]> bbs.cooldavid.org Git - net-next-2.6.git/blame_incremental - include/linux/sched.h
softlockup: allow panic on lockup
[net-next-2.6.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
91#include <asm/processor.h>
92
93struct mem_cgroup;
94struct exec_domain;
95struct futex_pi_state;
96struct robust_list_head;
97struct bio;
98
99/*
100 * List of flags we want to share for kernel threads,
101 * if only because they are not used by them anyway.
102 */
103#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
104
105/*
106 * These are the constant used to fake the fixed-point load-average
107 * counting. Some notes:
108 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
109 * a load-average precision of 10 bits integer + 11 bits fractional
110 * - if you want to count load-averages more often, you need more
111 * precision, or rounding will get you. With 2-second counting freq,
112 * the EXP_n values would be 1981, 2034 and 2043 if still using only
113 * 11 bit fractions.
114 */
115extern unsigned long avenrun[]; /* Load averages */
116
117#define FSHIFT 11 /* nr of bits of precision */
118#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
119#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
120#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
121#define EXP_5 2014 /* 1/exp(5sec/5min) */
122#define EXP_15 2037 /* 1/exp(5sec/15min) */
123
124#define CALC_LOAD(load,exp,n) \
125 load *= exp; \
126 load += n*(FIXED_1-exp); \
127 load >>= FSHIFT;
128
129extern unsigned long total_forks;
130extern int nr_threads;
131DECLARE_PER_CPU(unsigned long, process_counts);
132extern int nr_processes(void);
133extern unsigned long nr_running(void);
134extern unsigned long nr_uninterruptible(void);
135extern unsigned long nr_active(void);
136extern unsigned long nr_iowait(void);
137extern unsigned long weighted_cpuload(const int cpu);
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 cpumask_t nohz_cpu_mask;
250#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
251extern int select_nohz_load_balancer(int cpu);
252#else
253static inline int select_nohz_load_balancer(int cpu)
254{
255 return 0;
256}
257#endif
258
259extern unsigned long rt_needs_cpu(int cpu);
260
261/*
262 * Only dump TASK_* tasks. (0 for all tasks)
263 */
264extern void show_state_filter(unsigned long state_filter);
265
266static inline void show_state(void)
267{
268 show_state_filter(0);
269}
270
271extern void show_regs(struct pt_regs *);
272
273/*
274 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
275 * task), SP is the stack pointer of the first frame that should be shown in the back
276 * trace (or NULL if the entire call-chain of the task should be shown).
277 */
278extern void show_stack(struct task_struct *task, unsigned long *sp);
279
280void io_schedule(void);
281long io_schedule_timeout(long timeout);
282
283extern void cpu_init (void);
284extern void trap_init(void);
285extern void account_process_tick(struct task_struct *task, int user);
286extern void update_process_times(int user);
287extern void scheduler_tick(void);
288extern void hrtick_resched(void);
289
290extern void sched_show_task(struct task_struct *p);
291
292#ifdef CONFIG_DETECT_SOFTLOCKUP
293extern void softlockup_tick(void);
294extern void spawn_softlockup_task(void);
295extern void touch_softlockup_watchdog(void);
296extern void touch_all_softlockup_watchdogs(void);
297extern unsigned int softlockup_panic;
298extern unsigned long softlockup_thresh;
299extern unsigned long sysctl_hung_task_check_count;
300extern unsigned long sysctl_hung_task_timeout_secs;
301extern unsigned long sysctl_hung_task_warnings;
302#else
303static inline void softlockup_tick(void)
304{
305}
306static inline void spawn_softlockup_task(void)
307{
308}
309static inline void touch_softlockup_watchdog(void)
310{
311}
312static inline void touch_all_softlockup_watchdogs(void)
313{
314}
315#endif
316
317
318/* Attach to any functions which should be ignored in wchan output. */
319#define __sched __attribute__((__section__(".sched.text")))
320
321/* Linker adds these: start and end of __sched functions */
322extern char __sched_text_start[], __sched_text_end[];
323
324/* Is this address in the __sched functions? */
325extern int in_sched_functions(unsigned long addr);
326
327#define MAX_SCHEDULE_TIMEOUT LONG_MAX
328extern signed long schedule_timeout(signed long timeout);
329extern signed long schedule_timeout_interruptible(signed long timeout);
330extern signed long schedule_timeout_killable(signed long timeout);
331extern signed long schedule_timeout_uninterruptible(signed long timeout);
332asmlinkage void schedule(void);
333
334struct nsproxy;
335struct user_namespace;
336
337/* Maximum number of active map areas.. This is a random (large) number */
338#define DEFAULT_MAX_MAP_COUNT 65536
339
340extern int sysctl_max_map_count;
341
342#include <linux/aio.h>
343
344extern unsigned long
345arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
346 unsigned long, unsigned long);
347extern unsigned long
348arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
349 unsigned long len, unsigned long pgoff,
350 unsigned long flags);
351extern void arch_unmap_area(struct mm_struct *, unsigned long);
352extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
353
354#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
355/*
356 * The mm counters are not protected by its page_table_lock,
357 * so must be incremented atomically.
358 */
359#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
360#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
361#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
362#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
363#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
364
365#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
366/*
367 * The mm counters are protected by its page_table_lock,
368 * so can be incremented directly.
369 */
370#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
371#define get_mm_counter(mm, member) ((mm)->_##member)
372#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
373#define inc_mm_counter(mm, member) (mm)->_##member++
374#define dec_mm_counter(mm, member) (mm)->_##member--
375
376#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
377
378#define get_mm_rss(mm) \
379 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
380#define update_hiwater_rss(mm) do { \
381 unsigned long _rss = get_mm_rss(mm); \
382 if ((mm)->hiwater_rss < _rss) \
383 (mm)->hiwater_rss = _rss; \
384} while (0)
385#define update_hiwater_vm(mm) do { \
386 if ((mm)->hiwater_vm < (mm)->total_vm) \
387 (mm)->hiwater_vm = (mm)->total_vm; \
388} while (0)
389
390extern void set_dumpable(struct mm_struct *mm, int value);
391extern int get_dumpable(struct mm_struct *mm);
392
393/* mm flags */
394/* dumpable bits */
395#define MMF_DUMPABLE 0 /* core dump is permitted */
396#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
397#define MMF_DUMPABLE_BITS 2
398
399/* coredump filter bits */
400#define MMF_DUMP_ANON_PRIVATE 2
401#define MMF_DUMP_ANON_SHARED 3
402#define MMF_DUMP_MAPPED_PRIVATE 4
403#define MMF_DUMP_MAPPED_SHARED 5
404#define MMF_DUMP_ELF_HEADERS 6
405#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
406#define MMF_DUMP_FILTER_BITS 5
407#define MMF_DUMP_FILTER_MASK \
408 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
409#define MMF_DUMP_FILTER_DEFAULT \
410 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
411
412struct sighand_struct {
413 atomic_t count;
414 struct k_sigaction action[_NSIG];
415 spinlock_t siglock;
416 wait_queue_head_t signalfd_wqh;
417};
418
419struct pacct_struct {
420 int ac_flag;
421 long ac_exitcode;
422 unsigned long ac_mem;
423 cputime_t ac_utime, ac_stime;
424 unsigned long ac_minflt, ac_majflt;
425};
426
427/*
428 * NOTE! "signal_struct" does not have it's own
429 * locking, because a shared signal_struct always
430 * implies a shared sighand_struct, so locking
431 * sighand_struct is always a proper superset of
432 * the locking of signal_struct.
