a pid based check */
1885#define thread_group_leader(p) (p == p->group_leader)
1886
1887/* Do to the insanities of de_thread it is possible for a process
1888 * to have the pid of the thread group leader without actually being
1889 * the thread group leader. For iteration through the pids in proc
1890 * all we care about is that we have a task with the appropriate
1891 * pid, we don't actually care if we have the right task.
1892 */
1893static inline int has_group_leader_pid(struct task_struct *p)
1894{
1895 return p->pid == p->tgid;
1896}
1897
1898static inline
1899int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1900{
1901 return p1->tgid == p2->tgid;
1902}
1903
1904static inline struct task_struct *next_thread(const struct task_struct *p)
1905{
1906 return list_entry(rcu_dereference(p->thread_group.next),
1907 struct task_struct, thread_group);
1908}
1909
1910static inline int thread_group_empty(struct task_struct *p)
1911{
1912 return list_empty(&p->thread_group);
1913}
1914
1915#define delay_group_leader(p) \
1916 (thread_group_leader(p) && !thread_group_empty(p))
1917
1918/*
1919 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1920 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1921 * pins the final release of task.io_context. Also protects ->cpuset and
1922 * ->cgroup.subsys[].
1923 *
1924 * Nests both inside and outside of read_lock(&tasklist_lock).
1925 * It must not be nested with write_lock_irq(&tasklist_lock),
1926 * neither inside nor outside.
1927 */
1928static inline void task_lock(struct task_struct *p)
1929{
1930 spin_lock(&p->alloc_lock);
1931}
1932
1933static inline void task_unlock(struct task_struct *p)
1934{
1935 spin_unlock(&p->alloc_lock);
1936}
1937
1938extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1939 unsigned long *flags);
1940
1941static inline void unlock_task_sighand(struct task_struct *tsk,
1942 unsigned long *flags)
1943{
1944 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1945}
1946
1947#ifndef __HAVE_THREAD_FUNCTIONS
1948
1949#define task_thread_info(task) ((struct thread_info *)(task)->stack)
1950#define task_stack_page(task) ((task)->stack)
1951
1952static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1953{
1954 *task_thread_info(p) = *task_thread_info(org);
1955 task_thread_info(p)->task = p;
1956}
1957
1958static inline unsigned long *end_of_stack(struct task_struct *p)
1959{
1960 return (unsigned long *)(task_thread_info(p) + 1);
1961}
1962
1963#endif
1964
1965static inline int object_is_on_stack(void *obj)
1966{
1967 void *stack = task_stack_page(current);
1968
1969 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
1970}
1971
1972extern void thread_info_cache_init(void);
1973
1974/* set thread flags in other task's structures
1975 * - see asm/thread_info.h for TIF_xxxx flags available
1976 */
1977static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1978{
1979 set_ti_thread_flag(task_thread_info(tsk), flag);
1980}
1981
1982static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1983{
1984 clear_ti_thread_flag(task_thread_info(tsk), flag);
1985}
1986
1987static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1988{
1989 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1990}
1991
1992static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1993{
1994 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1995}
1996
1997static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1998{
1999 return test_ti_thread_flag(task_thread_info(tsk), flag);
2000}
2001
2002static inline void set_tsk_need_resched(struct task_struct *tsk)
2003{
2004 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2005}
2006
2007static inline void clear_tsk_need_resched(struct task_struct *tsk)
2008{
2009 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2010}
2011
2012static inline int test_tsk_need_resched(struct task_struct *tsk)
2013{
2014 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2015}
2016
2017static inline int signal_pending(struct task_struct *p)
2018{
2019 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2020}
2021
2022extern int __fatal_signal_pending(struct task_struct *p);
2023
2024static inline int fatal_signal_pending(struct task_struct *p)
2025{
2026 return signal_pending(p) && __fatal_signal_pending(p);
2027}
2028
2029static inline int signal_pending_state(long state, struct task_struct *p)
2030{
2031 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2032 return 0;
2033 if (!signal_pending(p))
2034 return 0;
2035
2036 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2037}
2038
2039static inline int need_resched(void)
2040{
2041 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2042}
2043
2044/*
2045 * cond_resched() and cond_resched_lock(): latency reduction via
2046 * explicit rescheduling in places that are safe. The return
2047 * value indicates whether a reschedule was done in fact.
2048 * cond_resched_lock() will drop the spinlock before scheduling,
2049 * cond_resched_softirq() will enable bhs before scheduling.
