blob: d04186d8cc685d9be0ddea01fc94a23b7a6828a7 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
4#include <asm/param.h> /* for HZ */
5
6#include <linux/config.h>
7#include <linux/capability.h>
8#include <linux/threads.h>
9#include <linux/kernel.h>
10#include <linux/types.h>
11#include <linux/timex.h>
12#include <linux/jiffies.h>
13#include <linux/rbtree.h>
14#include <linux/thread_info.h>
15#include <linux/cpumask.h>
16#include <linux/errno.h>
17#include <linux/nodemask.h>
18
19#include <asm/system.h>
20#include <asm/semaphore.h>
21#include <asm/page.h>
22#include <asm/ptrace.h>
23#include <asm/mmu.h>
24#include <asm/cputime.h>
25
26#include <linux/smp.h>
27#include <linux/sem.h>
28#include <linux/signal.h>
29#include <linux/securebits.h>
30#include <linux/fs_struct.h>
31#include <linux/compiler.h>
32#include <linux/completion.h>
33#include <linux/pid.h>
34#include <linux/percpu.h>
35#include <linux/topology.h>
36#include <linux/seccomp.h>
Ingo Molnare56d0902006-01-08 01:01:37 -080037#include <linux/rcupdate.h>
Ingo Molnar0771dfe2006-03-27 01:16:22 -080038#include <linux/futex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070039
H. J. Lu36d57ac2005-09-06 15:16:49 -070040#include <linux/auxvec.h> /* For AT_VECTOR_SIZE */
41
Linus Torvalds1da177e2005-04-16 15:20:36 -070042struct exec_domain;
43
44/*
45 * cloning flags:
46 */
47#define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
48#define CLONE_VM 0x00000100 /* set if VM shared between processes */
49#define CLONE_FS 0x00000200 /* set if fs info shared between processes */
50#define CLONE_FILES 0x00000400 /* set if open files shared between processes */
51#define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
52#define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
53#define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
54#define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
55#define CLONE_THREAD 0x00010000 /* Same thread group? */
56#define CLONE_NEWNS 0x00020000 /* New namespace group? */
57#define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
58#define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
59#define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
60#define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
61#define CLONE_DETACHED 0x00400000 /* Unused, ignored */
62#define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
63#define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
64#define CLONE_STOPPED 0x02000000 /* Start in stopped state */
65
66/*
67 * List of flags we want to share for kernel threads,
68 * if only because they are not used by them anyway.
69 */
70#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
71
72/*
73 * These are the constant used to fake the fixed-point load-average
74 * counting. Some notes:
75 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
76 * a load-average precision of 10 bits integer + 11 bits fractional
77 * - if you want to count load-averages more often, you need more
78 * precision, or rounding will get you. With 2-second counting freq,
79 * the EXP_n values would be 1981, 2034 and 2043 if still using only
80 * 11 bit fractions.
81 */
82extern unsigned long avenrun[]; /* Load averages */
83
84#define FSHIFT 11 /* nr of bits of precision */
85#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
86#define LOAD_FREQ (5*HZ) /* 5 sec intervals */
87#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
88#define EXP_5 2014 /* 1/exp(5sec/5min) */
89#define EXP_15 2037 /* 1/exp(5sec/15min) */
90
91#define CALC_LOAD(load,exp,n) \
92 load *= exp; \
93 load += n*(FIXED_1-exp); \
94 load >>= FSHIFT;
95
96extern unsigned long total_forks;
97extern int nr_threads;
98extern int last_pid;
99DECLARE_PER_CPU(unsigned long, process_counts);
100extern int nr_processes(void);
101extern unsigned long nr_running(void);
102extern unsigned long nr_uninterruptible(void);
103extern unsigned long nr_iowait(void);
104
105#include <linux/time.h>
106#include <linux/param.h>
107#include <linux/resource.h>
108#include <linux/timer.h>
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800109#include <linux/hrtimer.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700110
111#include <asm/processor.h>
112
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700113/*
114 * Task state bitmask. NOTE! These bits are also
115 * encoded in fs/proc/array.c: get_task_state().
116 *
117 * We have two separate sets of flags: task->state
118 * is about runnability, while task->exit_state are
119 * about the task exiting. Confusing, but this way
120 * modifying one set can't modify the other one by
121 * mistake.
122 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700123#define TASK_RUNNING 0
124#define TASK_INTERRUPTIBLE 1
125#define TASK_UNINTERRUPTIBLE 2
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700126#define TASK_STOPPED 4
127#define TASK_TRACED 8
128/* in tsk->exit_state */
129#define EXIT_ZOMBIE 16
130#define EXIT_DEAD 32
131/* in tsk->state again */
132#define TASK_NONINTERACTIVE 64
Linus Torvalds1da177e2005-04-16 15:20:36 -0700133
134#define __set_task_state(tsk, state_value) \
135 do { (tsk)->state = (state_value); } while (0)
136#define set_task_state(tsk, state_value) \
137 set_mb((tsk)->state, (state_value))
138
Andrew Morton498d0c52005-09-13 01:25:14 -0700139/*
140 * set_current_state() includes a barrier so that the write of current->state
141 * is correctly serialised wrt the caller's subsequent test of whether to
142 * actually sleep:
143 *
144 * set_current_state(TASK_UNINTERRUPTIBLE);
145 * if (do_i_need_to_sleep())
146 * schedule();
147 *
148 * If the caller does not need such serialisation then use __set_current_state()
149 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150#define __set_current_state(state_value) \
151 do { current->state = (state_value); } while (0)
152#define set_current_state(state_value) \
153 set_mb(current->state, (state_value))
154
155/* Task command name length */
156#define TASK_COMM_LEN 16
157
158/*
159 * Scheduling policies
160 */
161#define SCHED_NORMAL 0
162#define SCHED_FIFO 1
163#define SCHED_RR 2
Ingo Molnarb0a94992006-01-14 13:20:41 -0800164#define SCHED_BATCH 3
Linus Torvalds1da177e2005-04-16 15:20:36 -0700165
166struct sched_param {
167 int sched_priority;
168};
169
170#ifdef __KERNEL__
171
172#include <linux/spinlock.h>
173
174/*
175 * This serializes "schedule()" and also protects
176 * the run-queue from deletions/modifications (but
177 * _adding_ to the beginning of the run-queue has
178 * a separate lock).