433 */
434struct signal_struct {
435 atomic_t count;
436 atomic_t live;
437
438 wait_queue_head_t wait_chldexit; /* for wait4() */
439
440 /* current thread group signal load-balancing target: */
441 struct task_struct *curr_target;
442
443 /* shared signal handling: */
444 struct sigpending shared_pending;
445
446 /* thread group exit support */
447 int group_exit_code;
448 /* overloaded:
449 * - notify group_exit_task when ->count is equal to notify_count
450 * - everyone except group_exit_task is stopped during signal delivery
451 * of fatal signals, group_exit_task processes the signal.
452 */
453 struct task_struct *group_exit_task;
454 int notify_count;
455
456 /* thread group stop support, overloads group_exit_code too */
457 int group_stop_count;
458 unsigned int flags; /* see SIGNAL_* flags below */
459
460 /* POSIX.1b Interval Timers */
461 struct list_head posix_timers;
462
463 /* ITIMER_REAL timer for the process */
464 struct hrtimer real_timer;
465 struct pid *leader_pid;
466 ktime_t it_real_incr;
467
468 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
469 cputime_t it_prof_expires, it_virt_expires;
470 cputime_t it_prof_incr, it_virt_incr;
471
472 /* job control IDs */
473
474 /*
475 * pgrp and session fields are deprecated.
476 * use the task_session_Xnr and task_pgrp_Xnr routines below
477 */
478
479 union {
480 pid_t pgrp __deprecated;
481 pid_t __pgrp;
482 };
483
484 struct pid *tty_old_pgrp;
485
486 union {
487 pid_t session __deprecated;
488 pid_t __session;
489 };
490
491 /* boolean value for session group leader */
492 int leader;
493
494 struct tty_struct *tty; /* NULL if no tty */
495
496 /*
497 * Cumulative resource counters for dead threads in the group,
498 * and for reaped dead child processes forked by this group.
499 * Live threads maintain their own counters and add to these
500 * in __exit_signal, except for the group leader.
501 */
502 cputime_t utime, stime, cutime, cstime;
503 cputime_t gtime;
504 cputime_t cgtime;
505 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
506 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
507 unsigned long inblock, oublock, cinblock, coublock;
508
509 /*
510 * Cumulative ns of scheduled CPU time for dead threads in the
511 * group, not including a zombie group leader. (This only differs
512 * from jiffies_to_ns(utime + stime) if sched_clock uses something
513 * other than jiffies.)
514 */
515 unsigned long long sum_sched_runtime;
516
517 /*
518 * We don't bother to synchronize most readers of this at all,
519 * because there is no reader checking a limit that actually needs
520 * to get both rlim_cur and rlim_max atomically, and either one
521 * alone is a single word that can safely be read normally.
522 * getrlimit/setrlimit use task_lock(current->group_leader) to
523 * protect this instead of the siglock, because they really
524 * have no need to disable irqs.
525 */
526 struct rlimit rlim[RLIM_NLIMITS];
527
528 struct list_head cpu_timers[3];
529
530 /* keep the process-shared keyrings here so that they do the right
531 * thing in threads created with CLONE_THREAD */
532#ifdef CONFIG_KEYS
533 struct key *session_keyring; /* keyring inherited over fork */
534 struct key *process_keyring; /* keyring private to this process */
535#endif
536#ifdef CONFIG_BSD_PROCESS_ACCT
537 struct pacct_struct pacct; /* per-process accounting information */
538#endif
539#ifdef CONFIG_TASKSTATS
540 struct taskstats *stats;
541#endif
542#ifdef CONFIG_AUDIT
543 unsigned audit_tty;
544 struct tty_audit_buf *tty_audit_buf;
545#endif
546};
547
548/* Context switch must be unlocked if interrupts are to be enabled */
549#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
550# define __ARCH_WANT_UNLOCKED_CTXSW
551#endif
552
553/*
554 * Bits in flags field of signal_struct.
555 */
556#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
557#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
558#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
559#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
560/*
561 * Pending notifications to parent.
562 */
563#define SIGNAL_CLD_STOPPED 0x00000010
564#define SIGNAL_CLD_CONTINUED 0x00000020
565#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
566
567#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
568
569/* If true, all threads except ->group_exit_task have pending SIGKILL */
570static inline int signal_group_exit(const struct signal_struct *sig)
571{
572 return (sig->flags & SIGNAL_GROUP_EXIT) ||
573 (sig->group_exit_task != NULL);
574}
575
576/*
577 * Some day this will be a full-fledged user tracking system..
578 */
579struct user_struct {
580 atomic_t __count; /* reference count */
581 atomic_t processes; /* How many processes does this user have? */
582 atomic_t files; /* How many open files does this user have? */
583 atomic_t sigpending; /* How many pending signals does this user have? */
584#ifdef CONFIG_INOTIFY_USER
585 atomic_t inotify_watches; /* How many inotify watches does this user have? */
586 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
587#endif
588#ifdef CONFIG_POSIX_MQUEUE
589 /* protected by mq_lock */
590 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
591#endif
592 unsigned long locked_shm; /* How many pages of mlocked shm ? */
593
594#ifdef CONFIG_KEYS
595 struct key *uid_keyring; /* UID specific keyring */
596 struct key *session_keyring; /* UID's default session keyring */
597#endif
598
599 /* Hash table maintenance information */
600 struct hlist_node uidhash_node;
601 uid_t uid;
602
603#ifdef CONFIG_USER_SCHED
604 struct task_group *tg;
605#ifdef CONFIG_SYSFS
606 struct kobject kobj;
607 struct work_struct work;
608#endif
609#endif
610};
611
612extern int uids_sysfs_init(void);
613
614extern struct user_struct *find_user(uid_t);
615
616extern struct user_struct root_user;
617#define INIT_USER (&root_user)
618
619struct backing_dev_info;
620struct reclaim_state;
621
622#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
623struct sched_info {
624 /* cumulative counters */
625 unsigned long pcount; /* # of times run on this cpu */
626 unsigned long long cpu_time, /* time spent on the cpu */
627 run_delay; /* time spent waiting on a runqueue */
628
629 /* timestamps */
630 unsigned long long last_arrival,/* when we last ran on a cpu */
631 last_queued; /* when we were last queued to run */
632#ifdef CONFIG_SCHEDSTATS
633 /* BKL stats */
634 unsigned int bkl_count;
635#endif
636};
637#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
638
639#ifdef CONFIG_SCHEDSTATS
640extern const struct file_operations proc_schedstat_operations;
641#endif /* CONFIG_SCHEDSTATS */
642
643#ifdef CONFIG_TASK_DELAY_ACCT
644struct task_delay_info {
645 spinlock_t lock;
646 unsigned int flags; /* Private per-task flags */
647
648 /* For each stat XXX, add following, aligned appropriately
649 *
650 * struct timespec XXX_start, XXX_end;
651 * u64 XXX_delay;
652 * u32 XXX_count;
653 *
654 * Atomicity of updates to XXX_delay, XXX_count protected by
655 * single lock above (split into XXX_lock if contention is an issue).
