blob: 38b4791e6a5da17cd527054dcdc4e4cda000db5b [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
David Woodhouseb7b3c762006-04-27 00:12:56 +01004#include <linux/auxvec.h> /* For AT_VECTOR_SIZE */
5
6/*
7 * cloning flags:
8 */
9#define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
10#define CLONE_VM 0x00000100 /* set if VM shared between processes */
11#define CLONE_FS 0x00000200 /* set if fs info shared between processes */
12#define CLONE_FILES 0x00000400 /* set if open files shared between processes */
13#define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
14#define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
15#define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
16#define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
17#define CLONE_THREAD 0x00010000 /* Same thread group? */
18#define CLONE_NEWNS 0x00020000 /* New namespace group? */
19#define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
20#define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
21#define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
22#define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
23#define CLONE_DETACHED 0x00400000 /* Unused, ignored */
24#define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
25#define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
26#define CLONE_STOPPED 0x02000000 /* Start in stopped state */
27
28/*
29 * Scheduling policies
30 */
31#define SCHED_NORMAL 0
32#define SCHED_FIFO 1
33#define SCHED_RR 2
34#define SCHED_BATCH 3
35
David Woodhousea3b67142006-04-25 14:54:40 +010036#ifdef __KERNEL__
David Woodhouseb7b3c762006-04-27 00:12:56 +010037
38struct sched_param {
39 int sched_priority;
40};
41
Linus Torvalds1da177e2005-04-16 15:20:36 -070042#include <asm/param.h> /* for HZ */
43
Linus Torvalds1da177e2005-04-16 15:20:36 -070044#include <linux/capability.h>
45#include <linux/threads.h>
46#include <linux/kernel.h>
47#include <linux/types.h>
48#include <linux/timex.h>
49#include <linux/jiffies.h>
50#include <linux/rbtree.h>
51#include <linux/thread_info.h>
52#include <linux/cpumask.h>
53#include <linux/errno.h>
54#include <linux/nodemask.h>
55
56#include <asm/system.h>
57#include <asm/semaphore.h>
58#include <asm/page.h>
59#include <asm/ptrace.h>
60#include <asm/mmu.h>
61#include <asm/cputime.h>
62
63#include <linux/smp.h>
64#include <linux/sem.h>
65#include <linux/signal.h>
66#include <linux/securebits.h>
67#include <linux/fs_struct.h>
68#include <linux/compiler.h>
69#include <linux/completion.h>
70#include <linux/pid.h>
71#include <linux/percpu.h>
72#include <linux/topology.h>
73#include <linux/seccomp.h>
Ingo Molnare56d0902006-01-08 01:01:37 -080074#include <linux/rcupdate.h>
Ingo Molnar0771dfe2006-03-27 01:16:22 -080075#include <linux/futex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070076
David Woodhousea3b67142006-04-25 14:54:40 +010077#include <linux/time.h>
78#include <linux/param.h>
79#include <linux/resource.h>
80#include <linux/timer.h>
81#include <linux/hrtimer.h>
82
83#include <asm/processor.h>
H. J. Lu36d57ac2005-09-06 15:16:49 -070084
Linus Torvalds1da177e2005-04-16 15:20:36 -070085struct exec_domain;
86
87/*
Linus Torvalds1da177e2005-04-16 15:20:36 -070088 * List of flags we want to share for kernel threads,
89 * if only because they are not used by them anyway.
90 */
91#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
92
93/*
94 * These are the constant used to fake the fixed-point load-average
95 * counting. Some notes:
96 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
97 * a load-average precision of 10 bits integer + 11 bits fractional
98 * - if you want to count load-averages more often, you need more
99 * precision, or rounding will get you. With 2-second counting freq,
100 * the EXP_n values would be 1981, 2034 and 2043 if still using only
101 * 11 bit fractions.
102 */
103extern unsigned long avenrun[]; /* Load averages */
104
105#define FSHIFT 11 /* nr of bits of precision */
106#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
107#define LOAD_FREQ (5*HZ) /* 5 sec intervals */
108#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
109#define EXP_5 2014 /* 1/exp(5sec/5min) */
110#define EXP_15 2037 /* 1/exp(5sec/15min) */
111
112#define CALC_LOAD(load,exp,n) \
113 load *= exp; \
114 load += n*(FIXED_1-exp); \
115 load >>= FSHIFT;
116
117extern unsigned long total_forks;
118extern int nr_threads;
119extern int last_pid;
120DECLARE_PER_CPU(unsigned long, process_counts);
121extern int nr_processes(void);
122extern unsigned long nr_running(void);
123extern unsigned long nr_uninterruptible(void);
Jack Steinerdb1b1fe2006-03-31 02:31:21 -0800124extern unsigned long nr_active(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125extern unsigned long nr_iowait(void);
126
Linus Torvalds1da177e2005-04-16 15:20:36 -0700127
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700128/*
129 * Task state bitmask. NOTE! These bits are also
130 * encoded in fs/proc/array.c: get_task_state().
131 *
132 * We have two separate sets of flags: task->state
133 * is about runnability, while task->exit_state are
134 * about the task exiting. Confusing, but this way
135 * modifying one set can't modify the other one by
136 * mistake.
137 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700138#define TASK_RUNNING 0
139#define TASK_INTERRUPTIBLE 1
140#define TASK_UNINTERRUPTIBLE 2
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700141#define TASK_STOPPED 4
142#define TASK_TRACED 8
143/* in tsk->exit_state */
144#define EXIT_ZOMBIE 16
145#define EXIT_DEAD 32
146/* in tsk->state again */
147#define TASK_NONINTERACTIVE 64
Linus Torvalds1da177e2005-04-16 15:20:36 -0700148
149#define __set_task_state(tsk, state_value) \
150 do { (tsk)->state = (state_value); } while (0)
151#define set_task_state(tsk, state_value) \
152 set_mb((tsk)->state, (state_value))
153
Andrew Morton498d0c52005-09-13 01:25:14 -0700154/*
155 * set_current_state() includes a barrier so that the write of current->state
156 * is correctly serialised wrt the caller's subsequent test of whether to
157 * actually sleep:
158 *
159 * set_current_state(TASK_UNINTERRUPTIBLE);
160 * if (do_i_need_to_sleep())
161 * schedule();
162 *
163 * If the caller does not need such serialisation then use __set_current_state()
164 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700165#define __set_current_state(state_value) \
166 do { current->state = (state_value); } while (0)
167#define set_current_state(state_value) \
168 set_mb(current->state, (state_value))
169
170/* Task command name length */
171#define TASK_COMM_LEN 16
172
Linus Torvalds1da177e2005-04-16 15:20:36 -0700173#include <linux/spinlock.h>
174
175/*
176 * This serializes "schedule()" and also protects
177 * the run-queue from deletions/modifications (but
178 * _adding_ to the beginning of the run-queue has
179 * a separate lock).