179 */
180extern rwlock_t tasklist_lock;
181extern spinlock_t mmlist_lock;
182
183typedef struct task_struct task_t;
184
185extern void sched_init(void);
186extern void sched_init_smp(void);
187extern void init_idle(task_t *idle, int cpu);
188
189extern cpumask_t nohz_cpu_mask;
190
191extern void show_state(void);
192extern void show_regs(struct pt_regs *);
193
194/*
195 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
196 * task), SP is the stack pointer of the first frame that should be shown in the back
197 * trace (or NULL if the entire call-chain of the task should be shown).
198 */
199extern void show_stack(struct task_struct *task, unsigned long *sp);
200
201void io_schedule(void);
202long io_schedule_timeout(long timeout);
203
204extern void cpu_init (void);
205extern void trap_init(void);
206extern void update_process_times(int user);
207extern void scheduler_tick(void);
208
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700209#ifdef CONFIG_DETECT_SOFTLOCKUP
Ingo Molnar6687a972006-03-24 03:18:41 -0800210extern void softlockup_tick(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700211extern void spawn_softlockup_task(void);
212extern void touch_softlockup_watchdog(void);
213#else
Ingo Molnar6687a972006-03-24 03:18:41 -0800214static inline void softlockup_tick(void)
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700215{
216}
217static inline void spawn_softlockup_task(void)
218{
219}
220static inline void touch_softlockup_watchdog(void)
221{
222}
223#endif
224
225
Linus Torvalds1da177e2005-04-16 15:20:36 -0700226/* Attach to any functions which should be ignored in wchan output. */
227#define __sched __attribute__((__section__(".sched.text")))
228/* Is this address in the __sched functions? */
229extern int in_sched_functions(unsigned long addr);
230
231#define MAX_SCHEDULE_TIMEOUT LONG_MAX
232extern signed long FASTCALL(schedule_timeout(signed long timeout));
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700233extern signed long schedule_timeout_interruptible(signed long timeout);
234extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700235asmlinkage void schedule(void);
236
237struct namespace;
238
239/* Maximum number of active map areas.. This is a random (large) number */
240#define DEFAULT_MAX_MAP_COUNT 65536
241
242extern int sysctl_max_map_count;
243
244#include <linux/aio.h>
245
246extern unsigned long
247arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
248 unsigned long, unsigned long);
249extern unsigned long
250arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
251 unsigned long len, unsigned long pgoff,
252 unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700253extern void arch_unmap_area(struct mm_struct *, unsigned long);
254extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700255
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700256#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
257/*
258 * The mm counters are not protected by its page_table_lock,
259 * so must be incremented atomically.
260 */
Christoph Lameterd3cb4872006-01-06 00:11:20 -0800261#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
262#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
263#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
264#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
265#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
266typedef atomic_long_t mm_counter_t;
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700267
268#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
269/*
270 * The mm counters are protected by its page_table_lock,
271 * so can be incremented directly.
272 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700273#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
274#define get_mm_counter(mm, member) ((mm)->_##member)
275#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
276#define inc_mm_counter(mm, member) (mm)->_##member++
277#define dec_mm_counter(mm, member) (mm)->_##member--
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700278typedef unsigned long mm_counter_t;
Hugh Dickins42946212005-10-29 18:16:05 -0700279
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700280#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
281
282#define get_mm_rss(mm) \
283 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
Hugh Dickins365e9c872005-10-29 18:16:18 -0700284#define update_hiwater_rss(mm) do { \
285 unsigned long _rss = get_mm_rss(mm); \
286 if ((mm)->hiwater_rss < _rss) \
287 (mm)->hiwater_rss = _rss; \
288} while (0)
289#define update_hiwater_vm(mm) do { \
290 if ((mm)->hiwater_vm < (mm)->total_vm) \
291 (mm)->hiwater_vm = (mm)->total_vm; \
292} while (0)
293
Linus Torvalds1da177e2005-04-16 15:20:36 -0700294struct mm_struct {
295 struct vm_area_struct * mmap; /* list of VMAs */
296 struct rb_root mm_rb;
297 struct vm_area_struct * mmap_cache; /* last find_vma result */
298 unsigned long (*get_unmapped_area) (struct file *filp,
299 unsigned long addr, unsigned long len,
300 unsigned long pgoff, unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700301 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
Benjamin Herrenschmidt0551fbd2006-02-28 16:59:19 -0800302 unsigned long mmap_base; /* base of mmap area */
303 unsigned long task_size; /* size of task vm space */
304 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700305 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700306 pgd_t * pgd;
307 atomic_t mm_users; /* How many users with user space? */
308 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
309 int map_count; /* number of VMAs */
310 struct rw_semaphore mmap_sem;
311 spinlock_t page_table_lock; /* Protects page tables and some counters */
312
313 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
314 * together off init_mm.mmlist, and are protected
315 * by mmlist_lock
316 */
317
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700318 /* Special counters, in some configurations protected by the
319 * page_table_lock, in other configurations by being atomic.
320 */
Hugh Dickins42946212005-10-29 18:16:05 -0700321 mm_counter_t _file_rss;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700322 mm_counter_t _anon_rss;
323
Hugh Dickinsf4499522005-10-29 18:16:19 -0700324 unsigned long hiwater_rss; /* High-watermark of RSS usage */
325 unsigned long hiwater_vm; /* High-water virtual memory usage */
326
327 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
328 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
329 unsigned long start_code, end_code, start_data, end_data;
330 unsigned long start_brk, brk, start_stack;
331 unsigned long arg_start, arg_end, env_start, env_end;
332
H. J. Lu36d57ac2005-09-06 15:16:49 -0700333 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700334
Alan Coxd6e71142005-06-23 00:09:43 -0700335 unsigned dumpable:2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700336 cpumask_t cpu_vm_mask;
337
338 /* Architecture-specific MM context */
339 mm_context_t context;
340
341 /* Token based thrashing protection. */
342 unsigned long swap_token_time;
343 char recent_pagein;
344
345 /* coredumping support */
346 int core_waiters;
347 struct completion *core_startup_done, core_done;
348
349 /* aio bits */
350 rwlock_t ioctx_list_lock;
351 struct kioctx *ioctx_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700352};
353
354struct sighand_struct {
355 atomic_t count;
356 struct k_sigaction action[_NSIG];
357 spinlock_t siglock;
358};
359
360/*
361 * NOTE! "signal_struct" does not have it's own
362 * locking, because a shared signal_struct always
363 * implies a shared sighand_struct, so locking
364 * sighand_struct is always a proper superset of
365 * the locking of signal_struct.