656 */
657
658 /*
659 * XXX_count is incremented on every XXX operation, the delay
660 * associated with the operation is added to XXX_delay.
661 * XXX_delay contains the accumulated delay time in nanoseconds.
662 */
663 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
664 u64 blkio_delay; /* wait for sync block io completion */
665 u64 swapin_delay; /* wait for swapin block io completion */
666 u32 blkio_count; /* total count of the number of sync block */
667 /* io operations performed */
668 u32 swapin_count; /* total count of the number of swapin block */
669 /* io operations performed */
670};
671#endif /* CONFIG_TASK_DELAY_ACCT */
672
673static inline int sched_info_on(void)
674{
675#ifdef CONFIG_SCHEDSTATS
676 return 1;
677#elif defined(CONFIG_TASK_DELAY_ACCT)
678 extern int delayacct_on;
679 return delayacct_on;
680#else
681 return 0;
682#endif
683}
684
685enum cpu_idle_type {
686 CPU_IDLE,
687 CPU_NOT_IDLE,
688 CPU_NEWLY_IDLE,
689 CPU_MAX_IDLE_TYPES
690};
691
692/*
693 * sched-domains (multiprocessor balancing) declarations:
694 */
695
696/*
697 * Increase resolution of nice-level calculations:
698 */
699#define SCHED_LOAD_SHIFT 10
700#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
701
702#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
703
704#ifdef CONFIG_SMP
705#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
706#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
707#define SD_BALANCE_EXEC 4 /* Balance on exec */
708#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
709#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
710#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
711#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
712#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
713#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
714#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
715#define SD_SERIALIZE 1024 /* Only a single load balancing instance */
716#define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
717
718#define BALANCE_FOR_MC_POWER \
719 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
720
721#define BALANCE_FOR_PKG_POWER \
722 ((sched_mc_power_savings || sched_smt_power_savings) ? \
723 SD_POWERSAVINGS_BALANCE : 0)
724
725#define test_sd_parent(sd, flag) ((sd->parent && \
726 (sd->parent->flags & flag)) ? 1 : 0)
727
728
729struct sched_group {
730 struct sched_group *next; /* Must be a circular list */
731 cpumask_t cpumask;
732
733 /*
734 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
735 * single CPU. This is read only (except for setup, hotplug CPU).
736 * Note : Never change cpu_power without recompute its reciprocal
737 */
738 unsigned int __cpu_power;
739 /*
740 * reciprocal value of cpu_power to avoid expensive divides
741 * (see include/linux/reciprocal_div.h)
742 */
743 u32 reciprocal_cpu_power;
744};
745
746enum sched_domain_level {
747 SD_LV_NONE = 0,
748 SD_LV_SIBLING,
749 SD_LV_MC,
750 SD_LV_CPU,
751 SD_LV_NODE,
752 SD_LV_ALLNODES,
753 SD_LV_MAX
754};
755
756struct sched_domain_attr {
757 int relax_domain_level;
758};
759
760#define SD_ATTR_INIT (struct sched_domain_attr) { \
761 .relax_domain_level = -1, \
762}
763
764struct sched_domain {
765 /* These fields must be setup */
766 struct sched_domain *parent; /* top domain must be null terminated */
767 struct sched_domain *child; /* bottom domain must be null terminated */
768 struct sched_group *groups; /* the balancing groups of the domain */
769 cpumask_t span; /* span of all CPUs in this domain */
770 int first_cpu; /* cache of the first cpu in this domain */
771 unsigned long min_interval; /* Minimum balance interval ms */
772 unsigned long max_interval; /* Maximum balance interval ms */
773 unsigned int busy_factor; /* less balancing by factor if busy */
774 unsigned int imbalance_pct; /* No balance until over watermark */
775 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
776 unsigned int busy_idx;
777 unsigned int idle_idx;
778 unsigned int newidle_idx;
779 unsigned int wake_idx;
780 unsigned int forkexec_idx;
781 int flags; /* See SD_* */
782 enum sched_domain_level level;
783
784 /* Runtime fields. */
785 unsigned long last_balance; /* init to jiffies. units in jiffies */
786 unsigned int balance_interval; /* initialise to 1. units in ms. */
787 unsigned int nr_balance_failed; /* initialise to 0 */
788
789#ifdef CONFIG_SCHEDSTATS
790 /* load_balance() stats */
791 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
792 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
793 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
794 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
795 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
796 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
797 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
798 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
799
800 /* Active load balancing */
801 unsigned int alb_count;
802 unsigned int alb_failed;
803 unsigned int alb_pushed;
804
805 /* SD_BALANCE_EXEC stats */
806 unsigned int sbe_count;
807 unsigned int sbe_balanced;
808 unsigned int sbe_pushed;
809
810 /* SD_BALANCE_FORK stats */
811 unsigned int sbf_count;
812 unsigned int sbf_balanced;
813 unsigned int sbf_pushed;
814
815 /* try_to_wake_up() stats */
816 unsigned int ttwu_wake_remote;
817 unsigned int ttwu_move_affine;
818 unsigned int ttwu_move_balance;
819#endif
820};
821
822extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new,
823 struct sched_domain_attr *dattr_new);
824extern int arch_reinit_sched_domains(void);
825
826#endif /* CONFIG_SMP */
827
828/*
829 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
830 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
831 * task of nice 0 or enough lower priority tasks to bring up the
832 * weighted_cpuload
833 */
834static inline int above_background_load(void)
835{
836 unsigned long cpu;
837
838 for_each_online_cpu(cpu) {
839 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
840 return 1;
841 }
842 return 0;
843}
844
845struct io_context; /* See blkdev.h */
846#define NGROUPS_SMALL 32
847#define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
848struct group_info {
849 int ngroups;
850 atomic_t usage;
851 gid_t small_block[NGROUPS_SMALL];
852 int nblocks;
853 gid_t *blocks[0];
854};
855
856/*
857 * get_group_info() must be called with the owning task locked (via task_lock())
858 * when task != current. The reason being that the vast majority of callers are
859 * looking at current->group_info, which can not be changed except by the
860 * current task. Changing current->group_info requires the task lock, too.
