blob: e0054c1b9a098edb3c72e21cf5cd715bf07f8b6c [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>
Linus Torvalds1da177e2005-04-16 15:20:36 -070038
H. J. Lu36d57ac2005-09-06 15:16:49 -070039#include <linux/auxvec.h> /* For AT_VECTOR_SIZE */
40
Linus Torvalds1da177e2005-04-16 15:20:36 -070041struct exec_domain;
42
43/*
44 * cloning flags:
45 */
46#define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
47#define CLONE_VM 0x00000100 /* set if VM shared between processes */
48#define CLONE_FS 0x00000200 /* set if fs info shared between processes */
49#define CLONE_FILES 0x00000400 /* set if open files shared between processes */
50#define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
51#define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
52#define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
53#define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
54#define CLONE_THREAD 0x00010000 /* Same thread group? */
55#define CLONE_NEWNS 0x00020000 /* New namespace group? */
56#define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
57#define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
58#define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
59#define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
60#define CLONE_DETACHED 0x00400000 /* Unused, ignored */
61#define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
62#define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
63#define CLONE_STOPPED 0x02000000 /* Start in stopped state */
64
65/*
66 * List of flags we want to share for kernel threads,
67 * if only because they are not used by them anyway.
68 */
69#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
70
71/*
72 * These are the constant used to fake the fixed-point load-average
73 * counting. Some notes:
74 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
75 * a load-average precision of 10 bits integer + 11 bits fractional
76 * - if you want to count load-averages more often, you need more
77 * precision, or rounding will get you. With 2-second counting freq,
78 * the EXP_n values would be 1981, 2034 and 2043 if still using only
79 * 11 bit fractions.
80 */
81extern unsigned long avenrun[]; /* Load averages */
82
83#define FSHIFT 11 /* nr of bits of precision */
84#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
85#define LOAD_FREQ (5*HZ) /* 5 sec intervals */
86#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
87#define EXP_5 2014 /* 1/exp(5sec/5min) */
88#define EXP_15 2037 /* 1/exp(5sec/15min) */
89
90#define CALC_LOAD(load,exp,n) \
91 load *= exp; \
92 load += n*(FIXED_1-exp); \
93 load >>= FSHIFT;
94
95extern unsigned long total_forks;
96extern int nr_threads;
97extern int last_pid;
98DECLARE_PER_CPU(unsigned long, process_counts);
99extern int nr_processes(void);
100extern unsigned long nr_running(void);
101extern unsigned long nr_uninterruptible(void);
102extern unsigned long nr_iowait(void);
103
104#include <linux/time.h>
105#include <linux/param.h>
106#include <linux/resource.h>
107#include <linux/timer.h>
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800108#include <linux/hrtimer.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700109
110#include <asm/processor.h>
111
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700112/*
113 * Task state bitmask. NOTE! These bits are also
114 * encoded in fs/proc/array.c: get_task_state().
115 *
116 * We have two separate sets of flags: task->state
117 * is about runnability, while task->exit_state are
118 * about the task exiting. Confusing, but this way
119 * modifying one set can't modify the other one by
120 * mistake.
121 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700122#define TASK_RUNNING 0
123#define TASK_INTERRUPTIBLE 1
124#define TASK_UNINTERRUPTIBLE 2
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700125#define TASK_STOPPED 4
126#define TASK_TRACED 8
127/* in tsk->exit_state */
128#define EXIT_ZOMBIE 16
129#define EXIT_DEAD 32
130/* in tsk->state again */
131#define TASK_NONINTERACTIVE 64
Linus Torvalds1da177e2005-04-16 15:20:36 -0700132
133#define __set_task_state(tsk, state_value) \
134 do { (tsk)->state = (state_value); } while (0)
135#define set_task_state(tsk, state_value) \
136 set_mb((tsk)->state, (state_value))
137
Andrew Morton498d0c52005-09-13 01:25:14 -0700138/*
139 * set_current_state() includes a barrier so that the write of current->state
140 * is correctly serialised wrt the caller's subsequent test of whether to
141 * actually sleep:
142 *
143 * set_current_state(TASK_UNINTERRUPTIBLE);
144 * if (do_i_need_to_sleep())
145 * schedule();
146 *
147 * If the caller does not need such serialisation then use __set_current_state()
148 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700149#define __set_current_state(state_value) \
150 do { current->state = (state_value); } while (0)
151#define set_current_state(state_value) \
152 set_mb(current->state, (state_value))
153
154/* Task command name length */
155#define TASK_COMM_LEN 16
156
157/*
158 * Scheduling policies
159 */
160#define SCHED_NORMAL 0
161#define SCHED_FIFO 1
162#define SCHED_RR 2
Ingo Molnarb0a94992006-01-14 13:20:41 -0800163#define SCHED_BATCH 3
Linus Torvalds1da177e2005-04-16 15:20:36 -0700164
165struct sched_param {
166 int sched_priority;
167};
168
169#ifdef __KERNEL__
170
171#include <linux/spinlock.h>
172
173/*
174 * This serializes "schedule()" and also protects
175 * the run-queue from deletions/modifications (but
176 * _adding_ to the beginning of the run-queue has
177 * a separate lock).