366 */
367struct signal_struct {
368 atomic_t count;
369 atomic_t live;
370
371 wait_queue_head_t wait_chldexit; /* for wait4() */
372
373 /* current thread group signal load-balancing target: */
374 task_t *curr_target;
375
376 /* shared signal handling: */
377 struct sigpending shared_pending;
378
379 /* thread group exit support */
380 int group_exit_code;
381 /* overloaded:
382 * - notify group_exit_task when ->count is equal to notify_count
383 * - everyone except group_exit_task is stopped during signal delivery
384 * of fatal signals, group_exit_task processes the signal.
385 */
386 struct task_struct *group_exit_task;
387 int notify_count;
388
389 /* thread group stop support, overloads group_exit_code too */
390 int group_stop_count;
391 unsigned int flags; /* see SIGNAL_* flags below */
392
393 /* POSIX.1b Interval Timers */
394 struct list_head posix_timers;
395
396 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800397 struct hrtimer real_timer;
Roman Zippel05cfb612006-03-26 01:38:12 -0800398 struct task_struct *tsk;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800399 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700400
401 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
402 cputime_t it_prof_expires, it_virt_expires;
403 cputime_t it_prof_incr, it_virt_incr;
404
405 /* job control IDs */
406 pid_t pgrp;
407 pid_t tty_old_pgrp;
408 pid_t session;
409 /* boolean value for session group leader */
410 int leader;
411
412 struct tty_struct *tty; /* NULL if no tty */
413
414 /*
415 * Cumulative resource counters for dead threads in the group,
416 * and for reaped dead child processes forked by this group.
417 * Live threads maintain their own counters and add to these
418 * in __exit_signal, except for the group leader.
419 */
420 cputime_t utime, stime, cutime, cstime;
421 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
422 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
423
424 /*
425 * Cumulative ns of scheduled CPU time for dead threads in the
426 * group, not including a zombie group leader. (This only differs
427 * from jiffies_to_ns(utime + stime) if sched_clock uses something
428 * other than jiffies.)
429 */
430 unsigned long long sched_time;
431
432 /*
433 * We don't bother to synchronize most readers of this at all,
434 * because there is no reader checking a limit that actually needs
435 * to get both rlim_cur and rlim_max atomically, and either one
436 * alone is a single word that can safely be read normally.
437 * getrlimit/setrlimit use task_lock(current->group_leader) to
438 * protect this instead of the siglock, because they really
439 * have no need to disable irqs.
440 */
441 struct rlimit rlim[RLIM_NLIMITS];
442
443 struct list_head cpu_timers[3];
444
445 /* keep the process-shared keyrings here so that they do the right
446 * thing in threads created with CLONE_THREAD */
447#ifdef CONFIG_KEYS
448 struct key *session_keyring; /* keyring inherited over fork */
449 struct key *process_keyring; /* keyring private to this process */
450#endif
451};
452
Nick Piggin4866cde2005-06-25 14:57:23 -0700453/* Context switch must be unlocked if interrupts are to be enabled */
454#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
455# define __ARCH_WANT_UNLOCKED_CTXSW
456#endif
457
Linus Torvalds1da177e2005-04-16 15:20:36 -0700458/*
459 * Bits in flags field of signal_struct.
460 */
461#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
462#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
463#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
464#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
465
466
467/*
468 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
Ingo Molnarb0a94992006-01-14 13:20:41 -0800469 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
470 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
471 * values are inverted: lower p->prio value means higher priority.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700472 *
473 * The MAX_USER_RT_PRIO value allows the actual maximum
474 * RT priority to be separate from the value exported to
475 * user-space. This allows kernel threads to set their
476 * priority to a value higher than any user task. Note:
477 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
478 */
479
480#define MAX_USER_RT_PRIO 100
481#define MAX_RT_PRIO MAX_USER_RT_PRIO
482
483#define MAX_PRIO (MAX_RT_PRIO + 40)
484
485#define rt_task(p) (unlikely((p)->prio < MAX_RT_PRIO))
486
487/*
488 * Some day this will be a full-fledged user tracking system..
489 */
490struct user_struct {
491 atomic_t __count; /* reference count */
492 atomic_t processes; /* How many processes does this user have? */
493 atomic_t files; /* How many open files does this user have? */
494 atomic_t sigpending; /* How many pending signals does this user have? */
Robert Love0eeca282005-07-12 17:06:03 -0400495#ifdef CONFIG_INOTIFY
496 atomic_t inotify_watches; /* How many inotify watches does this user have? */
497 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
498#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700499 /* protected by mq_lock */
500 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
501 unsigned long locked_shm; /* How many pages of mlocked shm ? */
502
503#ifdef CONFIG_KEYS
504 struct key *uid_keyring; /* UID specific keyring */
505 struct key *session_keyring; /* UID's default session keyring */
506#endif
507
508 /* Hash table maintenance information */
509 struct list_head uidhash_list;
510 uid_t uid;
511};
512
513extern struct user_struct *find_user(uid_t);
514
515extern struct user_struct root_user;
516#define INIT_USER (&root_user)
517
518typedef struct prio_array prio_array_t;
519struct backing_dev_info;
520struct reclaim_state;
521
522#ifdef CONFIG_SCHEDSTATS
523struct sched_info {
524 /* cumulative counters */
525 unsigned long cpu_time, /* time spent on the cpu */
526 run_delay, /* time spent waiting on a runqueue */
527 pcnt; /* # of timeslices run on this cpu */
528
529 /* timestamps */
530 unsigned long last_arrival, /* when we last ran on a cpu */
531 last_queued; /* when we were last queued to run */
532};
533
534extern struct file_operations proc_schedstat_operations;
535#endif
536
537enum idle_type
538{
539 SCHED_IDLE,
540 NOT_IDLE,
541 NEWLY_IDLE,
542 MAX_IDLE_TYPES
543};
544
545/*
546 * sched-domains (multiprocessor balancing) declarations:
547 */
548#ifdef CONFIG_SMP
549#define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
550
551#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
552#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
553#define SD_BALANCE_EXEC 4 /* Balance on exec */
Nick Piggin147cbb42005-06-25 14:57:19 -0700554#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
555#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
556#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
557#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
558#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700559
560struct sched_group {
561 struct sched_group *next; /* Must be a circular list */
562 cpumask_t cpumask;
563
564 /*
565 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
566 * single CPU. This is read only (except for setup, hotplug CPU).
