blob: 6f167645e7e28e41c6b0188b13040edb9c579d2a [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);
Peter Williams2dd73a42006-06-27 02:54:34 -0700126extern unsigned long weighted_cpuload(const int cpu);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700127
Linus Torvalds1da177e2005-04-16 15:20:36 -0700128
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700129/*
130 * Task state bitmask. NOTE! These bits are also
131 * encoded in fs/proc/array.c: get_task_state().
132 *
133 * We have two separate sets of flags: task->state
134 * is about runnability, while task->exit_state are
135 * about the task exiting. Confusing, but this way
136 * modifying one set can't modify the other one by
137 * mistake.
138 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700139#define TASK_RUNNING 0
140#define TASK_INTERRUPTIBLE 1
141#define TASK_UNINTERRUPTIBLE 2
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700142#define TASK_STOPPED 4
143#define TASK_TRACED 8
144/* in tsk->exit_state */
145#define EXIT_ZOMBIE 16
146#define EXIT_DEAD 32
147/* in tsk->state again */
148#define TASK_NONINTERACTIVE 64
Linus Torvalds1da177e2005-04-16 15:20:36 -0700149
150#define __set_task_state(tsk, state_value) \
151 do { (tsk)->state = (state_value); } while (0)
152#define set_task_state(tsk, state_value) \
153 set_mb((tsk)->state, (state_value))
154
Andrew Morton498d0c52005-09-13 01:25:14 -0700155/*
156 * set_current_state() includes a barrier so that the write of current->state
157 * is correctly serialised wrt the caller's subsequent test of whether to
158 * actually sleep:
159 *
160 * set_current_state(TASK_UNINTERRUPTIBLE);
161 * if (do_i_need_to_sleep())
162 * schedule();
163 *
164 * If the caller does not need such serialisation then use __set_current_state()
165 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700166#define __set_current_state(state_value) \
167 do { current->state = (state_value); } while (0)
168#define set_current_state(state_value) \
169 set_mb(current->state, (state_value))
170
171/* Task command name length */
172#define TASK_COMM_LEN 16
173
Linus Torvalds1da177e2005-04-16 15:20:36 -0700174#include <linux/spinlock.h>
175
176/*
177 * This serializes "schedule()" and also protects
178 * the run-queue from deletions/modifications (but
179 * _adding_ to the beginning of the run-queue has
180 * a separate lock).
181 */
182extern rwlock_t tasklist_lock;
183extern spinlock_t mmlist_lock;
184
185typedef struct task_struct task_t;
186
187extern void sched_init(void);
188extern void sched_init_smp(void);
189extern void init_idle(task_t *idle, int cpu);
190
191extern cpumask_t nohz_cpu_mask;
192
193extern void show_state(void);
194extern void show_regs(struct pt_regs *);
195
196/*
197 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
198 * task), SP is the stack pointer of the first frame that should be shown in the back
199 * trace (or NULL if the entire call-chain of the task should be shown).