178 */
179extern rwlock_t tasklist_lock;
180extern spinlock_t mmlist_lock;
181
182typedef struct task_struct task_t;
183
184extern void sched_init(void);
185extern void sched_init_smp(void);
186extern void init_idle(task_t *idle, int cpu);
187
188extern cpumask_t nohz_cpu_mask;
189
190extern void show_state(void);
191extern void show_regs(struct pt_regs *);
192
193/*
194 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
195 * task), SP is the stack pointer of the first frame that should be shown in the back
196 * trace (or NULL if the entire call-chain of the task should be shown).
197 */
198extern void show_stack(struct task_struct *task, unsigned long *sp);
199
200void io_schedule(void);
201long io_schedule_timeout(long timeout);
202
203extern void cpu_init (void);
204extern void trap_init(void);
205extern void update_process_times(int user);
206extern void scheduler_tick(void);
207
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700208#ifdef CONFIG_DETECT_SOFTLOCKUP
Ingo Molnar6687a972006-03-24 03:18:41 -0800209extern void softlockup_tick(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700210extern void spawn_softlockup_task(void);
211extern void touch_softlockup_watchdog(void);
212#else
Ingo Molnar6687a972006-03-24 03:18:41 -0800213static inline void softlockup_tick(void)
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700214{
215}
216static inline void spawn_softlockup_task(void)
217{
218}
219static inline void touch_softlockup_watchdog(void)
220{
221}
222#endif
223
224
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225/* Attach to any functions which should be ignored in wchan output. */
226#define __sched __attribute__((__section__(".sched.text")))
227/* Is this address in the __sched functions? */
228extern int in_sched_functions(unsigned long addr);
229
230#define MAX_SCHEDULE_TIMEOUT LONG_MAX
231extern signed long FASTCALL(schedule_timeout(signed long timeout));
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700232extern signed long schedule_timeout_interruptible(signed long timeout);
233extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700234asmlinkage void schedule(void);
235
236struct namespace;
237
238/* Maximum number of active map areas.. This is a random (large) number */
239#define DEFAULT_MAX_MAP_COUNT 65536
240
241extern int sysctl_max_map_count;
242
243#include <linux/aio.h>
244
245extern unsigned long
246arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
247 unsigned long, unsigned long);
248extern unsigned long
249arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
250 unsigned long len, unsigned long pgoff,
251 unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700252extern void arch_unmap_area(struct mm_struct *, unsigned long);
253extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700254
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700255#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
256/*
257 * The mm counters are not protected by its page_table_lock,
258 * so must be incremented atomically.
259 */
Christoph Lameterd3cb4872006-01-06 00:11:20 -0800260#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
261#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
262#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
263#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
264#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
265typedef atomic_long_t mm_counter_t;
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700266
267#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
268/*
269 * The mm counters are protected by its page_table_lock,
270 * so can be incremented directly.
271 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700272#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
273#define get_mm_counter(mm, member) ((mm)->_##member)
274#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
275#define inc_mm_counter(mm, member) (mm)->_##member++
276#define dec_mm_counter(mm, member) (mm)->_##member--
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700277typedef unsigned long mm_counter_t;
Hugh Dickins42946212005-10-29 18:16:05 -0700278
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700279#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
280
281#define get_mm_rss(mm) \
282 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
Hugh Dickins365e9c872005-10-29 18:16:18 -0700283#define update_hiwater_rss(mm) do { \
284 unsigned long _rss = get_mm_rss(mm); \
285 if ((mm)->hiwater_rss < _rss) \
286 (mm)->hiwater_rss = _rss; \
287} while (0)
288#define update_hiwater_vm(mm) do { \
289 if ((mm)->hiwater_vm < (mm)->total_vm) \
290 (mm)->hiwater_vm = (mm)->total_vm; \
291} while (0)
292
Linus Torvalds1da177e2005-04-16 15:20:36 -0700293struct mm_struct {
294 struct vm_area_struct * mmap; /* list of VMAs */
295 struct rb_root mm_rb;
296 struct vm_area_struct * mmap_cache; /* last find_vma result */
297 unsigned long (*get_unmapped_area) (struct file *filp,
298 unsigned long addr, unsigned long len,
299 unsigned long pgoff, unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700300 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
Benjamin Herrenschmidt0551fbd2006-02-28 16:59:19 -0800301 unsigned long mmap_base; /* base of mmap area */
302 unsigned long task_size; /* size of task vm space */
303 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700304 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305 pgd_t * pgd;
306 atomic_t mm_users; /* How many users with user space? */
307 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
308 int map_count; /* number of VMAs */
309 struct rw_semaphore mmap_sem;
310 spinlock_t page_table_lock; /* Protects page tables and some counters */
311
312 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
313 * together off init_mm.mmlist, and are protected
314 * by mmlist_lock
315 */
316
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700317 /* Special counters, in some configurations protected by the
318 * page_table_lock, in other configurations by being atomic.