567 */
568 unsigned long cpu_power;
569};
570
571struct sched_domain {
572 /* These fields must be setup */
573 struct sched_domain *parent; /* top domain must be null terminated */
574 struct sched_group *groups; /* the balancing groups of the domain */
575 cpumask_t span; /* span of all CPUs in this domain */
576 unsigned long min_interval; /* Minimum balance interval ms */
577 unsigned long max_interval; /* Maximum balance interval ms */
578 unsigned int busy_factor; /* less balancing by factor if busy */
579 unsigned int imbalance_pct; /* No balance until over watermark */
580 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
581 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
582 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
Nick Piggin78979862005-06-25 14:57:13 -0700583 unsigned int busy_idx;
584 unsigned int idle_idx;
585 unsigned int newidle_idx;
586 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700587 unsigned int forkexec_idx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700588 int flags; /* See SD_* */
589
590 /* Runtime fields. */
591 unsigned long last_balance; /* init to jiffies. units in jiffies */
592 unsigned int balance_interval; /* initialise to 1. units in ms. */
593 unsigned int nr_balance_failed; /* initialise to 0 */
594
595#ifdef CONFIG_SCHEDSTATS
596 /* load_balance() stats */
597 unsigned long lb_cnt[MAX_IDLE_TYPES];
598 unsigned long lb_failed[MAX_IDLE_TYPES];
599 unsigned long lb_balanced[MAX_IDLE_TYPES];
600 unsigned long lb_imbalance[MAX_IDLE_TYPES];
601 unsigned long lb_gained[MAX_IDLE_TYPES];
602 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
603 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
604 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
605
606 /* Active load balancing */
607 unsigned long alb_cnt;
608 unsigned long alb_failed;
609 unsigned long alb_pushed;
610
Nick Piggin68767a02005-06-25 14:57:20 -0700611 /* SD_BALANCE_EXEC stats */
612 unsigned long sbe_cnt;
613 unsigned long sbe_balanced;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700614 unsigned long sbe_pushed;
615
Nick Piggin68767a02005-06-25 14:57:20 -0700616 /* SD_BALANCE_FORK stats */
617 unsigned long sbf_cnt;
618 unsigned long sbf_balanced;
619 unsigned long sbf_pushed;
620
Linus Torvalds1da177e2005-04-16 15:20:36 -0700621 /* try_to_wake_up() stats */
622 unsigned long ttwu_wake_remote;
623 unsigned long ttwu_move_affine;
624 unsigned long ttwu_move_balance;
625#endif
626};
627
Dinakar Guniguntala1a20ff22005-06-25 14:57:33 -0700628extern void partition_sched_domains(cpumask_t *partition1,
629 cpumask_t *partition2);
akpm@osdl.org198e2f12006-01-12 01:05:30 -0800630
631/*
632 * Maximum cache size the migration-costs auto-tuning code will
633 * search from:
634 */
635extern unsigned int max_cache_size;
636
637#endif /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700638
639
640struct io_context; /* See blkdev.h */
641void exit_io_context(void);
642struct cpuset;
643
644#define NGROUPS_SMALL 32
645#define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
646struct group_info {
647 int ngroups;
648 atomic_t usage;
649 gid_t small_block[NGROUPS_SMALL];
650 int nblocks;
651 gid_t *blocks[0];
652};
653
654/*
655 * get_group_info() must be called with the owning task locked (via task_lock())
656 * when task != current. The reason being that the vast majority of callers are
657 * looking at current->group_info, which can not be changed except by the
658 * current task. Changing current->group_info requires the task lock, too.
659 */
660#define get_group_info(group_info) do { \
661 atomic_inc(&(group_info)->usage); \
662} while (0)
663
664#define put_group_info(group_info) do { \
665 if (atomic_dec_and_test(&(group_info)->usage)) \
666 groups_free(group_info); \
667} while (0)
668
David Howells3e301482005-06-23 22:00:56 -0700669extern struct group_info *groups_alloc(int gidsetsize);
670extern void groups_free(struct group_info *group_info);
671extern int set_current_groups(struct group_info *group_info);
672extern int groups_search(struct group_info *group_info, gid_t grp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700673/* access the groups "array" with this macro */
674#define GROUP_AT(gi, i) \
675 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
676
Chen, Kenneth W383f2832005-09-09 13:02:02 -0700677#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
678extern void prefetch_stack(struct task_struct*);
679#else
680static inline void prefetch_stack(struct task_struct *t) { }
681#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700682
683struct audit_context; /* See audit.c */
684struct mempolicy;
685
686struct task_struct {
687 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
688 struct thread_info *thread_info;
689 atomic_t usage;
690 unsigned long flags; /* per process flags, defined below */
691 unsigned long ptrace;
692
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700693 int lock_depth; /* BKL lock depth */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700694
Chen, Kenneth Wd6077cb2006-02-14 13:53:10 -0800695#if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
Nick Piggin4866cde2005-06-25 14:57:23 -0700696 int oncpu;
697#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698 int prio, static_prio;
699 struct list_head run_list;
700 prio_array_t *array;
701
Jens Axboe22e2c502005-06-27 10:55:12 +0200702 unsigned short ioprio;
Jens Axboe2056a782006-03-23 20:00:26 +0100703 unsigned int btrace_seq;
Jens Axboe22e2c502005-06-27 10:55:12 +0200704
Linus Torvalds1da177e2005-04-16 15:20:36 -0700705 unsigned long sleep_avg;
706 unsigned long long timestamp, last_ran;
707 unsigned long long sched_time; /* sched_clock time spent running */
708 int activated;
709
710 unsigned long policy;
711 cpumask_t cpus_allowed;
712 unsigned int time_slice, first_time_slice;
713
714#ifdef CONFIG_SCHEDSTATS
715 struct sched_info sched_info;
716#endif
717
718 struct list_head tasks;
719 /*
720 * ptrace_list/ptrace_children forms the list of my children
721 * that were stolen by a ptracer.
722 */
723 struct list_head ptrace_children;
724 struct list_head ptrace_list;
725
726 struct mm_struct *mm, *active_mm;
727
728/* task state */
729 struct linux_binfmt *binfmt;
730 long exit_state;
731 int exit_code, exit_signal;
732 int pdeath_signal; /* The signal sent when the parent dies */
733 /* ??? */
734 unsigned long personality;
735 unsigned did_exec:1;
736 pid_t pid;
737 pid_t tgid;
738 /*
739 * pointers to (original) parent process, youngest child, younger sibling,
740 * older sibling, respectively. (p->father can be replaced with
741 * p->parent->pid)
742 */
743 struct task_struct *real_parent; /* real parent process (when being debugged) */
744 struct task_struct *parent; /* parent process */
745 /*
746 * children/sibling forms the list of my children plus the
747 * tasks I'm ptracing.