200 */
201extern void show_stack(struct task_struct *task, unsigned long *sp);
202
203void io_schedule(void);
204long io_schedule_timeout(long timeout);
205
206extern void cpu_init (void);
207extern void trap_init(void);
208extern void update_process_times(int user);
209extern void scheduler_tick(void);
210
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700211#ifdef CONFIG_DETECT_SOFTLOCKUP
Ingo Molnar6687a972006-03-24 03:18:41 -0800212extern void softlockup_tick(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700213extern void spawn_softlockup_task(void);
214extern void touch_softlockup_watchdog(void);
215#else
Ingo Molnar6687a972006-03-24 03:18:41 -0800216static inline void softlockup_tick(void)
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700217{
218}
219static inline void spawn_softlockup_task(void)
220{
221}
222static inline void touch_softlockup_watchdog(void)
223{
224}
225#endif
226
227
Linus Torvalds1da177e2005-04-16 15:20:36 -0700228/* Attach to any functions which should be ignored in wchan output. */
229#define __sched __attribute__((__section__(".sched.text")))
230/* Is this address in the __sched functions? */
231extern int in_sched_functions(unsigned long addr);
232
233#define MAX_SCHEDULE_TIMEOUT LONG_MAX
234extern signed long FASTCALL(schedule_timeout(signed long timeout));
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700235extern signed long schedule_timeout_interruptible(signed long timeout);
236extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700237asmlinkage void schedule(void);
238
239struct namespace;
240
241/* Maximum number of active map areas.. This is a random (large) number */
242#define DEFAULT_MAX_MAP_COUNT 65536
243
244extern int sysctl_max_map_count;
245
246#include <linux/aio.h>
247
248extern unsigned long
249arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
250 unsigned long, unsigned long);
251extern unsigned long
252arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
253 unsigned long len, unsigned long pgoff,
254 unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700255extern void arch_unmap_area(struct mm_struct *, unsigned long);
256extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700257
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700258#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
259/*
260 * The mm counters are not protected by its page_table_lock,
261 * so must be incremented atomically.
262 */
Christoph Lameterd3cb4872006-01-06 00:11:20 -0800263#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
264#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
265#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
266#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
267#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
268typedef atomic_long_t mm_counter_t;
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700269
270#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
271/*
272 * The mm counters are protected by its page_table_lock,
273 * so can be incremented directly.
274 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700275#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
276#define get_mm_counter(mm, member) ((mm)->_##member)
277#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
278#define inc_mm_counter(mm, member) (mm)->_##member++
279#define dec_mm_counter(mm, member) (mm)->_##member--
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700280typedef unsigned long mm_counter_t;
Hugh Dickins42946212005-10-29 18:16:05 -0700281
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700282#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
283
284#define get_mm_rss(mm) \
285 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
Hugh Dickins365e9c872005-10-29 18:16:18 -0700286#define update_hiwater_rss(mm) do { \
287 unsigned long _rss = get_mm_rss(mm); \
288 if ((mm)->hiwater_rss < _rss) \
289 (mm)->hiwater_rss = _rss; \
290} while (0)
291#define update_hiwater_vm(mm) do { \
292 if ((mm)->hiwater_vm < (mm)->total_vm) \
293 (mm)->hiwater_vm = (mm)->total_vm; \
294} while (0)
295
Linus Torvalds1da177e2005-04-16 15:20:36 -0700296struct mm_struct {
297 struct vm_area_struct * mmap; /* list of VMAs */
298 struct rb_root mm_rb;
299 struct vm_area_struct * mmap_cache; /* last find_vma result */
300 unsigned long (*get_unmapped_area) (struct file *filp,
301 unsigned long addr, unsigned long len,
302 unsigned long pgoff, unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700303 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
Benjamin Herrenschmidt0551fbd2006-02-28 16:59:19 -0800304 unsigned long mmap_base; /* base of mmap area */
305 unsigned long task_size; /* size of task vm space */
306 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700307 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700308 pgd_t * pgd;
309 atomic_t mm_users; /* How many users with user space? */
310 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
311 int map_count; /* number of VMAs */
312 struct rw_semaphore mmap_sem;
313 spinlock_t page_table_lock; /* Protects page tables and some counters */
314
315 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
316 * together off init_mm.mmlist, and are protected
317 * by mmlist_lock
318 */
319
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700320 /* Special counters, in some configurations protected by the
321 * page_table_lock, in other configurations by being atomic.