319 */
Hugh Dickins42946212005-10-29 18:16:05 -0700320 mm_counter_t _file_rss;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321 mm_counter_t _anon_rss;
322
Hugh Dickinsf4499522005-10-29 18:16:19 -0700323 unsigned long hiwater_rss; /* High-watermark of RSS usage */
324 unsigned long hiwater_vm; /* High-water virtual memory usage */
325
326 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
327 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
328 unsigned long start_code, end_code, start_data, end_data;
329 unsigned long start_brk, brk, start_stack;
330 unsigned long arg_start, arg_end, env_start, env_end;
331
H. J. Lu36d57ac2005-09-06 15:16:49 -0700332 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700333
Alan Coxd6e71142005-06-23 00:09:43 -0700334 unsigned dumpable:2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700335 cpumask_t cpu_vm_mask;
336
337 /* Architecture-specific MM context */
338 mm_context_t context;
339
340 /* Token based thrashing protection. */
341 unsigned long swap_token_time;
342 char recent_pagein;
343
344 /* coredumping support */
345 int core_waiters;
346 struct completion *core_startup_done, core_done;
347
348 /* aio bits */
349 rwlock_t ioctx_list_lock;
350 struct kioctx *ioctx_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700351};
352
353struct sighand_struct {
354 atomic_t count;
355 struct k_sigaction action[_NSIG];
356 spinlock_t siglock;
Ingo Molnare56d0902006-01-08 01:01:37 -0800357 struct rcu_head rcu;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700358};
359
Ingo Molnare56d0902006-01-08 01:01:37 -0800360extern void sighand_free_cb(struct rcu_head *rhp);
361
362static inline void sighand_free(struct sighand_struct *sp)
363{
364 call_rcu(&sp->rcu, sighand_free_cb);
365}
366
Linus Torvalds1da177e2005-04-16 15:20:36 -0700367/*
368 * NOTE! "signal_struct" does not have it's own
369 * locking, because a shared signal_struct always
370 * implies a shared sighand_struct, so locking
371 * sighand_struct is always a proper superset of
372 * the locking of signal_struct.
373 */
374struct signal_struct {
375 atomic_t count;
376 atomic_t live;
377
378 wait_queue_head_t wait_chldexit; /* for wait4() */
379
380 /* current thread group signal load-balancing target: */
381 task_t *curr_target;
382
383 /* shared signal handling: */
384 struct sigpending shared_pending;
385
386 /* thread group exit support */
387 int group_exit_code;
388 /* overloaded:
389 * - notify group_exit_task when ->count is equal to notify_count
390 * - everyone except group_exit_task is stopped during signal delivery
391 * of fatal signals, group_exit_task processes the signal.
392 */
393 struct task_struct *group_exit_task;
394 int notify_count;
395
396 /* thread group stop support, overloads group_exit_code too */
397 int group_stop_count;
398 unsigned int flags; /* see SIGNAL_* flags below */
399
400 /* POSIX.1b Interval Timers */
401 struct list_head posix_timers;
402
403 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800404 struct hrtimer real_timer;
405 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700406
407 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
408 cputime_t it_prof_expires, it_virt_expires;
409 cputime_t it_prof_incr, it_virt_incr;
410
411 /* job control IDs */
412 pid_t pgrp;
413 pid_t tty_old_pgrp;
414 pid_t session;
415 /* boolean value for session group leader */
416 int leader;
417
418 struct tty_struct *tty; /* NULL if no tty */
419
420 /*
421 * Cumulative resource counters for dead threads in the group,
422 * and for reaped dead child processes forked by this group.
423 * Live threads maintain their own counters and add to these
424 * in __exit_signal, except for the group leader.
425 */
426 cputime_t utime, stime, cutime, cstime;
427 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
428 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
429
430 /*
431 * Cumulative ns of scheduled CPU time for dead threads in the
432 * group, not including a zombie group leader. (This only differs
433 * from jiffies_to_ns(utime + stime) if sched_clock uses something
434 * other than jiffies.)
435 */
436 unsigned long long sched_time;
437
438 /*
439 * We don't bother to synchronize most readers of this at all,
440 * because there is no reader checking a limit that actually needs
441 * to get both rlim_cur and rlim_max atomically, and either one
442 * alone is a single word that can safely be read normally.
443 * getrlimit/setrlimit use task_lock(current->group_leader) to
444 * protect this instead of the siglock, because they really
445 * have no need to disable irqs.
446 */
447 struct rlimit rlim[RLIM_NLIMITS];
448
449 struct list_head cpu_timers[3];
450
451 /* keep the process-shared keyrings here so that they do the right
452 * thing in threads created with CLONE_THREAD */
453#ifdef CONFIG_KEYS
454 struct key *session_keyring; /* keyring inherited over fork */
455 struct key *process_keyring; /* keyring private to this process */
456#endif
457};
458
Nick Piggin4866cde2005-06-25 14:57:23 -0700459/* Context switch must be unlocked if interrupts are to be enabled */
460#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
461# define __ARCH_WANT_UNLOCKED_CTXSW
462#endif
463
Linus Torvalds1da177e2005-04-16 15:20:36 -0700464/*
465 * Bits in flags field of signal_struct.
466 */
467#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
468#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
469#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
470#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
471
472
473/*
474 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
Ingo Molnarb0a94992006-01-14 13:20:41 -0800475 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
476 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
477 * values are inverted: lower p->prio value means higher priority.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700478 *
479 * The MAX_USER_RT_PRIO value allows the actual maximum
480 * RT priority to be separate from the value exported to
481 * user-space. This allows kernel threads to set their
482 * priority to a value higher than any user task. Note:
483 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
484 */
485
486#define MAX_USER_RT_PRIO 100
487#define MAX_RT_PRIO MAX_USER_RT_PRIO
488
489#define MAX_PRIO (MAX_RT_PRIO + 40)
490
491#define rt_task(p) (unlikely((p)->prio < MAX_RT_PRIO))
492
493/*
494 * Some day this will be a full-fledged user tracking system..