748 */
749 struct list_head children; /* list of my children */
750 struct list_head sibling; /* linkage in my parent's children list */
751 struct task_struct *group_leader; /* threadgroup leader */
752
753 /* PID/PID hash table linkage. */
754 struct pid pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -0800755 struct list_head thread_group;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700756
757 struct completion *vfork_done; /* for vfork() */
758 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
759 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
760
761 unsigned long rt_priority;
762 cputime_t utime, stime;
763 unsigned long nvcsw, nivcsw; /* context switch counts */
764 struct timespec start_time;
765/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
766 unsigned long min_flt, maj_flt;
767
768 cputime_t it_prof_expires, it_virt_expires;
769 unsigned long long it_sched_expires;
770 struct list_head cpu_timers[3];
771
772/* process credentials */
773 uid_t uid,euid,suid,fsuid;
774 gid_t gid,egid,sgid,fsgid;
775 struct group_info *group_info;
776 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
777 unsigned keep_capabilities:1;
778 struct user_struct *user;
779#ifdef CONFIG_KEYS
David Howellsb5f545c2006-01-08 01:02:47 -0800780 struct key *request_key_auth; /* assumed request_key authority */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700781 struct key *thread_keyring; /* keyring private to this thread */
David Howells3e301482005-06-23 22:00:56 -0700782 unsigned char jit_keyring; /* default keyring to attach requested keys to */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700783#endif
784 int oomkilladj; /* OOM kill score adjustment (bit shift). */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700785 char comm[TASK_COMM_LEN]; /* executable name excluding path
786 - access with [gs]et_task_comm (which lock
787 it with task_lock())
788 - initialized normally by flush_old_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700789/* file system info */
790 int link_count, total_link_count;
791/* ipc stuff */
792 struct sysv_sem sysvsem;
793/* CPU-specific state of this task */
794 struct thread_struct thread;
795/* filesystem information */
796 struct fs_struct *fs;
797/* open file information */
798 struct files_struct *files;
799/* namespace */
800 struct namespace *namespace;
801/* signal handlers */
802 struct signal_struct *signal;
803 struct sighand_struct *sighand;
804
805 sigset_t blocked, real_blocked;
David Woodhouse150256d2006-01-18 17:43:57 -0800806 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700807 struct sigpending pending;
808
809 unsigned long sas_ss_sp;
810 size_t sas_ss_size;
811 int (*notifier)(void *priv);
812 void *notifier_data;
813 sigset_t *notifier_mask;
814
815 void *security;
816 struct audit_context *audit_context;
817 seccomp_t seccomp;
818
819/* Thread group tracking */
820 u32 parent_exec_id;
821 u32 self_exec_id;
822/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
823 spinlock_t alloc_lock;
824/* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
825 spinlock_t proc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700826
Ingo Molnar408894e2006-01-09 15:59:20 -0800827#ifdef CONFIG_DEBUG_MUTEXES
828 /* mutex deadlock detection */
829 struct mutex_waiter *blocked_on;
830#endif
831
Linus Torvalds1da177e2005-04-16 15:20:36 -0700832/* journalling filesystem info */
833 void *journal_info;
834
835/* VM state */
836 struct reclaim_state *reclaim_state;
837
838 struct dentry *proc_dentry;
839 struct backing_dev_info *backing_dev_info;
840
841 struct io_context *io_context;
842
843 unsigned long ptrace_message;
844 siginfo_t *last_siginfo; /* For ptrace use. */
845/*
846 * current io wait handle: wait queue entry to use for io waits
847 * If this thread is processing aio, this points at the waitqueue
848 * inside the currently handled kiocb. It may be NULL (i.e. default
849 * to a stack based synchronous wait) if its doing sync IO.
850 */
851 wait_queue_t *io_wait;
852/* i/o counters(bytes read/written, #syscalls */
853 u64 rchar, wchar, syscr, syscw;
854#if defined(CONFIG_BSD_PROCESS_ACCT)
855 u64 acct_rss_mem1; /* accumulated rss usage */
856 u64 acct_vm_mem1; /* accumulated virtual memory usage */
857 clock_t acct_stimexpd; /* clock_t-converted stime since last update */
858#endif
859#ifdef CONFIG_NUMA
860 struct mempolicy *mempolicy;
861 short il_next;
862#endif
863#ifdef CONFIG_CPUSETS
864 struct cpuset *cpuset;
865 nodemask_t mems_allowed;
866 int cpuset_mems_generation;
Paul Jackson825a46a2006-03-24 03:16:03 -0800867 int cpuset_mem_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700868#endif
Ingo Molnar0771dfe2006-03-27 01:16:22 -0800869 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -0800870#ifdef CONFIG_COMPAT
871 struct compat_robust_list_head __user *compat_robust_list;
872#endif
Ingo Molnar0771dfe2006-03-27 01:16:22 -0800873
Jens Axboe22e2c502005-06-27 10:55:12 +0200874 atomic_t fs_excl; /* holding fs exclusive resources */
Ingo Molnare56d0902006-01-08 01:01:37 -0800875 struct rcu_head rcu;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700876};
877
878static inline pid_t process_group(struct task_struct *tsk)
879{
880 return tsk->signal->pgrp;
881}
882
883/**
884 * pid_alive - check that a task structure is not stale
885 * @p: Task structure to be checked.
886 *
887 * Test if a process is not yet dead (at most zombie state)
888 * If pid_alive fails, then pointers within the task structure
889 * can be stale and must not be dereferenced.