861 */
862#define get_group_info(group_info) do { \
863 atomic_inc(&(group_info)->usage); \
864} while (0)
865
866#define put_group_info(group_info) do { \
867 if (atomic_dec_and_test(&(group_info)->usage)) \
868 groups_free(group_info); \
869} while (0)
870
871extern struct group_info *groups_alloc(int gidsetsize);
872extern void groups_free(struct group_info *group_info);
873extern int set_current_groups(struct group_info *group_info);
874extern int groups_search(struct group_info *group_info, gid_t grp);
875/* access the groups "array" with this macro */
876#define GROUP_AT(gi, i) \
877 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
878
879#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
880extern void prefetch_stack(struct task_struct *t);
881#else
882static inline void prefetch_stack(struct task_struct *t) { }
883#endif
884
885struct audit_context; /* See audit.c */
886struct mempolicy;
887struct pipe_inode_info;
888struct uts_namespace;
889
890struct rq;
891struct sched_domain;
892
893struct sched_class {
894 const struct sched_class *next;
895
896 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
897 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
898 void (*yield_task) (struct rq *rq);
899 int (*select_task_rq)(struct task_struct *p, int sync);
900
901 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
902
903 struct task_struct * (*pick_next_task) (struct rq *rq);
904 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
905
906#ifdef CONFIG_SMP
907 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
908 struct rq *busiest, unsigned long max_load_move,
909 struct sched_domain *sd, enum cpu_idle_type idle,
910 int *all_pinned, int *this_best_prio);
911
912 int (*move_one_task) (struct rq *this_rq, int this_cpu,
913 struct rq *busiest, struct sched_domain *sd,
914 enum cpu_idle_type idle);
915 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
916 void (*post_schedule) (struct rq *this_rq);
917 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
918#endif
919
920 void (*set_curr_task) (struct rq *rq);
921 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
922 void (*task_new) (struct rq *rq, struct task_struct *p);
923 void (*set_cpus_allowed)(struct task_struct *p,
924 const cpumask_t *newmask);
925
926 void (*join_domain)(struct rq *rq);
927 void (*leave_domain)(struct rq *rq);
928
929 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
930 int running);
931 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
932 int running);
933 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
934 int oldprio, int running);
935
936#ifdef CONFIG_FAIR_GROUP_SCHED
937 void (*moved_group) (struct task_struct *p);
938#endif
939};
940
941struct load_weight {
942 unsigned long weight, inv_weight;
943};
944
945/*
946 * CFS stats for a schedulable entity (task, task-group etc)
947 *
948 * Current field usage histogram:
949 *
950 * 4 se->block_start
951 * 4 se->run_node
952 * 4 se->sleep_start
953 * 6 se->load.weight
954 */
955struct sched_entity {
956 struct load_weight load; /* for load-balancing */
957 struct rb_node run_node;
958 struct list_head group_node;
959 unsigned int on_rq;
960
961 u64 exec_start;
962 u64 sum_exec_runtime;
963 u64 vruntime;
964 u64 prev_sum_exec_runtime;
965
966 u64 last_wakeup;
967 u64 avg_overlap;
968
969#ifdef CONFIG_SCHEDSTATS
970 u64 wait_start;
971 u64 wait_max;
972 u64 wait_count;
973 u64 wait_sum;
974
975 u64 sleep_start;
976 u64 sleep_max;
977 s64 sum_sleep_runtime;
978
979 u64 block_start;
980 u64 block_max;
981 u64 exec_max;
982 u64 slice_max;
983
984 u64 nr_migrations;
985 u64 nr_migrations_cold;
986 u64 nr_failed_migrations_affine;
987 u64 nr_failed_migrations_running;
988 u64 nr_failed_migrations_hot;
989 u64 nr_forced_migrations;
990 u64 nr_forced2_migrations;
991
992 u64 nr_wakeups;
993 u64 nr_wakeups_sync;
994 u64 nr_wakeups_migrate;
995 u64 nr_wakeups_local;
996 u64 nr_wakeups_remote;
997 u64 nr_wakeups_affine;
998 u64 nr_wakeups_affine_attempts;
999 u64 nr_wakeups_passive;
1000 u64 nr_wakeups_idle;
1001#endif
1002
1003#ifdef CONFIG_FAIR_GROUP_SCHED
1004 struct sched_entity *parent;
1005 /* rq on which this entity is (to be) queued: */
1006 struct cfs_rq *cfs_rq;
1007 /* rq "owned" by this entity/group: */
1008 struct cfs_rq *my_q;
1009#endif
1010};
1011
1012struct sched_rt_entity {
1013 struct list_head run_list;
1014 unsigned int time_slice;
1015 unsigned long timeout;
1016 int nr_cpus_allowed;
1017
1018 struct sched_rt_entity *back;
1019#ifdef CONFIG_RT_GROUP_SCHED
1020 struct sched_rt_entity *parent;
1021 /* rq on which this entity is (to be) queued: */
1022 struct rt_rq *rt_rq;
1023 /* rq "owned" by this entity/group: */
1024 struct rt_rq *my_q;
1025#endif
1026};
1027
1028struct task_struct {
1029 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1030 void *stack;
1031 atomic_t usage;
1032 unsigned int flags; /* per process flags, defined below */
1033 unsigned int ptrace;
1034
1035 int lock_depth; /* BKL lock depth */
1036
1037#ifdef CONFIG_SMP
1038#ifdef __ARCH_WANT_UNLOCKED_CTXSW
1039 int oncpu;
1040#endif
1041#endif
1042
1043 int prio, static_prio, normal_prio;
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 * ptrace_list/ptrace_children forms the list of my children
1082 * that were stolen by a ptracer.
1083 */
1084 struct list_head ptrace_children;
1085 struct list_head ptrace_list;
1086
1087 struct mm_struct *mm, *active_mm;
1088
1089/* task state */
1090 struct linux_binfmt *binfmt;
1091 int exit_state;
1092 int exit_code, exit_signal;
1093 int pdeath_signal; /* The signal sent when the parent dies */
1094 /* ??? */
1095 unsigned int personality;
1096 unsigned did_exec:1;
1097 pid_t pid;
1098 pid_t tgid;
1099
1100#ifdef CONFIG_CC_STACKPROTECTOR
1101 /* Canary value for the -fstack-protector gcc feature */
1102 unsigned long stack_canary;
1103#endif
1104 /*
1105 * pointers to (original) parent process, youngest child, younger sibling,
1106 * older sibling, respectively. (p->father can be replaced with
1107 * p->parent->pid)
1108 */
1109 struct task_struct *real_parent; /* real parent process (when being debugged) */
1110 struct task_struct *parent; /* parent process */
1111 /*
1112 * children/sibling forms the list of my children plus the
1113 * tasks I'm ptracing.