495 */
496struct user_struct {
497 atomic_t __count; /* reference count */
498 atomic_t processes; /* How many processes does this user have? */
499 atomic_t files; /* How many open files does this user have? */
500 atomic_t sigpending; /* How many pending signals does this user have? */
Robert Love0eeca282005-07-12 17:06:03 -0400501#ifdef CONFIG_INOTIFY
502 atomic_t inotify_watches; /* How many inotify watches does this user have? */
503 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
504#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700505 /* protected by mq_lock */
506 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
507 unsigned long locked_shm; /* How many pages of mlocked shm ? */
508
509#ifdef CONFIG_KEYS
510 struct key *uid_keyring; /* UID specific keyring */
511 struct key *session_keyring; /* UID's default session keyring */
512#endif
513
514 /* Hash table maintenance information */
515 struct list_head uidhash_list;
516 uid_t uid;
517};
518
519extern struct user_struct *find_user(uid_t);
520
521extern struct user_struct root_user;
522#define INIT_USER (&root_user)
523
524typedef struct prio_array prio_array_t;
525struct backing_dev_info;
526struct reclaim_state;
527
528#ifdef CONFIG_SCHEDSTATS
529struct sched_info {
530 /* cumulative counters */
531 unsigned long cpu_time, /* time spent on the cpu */
532 run_delay, /* time spent waiting on a runqueue */
533 pcnt; /* # of timeslices run on this cpu */
534
535 /* timestamps */
536 unsigned long last_arrival, /* when we last ran on a cpu */
537 last_queued; /* when we were last queued to run */
538};
539
540extern struct file_operations proc_schedstat_operations;
541#endif
542
543enum idle_type
544{
545 SCHED_IDLE,
546 NOT_IDLE,
547 NEWLY_IDLE,
548 MAX_IDLE_TYPES
549};
550
551/*
552 * sched-domains (multiprocessor balancing) declarations:
553 */
554#ifdef CONFIG_SMP
555#define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
556
557#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
558#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
559#define SD_BALANCE_EXEC 4 /* Balance on exec */
Nick Piggin147cbb42005-06-25 14:57:19 -0700560#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
561#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
562#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
563#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
564#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700565
566struct sched_group {
567 struct sched_group *next; /* Must be a circular list */
568 cpumask_t cpumask;
569
570 /*
571 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
572 * single CPU. This is read only (except for setup, hotplug CPU).
573 */
574 unsigned long cpu_power;
575};
576
577struct sched_domain {
578 /* These fields must be setup */
579 struct sched_domain *parent; /* top domain must be null terminated */
580 struct sched_group *groups; /* the balancing groups of the domain */
581 cpumask_t span; /* span of all CPUs in this domain */
582 unsigned long min_interval; /* Minimum balance interval ms */
583 unsigned long max_interval; /* Maximum balance interval ms */
584 unsigned int busy_factor; /* less balancing by factor if busy */
585 unsigned int imbalance_pct; /* No balance until over watermark */
586 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
587 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
588 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
Nick Piggin78979862005-06-25 14:57:13 -0700589 unsigned int busy_idx;
590 unsigned int idle_idx;
591 unsigned int newidle_idx;
592 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700593 unsigned int forkexec_idx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700594 int flags; /* See SD_* */
595
596 /* Runtime fields. */
597 unsigned long last_balance; /* init to jiffies. units in jiffies */
598 unsigned int balance_interval; /* initialise to 1. units in ms. */
599 unsigned int nr_balance_failed; /* initialise to 0 */
600
601#ifdef CONFIG_SCHEDSTATS
602 /* load_balance() stats */
603 unsigned long lb_cnt[MAX_IDLE_TYPES];
604 unsigned long lb_failed[MAX_IDLE_TYPES];
605 unsigned long lb_balanced[MAX_IDLE_TYPES];
606 unsigned long lb_imbalance[MAX_IDLE_TYPES];
607 unsigned long lb_gained[MAX_IDLE_TYPES];
608 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
609 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
610 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
611
612 /* Active load balancing */
613 unsigned long alb_cnt;
614 unsigned long alb_failed;
615 unsigned long alb_pushed;
616
Nick Piggin68767a02005-06-25 14:57:20 -0700617 /* SD_BALANCE_EXEC stats */
618 unsigned long sbe_cnt;
619 unsigned long sbe_balanced;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700620 unsigned long sbe_pushed;
621
Nick Piggin68767a02005-06-25 14:57:20 -0700622 /* SD_BALANCE_FORK stats */
623 unsigned long sbf_cnt;
624 unsigned long sbf_balanced;
625 unsigned long sbf_pushed;
626
Linus Torvalds1da177e2005-04-16 15:20:36 -0700627 /* try_to_wake_up() stats */
628 unsigned long ttwu_wake_remote;
629 unsigned long ttwu_move_affine;
630 unsigned long ttwu_move_balance;
631#endif
632};
633
Dinakar Guniguntala1a20ff22005-06-25 14:57:33 -0700634extern void partition_sched_domains(cpumask_t *partition1,
635 cpumask_t *partition2);
akpm@osdl.org198e2f12006-01-12 01:05:30 -0800636
637/*
638 * Maximum cache size the migration-costs auto-tuning code will
639 * search from:
640 */
641extern unsigned int max_cache_size;
642
643#endif /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700644
645
646struct io_context; /* See blkdev.h */
647void exit_io_context(void);
648struct cpuset;
649
650#define NGROUPS_SMALL 32
651#define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
652struct group_info {
653 int ngroups;
654 atomic_t usage;
655 gid_t small_block[NGROUPS_SMALL];
656 int nblocks;
657 gid_t *blocks[0];
658};
659
660/*
661 * get_group_info() must be called with the owning task locked (via task_lock())
662 * when task != current. The reason being that the vast majority of callers are
663 * looking at current->group_info, which can not be changed except by the
664 * current task. Changing current->group_info requires the task lock, too.