890 */
891static inline int pid_alive(struct task_struct *p)
892{
893 return p->pids[PIDTYPE_PID].nr != 0;
894}
895
896extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700897#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -0800898
Ingo Molnare56d0902006-01-08 01:01:37 -0800899extern void __put_task_struct_cb(struct rcu_head *rhp);
900
901static inline void put_task_struct(struct task_struct *t)
902{
903 if (atomic_dec_and_test(&t->usage))
904 call_rcu(&t->rcu, __put_task_struct_cb);
905}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700906
907/*
908 * Per process flags
909 */
910#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
911 /* Not implemented yet, only for 486*/
912#define PF_STARTING 0x00000002 /* being created */
913#define PF_EXITING 0x00000004 /* getting shut down */
914#define PF_DEAD 0x00000008 /* Dead */
915#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
916#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
917#define PF_DUMPCORE 0x00000200 /* dumped core */
918#define PF_SIGNALED 0x00000400 /* killed by a signal */
919#define PF_MEMALLOC 0x00000800 /* Allocating memory */
920#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
921#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
922#define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */
923#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
924#define PF_FROZEN 0x00010000 /* frozen for system suspend */
925#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
926#define PF_KSWAPD 0x00040000 /* I am kswapd */
927#define PF_SWAPOFF 0x00080000 /* I am in swapoff */
928#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
929#define PF_SYNCWRITE 0x00200000 /* I am doing a sync write */
930#define PF_BORROWED_MM 0x00400000 /* I am a kthread doing use_mm */
931#define PF_RANDOMIZE 0x00800000 /* randomize virtual address space */
Christoph Lameter930d9152006-01-08 01:00:47 -0800932#define PF_SWAPWRITE 0x01000000 /* Allowed to write to swap */
Paul Jackson825a46a2006-03-24 03:16:03 -0800933#define PF_SPREAD_PAGE 0x04000000 /* Spread page cache over cpuset */
934#define PF_SPREAD_SLAB 0x08000000 /* Spread some slab caches over cpuset */
Paul Jacksonc61afb12006-03-24 03:16:08 -0800935#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700936
937/*
938 * Only the _current_ task can read/write to tsk->flags, but other
939 * tasks can access tsk->flags in readonly mode for example
940 * with tsk_used_math (like during threaded core dumping).
941 * There is however an exception to this rule during ptrace
942 * or during fork: the ptracer task is allowed to write to the
943 * child->flags of its traced child (same goes for fork, the parent
944 * can write to the child->flags), because we're guaranteed the
945 * child is not running and in turn not changing child->flags
946 * at the same time the parent does it.
947 */
948#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
949#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
950#define clear_used_math() clear_stopped_child_used_math(current)
951#define set_used_math() set_stopped_child_used_math(current)
952#define conditional_stopped_child_used_math(condition, child) \
953 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
954#define conditional_used_math(condition) \
955 conditional_stopped_child_used_math(condition, current)
956#define copy_to_stopped_child_used_math(child) \
957 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
958/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
959#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
960#define used_math() tsk_used_math(current)
961
962#ifdef CONFIG_SMP
963extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
964#else
965static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
966{
Paul Jackson4098f992005-10-30 15:03:21 -0800967 if (!cpu_isset(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700968 return -EINVAL;
969 return 0;
970}
971#endif
972
973extern unsigned long long sched_clock(void);
974extern unsigned long long current_sched_time(const task_t *current_task);
975
976/* sched_exec is called by processes performing an exec */
977#ifdef CONFIG_SMP
978extern void sched_exec(void);
979#else
980#define sched_exec() {}
981#endif
982
983#ifdef CONFIG_HOTPLUG_CPU
984extern void idle_task_exit(void);
985#else
986static inline void idle_task_exit(void) {}
987#endif
988
989extern void sched_idle_next(void);
990extern void set_user_nice(task_t *p, long nice);
991extern int task_prio(const task_t *p);
992extern int task_nice(const task_t *p);
Matt Mackalle43379f2005-05-01 08:59:00 -0700993extern int can_nice(const task_t *p, const int nice);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700994extern int task_curr(const task_t *p);
995extern int idle_cpu(int cpu);
996extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
997extern task_t *idle_task(int cpu);
Keith Owensa2a97982005-09-11 17:19:06 +1000998extern task_t *curr_task(int cpu);
999extern void set_curr_task(int cpu, task_t *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001000
1001void yield(void);
1002
1003/*
1004 * The default (Linux) execution domain.
1005 */
1006extern struct exec_domain default_exec_domain;
1007
1008union thread_union {
1009 struct thread_info thread_info;
1010 unsigned long stack[THREAD_SIZE/sizeof(long)];
1011};
1012
1013#ifndef __HAVE_ARCH_KSTACK_END
1014static inline int kstack_end(void *addr)
1015{
1016 /* Reliable end of stack detection:
1017 * Some APM bios versions misalign the stack
1018 */
1019 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1020}
1021#endif
1022
1023extern union thread_union init_thread_union;
1024extern struct task_struct init_task;
1025
1026extern struct mm_struct init_mm;
1027
1028#define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1029extern struct task_struct *find_task_by_pid_type(int type, int pid);
1030extern void set_special_pids(pid_t session, pid_t pgrp);
1031extern void __set_special_pids(pid_t session, pid_t pgrp);
1032
1033/* per-UID process charging. */
1034extern struct user_struct * alloc_uid(uid_t);
1035static inline struct user_struct *get_uid(struct user_struct *u)
1036{
1037 atomic_inc(&u->__count);
1038 return u;
1039}
1040extern void free_uid(struct user_struct *);
1041extern void switch_uid(struct user_struct *);
1042
1043#include <asm/current.h>
1044
1045extern void do_timer(struct pt_regs *);
1046
1047extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1048extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1049extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1050 unsigned long clone_flags));
1051#ifdef CONFIG_SMP
1052 extern void kick_process(struct task_struct *tsk);
1053#else
1054 static inline void kick_process(struct task_struct *tsk) { }
1055#endif
Nick Piggin476d1392005-06-25 14:57:29 -07001056extern void FASTCALL(sched_fork(task_t * p, int clone_flags));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001057extern void FASTCALL(sched_exit(task_t * p));