1114 */
1115 struct list_head children; /* list of my children */
1116 struct list_head sibling; /* linkage in my parent's children list */
1117 struct task_struct *group_leader; /* threadgroup leader */
1118
1119 /* PID/PID hash table linkage. */
1120 struct pid_link pids[PIDTYPE_MAX];
1121 struct list_head thread_group;
1122
1123 struct completion *vfork_done; /* for vfork() */
1124 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1125 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1126
1127 unsigned int rt_priority;
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 unsigned securebits;
1147 struct user_struct *user;
1148#ifdef CONFIG_KEYS
1149 struct key *request_key_auth; /* assumed request_key authority */
1150 struct key *thread_keyring; /* keyring private to this thread */
1151 unsigned char jit_keyring; /* default keyring to attach requested keys to */
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 struct held_lock held_locks[MAX_LOCK_DEPTH];
1239 unsigned int lockdep_recursion;
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#ifdef CONFIG_TASK_XACCT
1258/* i/o counters(bytes read/written, #syscalls */
1259 u64 rchar, wchar, syscr, syscw;
1260#endif
1261 struct task_io_accounting ioac;
1262#if defined(CONFIG_TASK_XACCT)
1263 u64 acct_rss_mem1; /* accumulated rss usage */
1264 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1265 cputime_t acct_stimexpd;/* stime since last update */
1266#endif
1267#ifdef CONFIG_NUMA
1268 struct mempolicy *mempolicy;
1269 short il_next;
1270#endif
1271#ifdef CONFIG_CPUSETS
1272 nodemask_t mems_allowed;
1273 int cpuset_mems_generation;
1274 int cpuset_mem_spread_rotor;
1275#endif
1276#ifdef CONFIG_CGROUPS
1277 /* Control Group info protected by css_set_lock */
1278 struct css_set *cgroups;
1279 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1280 struct list_head cg_list;
1281#endif
1282#ifdef CONFIG_FUTEX
1283 struct robust_list_head __user *robust_list;
1284#ifdef CONFIG_COMPAT
1285 struct compat_robust_list_head __user *compat_robust_list;
1286#endif
1287 struct list_head pi_state_list;
1288 struct futex_pi_state *pi_state_cache;
1289#endif
1290 atomic_t fs_excl; /* holding fs exclusive resources */
1291 struct rcu_head rcu;
1292
1293 /*
1294 * cache last used pipe for splice
1295 */
1296 struct pipe_inode_info *splice_pipe;
1297#ifdef CONFIG_TASK_DELAY_ACCT
1298 struct task_delay_info *delays;
1299#endif
1300#ifdef CONFIG_FAULT_INJECTION
1301 int make_it_fail;
1302#endif
1303 struct prop_local_single dirties;
1304#ifdef CONFIG_LATENCYTOP
1305 int latency_record_count;
1306 struct latency_record latency_record[LT_SAVECOUNT];
1307#endif
1308};
1309
1310/*
1311 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1312 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1313 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1314 * values are inverted: lower p->prio value means higher priority.
1315 *
1316 * The MAX_USER_RT_PRIO value allows the actual maximum
1317 * RT priority to be separate from the value exported to
1318 * user-space. This allows kernel threads to set their
1319 * priority to a value higher than any user task. Note:
1320 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1321 */
1322
1323#define MAX_USER_RT_PRIO 100
1324#define MAX_RT_PRIO MAX_USER_RT_PRIO
1325
1326#define MAX_PRIO (MAX_RT_PRIO + 40)
1327#define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1328
1329static inline int rt_prio(int prio)
1330{
1331 if (unlikely(prio < MAX_RT_PRIO))
1332 return 1;
1333 return 0;
1334}
1335
1336static inline int rt_task(struct task_struct *p)
1337{
1338 return rt_prio(p->prio);
1339}
1340
1341static inline void set_task_session(struct task_struct *tsk, pid_t session)
1342{
1343 tsk->signal->__session = session;
1344}
1345
1346static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1347{
1348 tsk->signal->__pgrp = pgrp;
1349}
1350
1351static inline struct pid *task_pid(struct task_struct *task)
1352{
1353 return task->pids[PIDTYPE_PID].pid;
1354}
1355
1356static inline struct pid *task_tgid(struct task_struct *task)
1357{
1358 return task->group_leader->pids[PIDTYPE_PID].pid;
1359}
1360
1361static inline struct pid *task_pgrp(struct task_struct *task)
1362{
1363 return task->group_leader->pids[PIDTYPE_PGID].pid;
1364}
1365
1366static inline struct pid *task_session(struct task_struct *task)
1367{
1368 return task->group_leader->pids[PIDTYPE_SID].pid;
1369}
1370
1371struct pid_namespace;
1372
1373/*
1374 * the helpers to get the task's different pids as they are seen
1375 * from various namespaces
1376 *
1377 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1378 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1379 * current.
1380 * task_xid_nr_ns() : id seen from the ns specified;
1381 *
1382 * set_task_vxid() : assigns a virtual id to a task;
1383 *
1384 * see also pid_nr() etc in include/linux/pid.h
1385 */
1386
1387static inline pid_t task_pid_nr(struct task_struct *tsk)
1388{
1389 return tsk->pid;
1390}
1391
1392pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1393
1394static inline pid_t task_pid_vnr(struct task_struct *tsk)
1395{
1396 return pid_vnr(task_pid(tsk));
1397}
1398
1399
1400static inline pid_t task_tgid_nr(struct task_struct *tsk)
1401{
1402 return tsk->tgid;
1403}
1404
1405pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1406
1407static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1408{
1409 return pid_vnr(task_tgid(tsk));
1410}
1411
1412
1413static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1414{
1415 return tsk->signal->__pgrp;
1416}
1417
1418pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1419
1420static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1421{
1422 return pid_vnr(task_pgrp(tsk));
1423}
1424
1425
1426static inline pid_t task_session_nr(struct task_struct *tsk)
1427{
1428 return tsk->signal->__session;
1429}
1430
1431pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1432
1433static inline pid_t task_session_vnr(struct task_struct *tsk)
1434{
1435 return pid_vnr(task_session(tsk));
1436}
1437
1438
1439/**
1440 * pid_alive - check that a task structure is not stale
1441 * @p: Task structure to be checked.
1442 *
1443 * Test if a process is not yet dead (at most zombie state)
1444 * If pid_alive fails, then pointers within the task structure
1445 * can be stale and must not be dereferenced.
1446 */
1447static inline int pid_alive(struct task_struct *p)
1448{
1449 return p->pids[PIDTYPE_PID].pid != NULL;
1450}
1451
1452/**
1453 * is_global_init - check if a task structure is init
1454 * @tsk: Task structure to be checked.
1455 *
1456 * Check if a task structure is the first user space task the kernel created.
1457 */
1458static inline int is_global_init(struct task_struct *tsk)
1459{
1460 return tsk->pid == 1;
1461}
1462
1463/*
1464 * is_container_init:
1465 * check whether in the task is init in its own pid namespace.