180 */
181extern rwlock_t tasklist_lock;
182extern spinlock_t mmlist_lock;
183
184typedef struct task_struct task_t;
185
186extern void sched_init(void);
187extern void sched_init_smp(void);
188extern void init_idle(task_t *idle, int cpu);
189
190extern cpumask_t nohz_cpu_mask;
191
192extern void show_state(void);
193extern void show_regs(struct pt_regs *);
194
195/*
196 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
197 * task), SP is the stack pointer of the first frame that should be shown in the back
198 * trace (or NULL if the entire call-chain of the task should be shown).
199 */
200extern void show_stack(struct task_struct *task, unsigned long *sp);
201
202void io_schedule(void);
203long io_schedule_timeout(long timeout);
204
205extern void cpu_init (void);
206extern void trap_init(void);
207extern void update_process_times(int user);
208extern void scheduler_tick(void);
209
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700210#ifdef CONFIG_DETECT_SOFTLOCKUP
Ingo Molnar6687a972006-03-24 03:18:41 -0800211extern void softlockup_tick(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700212extern void spawn_softlockup_task(void);
213extern void touch_softlockup_watchdog(void);
214#else
Ingo Molnar6687a972006-03-24 03:18:41 -0800215static inline void softlockup_tick(void)
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700216{
217}
218static inline void spawn_softlockup_task(void)
219{
220}
221static inline void touch_softlockup_watchdog(void)
222{
223}
224#endif
225
226
Linus Torvalds1da177e2005-04-16 15:20:36 -0700227/* Attach to any functions which should be ignored in wchan output. */
228#define __sched __attribute__((__section__(".sched.text")))
229/* Is this address in the __sched functions? */
230extern int in_sched_functions(unsigned long addr);
231
232#define MAX_SCHEDULE_TIMEOUT LONG_MAX
233extern signed long FASTCALL(schedule_timeout(signed long timeout));
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700234extern signed long schedule_timeout_interruptible(signed long timeout);
235extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700236asmlinkage void schedule(void);
237
238struct namespace;
239
240/* Maximum number of active map areas.. This is a random (large) number */
241#define DEFAULT_MAX_MAP_COUNT 65536
242
243extern int sysctl_max_map_count;
244
245#include <linux/aio.h>
246
247extern unsigned long
248arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
249 unsigned long, unsigned long);
250extern unsigned long
251arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
252 unsigned long len, unsigned long pgoff,
253 unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700254extern void arch_unmap_area(struct mm_struct *, unsigned long);
255extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700256
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700257#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
258/*
259 * The mm counters are not protected by its page_table_lock,
260 * so must be incremented atomically.
261 */
Christoph Lameterd3cb4872006-01-06 00:11:20 -0800262#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
263#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
264#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
265#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
266#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
267typedef atomic_long_t mm_counter_t;
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700268
269#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
270/*
271 * The mm counters are protected by its page_table_lock,
272 * so can be incremented directly.
273 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700274#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
275#define get_mm_counter(mm, member) ((mm)->_##member)
276#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
277#define inc_mm_counter(mm, member) (mm)->_##member++
278#define dec_mm_counter(mm, member) (mm)->_##member--
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700279typedef unsigned long mm_counter_t;
Hugh Dickins42946212005-10-29 18:16:05 -0700280
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700281#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
282
283#define get_mm_rss(mm) \
284 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
Hugh Dickins365e9c872005-10-29 18:16:18 -0700285#define update_hiwater_rss(mm) do { \
286 unsigned long _rss = get_mm_rss(mm); \
287 if ((mm)->hiwater_rss < _rss) \
288 (mm)->hiwater_rss = _rss; \
289} while (0)
290#define update_hiwater_vm(mm) do { \
291 if ((mm)->hiwater_vm < (mm)->total_vm) \
292 (mm)->hiwater_vm = (mm)->total_vm; \
293} while (0)
294
Linus Torvalds1da177e2005-04-16 15:20:36 -0700295struct mm_struct {
296 struct vm_area_struct * mmap; /* list of VMAs */
297 struct rb_root mm_rb;
298 struct vm_area_struct * mmap_cache; /* last find_vma result */
299 unsigned long (*get_unmapped_area) (struct file *filp,
300 unsigned long addr, unsigned long len,
301 unsigned long pgoff, unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700302 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
Benjamin Herrenschmidt0551fbd2006-02-28 16:59:19 -0800303 unsigned long mmap_base; /* base of mmap area */
304 unsigned long task_size; /* size of task vm space */
305 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700306 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700307 pgd_t * pgd;
308 atomic_t mm_users; /* How many users with user space? */
309 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
310 int map_count; /* number of VMAs */
311 struct rw_semaphore mmap_sem;
312 spinlock_t page_table_lock; /* Protects page tables and some counters */
313
314 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
315 * together off init_mm.mmlist, and are protected
316 * by mmlist_lock
317 */
318
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700319 /* Special counters, in some configurations protected by the
320 * page_table_lock, in other configurations by being atomic.
321 */
Hugh Dickins42946212005-10-29 18:16:05 -0700322 mm_counter_t _file_rss;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700323 mm_counter_t _anon_rss;
324
Hugh Dickinsf4499522005-10-29 18:16:19 -0700325 unsigned long hiwater_rss; /* High-watermark of RSS usage */
326 unsigned long hiwater_vm; /* High-water virtual memory usage */
327
328 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
329 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
330 unsigned long start_code, end_code, start_data, end_data;
331 unsigned long start_brk, brk, start_stack;
332 unsigned long arg_start, arg_end, env_start, env_end;
333
H. J. Lu36d57ac2005-09-06 15:16:49 -0700334 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700335
Alan Coxd6e71142005-06-23 00:09:43 -0700336 unsigned dumpable:2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700337 cpumask_t cpu_vm_mask;
338
339 /* Architecture-specific MM context */
340 mm_context_t context;
341
342 /* Token based thrashing protection. */
343 unsigned long swap_token_time;
344 char recent_pagein;
345
346 /* coredumping support */
347 int core_waiters;
348 struct completion *core_startup_done, core_done;
349
350 /* aio bits */
351 rwlock_t ioctx_list_lock;
352 struct kioctx *ioctx_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700353};
354
355struct sighand_struct {
356 atomic_t count;
357 struct k_sigaction action[_NSIG];
358 spinlock_t siglock;
359};
360
361/*
362 * NOTE! "signal_struct" does not have it's own
363 * locking, because a shared signal_struct always
364 * implies a shared sighand_struct, so locking
365 * sighand_struct is always a proper superset of
366 * the locking of signal_struct.