665 */
666#define get_group_info(group_info) do { \
667 atomic_inc(&(group_info)->usage); \
668} while (0)
669
670#define put_group_info(group_info) do { \
671 if (atomic_dec_and_test(&(group_info)->usage)) \
672 groups_free(group_info); \
673} while (0)
674
David Howells3e301482005-06-23 22:00:56 -0700675extern struct group_info *groups_alloc(int gidsetsize);
676extern void groups_free(struct group_info *group_info);
677extern int set_current_groups(struct group_info *group_info);
678extern int groups_search(struct group_info *group_info, gid_t grp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700679/* access the groups "array" with this macro */
680#define GROUP_AT(gi, i) \
681 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
682
Chen, Kenneth W383f2832005-09-09 13:02:02 -0700683#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
684extern void prefetch_stack(struct task_struct*);
685#else
686static inline void prefetch_stack(struct task_struct *t) { }
687#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700688
689struct audit_context; /* See audit.c */
690struct mempolicy;
691
692struct task_struct {
693 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
694 struct thread_info *thread_info;
695 atomic_t usage;
696 unsigned long flags; /* per process flags, defined below */
697 unsigned long ptrace;
698
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700699 int lock_depth; /* BKL lock depth */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700700
Chen, Kenneth Wd6077cb2006-02-14 13:53:10 -0800701#if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
Nick Piggin4866cde2005-06-25 14:57:23 -0700702 int oncpu;
703#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700704 int prio, static_prio;
705 struct list_head run_list;
706 prio_array_t *array;
707
Jens Axboe22e2c502005-06-27 10:55:12 +0200708 unsigned short ioprio;
Jens Axboe2056a782006-03-23 20:00:26 +0100709 unsigned int btrace_seq;
Jens Axboe22e2c502005-06-27 10:55:12 +0200710
Linus Torvalds1da177e2005-04-16 15:20:36 -0700711 unsigned long sleep_avg;
712 unsigned long long timestamp, last_ran;
713 unsigned long long sched_time; /* sched_clock time spent running */
714 int activated;
715
716 unsigned long policy;
717 cpumask_t cpus_allowed;
718 unsigned int time_slice, first_time_slice;
719
720#ifdef CONFIG_SCHEDSTATS
721 struct sched_info sched_info;
722#endif
723
724 struct list_head tasks;
725 /*
726 * ptrace_list/ptrace_children forms the list of my children
727 * that were stolen by a ptracer.
728 */
729 struct list_head ptrace_children;
730 struct list_head ptrace_list;
731
732 struct mm_struct *mm, *active_mm;
733
734/* task state */
735 struct linux_binfmt *binfmt;
736 long exit_state;
737 int exit_code, exit_signal;
738 int pdeath_signal; /* The signal sent when the parent dies */
739 /* ??? */
740 unsigned long personality;
741 unsigned did_exec:1;
742 pid_t pid;
743 pid_t tgid;
744 /*
745 * pointers to (original) parent process, youngest child, younger sibling,
746 * older sibling, respectively. (p->father can be replaced with
747 * p->parent->pid)
748 */
749 struct task_struct *real_parent; /* real parent process (when being debugged) */
750 struct task_struct *parent; /* parent process */
751 /*
752 * children/sibling forms the list of my children plus the
753 * tasks I'm ptracing.