1466 */
1467extern int is_container_init(struct task_struct *tsk);
1468
1469extern struct pid *cad_pid;
1470
1471extern void free_task(struct task_struct *tsk);
1472#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1473
1474extern void __put_task_struct(struct task_struct *t);
1475
1476static inline void put_task_struct(struct task_struct *t)
1477{
1478 if (atomic_dec_and_test(&t->usage))
1479 __put_task_struct(t);
1480}
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_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
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_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1510#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1511#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1512
1513/*
1514 * Only the _current_ task can read/write to tsk->flags, but other
1515 * tasks can access tsk->flags in readonly mode for example
1516 * with tsk_used_math (like during threaded core dumping).
1517 * There is however an exception to this rule during ptrace
1518 * or during fork: the ptracer task is allowed to write to the
1519 * child->flags of its traced child (same goes for fork, the parent
1520 * can write to the child->flags), because we're guaranteed the
1521 * child is not running and in turn not changing child->flags
1522 * at the same time the parent does it.
1523 */
1524#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1525#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1526#define clear_used_math() clear_stopped_child_used_math(current)
1527#define set_used_math() set_stopped_child_used_math(current)
1528#define conditional_stopped_child_used_math(condition, child) \
1529 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1530#define conditional_used_math(condition) \
1531 conditional_stopped_child_used_math(condition, current)
1532#define copy_to_stopped_child_used_math(child) \
1533 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1534/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1535#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1536#define used_math() tsk_used_math(current)
1537
1538#ifdef CONFIG_SMP
1539extern int set_cpus_allowed_ptr(struct task_struct *p,
1540 const cpumask_t *new_mask);
1541#else
1542static inline int set_cpus_allowed_ptr(struct task_struct *p,
1543 const cpumask_t *new_mask)
1544{
1545 if (!cpu_isset(0, *new_mask))
1546 return -EINVAL;
1547 return 0;
1548}
1549#endif
1550static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1551{
1552 return set_cpus_allowed_ptr(p, &new_mask);
1553}
1554
1555extern unsigned long long sched_clock(void);
1556
1557#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1558static inline void sched_clock_init(void)
1559{
1560}
1561
1562static inline u64 sched_clock_cpu(int cpu)
1563{
1564 return sched_clock();
1565}
1566
1567static inline void sched_clock_tick(void)
1568{
1569}
1570
1571static inline void sched_clock_idle_sleep_event(void)
1572{
1573}
1574
1575static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1576{
1577}
1578#else
1579extern void sched_clock_init(void);
1580extern u64 sched_clock_cpu(int cpu);
1581extern void sched_clock_tick(void);
1582extern void sched_clock_idle_sleep_event(void);
1583extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1584#endif
1585
1586/*
1587 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1588 * clock constructed from sched_clock():
1589 */
1590extern unsigned long long cpu_clock(int cpu);
1591
1592extern unsigned long long
1593task_sched_runtime(struct task_struct *task);
1594
1595/* sched_exec is called by processes performing an exec */
1596#ifdef CONFIG_SMP
1597extern void sched_exec(void);
1598#else
1599#define sched_exec() {}
1600#endif
1601
1602extern void sched_clock_idle_sleep_event(void);
1603extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1604
1605#ifdef CONFIG_HOTPLUG_CPU
1606extern void idle_task_exit(void);
1607#else
1608static inline void idle_task_exit(void) {}
1609#endif
1610
1611extern void sched_idle_next(void);
1612
1613#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1614extern void wake_up_idle_cpu(int cpu);
1615#else
1616static inline void wake_up_idle_cpu(int cpu) { }
1617#endif
1618
1619#ifdef CONFIG_SCHED_DEBUG
1620extern unsigned int sysctl_sched_latency;
1621extern unsigned int sysctl_sched_min_granularity;
1622extern unsigned int sysctl_sched_wakeup_granularity;
1623extern unsigned int sysctl_sched_child_runs_first;
1624extern unsigned int sysctl_sched_features;
1625extern unsigned int sysctl_sched_migration_cost;
1626extern unsigned int sysctl_sched_nr_migrate;
1627
1628int sched_nr_latency_handler(struct ctl_table *table, int write,
1629 struct file *file, void __user *buffer, size_t *length,
1630 loff_t *ppos);
1631#endif
1632extern unsigned int sysctl_sched_rt_period;
1633extern int sysctl_sched_rt_runtime;
1634
1635int sched_rt_handler(struct ctl_table *table, int write,
1636 struct file *filp, void __user *buffer, size_t *lenp,
1637 loff_t *ppos);
1638
1639extern unsigned int sysctl_sched_compat_yield;
1640
1641#ifdef CONFIG_RT_MUTEXES
1642extern int rt_mutex_getprio(struct task_struct *p);
1643extern void rt_mutex_setprio(struct task_struct *p, int prio);
1644extern void rt_mutex_adjust_pi(struct task_struct *p);
1645#else
1646static inline int rt_mutex_getprio(struct task_struct *p)
1647{
1648 return p->normal_prio;
1649}
1650# define rt_mutex_adjust_pi(p) do { } while (0)
1651#endif
1652
1653extern void set_user_nice(struct task_struct *p, long nice);
1654extern int task_prio(const struct task_struct *p);
1655extern int task_nice(const struct task_struct *p);
1656extern int can_nice(const struct task_struct *p, const int nice);
1657extern int task_curr(const struct task_struct *p);
1658extern int idle_cpu(int cpu);
1659extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1660extern struct task_struct *idle_task(int cpu);
1661extern struct task_struct *curr_task(int cpu);
1662extern void set_curr_task(int cpu, struct task_struct *p);
1663
1664void yield(void);
1665
1666/*
1667 * The default (Linux) execution domain.