367 */
368struct signal_struct {
369 atomic_t count;
370 atomic_t live;
371
372 wait_queue_head_t wait_chldexit; /* for wait4() */
373
374 /* current thread group signal load-balancing target: */
375 task_t *curr_target;
376
377 /* shared signal handling: */
378 struct sigpending shared_pending;
379
380 /* thread group exit support */
381 int group_exit_code;
382 /* overloaded:
383 * - notify group_exit_task when ->count is equal to notify_count
384 * - everyone except group_exit_task is stopped during signal delivery
385 * of fatal signals, group_exit_task processes the signal.
386 */
387 struct task_struct *group_exit_task;
388 int notify_count;
389
390 /* thread group stop support, overloads group_exit_code too */
391 int group_stop_count;
392 unsigned int flags; /* see SIGNAL_* flags below */
393
394 /* POSIX.1b Interval Timers */
395 struct list_head posix_timers;
396
397 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800398 struct hrtimer real_timer;
Roman Zippel05cfb612006-03-26 01:38:12 -0800399 struct task_struct *tsk;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800400 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700401
402 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
403 cputime_t it_prof_expires, it_virt_expires;
404 cputime_t it_prof_incr, it_virt_incr;
405
406 /* job control IDs */
407 pid_t pgrp;
408 pid_t tty_old_pgrp;
409 pid_t session;
410 /* boolean value for session group leader */
411 int leader;
412
413 struct tty_struct *tty; /* NULL if no tty */
414
415 /*
416 * Cumulative resource counters for dead threads in the group,
417 * and for reaped dead child processes forked by this group.
418 * Live threads maintain their own counters and add to these
419 * in __exit_signal, except for the group leader.
420 */
421 cputime_t utime, stime, cutime, cstime;
422 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
423 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
424
425 /*
426 * Cumulative ns of scheduled CPU time for dead threads in the
427 * group, not including a zombie group leader. (This only differs
428 * from jiffies_to_ns(utime + stime) if sched_clock uses something
429 * other than jiffies.)
430 */
431 unsigned long long sched_time;
432
433 /*
434 * We don't bother to synchronize most readers of this at all,
435 * because there is no reader checking a limit that actually needs
436 * to get both rlim_cur and rlim_max atomically, and either one
437 * alone is a single word that can safely be read normally.
438 * getrlimit/setrlimit use task_lock(current->group_leader) to
439 * protect this instead of the siglock, because they really
440 * have no need to disable irqs.
441 */
442 struct rlimit rlim[RLIM_NLIMITS];
443
444 struct list_head cpu_timers[3];
445
446 /* keep the process-shared keyrings here so that they do the right
447 * thing in threads created with CLONE_THREAD */
448#ifdef CONFIG_KEYS
449 struct key *session_keyring; /* keyring inherited over fork */
450 struct key *process_keyring; /* keyring private to this process */
451#endif
452};
453
Nick Piggin4866cde2005-06-25 14:57:23 -0700454/* Context switch must be unlocked if interrupts are to be enabled */
455#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
456# define __ARCH_WANT_UNLOCKED_CTXSW
457#endif
458
Linus Torvalds1da177e2005-04-16 15:20:36 -0700459/*
460 * Bits in flags field of signal_struct.
461 */
462#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
463#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
464#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
465#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
466
467
468/*
469 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
Ingo Molnarb0a94992006-01-14 13:20:41 -0800470 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
471 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
472 * values are inverted: lower p->prio value means higher priority.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700473 *
474 * The MAX_USER_RT_PRIO value allows the actual maximum
475 * RT priority to be separate from the value exported to
476 * user-space. This allows kernel threads to set their
477 * priority to a value higher than any user task. Note:
478 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
479 */
480
481#define MAX_USER_RT_PRIO 100
482#define MAX_RT_PRIO MAX_USER_RT_PRIO
483
484#define MAX_PRIO (MAX_RT_PRIO + 40)
485
486#define rt_task(p) (unlikely((p)->prio < MAX_RT_PRIO))
Con Kolivasd425b272006-03-31 02:31:29 -0800487#define batch_task(p) (unlikely((p)->policy == SCHED_BATCH))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700488
489/*
490 * Some day this will be a full-fledged user tracking system..
491 */
492struct user_struct {
493 atomic_t __count; /* reference count */
494 atomic_t processes; /* How many processes does this user have? */
495 atomic_t files; /* How many open files does this user have? */
496 atomic_t sigpending; /* How many pending signals does this user have? */
Amy Griffis2d9048e2006-06-01 13:10:59 -0700497#ifdef CONFIG_INOTIFY_USER
Robert Love0eeca282005-07-12 17:06:03 -0400498 atomic_t inotify_watches; /* How many inotify watches does this user have? */
499 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
500#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700501 /* protected by mq_lock */
502 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
503 unsigned long locked_shm; /* How many pages of mlocked shm ? */
504
505#ifdef CONFIG_KEYS
506 struct key *uid_keyring; /* UID specific keyring */
507 struct key *session_keyring; /* UID's default session keyring */
508#endif
509
510 /* Hash table maintenance information */
511 struct list_head uidhash_list;
512 uid_t uid;
513};
514
515extern struct user_struct *find_user(uid_t);
516
517extern struct user_struct root_user;
518#define INIT_USER (&root_user)
519
520typedef struct prio_array prio_array_t;
521struct backing_dev_info;
522struct reclaim_state;
523
524#ifdef CONFIG_SCHEDSTATS
525struct sched_info {
526 /* cumulative counters */
527 unsigned long cpu_time, /* time spent on the cpu */
528 run_delay, /* time spent waiting on a runqueue */
529 pcnt; /* # of timeslices run on this cpu */
530
531 /* timestamps */
532 unsigned long last_arrival, /* when we last ran on a cpu */
533 last_queued; /* when we were last queued to run */
534};
535
536extern struct file_operations proc_schedstat_operations;
537#endif
538
539enum idle_type
540{
541 SCHED_IDLE,
542 NOT_IDLE,
543 NEWLY_IDLE,
544 MAX_IDLE_TYPES
545};
546
547/*
548 * sched-domains (multiprocessor balancing) declarations:
549 */
550#ifdef CONFIG_SMP
551#define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
552
553#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
554#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
555#define SD_BALANCE_EXEC 4 /* Balance on exec */
Nick Piggin147cbb42005-06-25 14:57:19 -0700556#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
557#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
558#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
559#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
560#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700561
562struct sched_group {
563 struct sched_group *next; /* Must be a circular list */
564 cpumask_t cpumask;
565
566 /*
567 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
568 * single CPU. This is read only (except for setup, hotplug CPU).