754 */
755 struct list_head children; /* list of my children */
756 struct list_head sibling; /* linkage in my parent's children list */
757 struct task_struct *group_leader; /* threadgroup leader */
758
759 /* PID/PID hash table linkage. */
760 struct pid pids[PIDTYPE_MAX];
761
762 struct completion *vfork_done; /* for vfork() */
763 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
764 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
765
766 unsigned long rt_priority;
767 cputime_t utime, stime;
768 unsigned long nvcsw, nivcsw; /* context switch counts */
769 struct timespec start_time;
770/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
771 unsigned long min_flt, maj_flt;
772
773 cputime_t it_prof_expires, it_virt_expires;
774 unsigned long long it_sched_expires;
775 struct list_head cpu_timers[3];
776
777/* process credentials */
778 uid_t uid,euid,suid,fsuid;
779 gid_t gid,egid,sgid,fsgid;
780 struct group_info *group_info;
781 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
782 unsigned keep_capabilities:1;
783 struct user_struct *user;
784#ifdef CONFIG_KEYS
David Howellsb5f545c2006-01-08 01:02:47 -0800785 struct key *request_key_auth; /* assumed request_key authority */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786 struct key *thread_keyring; /* keyring private to this thread */
David Howells3e301482005-06-23 22:00:56 -0700787 unsigned char jit_keyring; /* default keyring to attach requested keys to */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700788#endif
789 int oomkilladj; /* OOM kill score adjustment (bit shift). */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700790 char comm[TASK_COMM_LEN]; /* executable name excluding path
791 - access with [gs]et_task_comm (which lock
792 it with task_lock())
793 - initialized normally by flush_old_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700794/* file system info */
795 int link_count, total_link_count;
796/* ipc stuff */
797 struct sysv_sem sysvsem;
798/* CPU-specific state of this task */
799 struct thread_struct thread;
800/* filesystem information */
801 struct fs_struct *fs;
802/* open file information */
803 struct files_struct *files;
804/* namespace */
805 struct namespace *namespace;
806/* signal handlers */
807 struct signal_struct *signal;
808 struct sighand_struct *sighand;
809
810 sigset_t blocked, real_blocked;
David Woodhouse150256d2006-01-18 17:43:57 -0800811 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700812 struct sigpending pending;
813
814 unsigned long sas_ss_sp;
815 size_t sas_ss_size;
816 int (*notifier)(void *priv);
817 void *notifier_data;
818 sigset_t *notifier_mask;
819
820 void *security;
821 struct audit_context *audit_context;
822 seccomp_t seccomp;
823
824/* Thread group tracking */
825 u32 parent_exec_id;
826 u32 self_exec_id;
827/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
828 spinlock_t alloc_lock;
829/* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
830 spinlock_t proc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700831
Ingo Molnar408894e2006-01-09 15:59:20 -0800832#ifdef CONFIG_DEBUG_MUTEXES
833 /* mutex deadlock detection */
834 struct mutex_waiter *blocked_on;
835#endif
836
Linus Torvalds1da177e2005-04-16 15:20:36 -0700837/* journalling filesystem info */
838 void *journal_info;
839
840/* VM state */
841 struct reclaim_state *reclaim_state;
842
843 struct dentry *proc_dentry;
844 struct backing_dev_info *backing_dev_info;
845
846 struct io_context *io_context;
847
848 unsigned long ptrace_message;
849 siginfo_t *last_siginfo; /* For ptrace use. */
850/*
851 * current io wait handle: wait queue entry to use for io waits
852 * If this thread is processing aio, this points at the waitqueue
853 * inside the currently handled kiocb. It may be NULL (i.e. default
854 * to a stack based synchronous wait) if its doing sync IO.
855 */
856 wait_queue_t *io_wait;
857/* i/o counters(bytes read/written, #syscalls */
858 u64 rchar, wchar, syscr, syscw;
859#if defined(CONFIG_BSD_PROCESS_ACCT)
860 u64 acct_rss_mem1; /* accumulated rss usage */
861 u64 acct_vm_mem1; /* accumulated virtual memory usage */
862 clock_t acct_stimexpd; /* clock_t-converted stime since last update */
863#endif
864#ifdef CONFIG_NUMA
865 struct mempolicy *mempolicy;
866 short il_next;
867#endif
868#ifdef CONFIG_CPUSETS
869 struct cpuset *cpuset;
870 nodemask_t mems_allowed;
871 int cpuset_mems_generation;
Paul Jackson825a46a2006-03-24 03:16:03 -0800872 int cpuset_mem_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700873#endif
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);
1097extern int kill_sl(pid_t, int, int);
1098extern int kill_proc(pid_t, int, int);
1099extern struct sigqueue *sigqueue_alloc(void);
1100extern void sigqueue_free(struct sigqueue *);
1101extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1102extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03001103extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001104extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1105
1106/* These can be the second arg to send_sig_info/send_group_sig_info. */
1107#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1108#define SEND_SIG_PRIV ((struct siginfo *) 1)
1109#define SEND_SIG_FORCED ((struct siginfo *) 2)
1110
Oleg Nesterov621d3122005-10-30 15:03:45 -08001111static inline int is_si_special(const struct siginfo *info)
1112{
1113 return info <= SEND_SIG_FORCED;
1114}
1115
Linus Torvalds1da177e2005-04-16 15:20:36 -07001116/* True if we are on the alternate signal stack. */
1117
1118static inline int on_sig_stack(unsigned long sp)
1119{
1120 return (sp - current->sas_ss_sp < current->sas_ss_size);
1121}
1122
1123static inline int sas_ss_flags(unsigned long sp)
1124{
1125 return (current->sas_ss_size == 0 ? SS_DISABLE
1126 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1127}
1128
Linus Torvalds1da177e2005-04-16 15:20:36 -07001129/*
1130 * Routines for handling mm_structs
1131 */
1132extern struct mm_struct * mm_alloc(void);
1133
1134/* mmdrop drops the mm and the page tables */
1135extern void FASTCALL(__mmdrop(struct mm_struct *));
1136static inline void mmdrop(struct mm_struct * mm)
1137{
1138 if (atomic_dec_and_test(&mm->mm_count))
1139 __mmdrop(mm);
1140}
1141