1668 */
1669extern struct exec_domain default_exec_domain;
1670
1671union thread_union {
1672 struct thread_info thread_info;
1673 unsigned long stack[THREAD_SIZE/sizeof(long)];
1674};
1675
1676#ifndef __HAVE_ARCH_KSTACK_END
1677static inline int kstack_end(void *addr)
1678{
1679 /* Reliable end of stack detection:
1680 * Some APM bios versions misalign the stack
1681 */
1682 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1683}
1684#endif
1685
1686extern union thread_union init_thread_union;
1687extern struct task_struct init_task;
1688
1689extern struct mm_struct init_mm;
1690
1691extern struct pid_namespace init_pid_ns;
1692
1693/*
1694 * find a task by one of its numerical ids
1695 *
1696 * find_task_by_pid_type_ns():
1697 * it is the most generic call - it finds a task by all id,
1698 * type and namespace specified
1699 * find_task_by_pid_ns():
1700 * finds a task by its pid in the specified namespace
1701 * find_task_by_vpid():
1702 * finds a task by its virtual pid
1703 * find_task_by_pid():
1704 * finds a task by its global pid
1705 *
1706 * see also find_pid() etc in include/linux/pid.h
1707 */
1708
1709extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1710 struct pid_namespace *ns);
1711
1712static inline struct task_struct *__deprecated find_task_by_pid(pid_t nr)
1713{
1714 return find_task_by_pid_type_ns(PIDTYPE_PID, nr, &init_pid_ns);
1715}
1716extern struct task_struct *find_task_by_vpid(pid_t nr);
1717extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1718 struct pid_namespace *ns);
1719
1720extern void __set_special_pids(struct pid *pid);
1721
1722/* per-UID process charging. */
1723extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1724static inline struct user_struct *get_uid(struct user_struct *u)
1725{
1726 atomic_inc(&u->__count);
1727 return u;
1728}
1729extern void free_uid(struct user_struct *);
1730extern void switch_uid(struct user_struct *);
1731extern void release_uids(struct user_namespace *ns);
1732
1733#include <asm/current.h>
1734
1735extern void do_timer(unsigned long ticks);
1736
1737extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1738extern int wake_up_process(struct task_struct *tsk);
1739extern void wake_up_new_task(struct task_struct *tsk,
1740 unsigned long clone_flags);
1741#ifdef CONFIG_SMP
1742 extern void kick_process(struct task_struct *tsk);
1743#else
1744 static inline void kick_process(struct task_struct *tsk) { }
1745#endif
1746extern void sched_fork(struct task_struct *p, int clone_flags);
1747extern void sched_dead(struct task_struct *p);
1748
1749extern int in_group_p(gid_t);
1750extern int in_egroup_p(gid_t);
1751
1752extern void proc_caches_init(void);
1753extern void flush_signals(struct task_struct *);
1754extern void ignore_signals(struct task_struct *);
1755extern void flush_signal_handlers(struct task_struct *, int force_default);
1756extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1757
1758static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1759{
1760 unsigned long flags;
1761 int ret;
1762
1763 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1764 ret = dequeue_signal(tsk, mask, info);
1765 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1766
1767 return ret;
1768}
1769
1770extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1771 sigset_t *mask);
1772extern void unblock_all_signals(void);
1773extern void release_task(struct task_struct * p);
1774extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1775extern int force_sigsegv(int, struct task_struct *);
1776extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1777extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1778extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1779extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1780extern int kill_pgrp(struct pid *pid, int sig, int priv);
1781extern int kill_pid(struct pid *pid, int sig, int priv);
1782extern int kill_proc_info(int, struct siginfo *, pid_t);
1783extern void do_notify_parent(struct task_struct *, int);
1784extern void force_sig(int, struct task_struct *);
1785extern void force_sig_specific(int, struct task_struct *);
1786extern int send_sig(int, struct task_struct *, int);
1787extern void zap_other_threads(struct task_struct *p);
1788extern int kill_proc(pid_t, int, int);
1789extern struct sigqueue *sigqueue_alloc(void);
1790extern void sigqueue_free(struct sigqueue *);
1791extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1792extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1793extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1794
1795static inline int kill_cad_pid(int sig, int priv)
1796{
1797 return kill_pid(cad_pid, sig, priv);
1798}
1799
1800/* These can be the second arg to send_sig_info/send_group_sig_info. */
1801#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1802#define SEND_SIG_PRIV ((struct siginfo *) 1)
1803#define SEND_SIG_FORCED ((struct siginfo *) 2)
1804
1805static inline int is_si_special(const struct siginfo *info)
1806{
1807 return info <= SEND_SIG_FORCED;
1808}
1809
1810/* True if we are on the alternate signal stack. */
1811
1812static inline int on_sig_stack(unsigned long sp)
1813{
1814 return (sp - current->sas_ss_sp < current->sas_ss_size);
1815}
1816
1817static inline int sas_ss_flags(unsigned long sp)
1818{
1819 return (current->sas_ss_size == 0 ? SS_DISABLE
1820 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1821}
1822
1823/*
1824 * Routines for handling mm_structs
1825 */
1826extern struct mm_struct * mm_alloc(void);
1827
1828/* mmdrop drops the mm and the page tables */
1829extern void __mmdrop(struct mm_struct *);
1830static inline void mmdrop(struct mm_struct * mm)
1831{
1832 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1833 __mmdrop(mm);
1834}
1835
1836/* mmput gets rid of the mappings and all user-space */
1837extern void mmput(struct mm_struct *);
1838/* Grab a reference to a task's mm, if it is not already going away */
1839extern struct mm_struct *get_task_mm(struct task_struct *task);
1840/* Remove the current tasks stale references to the old mm_struct */
1841extern void mm_release(struct task_struct *, struct mm_struct *);
1842/* Allocate a new mm structure and copy contents from tsk->mm */
1843extern struct mm_struct *dup_mm(struct task_struct *tsk);
1844
1845extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1846extern void flush_thread(void);
1847extern void exit_thread(void);
1848
1849extern void exit_files(struct task_struct *);
1850extern void __cleanup_signal(struct signal_struct *);
1851extern void __cleanup_sighand(struct sighand_struct *);
1852extern void exit_itimers(struct signal_struct *);
1853
1854extern NORET_TYPE void do_group_exit(int);
1855
1856extern void daemonize(const char *, ...);
1857extern int allow_signal(int);
1858extern int disallow_signal(int);
1859
1860extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1861extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1862struct task_struct *fork_idle(int);
1863
1864extern void set_task_comm(struct task_struct *tsk, char *from);
1865extern char *get_task_comm(char *to, struct task_struct *tsk);
1866
1867#ifdef CONFIG_SMP
1868extern void wait_task_inactive(struct task_struct * p);
1869#else
1870#define wait_task_inactive(p) do { } while (0)
1871#endif
1872
1873#define remove_parent(p) list_del_init(&(p)->sibling)
1874#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1875
1876#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1877
1878#define for_each_process(p) \
1879 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1880
1881/*
1882 * Careful: do_each_thread/while_each_thread is a double loop so
1883 * 'break' will not work as expected - use goto instead.
1884 */
1885#define do_each_thread(g, t) \
1886 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1887
1888#define while_each_thread(g, t) \
1889 while ((t = next_thread(t)) != g)
1890
1891/* de_thread depends on thread_group_leader not being a pid based check */
1892#define thread_group_leader(p) (p == p->group_leader)
1893
1894/* Do to the insanities of de_thread it is possible for a process
1895 * to have the pid of the thread group leader without actually being
1896 * the thread group leader. For iteration through the pids in proc
1897 * all we care about is that we have a task with the appropriate
1898 * pid, we don't actually care if we have the right task.