569 */
570 unsigned long cpu_power;
571};
572
573struct sched_domain {
574 /* These fields must be setup */
575 struct sched_domain *parent; /* top domain must be null terminated */
576 struct sched_group *groups; /* the balancing groups of the domain */
577 cpumask_t span; /* span of all CPUs in this domain */
578 unsigned long min_interval; /* Minimum balance interval ms */
579 unsigned long max_interval; /* Maximum balance interval ms */
580 unsigned int busy_factor; /* less balancing by factor if busy */
581 unsigned int imbalance_pct; /* No balance until over watermark */
582 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
583 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
584 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
Nick Piggin78979862005-06-25 14:57:13 -0700585 unsigned int busy_idx;
586 unsigned int idle_idx;
587 unsigned int newidle_idx;
588 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700589 unsigned int forkexec_idx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590 int flags; /* See SD_* */
591
592 /* Runtime fields. */
593 unsigned long last_balance; /* init to jiffies. units in jiffies */
594 unsigned int balance_interval; /* initialise to 1. units in ms. */
595 unsigned int nr_balance_failed; /* initialise to 0 */
596
597#ifdef CONFIG_SCHEDSTATS
598 /* load_balance() stats */
599 unsigned long lb_cnt[MAX_IDLE_TYPES];
600 unsigned long lb_failed[MAX_IDLE_TYPES];
601 unsigned long lb_balanced[MAX_IDLE_TYPES];
602 unsigned long lb_imbalance[MAX_IDLE_TYPES];
603 unsigned long lb_gained[MAX_IDLE_TYPES];
604 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
605 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
606 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
607
608 /* Active load balancing */
609 unsigned long alb_cnt;
610 unsigned long alb_failed;
611 unsigned long alb_pushed;
612
Nick Piggin68767a02005-06-25 14:57:20 -0700613 /* SD_BALANCE_EXEC stats */
614 unsigned long sbe_cnt;
615 unsigned long sbe_balanced;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700616 unsigned long sbe_pushed;
617
Nick Piggin68767a02005-06-25 14:57:20 -0700618 /* SD_BALANCE_FORK stats */
619 unsigned long sbf_cnt;
620 unsigned long sbf_balanced;
621 unsigned long sbf_pushed;
622
Linus Torvalds1da177e2005-04-16 15:20:36 -0700623 /* try_to_wake_up() stats */
624 unsigned long ttwu_wake_remote;
625 unsigned long ttwu_move_affine;
626 unsigned long ttwu_move_balance;
627#endif
628};
629
Dinakar Guniguntala1a20ff22005-06-25 14:57:33 -0700630extern void partition_sched_domains(cpumask_t *partition1,
631 cpumask_t *partition2);
akpm@osdl.org198e2f12006-01-12 01:05:30 -0800632
633/*
634 * Maximum cache size the migration-costs auto-tuning code will
635 * search from:
636 */
637extern unsigned int max_cache_size;
638
639#endif /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700640
641
642struct io_context; /* See blkdev.h */
643void exit_io_context(void);
644struct cpuset;
645
646#define NGROUPS_SMALL 32
647#define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
648struct group_info {
649 int ngroups;
650 atomic_t usage;
651 gid_t small_block[NGROUPS_SMALL];
652 int nblocks;
653 gid_t *blocks[0];
654};
655
656/*
657 * get_group_info() must be called with the owning task locked (via task_lock())
658 * when task != current. The reason being that the vast majority of callers are
659 * looking at current->group_info, which can not be changed except by the
660 * current task. Changing current->group_info requires the task lock, too.
661 */
662#define get_group_info(group_info) do { \
663 atomic_inc(&(group_info)->usage); \
664} while (0)
665
666#define put_group_info(group_info) do { \
667 if (atomic_dec_and_test(&(group_info)->usage)) \
668 groups_free(group_info); \
669} while (0)
670
David Howells3e301482005-06-23 22:00:56 -0700671extern struct group_info *groups_alloc(int gidsetsize);
672extern void groups_free(struct group_info *group_info);
673extern int set_current_groups(struct group_info *group_info);
674extern int groups_search(struct group_info *group_info, gid_t grp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700675/* access the groups "array" with this macro */
676#define GROUP_AT(gi, i) \
677 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
678
Chen, Kenneth W383f2832005-09-09 13:02:02 -0700679#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
680extern void prefetch_stack(struct task_struct*);
681#else
682static inline void prefetch_stack(struct task_struct *t) { }
683#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700684
685struct audit_context; /* See audit.c */
686struct mempolicy;
Jens Axboeb92ce552006-04-11 13:52:07 +0200687struct pipe_inode_info;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700688
Con Kolivas3dee3862006-03-31 02:31:23 -0800689enum sleep_type {
690 SLEEP_NORMAL,
691 SLEEP_NONINTERACTIVE,
692 SLEEP_INTERACTIVE,
693 SLEEP_INTERRUPTED,
694};
695
Linus Torvalds1da177e2005-04-16 15:20:36 -0700696struct task_struct {
697 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
698 struct thread_info *thread_info;
699 atomic_t usage;
700 unsigned long flags; /* per process flags, defined below */
701 unsigned long ptrace;
702
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700703 int lock_depth; /* BKL lock depth */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700704
Chen, Kenneth Wd6077cb2006-02-14 13:53:10 -0800705#if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
Nick Piggin4866cde2005-06-25 14:57:23 -0700706 int oncpu;
707#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700708 int prio, static_prio;
709 struct list_head run_list;
710 prio_array_t *array;
711
Jens Axboe22e2c502005-06-27 10:55:12 +0200712 unsigned short ioprio;
Jens Axboe2056a782006-03-23 20:00:26 +0100713 unsigned int btrace_seq;
Jens Axboe22e2c502005-06-27 10:55:12 +0200714
Linus Torvalds1da177e2005-04-16 15:20:36 -0700715 unsigned long sleep_avg;
716 unsigned long long timestamp, last_ran;
717 unsigned long long sched_time; /* sched_clock time spent running */
Con Kolivas3dee3862006-03-31 02:31:23 -0800718 enum sleep_type sleep_type;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700719
720 unsigned long policy;
721 cpumask_t cpus_allowed;
722 unsigned int time_slice, first_time_slice;
723
724#ifdef CONFIG_SCHEDSTATS
725 struct sched_info sched_info;
726#endif
727
728 struct list_head tasks;
729 /*
730 * ptrace_list/ptrace_children forms the list of my children
731 * that were stolen by a ptracer.