2050 */
2051extern int _cond_resched(void);
2052#ifdef CONFIG_PREEMPT_BKL
2053static inline int cond_resched(void)
2054{
2055 return 0;
2056}
2057#else
2058static inline int cond_resched(void)
2059{
2060 return _cond_resched();
2061}
2062#endif
2063extern int cond_resched_lock(spinlock_t * lock);
2064extern int cond_resched_softirq(void);
2065static inline int cond_resched_bkl(void)
2066{
2067 return _cond_resched();
2068}
2069
2070/*
2071 * Does a critical section need to be broken due to another
2072 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2073 * but a general need for low latency)
2074 */
2075static inline int spin_needbreak(spinlock_t *lock)
2076{
2077#ifdef CONFIG_PREEMPT
2078 return spin_is_contended(lock);
2079#else
2080 return 0;
2081#endif
2082}
2083
2084/*
2085 * Reevaluate whether the task has signals pending delivery.
2086 * Wake the task if so.
2087 * This is required every time the blocked sigset_t changes.
2088 * callers must hold sighand->siglock.
2089 */
2090extern void recalc_sigpending_and_wake(struct task_struct *t);
2091extern void recalc_sigpending(void);
2092
2093extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2094
2095/*
2096 * Wrappers for p->thread_info->cpu access. No-op on UP.
2097 */
2098#ifdef CONFIG_SMP
2099
2100static inline unsigned int task_cpu(const struct task_struct *p)
2101{
2102 return task_thread_info(p)->cpu;
2103}
2104
2105extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2106
2107#else
2108
2109static inline unsigned int task_cpu(const struct task_struct *p)
2110{
2111 return 0;
2112}
2113
2114static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2115{
2116}
2117
2118#endif /* CONFIG_SMP */
2119
2120extern void arch_pick_mmap_layout(struct mm_struct *mm);
2121
2122#ifdef CONFIG_TRACING
2123extern void
2124__trace_special(void *__tr, void *__data,
2125 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2126#else
2127static inline void
2128__trace_special(void *__tr, void *__data,
2129 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2130{
2131}
2132#endif
2133
2134extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask);
2135extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2136
2137extern int sched_mc_power_savings, sched_smt_power_savings;
2138
2139extern void normalize_rt_tasks(void);
2140
2141#ifdef CONFIG_GROUP_SCHED
2142
2143extern struct task_group init_task_group;
2144#ifdef CONFIG_USER_SCHED
2145extern struct task_group root_task_group;
2146#endif
2147
2148extern struct task_group *sched_create_group(struct task_group *parent);
2149extern void sched_destroy_group(struct task_group *tg);
2150extern void sched_move_task(struct task_struct *tsk);
2151#ifdef CONFIG_FAIR_GROUP_SCHED
2152extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2153extern unsigned long sched_group_shares(struct task_group *tg);
2154#endif
2155#ifdef CONFIG_RT_GROUP_SCHED
2156extern int sched_group_set_rt_runtime(struct task_group *tg,
2157 long rt_runtime_us);
2158extern long sched_group_rt_runtime(struct task_group *tg);
2159extern int sched_group_set_rt_period(struct task_group *tg,
2160 long rt_period_us);
2161extern long sched_group_rt_period(struct task_group *tg);
2162#endif
2163#endif
2164
2165#ifdef CONFIG_TASK_XACCT
2166static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2167{
2168 tsk->ioac.rchar += amt;
2169}
2170
2171static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2172{
2173 tsk->ioac.wchar += amt;
2174}
2175
2176static inline void inc_syscr(struct task_struct *tsk)
2177{
2178 tsk->ioac.syscr++;
2179}
2180
2181static inline void inc_syscw(struct task_struct *tsk)
2182{
2183 tsk->ioac.syscw++;
2184}
2185#else
2186static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2187{
2188}
2189
2190static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2191{
2192}
2193
2194static inline void inc_syscr(struct task_struct *tsk)
2195{
2196}
2197
2198static inline void inc_syscw(struct task_struct *tsk)
2199{
2200}
2201#endif
2202
2203#ifndef TASK_SIZE_OF
2204#define TASK_SIZE_OF(tsk) TASK_SIZE
2205#endif
2206
2207#ifdef CONFIG_MM_OWNER
2208extern void mm_update_next_owner(struct mm_struct *mm);
2209extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2210#else
2211static inline void mm_update_next_owner(struct mm_struct *mm)
2212{
2213}
2214
2215static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2216{
2217}
2218#endif /* CONFIG_MM_OWNER */
2219
2220#define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2221
2222#endif /* __KERNEL__ */
2223
2224#endif