1142/* mmput gets rid of the mappings and all user-space */
1143extern void mmput(struct mm_struct *);
1144/* Grab a reference to a task's mm, if it is not already going away */
1145extern struct mm_struct *get_task_mm(struct task_struct *task);
1146/* Remove the current tasks stale references to the old mm_struct */
1147extern void mm_release(struct task_struct *, struct mm_struct *);
1148
1149extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1150extern void flush_thread(void);
1151extern void exit_thread(void);
1152
Linus Torvalds1da177e2005-04-16 15:20:36 -07001153extern void exit_files(struct task_struct *);
1154extern void exit_signal(struct task_struct *);
1155extern void __exit_signal(struct task_struct *);
1156extern void exit_sighand(struct task_struct *);
1157extern void __exit_sighand(struct task_struct *);
1158extern void exit_itimers(struct signal_struct *);
1159
1160extern NORET_TYPE void do_group_exit(int);
1161
Linus Torvalds1da177e2005-04-16 15:20:36 -07001162extern void daemonize(const char *, ...);
1163extern int allow_signal(int);
1164extern int disallow_signal(int);
1165extern task_t *child_reaper;
1166
1167extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1168extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1169task_t *fork_idle(int);
1170
1171extern void set_task_comm(struct task_struct *tsk, char *from);
1172extern void get_task_comm(char *to, struct task_struct *tsk);
1173
1174#ifdef CONFIG_SMP
1175extern void wait_task_inactive(task_t * p);
1176#else
1177#define wait_task_inactive(p) do { } while (0)
1178#endif
1179
1180#define remove_parent(p) list_del_init(&(p)->sibling)
1181#define add_parent(p, parent) list_add_tail(&(p)->sibling,&(parent)->children)
1182
1183#define REMOVE_LINKS(p) do { \
1184 if (thread_group_leader(p)) \
1185 list_del_init(&(p)->tasks); \
1186 remove_parent(p); \
1187 } while (0)
1188
1189#define SET_LINKS(p) do { \
1190 if (thread_group_leader(p)) \
1191 list_add_tail(&(p)->tasks,&init_task.tasks); \
1192 add_parent(p, (p)->parent); \
1193 } while (0)
1194
1195#define next_task(p) list_entry((p)->tasks.next, struct task_struct, tasks)
1196#define prev_task(p) list_entry((p)->tasks.prev, struct task_struct, tasks)
1197
1198#define for_each_process(p) \
1199 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1200
1201/*
1202 * Careful: do_each_thread/while_each_thread is a double loop so
1203 * 'break' will not work as expected - use goto instead.
1204 */
1205#define do_each_thread(g, t) \
1206 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1207
1208#define while_each_thread(g, t) \
1209 while ((t = next_thread(t)) != g)
1210
1211extern task_t * FASTCALL(next_thread(const task_t *p));
1212
1213#define thread_group_leader(p) (p->pid == p->tgid)
1214
1215static inline int thread_group_empty(task_t *p)
1216{
1217 return list_empty(&p->pids[PIDTYPE_TGID].pid_list);
1218}
1219
1220#define delay_group_leader(p) \
1221 (thread_group_leader(p) && !thread_group_empty(p))
1222
1223extern void unhash_process(struct task_struct *p);
1224
1225/*
1226 * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02001227 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Jackson053199e2005-10-30 15:02:30 -08001228 * pins the final release of task.io_context. Also protects ->cpuset.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001229 *
1230 * Nests both inside and outside of read_lock(&tasklist_lock).
1231 * It must not be nested with write_lock_irq(&tasklist_lock),
1232 * neither inside nor outside.
1233 */
1234static inline void task_lock(struct task_struct *p)
1235{
1236 spin_lock(&p->alloc_lock);
1237}
1238
1239static inline void task_unlock(struct task_struct *p)
1240{
1241 spin_unlock(&p->alloc_lock);
1242}
1243
Al Virof0373602005-11-13 16:06:57 -08001244#ifndef __HAVE_THREAD_FUNCTIONS
1245
Al Viroa1261f52005-11-13 16:06:55 -08001246#define task_thread_info(task) (task)->thread_info
Al Viro9fc65872006-01-12 01:05:34 -08001247#define task_stack_page(task) ((void*)((task)->thread_info))
Al Viroa1261f52005-11-13 16:06:55 -08001248
Al Viro10ebffd2005-11-13 16:06:56 -08001249static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1250{
1251 *task_thread_info(p) = *task_thread_info(org);
1252 task_thread_info(p)->task = p;
1253}
1254
1255static inline unsigned long *end_of_stack(struct task_struct *p)
1256{
1257 return (unsigned long *)(p->thread_info + 1);
1258}
1259
Al Virof0373602005-11-13 16:06:57 -08001260#endif
1261
Linus Torvalds1da177e2005-04-16 15:20:36 -07001262/* set thread flags in other task's structures
1263 * - see asm/thread_info.h for TIF_xxxx flags available
1264 */
1265static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1266{
Al Viroa1261f52005-11-13 16:06:55 -08001267 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001268}
1269
1270static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1271{
Al Viroa1261f52005-11-13 16:06:55 -08001272 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001273}
1274
1275static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1276{
Al Viroa1261f52005-11-13 16:06:55 -08001277 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001278}
1279
1280static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1281{
Al Viroa1261f52005-11-13 16:06:55 -08001282 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001283}
1284
1285static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1286{
Al Viroa1261f52005-11-13 16:06:55 -08001287 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001288}
1289
1290static inline void set_tsk_need_resched(struct task_struct *tsk)
1291{
1292 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1293}
1294
1295static inline void clear_tsk_need_resched(struct task_struct *tsk)
1296{
1297 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1298}
1299
1300static inline int signal_pending(struct task_struct *p)
1301{
1302 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1303}
1304
1305static inline int need_resched(void)
1306{
1307 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1308}
1309
1310/*
1311 * cond_resched() and cond_resched_lock(): latency reduction via
1312 * explicit rescheduling in places that are safe. The return
1313 * value indicates whether a reschedule was done in fact.