732 */
733 struct list_head ptrace_children;
734 struct list_head ptrace_list;
735
736 struct mm_struct *mm, *active_mm;
737
738/* task state */
739 struct linux_binfmt *binfmt;
740 long exit_state;
741 int exit_code, exit_signal;
742 int pdeath_signal; /* The signal sent when the parent dies */
743 /* ??? */
744 unsigned long personality;
745 unsigned did_exec:1;
746 pid_t pid;
747 pid_t tgid;
748 /*
749 * pointers to (original) parent process, youngest child, younger sibling,
750 * older sibling, respectively. (p->father can be replaced with
751 * p->parent->pid)
752 */
753 struct task_struct *real_parent; /* real parent process (when being debugged) */
754 struct task_struct *parent; /* parent process */
755 /*
756 * children/sibling forms the list of my children plus the
757 * tasks I'm ptracing.
758 */
759 struct list_head children; /* list of my children */
760 struct list_head sibling; /* linkage in my parent's children list */
761 struct task_struct *group_leader; /* threadgroup leader */
762
763 /* PID/PID hash table linkage. */
Eric W. Biederman92476d72006-03-31 02:31:42 -0800764 struct pid_link pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -0800765 struct list_head thread_group;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700766
767 struct completion *vfork_done; /* for vfork() */
768 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
769 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
770
771 unsigned long rt_priority;
772 cputime_t utime, stime;
773 unsigned long nvcsw, nivcsw; /* context switch counts */
774 struct timespec start_time;
775/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
776 unsigned long min_flt, maj_flt;
777
778 cputime_t it_prof_expires, it_virt_expires;
779 unsigned long long it_sched_expires;
780 struct list_head cpu_timers[3];
781
782/* process credentials */
783 uid_t uid,euid,suid,fsuid;
784 gid_t gid,egid,sgid,fsgid;
785 struct group_info *group_info;
786 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
787 unsigned keep_capabilities:1;
788 struct user_struct *user;
789#ifdef CONFIG_KEYS
David Howellsb5f545c2006-01-08 01:02:47 -0800790 struct key *request_key_auth; /* assumed request_key authority */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700791 struct key *thread_keyring; /* keyring private to this thread */
David Howells3e301482005-06-23 22:00:56 -0700792 unsigned char jit_keyring; /* default keyring to attach requested keys to */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700793#endif
794 int oomkilladj; /* OOM kill score adjustment (bit shift). */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700795 char comm[TASK_COMM_LEN]; /* executable name excluding path
796 - access with [gs]et_task_comm (which lock
797 it with task_lock())
798 - initialized normally by flush_old_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700799/* file system info */
800 int link_count, total_link_count;
801/* ipc stuff */
802 struct sysv_sem sysvsem;
803/* CPU-specific state of this task */
804 struct thread_struct thread;
805/* filesystem information */
806 struct fs_struct *fs;
807/* open file information */
808 struct files_struct *files;
809/* namespace */
810 struct namespace *namespace;
811/* signal handlers */
812 struct signal_struct *signal;
813 struct sighand_struct *sighand;
814
815 sigset_t blocked, real_blocked;
David Woodhouse150256d2006-01-18 17:43:57 -0800816 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700817 struct sigpending pending;
818
819 unsigned long sas_ss_sp;
820 size_t sas_ss_size;
821 int (*notifier)(void *priv);
822 void *notifier_data;
823 sigset_t *notifier_mask;
824
825 void *security;
826 struct audit_context *audit_context;
827 seccomp_t seccomp;
828
829/* Thread group tracking */
830 u32 parent_exec_id;
831 u32 self_exec_id;
832/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
833 spinlock_t alloc_lock;
834/* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
835 spinlock_t proc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700836
Ingo Molnar408894e2006-01-09 15:59:20 -0800837#ifdef CONFIG_DEBUG_MUTEXES
838 /* mutex deadlock detection */
839 struct mutex_waiter *blocked_on;
840#endif
841
Linus Torvalds1da177e2005-04-16 15:20:36 -0700842/* journalling filesystem info */
843 void *journal_info;
844
845/* VM state */
846 struct reclaim_state *reclaim_state;
847
848 struct dentry *proc_dentry;
849 struct backing_dev_info *backing_dev_info;
850
851 struct io_context *io_context;
852
853 unsigned long ptrace_message;
854 siginfo_t *last_siginfo; /* For ptrace use. */
855/*
856 * current io wait handle: wait queue entry to use for io waits
857 * If this thread is processing aio, this points at the waitqueue
858 * inside the currently handled kiocb. It may be NULL (i.e. default
859 * to a stack based synchronous wait) if its doing sync IO.
860 */
861 wait_queue_t *io_wait;
862/* i/o counters(bytes read/written, #syscalls */
863 u64 rchar, wchar, syscr, syscw;
864#if defined(CONFIG_BSD_PROCESS_ACCT)
865 u64 acct_rss_mem1; /* accumulated rss usage */
866 u64 acct_vm_mem1; /* accumulated virtual memory usage */
867 clock_t acct_stimexpd; /* clock_t-converted stime since last update */
868#endif
869#ifdef CONFIG_NUMA
870 struct mempolicy *mempolicy;
871 short il_next;
872#endif
873#ifdef CONFIG_CPUSETS
874 struct cpuset *cpuset;
875 nodemask_t mems_allowed;
876 int cpuset_mems_generation;
Paul Jackson825a46a2006-03-24 03:16:03 -0800877 int cpuset_mem_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700878#endif
Ingo Molnar0771dfe2006-03-27 01:16:22 -0800879 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -0800880#ifdef CONFIG_COMPAT
881 struct compat_robust_list_head __user *compat_robust_list;
882#endif
Ingo Molnar0771dfe2006-03-27 01:16:22 -0800883
Jens Axboe22e2c502005-06-27 10:55:12 +0200884 atomic_t fs_excl; /* holding fs exclusive resources */
Ingo Molnare56d0902006-01-08 01:01:37 -0800885 struct rcu_head rcu;
Jens Axboeb92ce552006-04-11 13:52:07 +0200886
887 /*
888 * cache last used pipe for splice
889 */
890 struct pipe_inode_info *splice_pipe;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700891};
892
893static inline pid_t process_group(struct task_struct *tsk)
894{
895 return tsk->signal->pgrp;
896}
897
898/**
899 * pid_alive - check that a task structure is not stale
900 * @p: Task structure to be checked.