1058
1059extern int in_group_p(gid_t);
1060extern int in_egroup_p(gid_t);
1061
1062extern void proc_caches_init(void);
1063extern void flush_signals(struct task_struct *);
1064extern void flush_signal_handlers(struct task_struct *, int force_default);
1065extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1066
1067static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1068{
1069 unsigned long flags;
1070 int ret;
1071
1072 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1073 ret = dequeue_signal(tsk, mask, info);
1074 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1075
1076 return ret;
1077}
1078
1079extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1080 sigset_t *mask);
1081extern void unblock_all_signals(void);
1082extern void release_task(struct task_struct * p);
1083extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1084extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1085extern int force_sigsegv(int, struct task_struct *);
1086extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1087extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1088extern int kill_pg_info(int, struct siginfo *, pid_t);
1089extern int kill_proc_info(int, struct siginfo *, pid_t);
Harald Welte46113832005-10-10 19:44:29 +02001090extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001091extern void do_notify_parent(struct task_struct *, int);
1092extern void force_sig(int, struct task_struct *);
1093extern void force_sig_specific(int, struct task_struct *);
1094extern int send_sig(int, struct task_struct *, int);
1095extern void zap_other_threads(struct task_struct *p);
1096extern int kill_pg(pid_t, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001097extern int kill_proc(pid_t, int, int);
1098extern struct sigqueue *sigqueue_alloc(void);
1099extern void sigqueue_free(struct sigqueue *);
1100extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1101extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03001102extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001103extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1104
1105/* These can be the second arg to send_sig_info/send_group_sig_info. */
1106#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1107#define SEND_SIG_PRIV ((struct siginfo *) 1)
1108#define SEND_SIG_FORCED ((struct siginfo *) 2)
1109
Oleg Nesterov621d3122005-10-30 15:03:45 -08001110static inline int is_si_special(const struct siginfo *info)
1111{
1112 return info <= SEND_SIG_FORCED;
1113}
1114
Linus Torvalds1da177e2005-04-16 15:20:36 -07001115/* True if we are on the alternate signal stack. */
1116
1117static inline int on_sig_stack(unsigned long sp)
1118{
1119 return (sp - current->sas_ss_sp < current->sas_ss_size);
1120}
1121
1122static inline int sas_ss_flags(unsigned long sp)
1123{
1124 return (current->sas_ss_size == 0 ? SS_DISABLE
1125 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1126}
1127
Linus Torvalds1da177e2005-04-16 15:20:36 -07001128/*
1129 * Routines for handling mm_structs
1130 */
1131extern struct mm_struct * mm_alloc(void);
1132
1133/* mmdrop drops the mm and the page tables */
1134extern void FASTCALL(__mmdrop(struct mm_struct *));
1135static inline void mmdrop(struct mm_struct * mm)
1136{
1137 if (atomic_dec_and_test(&mm->mm_count))
1138 __mmdrop(mm);
1139}
1140
1141/* mmput gets rid of the mappings and all user-space */
1142extern void mmput(struct mm_struct *);
1143/* Grab a reference to a task's mm, if it is not already going away */
1144extern struct mm_struct *get_task_mm(struct task_struct *task);
1145/* Remove the current tasks stale references to the old mm_struct */
1146extern void mm_release(struct task_struct *, struct mm_struct *);
1147
1148extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1149extern void flush_thread(void);
1150extern void exit_thread(void);
1151
Linus Torvalds1da177e2005-04-16 15:20:36 -07001152extern void exit_files(struct task_struct *);
Oleg Nesterov6b3934e2006-03-28 16:11:16 -08001153extern void __cleanup_signal(struct signal_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08001154extern void __cleanup_sighand(struct sighand_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001155extern void exit_itimers(struct signal_struct *);
1156
1157extern NORET_TYPE void do_group_exit(int);
1158
Linus Torvalds1da177e2005-04-16 15:20:36 -07001159extern void daemonize(const char *, ...);
1160extern int allow_signal(int);
1161extern int disallow_signal(int);
1162extern task_t *child_reaper;
1163
1164extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1165extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1166task_t *fork_idle(int);
1167
1168extern void set_task_comm(struct task_struct *tsk, char *from);
1169extern void get_task_comm(char *to, struct task_struct *tsk);
1170
1171#ifdef CONFIG_SMP
1172extern void wait_task_inactive(task_t * p);
1173#else
1174#define wait_task_inactive(p) do { } while (0)
1175#endif
1176
1177#define remove_parent(p) list_del_init(&(p)->sibling)
Oleg Nesterov8fafabd2006-03-28 16:11:05 -08001178#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001179
Linus Torvalds1da177e2005-04-16 15:20:36 -07001180#define next_task(p) list_entry((p)->tasks.next, struct task_struct, tasks)
1181#define prev_task(p) list_entry((p)->tasks.prev, struct task_struct, tasks)
1182
1183#define for_each_process(p) \
1184 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1185
1186/*
1187 * Careful: do_each_thread/while_each_thread is a double loop so
1188 * 'break' will not work as expected - use goto instead.
1189 */
1190#define do_each_thread(g, t) \
1191 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1192
1193#define while_each_thread(g, t) \
1194 while ((t = next_thread(t)) != g)
1195
Linus Torvalds1da177e2005-04-16 15:20:36 -07001196#define thread_group_leader(p) (p->pid == p->tgid)
1197
Oleg Nesterov47e65322006-03-28 16:11:25 -08001198static inline task_t *next_thread(task_t *p)
1199{
1200 return list_entry(rcu_dereference(p->thread_group.next),
1201 task_t, thread_group);
1202}
1203
Linus Torvalds1da177e2005-04-16 15:20:36 -07001204static inline int thread_group_empty(task_t *p)
1205{
Oleg Nesterov47e65322006-03-28 16:11:25 -08001206 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001207}
1208
1209#define delay_group_leader(p) \
1210 (thread_group_leader(p) && !thread_group_empty(p))
1211
Linus Torvalds1da177e2005-04-16 15:20:36 -07001212/*
1213 * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02001214 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Jackson053199e2005-10-30 15:02:30 -08001215 * pins the final release of task.io_context. Also protects ->cpuset.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001216 *
1217 * Nests both inside and outside of read_lock(&tasklist_lock).
1218 * It must not be nested with write_lock_irq(&tasklist_lock),
1219 * neither inside nor outside.