1899 */
1900static inline int has_group_leader_pid(struct task_struct *p)
1901{
1902 return p->pid == p->tgid;
1903}
1904
1905static inline
1906int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1907{
1908 return p1->tgid == p2->tgid;
1909}
1910
1911static inline struct task_struct *next_thread(const struct task_struct *p)
1912{
1913 return list_entry(rcu_dereference(p->thread_group.next),
1914 struct task_struct, thread_group);
1915}
1916
1917static inline int thread_group_empty(struct task_struct *p)
1918{
1919 return list_empty(&p->thread_group);
1920}
1921
1922#define delay_group_leader(p) \
1923 (thread_group_leader(p) && !thread_group_empty(p))
1924
1925/*
1926 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1927 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1928 * pins the final release of task.io_context. Also protects ->cpuset and
1929 * ->cgroup.subsys[].
1930 *
1931 * Nests both inside and outside of read_lock(&tasklist_lock).
1932 * It must not be nested with write_lock_irq(&tasklist_lock),
1933 * neither inside nor outside.
1934 */
1935static inline void task_lock(struct task_struct *p)
1936{
1937 spin_lock(&p->alloc_lock);
1938}
1939
1940static inline void task_unlock(struct task_struct *p)
1941{
1942 spin_unlock(&p->alloc_lock);
1943}
1944
1945extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1946 unsigned long *flags);
1947
1948static inline void unlock_task_sighand(struct task_struct *tsk,
1949 unsigned long *flags)
1950{
1951 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1952}
1953
1954#ifndef __HAVE_THREAD_FUNCTIONS
1955
1956#define task_thread_info(task) ((struct thread_info *)(task)->stack)
1957#define task_stack_page(task) ((task)->stack)
1958
1959static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1960{
1961 *task_thread_info(p) = *task_thread_info(org);
1962 task_thread_info(p)->task = p;
1963}
1964
1965static inline unsigned long *end_of_stack(struct task_struct *p)
1966{
1967 return (unsigned long *)(task_thread_info(p) + 1);
1968}
1969
1970#endif
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 need_resched(void)
2030{
2031 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2032}
2033
2034/*
2035 * cond_resched() and cond_resched_lock(): latency reduction via
2036 * explicit rescheduling in places that are safe. The return
2037 * value indicates whether a reschedule was done in fact.
2038 * cond_resched_lock() will drop the spinlock before scheduling,
2039 * cond_resched_softirq() will enable bhs before scheduling.
2040 */
2041extern int _cond_resched(void);
2042#ifdef CONFIG_PREEMPT_BKL
2043static inline int cond_resched(void)
2044{
2045 return 0;
2046}
2047#else
2048static inline int cond_resched(void)
2049{
2050 return _cond_resched();
2051}
2052#endif
2053extern int cond_resched_lock(spinlock_t * lock);
2054extern int cond_resched_softirq(void);
2055static inline int cond_resched_bkl(void)
2056{
2057 return _cond_resched();
2058}
2059
2060/*
2061 * Does a critical section need to be broken due to another
2062 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2063 * but a general need for low latency)
2064 */
2065static inline int spin_needbreak(spinlock_t *lock)
2066{
2067#ifdef CONFIG_PREEMPT
2068 return spin_is_contended(lock);
2069#else
2070 return 0;
2071#endif
2072}
2073
2074/*
2075 * Reevaluate whether the task has signals pending delivery.
2076 * Wake the task if so.
2077 * This is required every time the blocked sigset_t changes.
2078 * callers must hold sighand->siglock.
2079 */
2080extern void recalc_sigpending_and_wake(struct task_struct *t);
2081extern void recalc_sigpending(void);
2082
2083extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2084
2085/*
2086 * Wrappers for p->thread_info->cpu access. No-op on UP.
2087 */
2088#ifdef CONFIG_SMP
2089
2090static inline unsigned int task_cpu(const struct task_struct *p)
2091{
2092 return task_thread_info(p)->cpu;
2093}
2094
2095extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2096
2097#else
2098
2099static inline unsigned int task_cpu(const struct task_struct *p)
2100{
2101 return 0;
2102}
2103
2104static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2105{
2106}
2107
2108#endif /* CONFIG_SMP */
2109
2110#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2111extern void arch_pick_mmap_layout(struct mm_struct *mm);
2112#else
2113static inline void arch_pick_mmap_layout(struct mm_struct *mm)
2114{
2115 mm->mmap_base = TASK_UNMAPPED_BASE;
2116 mm->get_unmapped_area = arch_get_unmapped_area;
2117 mm->unmap_area = arch_unmap_area;
2118}
2119#endif
2120
2121extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask);
2122extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2123
2124extern int sched_mc_power_savings, sched_smt_power_savings;
2125
2126extern void normalize_rt_tasks(void);
2127
2128#ifdef CONFIG_GROUP_SCHED
2129
2130extern struct task_group init_task_group;
2131#ifdef CONFIG_USER_SCHED
2132extern struct task_group root_task_group;
2133#endif
2134
2135extern struct task_group *sched_create_group(struct task_group *parent);
2136extern void sched_destroy_group(struct task_group *tg);
2137extern void sched_move_task(struct task_struct *tsk);
2138#ifdef CONFIG_FAIR_GROUP_SCHED
2139extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2140extern unsigned long sched_group_shares(struct task_group *tg);
2141#endif
2142#ifdef CONFIG_RT_GROUP_SCHED
2143extern int sched_group_set_rt_runtime(struct task_group *tg,
2144 long rt_runtime_us);
2145extern long sched_group_rt_runtime(struct task_group *tg);
2146extern int sched_group_set_rt_period(struct task_group *tg,
2147 long rt_period_us);
2148extern long sched_group_rt_period(struct task_group *tg);
2149#endif
2150#endif
2151
2152#ifdef CONFIG_TASK_XACCT
2153static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2154{
2155 tsk->rchar += amt;
2156}
2157
2158static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2159{
2160 tsk->wchar += amt;
2161}
2162
2163static inline void inc_syscr(struct task_struct *tsk)
2164{
2165 tsk->syscr++;
2166}
2167
2168static inline void inc_syscw(struct task_struct *tsk)
2169{
2170 tsk->syscw++;
2171}
2172#else
2173static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2174{
2175}
2176
2177static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2178{
2179}
2180
2181static inline void inc_syscr(struct task_struct *tsk)
2182{
2183}
2184
2185static inline void inc_syscw(struct task_struct *tsk)
2186{
2187}
2188#endif
2189
2190#ifdef CONFIG_SMP
2191void migration_init(void);
2192#else
2193static inline void migration_init(void)
2194{
2195}
2196#endif
2197
2198#ifndef TASK_SIZE_OF
2199#define TASK_SIZE_OF(tsk) TASK_SIZE
2200#endif
2201
2202#ifdef CONFIG_MM_OWNER
2203extern void mm_update_next_owner(struct mm_struct *mm);
2204extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2205#else
2206static inline void mm_update_next_owner(struct mm_struct *mm)
2207{
2208}
2209
2210static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2211{
2212}
2213#endif /* CONFIG_MM_OWNER */
2214
2215#endif /* __KERNEL__ */
2216
2217#endif