901 *
902 * Test if a process is not yet dead (at most zombie state)
903 * If pid_alive fails, then pointers within the task structure
904 * can be stale and must not be dereferenced.
905 */
906static inline int pid_alive(struct task_struct *p)
907{
Eric W. Biederman92476d72006-03-31 02:31:42 -0800908 return p->pids[PIDTYPE_PID].pid != NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700909}
910
911extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700912#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -0800913
Andrew Morton158d9eb2006-03-31 02:31:34 -0800914extern void __put_task_struct(struct task_struct *t);
Ingo Molnare56d0902006-01-08 01:01:37 -0800915
916static inline void put_task_struct(struct task_struct *t)
917{
918 if (atomic_dec_and_test(&t->usage))
Eric W. Biederman8c7904a2006-03-31 02:31:37 -0800919 __put_task_struct(t);
Ingo Molnare56d0902006-01-08 01:01:37 -0800920}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700921
922/*
923 * Per process flags
924 */
925#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
926 /* Not implemented yet, only for 486*/
927#define PF_STARTING 0x00000002 /* being created */
928#define PF_EXITING 0x00000004 /* getting shut down */
929#define PF_DEAD 0x00000008 /* Dead */
930#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
931#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
932#define PF_DUMPCORE 0x00000200 /* dumped core */
933#define PF_SIGNALED 0x00000400 /* killed by a signal */
934#define PF_MEMALLOC 0x00000800 /* Allocating memory */
935#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
936#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
937#define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */
938#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
939#define PF_FROZEN 0x00010000 /* frozen for system suspend */
940#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
941#define PF_KSWAPD 0x00040000 /* I am kswapd */
942#define PF_SWAPOFF 0x00080000 /* I am in swapoff */
943#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
Jens Axboeb31dc662006-06-13 08:26:10 +0200944#define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
945#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
946#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
947#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
948#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
Paul Jacksonc61afb12006-03-24 03:16:08 -0800949#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700950
951/*
952 * Only the _current_ task can read/write to tsk->flags, but other
953 * tasks can access tsk->flags in readonly mode for example
954 * with tsk_used_math (like during threaded core dumping).
955 * There is however an exception to this rule during ptrace
956 * or during fork: the ptracer task is allowed to write to the
957 * child->flags of its traced child (same goes for fork, the parent
958 * can write to the child->flags), because we're guaranteed the
959 * child is not running and in turn not changing child->flags
960 * at the same time the parent does it.
961 */
962#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
963#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
964#define clear_used_math() clear_stopped_child_used_math(current)
965#define set_used_math() set_stopped_child_used_math(current)
966#define conditional_stopped_child_used_math(condition, child) \
967 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
968#define conditional_used_math(condition) \
969 conditional_stopped_child_used_math(condition, current)
970#define copy_to_stopped_child_used_math(child) \
971 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
972/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
973#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
974#define used_math() tsk_used_math(current)
975
976#ifdef CONFIG_SMP
977extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
978#else
979static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
980{
Paul Jackson4098f992005-10-30 15:03:21 -0800981 if (!cpu_isset(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700982 return -EINVAL;
983 return 0;
984}
985#endif
986
987extern unsigned long long sched_clock(void);
988extern unsigned long long current_sched_time(const task_t *current_task);
989
990/* sched_exec is called by processes performing an exec */
991#ifdef CONFIG_SMP
992extern void sched_exec(void);
993#else
994#define sched_exec() {}
995#endif
996
997#ifdef CONFIG_HOTPLUG_CPU
998extern void idle_task_exit(void);
999#else
1000static inline void idle_task_exit(void) {}
1001#endif
1002
1003extern void sched_idle_next(void);
1004extern void set_user_nice(task_t *p, long nice);
1005extern int task_prio(const task_t *p);
1006extern int task_nice(const task_t *p);
Matt Mackalle43379f2005-05-01 08:59:00 -07001007extern int can_nice(const task_t *p, const int nice);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001008extern int task_curr(const task_t *p);
1009extern int idle_cpu(int cpu);
1010extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1011extern task_t *idle_task(int cpu);
Keith Owensa2a97982005-09-11 17:19:06 +10001012extern task_t *curr_task(int cpu);
1013extern void set_curr_task(int cpu, task_t *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001014
1015void yield(void);
1016
1017/*
1018 * The default (Linux) execution domain.
1019 */
1020extern struct exec_domain default_exec_domain;
1021
1022union thread_union {
1023 struct thread_info thread_info;
1024 unsigned long stack[THREAD_SIZE/sizeof(long)];
1025};
1026
1027#ifndef __HAVE_ARCH_KSTACK_END
1028static inline int kstack_end(void *addr)
1029{
1030 /* Reliable end of stack detection:
1031 * Some APM bios versions misalign the stack
1032 */
1033 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1034}
1035#endif
1036
1037extern union thread_union init_thread_union;
1038extern struct task_struct init_task;
1039
1040extern struct mm_struct init_mm;
1041
1042#define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1043extern struct task_struct *find_task_by_pid_type(int type, int pid);
1044extern void set_special_pids(pid_t session, pid_t pgrp);
1045extern void __set_special_pids(pid_t session, pid_t pgrp);
1046
1047/* per-UID process charging. */
1048extern struct user_struct * alloc_uid(uid_t);
1049static inline struct user_struct *get_uid(struct user_struct *u)
1050{
1051 atomic_inc(&u->__count);
1052 return u;
1053}
1054extern void free_uid(struct user_struct *);
1055extern void switch_uid(struct user_struct *);
1056
1057#include <asm/current.h>
1058
1059extern void do_timer(struct pt_regs *);
1060
1061extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1062extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1063extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1064 unsigned long clone_flags));
1065#ifdef CONFIG_SMP
1066 extern void kick_process(struct task_struct *tsk);
1067#else
1068 static inline void kick_process(struct task_struct *tsk) { }
1069#endif