1220 */
1221static inline void task_lock(struct task_struct *p)
1222{
1223 spin_lock(&p->alloc_lock);
1224}
1225
1226static inline void task_unlock(struct task_struct *p)
1227{
1228 spin_unlock(&p->alloc_lock);
1229}
1230
Oleg Nesterovf63ee722006-03-28 16:11:13 -08001231extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1232 unsigned long *flags);
1233
1234static inline void unlock_task_sighand(struct task_struct *tsk,
1235 unsigned long *flags)
1236{
1237 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1238}
1239
Al Virof0373602005-11-13 16:06:57 -08001240#ifndef __HAVE_THREAD_FUNCTIONS
1241
Al Viroa1261f52005-11-13 16:06:55 -08001242#define task_thread_info(task) (task)->thread_info
Al Viro9fc65872006-01-12 01:05:34 -08001243#define task_stack_page(task) ((void*)((task)->thread_info))
Al Viroa1261f52005-11-13 16:06:55 -08001244
Al Viro10ebffd2005-11-13 16:06:56 -08001245static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1246{
1247 *task_thread_info(p) = *task_thread_info(org);
1248 task_thread_info(p)->task = p;
1249}
1250
1251static inline unsigned long *end_of_stack(struct task_struct *p)
1252{
1253 return (unsigned long *)(p->thread_info + 1);
1254}
1255
Al Virof0373602005-11-13 16:06:57 -08001256#endif
1257
Linus Torvalds1da177e2005-04-16 15:20:36 -07001258/* set thread flags in other task's structures
1259 * - see asm/thread_info.h for TIF_xxxx flags available
1260 */
1261static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1262{
Al Viroa1261f52005-11-13 16:06:55 -08001263 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001264}
1265
1266static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1267{
Al Viroa1261f52005-11-13 16:06:55 -08001268 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001269}
1270
1271static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1272{
Al Viroa1261f52005-11-13 16:06:55 -08001273 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001274}
1275
1276static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1277{
Al Viroa1261f52005-11-13 16:06:55 -08001278 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001279}
1280
1281static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1282{
Al Viroa1261f52005-11-13 16:06:55 -08001283 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001284}
1285
1286static inline void set_tsk_need_resched(struct task_struct *tsk)
1287{
1288 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1289}
1290
1291static inline void clear_tsk_need_resched(struct task_struct *tsk)
1292{
1293 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1294}
1295
1296static inline int signal_pending(struct task_struct *p)
1297{
1298 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1299}
1300
1301static inline int need_resched(void)
1302{
1303 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1304}
1305
1306/*
1307 * cond_resched() and cond_resched_lock(): latency reduction via
1308 * explicit rescheduling in places that are safe. The return
1309 * value indicates whether a reschedule was done in fact.
1310 * cond_resched_lock() will drop the spinlock before scheduling,
1311 * cond_resched_softirq() will enable bhs before scheduling.
1312 */
1313extern int cond_resched(void);
1314extern int cond_resched_lock(spinlock_t * lock);
1315extern int cond_resched_softirq(void);
1316
1317/*
1318 * Does a critical section need to be broken due to another
1319 * task waiting?:
1320 */
1321#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1322# define need_lockbreak(lock) ((lock)->break_lock)
1323#else
1324# define need_lockbreak(lock) 0
1325#endif
1326
1327/*
1328 * Does a critical section need to be broken due to another
1329 * task waiting or preemption being signalled:
1330 */
1331static inline int lock_need_resched(spinlock_t *lock)
1332{
1333 if (need_lockbreak(lock) || need_resched())
1334 return 1;
1335 return 0;
1336}
1337
1338/* Reevaluate whether the task has signals pending delivery.
1339 This is required every time the blocked sigset_t changes.
1340 callers must hold sighand->siglock. */
1341
1342extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1343extern void recalc_sigpending(void);
1344
1345extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1346
1347/*
1348 * Wrappers for p->thread_info->cpu access. No-op on UP.
1349 */
1350#ifdef CONFIG_SMP
1351
1352static inline unsigned int task_cpu(const struct task_struct *p)
1353{
Al Viroa1261f52005-11-13 16:06:55 -08001354 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001355}
1356
1357static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1358{
Al Viroa1261f52005-11-13 16:06:55 -08001359 task_thread_info(p)->cpu = cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001360}
1361
1362#else
1363
1364static inline unsigned int task_cpu(const struct task_struct *p)
1365{
1366 return 0;
1367}
1368
1369static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1370{
1371}
1372
1373#endif /* CONFIG_SMP */
1374
1375#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1376extern void arch_pick_mmap_layout(struct mm_struct *mm);
1377#else
1378static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1379{
1380 mm->mmap_base = TASK_UNMAPPED_BASE;
1381 mm->get_unmapped_area = arch_get_unmapped_area;
1382 mm->unmap_area = arch_unmap_area;
1383}
1384#endif
1385
1386extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1387extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1388
Linus Torvalds1da177e2005-04-16 15:20:36 -07001389extern void normalize_rt_tasks(void);
1390
Linus Torvalds1da177e2005-04-16 15:20:36 -07001391#ifdef CONFIG_PM
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001392/*
1393 * Check if a process has been frozen
1394 */
1395static inline int frozen(struct task_struct *p)
1396{
1397 return p->flags & PF_FROZEN;
1398}
1399
1400/*
1401 * Check if there is a request to freeze a process
1402 */
1403static inline int freezing(struct task_struct *p)
1404{
1405 return p->flags & PF_FREEZE;
1406}
1407
1408/*
1409 * Request that a process be frozen
1410 * FIXME: SMP problem. We may not modify other process' flags!
1411 */
1412static inline void freeze(struct task_struct *p)
1413{
1414 p->flags |= PF_FREEZE;
1415}
1416
1417/*
1418 * Wake up a frozen process
1419 */
1420static inline int thaw_process(struct task_struct *p)
1421{
1422 if (frozen(p)) {
1423 p->flags &= ~PF_FROZEN;
1424 wake_up_process(p);
1425 return 1;
1426 }
1427 return 0;
1428}
1429
1430/*
1431 * freezing is complete, mark process as frozen
1432 */
1433static inline void frozen_process(struct task_struct *p)
1434{
1435 p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1436}
1437
1438extern void refrigerator(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001439extern int freeze_processes(void);
1440extern void thaw_processes(void);
1441
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001442static inline int try_to_freeze(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001443{
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001444 if (freezing(current)) {
1445 refrigerator();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001446 return 1;
1447 } else
1448 return 0;
1449}
1450#else
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001451static inline int frozen(struct task_struct *p) { return 0; }
1452static inline int freezing(struct task_struct *p) { return 0; }
1453static inline void freeze(struct task_struct *p) { BUG(); }
1454static inline int thaw_process(struct task_struct *p) { return 1; }
1455static inline void frozen_process(struct task_struct *p) { BUG(); }
1456
1457static inline void refrigerator(void) {}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001458static inline int freeze_processes(void) { BUG(); return 0; }
1459static inline void thaw_processes(void) {}
1460
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001461static inline int try_to_freeze(void) { return 0; }
1462
Linus Torvalds1da177e2005-04-16 15:20:36 -07001463#endif /* CONFIG_PM */
1464#endif /* __KERNEL__ */
1465
1466#endif