1314 * cond_resched_lock() will drop the spinlock before scheduling,
1315 * cond_resched_softirq() will enable bhs before scheduling.
1316 */
1317extern int cond_resched(void);
1318extern int cond_resched_lock(spinlock_t * lock);
1319extern int cond_resched_softirq(void);
1320
1321/*
1322 * Does a critical section need to be broken due to another
1323 * task waiting?:
1324 */
1325#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1326# define need_lockbreak(lock) ((lock)->break_lock)
1327#else
1328# define need_lockbreak(lock) 0
1329#endif
1330
1331/*
1332 * Does a critical section need to be broken due to another
1333 * task waiting or preemption being signalled:
1334 */
1335static inline int lock_need_resched(spinlock_t *lock)
1336{
1337 if (need_lockbreak(lock) || need_resched())
1338 return 1;
1339 return 0;
1340}
1341
1342/* Reevaluate whether the task has signals pending delivery.
1343 This is required every time the blocked sigset_t changes.
1344 callers must hold sighand->siglock. */
1345
1346extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1347extern void recalc_sigpending(void);
1348
1349extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1350
1351/*
1352 * Wrappers for p->thread_info->cpu access. No-op on UP.
1353 */
1354#ifdef CONFIG_SMP
1355
1356static inline unsigned int task_cpu(const struct task_struct *p)
1357{
Al Viroa1261f52005-11-13 16:06:55 -08001358 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001359}
1360
1361static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1362{
Al Viroa1261f52005-11-13 16:06:55 -08001363 task_thread_info(p)->cpu = cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001364}
1365
1366#else
1367
1368static inline unsigned int task_cpu(const struct task_struct *p)
1369{
1370 return 0;
1371}
1372
1373static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1374{
1375}
1376
1377#endif /* CONFIG_SMP */
1378
1379#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1380extern void arch_pick_mmap_layout(struct mm_struct *mm);
1381#else
1382static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1383{
1384 mm->mmap_base = TASK_UNMAPPED_BASE;
1385 mm->get_unmapped_area = arch_get_unmapped_area;
1386 mm->unmap_area = arch_unmap_area;
1387}
1388#endif
1389
1390extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1391extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1392
Linus Torvalds1da177e2005-04-16 15:20:36 -07001393extern void normalize_rt_tasks(void);
1394
Linus Torvalds1da177e2005-04-16 15:20:36 -07001395#ifdef CONFIG_PM
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001396/*
1397 * Check if a process has been frozen
1398 */
1399static inline int frozen(struct task_struct *p)
1400{
1401 return p->flags & PF_FROZEN;
1402}
1403
1404/*
1405 * Check if there is a request to freeze a process
1406 */
1407static inline int freezing(struct task_struct *p)
1408{
1409 return p->flags & PF_FREEZE;
1410}
1411
1412/*
1413 * Request that a process be frozen
1414 * FIXME: SMP problem. We may not modify other process' flags!
1415 */
1416static inline void freeze(struct task_struct *p)
1417{
1418 p->flags |= PF_FREEZE;
1419}
1420
1421/*
1422 * Wake up a frozen process
1423 */
1424static inline int thaw_process(struct task_struct *p)
1425{
1426 if (frozen(p)) {
1427 p->flags &= ~PF_FROZEN;
1428 wake_up_process(p);
1429 return 1;
1430 }
1431 return 0;
1432}
1433
1434/*
1435 * freezing is complete, mark process as frozen
1436 */
1437static inline void frozen_process(struct task_struct *p)
1438{
1439 p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1440}
1441
1442extern void refrigerator(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001443extern int freeze_processes(void);
1444extern void thaw_processes(void);
1445
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001446static inline int try_to_freeze(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001447{
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001448 if (freezing(current)) {
1449 refrigerator();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001450 return 1;
1451 } else
1452 return 0;
1453}
1454#else
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001455static inline int frozen(struct task_struct *p) { return 0; }
1456static inline int freezing(struct task_struct *p) { return 0; }
1457static inline void freeze(struct task_struct *p) { BUG(); }
1458static inline int thaw_process(struct task_struct *p) { return 1; }
1459static inline void frozen_process(struct task_struct *p) { BUG(); }
1460
1461static inline void refrigerator(void) {}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001462static inline int freeze_processes(void) { BUG(); return 0; }
1463static inline void thaw_processes(void) {}
1464
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001465static inline int try_to_freeze(void) { return 0; }
1466
Linus Torvalds1da177e2005-04-16 15:20:36 -07001467#endif /* CONFIG_PM */
1468#endif /* __KERNEL__ */
1469
1470#endif