Nick Piggin476d1392005-06-25 14:57:29 -07001070extern void FASTCALL(sched_fork(task_t * p, int clone_flags));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001071extern void FASTCALL(sched_exit(task_t * p));
1072
1073extern int in_group_p(gid_t);
1074extern int in_egroup_p(gid_t);
1075
1076extern void proc_caches_init(void);
1077extern void flush_signals(struct task_struct *);
1078extern void flush_signal_handlers(struct task_struct *, int force_default);
1079extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1080
1081static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1082{
1083 unsigned long flags;
1084 int ret;
1085
1086 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1087 ret = dequeue_signal(tsk, mask, info);
1088 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1089
1090 return ret;
1091}
1092
1093extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1094 sigset_t *mask);
1095extern void unblock_all_signals(void);
1096extern void release_task(struct task_struct * p);
1097extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1098extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1099extern int force_sigsegv(int, struct task_struct *);
1100extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1101extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1102extern int kill_pg_info(int, struct siginfo *, pid_t);
1103extern int kill_proc_info(int, struct siginfo *, pid_t);
Harald Welte46113832005-10-10 19:44:29 +02001104extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001105extern void do_notify_parent(struct task_struct *, int);
1106extern void force_sig(int, struct task_struct *);
1107extern void force_sig_specific(int, struct task_struct *);
1108extern int send_sig(int, struct task_struct *, int);
1109extern void zap_other_threads(struct task_struct *p);
1110extern int kill_pg(pid_t, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001111extern int kill_proc(pid_t, int, int);
1112extern struct sigqueue *sigqueue_alloc(void);
1113extern void sigqueue_free(struct sigqueue *);
1114extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1115extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03001116extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001117extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1118
1119/* These can be the second arg to send_sig_info/send_group_sig_info. */
1120#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1121#define SEND_SIG_PRIV ((struct siginfo *) 1)
1122#define SEND_SIG_FORCED ((struct siginfo *) 2)
1123
Oleg Nesterov621d3122005-10-30 15:03:45 -08001124static inline int is_si_special(const struct siginfo *info)
1125{
1126 return info <= SEND_SIG_FORCED;
1127}
1128
Linus Torvalds1da177e2005-04-16 15:20:36 -07001129/* True if we are on the alternate signal stack. */
1130
1131static inline int on_sig_stack(unsigned long sp)
1132{
1133 return (sp - current->sas_ss_sp < current->sas_ss_size);
1134}
1135
1136static inline int sas_ss_flags(unsigned long sp)
1137{
1138 return (current->sas_ss_size == 0 ? SS_DISABLE
1139 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1140}
1141
Linus Torvalds1da177e2005-04-16 15:20:36 -07001142/*
1143 * Routines for handling mm_structs
1144 */
1145extern struct mm_struct * mm_alloc(void);
1146
1147/* mmdrop drops the mm and the page tables */
1148extern void FASTCALL(__mmdrop(struct mm_struct *));
1149static inline void mmdrop(struct mm_struct * mm)
1150{
1151 if (atomic_dec_and_test(&mm->mm_count))
1152 __mmdrop(mm);
1153}
1154
1155/* mmput gets rid of the mappings and all user-space */
1156extern void mmput(struct mm_struct *);
1157/* Grab a reference to a task's mm, if it is not already going away */
1158extern struct mm_struct *get_task_mm(struct task_struct *task);
1159/* Remove the current tasks stale references to the old mm_struct */
1160extern void mm_release(struct task_struct *, struct mm_struct *);
1161
1162extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1163extern void flush_thread(void);
1164extern void exit_thread(void);
1165
Linus Torvalds1da177e2005-04-16 15:20:36 -07001166extern void exit_files(struct task_struct *);
Oleg Nesterov6b3934e2006-03-28 16:11:16 -08001167extern void __cleanup_signal(struct signal_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08001168extern void __cleanup_sighand(struct sighand_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001169extern void exit_itimers(struct signal_struct *);
1170
1171extern NORET_TYPE void do_group_exit(int);
1172
Linus Torvalds1da177e2005-04-16 15:20:36 -07001173extern void daemonize(const char *, ...);
1174extern int allow_signal(int);
1175extern int disallow_signal(int);
1176extern task_t *child_reaper;
1177
1178extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1179extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1180task_t *fork_idle(int);
1181
1182extern void set_task_comm(struct task_struct *tsk, char *from);
1183extern void get_task_comm(char *to, struct task_struct *tsk);
1184
1185#ifdef CONFIG_SMP
1186extern void wait_task_inactive(task_t * p);
1187#else
1188#define wait_task_inactive(p) do { } while (0)
1189#endif
1190
1191#define remove_parent(p) list_del_init(&(p)->sibling)
Oleg Nesterov8fafabd2006-03-28 16:11:05 -08001192#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001193
Eric W. Biederman5e85d4a2006-04-18 22:20:16 -07001194#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001195
1196#define for_each_process(p) \
1197 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1198
1199/*
1200 * Careful: do_each_thread/while_each_thread is a double loop so
1201 * 'break' will not work as expected - use goto instead.
1202 */
1203#define do_each_thread(g, t) \
1204 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1205
1206#define while_each_thread(g, t) \
1207 while ((t = next_thread(t)) != g)
1208
Eric W. Biedermande12a782006-04-10 17:16:49 -06001209/* de_thread depends on thread_group_leader not being a pid based check */
1210#define thread_group_leader(p) (p == p->group_leader)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001211
Keith Owensa9cdf412006-04-10 22:54:07 -07001212static inline task_t *next_thread(const task_t *p)
Oleg Nesterov47e65322006-03-28 16:11:25 -08001213{
1214 return list_entry(rcu_dereference(p->thread_group.next),
1215 task_t, thread_group);
1216}
1217
Linus Torvalds1da177e2005-04-16 15:20:36 -07001218static inline int thread_group_empty(task_t *p)
1219{
Oleg Nesterov47e65322006-03-28 16:11:25 -08001220 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001221}
1222
1223#define delay_group_leader(p) \
1224 (thread_group_leader(p) && !thread_group_empty(p))
1225
Linus Torvalds1da177e2005-04-16 15:20:36 -07001226/*
Eric W. Biederman260ea102006-06-23 02:05:18 -07001227 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02001228 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Jackson053199e2005-10-30 15:02:30 -08001229 * pins the final release of task.io_context. Also protects ->cpuset.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001230 *
1231 * Nests both inside and outside of read_lock(&tasklist_lock).
1232 * It must not be nested with write_lock_irq(&tasklist_lock),
1233 * neither inside nor outside.
1234 */
1235static inline void task_lock(struct task_struct *p)
1236{
1237 spin_lock(&p->alloc_lock);
1238}
1239
1240static inline void task_unlock(struct task_struct *p)
1241{
1242 spin_unlock(&p->alloc_lock);
1243}
1244
Oleg Nesterovf63ee722006-03-28 16:11:13 -08001245extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1246 unsigned long *flags);
1247
1248static inline void unlock_task_sighand(struct task_struct *tsk,
1249 unsigned long *flags)
1250{
1251 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1252}
1253
Al Virof0373602005-11-13 16:06:57 -08001254#ifndef __HAVE_THREAD_FUNCTIONS
1255
Al Viroa1261f52005-11-13 16:06:55 -08001256#define task_thread_info(task) (task)->thread_info
Al Viro9fc65872006-01-12 01:05:34 -08001257#define task_stack_page(task) ((void*)((task)->thread_info))
Al Viroa1261f52005-11-13 16:06:55 -08001258
Al Viro10ebffd2005-11-13 16:06:56 -08001259static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1260{
1261 *task_thread_info(p) = *task_thread_info(org);
1262 task_thread_info(p)->task = p;
1263}
1264
1265static inline unsigned long *end_of_stack(struct task_struct *p)
1266{
1267 return (unsigned long *)(p->thread_info + 1);
1268}
1269
Al Virof0373602005-11-13 16:06:57 -08001270#endif
1271
Linus Torvalds1da177e2005-04-16 15:20:36 -07001272/* set thread flags in other task's structures
1273 * - see asm/thread_info.h for TIF_xxxx flags available
1274 */
1275static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1276{
Al Viroa1261f52005-11-13 16:06:55 -08001277 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001278}
1279
1280static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1281{
Al Viroa1261f52005-11-13 16:06:55 -08001282 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001283}
1284
1285static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1286{
Al Viroa1261f52005-11-13 16:06:55 -08001287 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001288}
1289
1290static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1291{
Al Viroa1261f52005-11-13 16:06:55 -08001292 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001293}
1294
1295static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1296{
Al Viroa1261f52005-11-13 16:06:55 -08001297 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001298}
1299
1300static inline void set_tsk_need_resched(struct task_struct *tsk)
1301{
1302 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1303}
1304
1305static inline void clear_tsk_need_resched(struct task_struct *tsk)
1306{
1307 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1308}
1309
1310static inline int signal_pending(struct task_struct *p)
1311{
1312 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1313}
1314
1315static inline int need_resched(void)
1316{
1317 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1318}
1319
1320/*
1321 * cond_resched() and cond_resched_lock(): latency reduction via
1322 * explicit rescheduling in places that are safe. The return
1323 * value indicates whether a reschedule was done in fact.
1324 * cond_resched_lock() will drop the spinlock before scheduling,
1325 * cond_resched_softirq() will enable bhs before scheduling.
1326 */
1327extern int cond_resched(void);
1328extern int cond_resched_lock(spinlock_t * lock);
1329extern int cond_resched_softirq(void);
1330
1331/*
1332 * Does a critical section need to be broken due to another
1333 * task waiting?:
1334 */
1335#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1336# define need_lockbreak(lock) ((lock)->break_lock)
1337#else
1338# define need_lockbreak(lock) 0
1339#endif
1340
1341/*
1342 * Does a critical section need to be broken due to another
1343 * task waiting or preemption being signalled:
1344 */
1345static inline int lock_need_resched(spinlock_t *lock)
1346{
1347 if (need_lockbreak(lock) || need_resched())
1348 return 1;
1349 return 0;
1350}
1351
1352/* Reevaluate whether the task has signals pending delivery.
1353 This is required every time the blocked sigset_t changes.
1354 callers must hold sighand->siglock. */
1355
1356extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1357extern void recalc_sigpending(void);
1358
1359extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1360
1361/*
1362 * Wrappers for p->thread_info->cpu access. No-op on UP.
1363 */
1364#ifdef CONFIG_SMP
1365
1366static inline unsigned int task_cpu(const struct task_struct *p)
1367{
Al Viroa1261f52005-11-13 16:06:55 -08001368 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001369}
1370
1371static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1372{
Al Viroa1261f52005-11-13 16:06:55 -08001373 task_thread_info(p)->cpu = cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001374}
1375
1376#else
1377
1378static inline unsigned int task_cpu(const struct task_struct *p)
1379{
1380 return 0;
1381}
1382
1383static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1384{
1385}
1386
1387#endif /* CONFIG_SMP */
1388
1389#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1390extern void arch_pick_mmap_layout(struct mm_struct *mm);
1391#else
1392static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1393{
1394 mm->mmap_base = TASK_UNMAPPED_BASE;
1395 mm->get_unmapped_area = arch_get_unmapped_area;
1396 mm->unmap_area = arch_unmap_area;
1397}
1398#endif
1399
1400extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1401extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1402
Linus Torvalds1da177e2005-04-16 15:20:36 -07001403extern void normalize_rt_tasks(void);
1404
Linus Torvalds1da177e2005-04-16 15:20:36 -07001405#ifdef CONFIG_PM
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001406/*
1407 * Check if a process has been frozen
1408 */
1409static inline int frozen(struct task_struct *p)
1410{
1411 return p->flags & PF_FROZEN;
1412}
1413
1414/*
1415 * Check if there is a request to freeze a process
1416 */
1417static inline int freezing(struct task_struct *p)
1418{
1419 return p->flags & PF_FREEZE;
1420}
1421
1422/*
1423 * Request that a process be frozen
1424 * FIXME: SMP problem. We may not modify other process' flags!
1425 */
1426static inline void freeze(struct task_struct *p)
1427{
1428 p->flags |= PF_FREEZE;
1429}
1430
1431/*
1432 * Wake up a frozen process
1433 */
1434static inline int thaw_process(struct task_struct *p)
1435{
1436 if (frozen(p)) {
1437 p->flags &= ~PF_FROZEN;
1438 wake_up_process(p);
1439 return 1;
1440 }
1441 return 0;
1442}
1443
1444/*
1445 * freezing is complete, mark process as frozen
1446 */
1447static inline void frozen_process(struct task_struct *p)
1448{
1449 p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1450}
1451
1452extern void refrigerator(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001453extern int freeze_processes(void);
1454extern void thaw_processes(void);
1455
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001456static inline int try_to_freeze(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001457{
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001458 if (freezing(current)) {
1459 refrigerator();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001460 return 1;
1461 } else
1462 return 0;
1463}
1464#else
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001465static inline int frozen(struct task_struct *p) { return 0; }
1466static inline int freezing(struct task_struct *p) { return 0; }
1467static inline void freeze(struct task_struct *p) { BUG(); }
1468static inline int thaw_process(struct task_struct *p) { return 1; }
1469static inline void frozen_process(struct task_struct *p) { BUG(); }
1470
1471static inline void refrigerator(void) {}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001472static inline int freeze_processes(void) { BUG(); return 0; }
1473static inline void thaw_processes(void) {}
1474
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001475static inline int try_to_freeze(void) { return 0; }
1476
Linus Torvalds1da177e2005-04-16 15:20:36 -07001477#endif /* CONFIG_PM */
1478#endif /* __KERNEL__ */
1479
1480#endif