322 */
Hugh Dickins42946212005-10-29 18:16:05 -0700323 mm_counter_t _file_rss;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700324 mm_counter_t _anon_rss;
325
Hugh Dickinsf4499522005-10-29 18:16:19 -0700326 unsigned long hiwater_rss; /* High-watermark of RSS usage */
327 unsigned long hiwater_vm; /* High-water virtual memory usage */
328
329 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
330 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
331 unsigned long start_code, end_code, start_data, end_data;
332 unsigned long start_brk, brk, start_stack;
333 unsigned long arg_start, arg_end, env_start, env_end;
334
H. J. Lu36d57ac2005-09-06 15:16:49 -0700335 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700336
Alan Coxd6e71142005-06-23 00:09:43 -0700337 unsigned dumpable:2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700338 cpumask_t cpu_vm_mask;
339
340 /* Architecture-specific MM context */
341 mm_context_t context;
342
343 /* Token based thrashing protection. */
344 unsigned long swap_token_time;
345 char recent_pagein;
346
347 /* coredumping support */
348 int core_waiters;
349 struct completion *core_startup_done, core_done;
350
351 /* aio bits */
352 rwlock_t ioctx_list_lock;
353 struct kioctx *ioctx_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354};
355
356struct sighand_struct {
357 atomic_t count;
358 struct k_sigaction action[_NSIG];
359 spinlock_t siglock;
360};
361
KaiGai Kohei0e464812006-06-25 05:49:24 -0700362struct pacct_struct {
KaiGai Koheif6ec29a2006-06-25 05:49:25 -0700363 int ac_flag;
364 long ac_exitcode;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700365 unsigned long ac_mem;
KaiGai Kohei77787bf2006-06-25 05:49:26 -0700366 cputime_t ac_utime, ac_stime;
367 unsigned long ac_minflt, ac_majflt;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700368};
369
Linus Torvalds1da177e2005-04-16 15:20:36 -0700370/*
371 * NOTE! "signal_struct" does not have it's own
372 * locking, because a shared signal_struct always
373 * implies a shared sighand_struct, so locking
374 * sighand_struct is always a proper superset of
375 * the locking of signal_struct.
376 */
377struct signal_struct {
378 atomic_t count;
379 atomic_t live;
380
381 wait_queue_head_t wait_chldexit; /* for wait4() */
382
383 /* current thread group signal load-balancing target: */
384 task_t *curr_target;
385
386 /* shared signal handling: */
387 struct sigpending shared_pending;
388
389 /* thread group exit support */
390 int group_exit_code;
391 /* overloaded:
392 * - notify group_exit_task when ->count is equal to notify_count
393 * - everyone except group_exit_task is stopped during signal delivery
394 * of fatal signals, group_exit_task processes the signal.
395 */
396 struct task_struct *group_exit_task;
397 int notify_count;
398
399 /* thread group stop support, overloads group_exit_code too */
400 int group_stop_count;
401 unsigned int flags; /* see SIGNAL_* flags below */
402
403 /* POSIX.1b Interval Timers */
404 struct list_head posix_timers;
405
406 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800407 struct hrtimer real_timer;
Roman Zippel05cfb612006-03-26 01:38:12 -0800408 struct task_struct *tsk;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800409 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410
411 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
412 cputime_t it_prof_expires, it_virt_expires;
413 cputime_t it_prof_incr, it_virt_incr;
414
415 /* job control IDs */
416 pid_t pgrp;
417 pid_t tty_old_pgrp;
418 pid_t session;
419 /* boolean value for session group leader */
420 int leader;
421
422 struct tty_struct *tty; /* NULL if no tty */
423
424 /*
425 * Cumulative resource counters for dead threads in the group,
426 * and for reaped dead child processes forked by this group.
427 * Live threads maintain their own counters and add to these
428 * in __exit_signal, except for the group leader.
429 */
430 cputime_t utime, stime, cutime, cstime;
431 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
432 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
433
434 /*
435 * Cumulative ns of scheduled CPU time for dead threads in the
436 * group, not including a zombie group leader. (This only differs
437 * from jiffies_to_ns(utime + stime) if sched_clock uses something
438 * other than jiffies.)
439 */
440 unsigned long long sched_time;
441
442 /*
443 * We don't bother to synchronize most readers of this at all,
444 * because there is no reader checking a limit that actually needs
445 * to get both rlim_cur and rlim_max atomically, and either one
446 * alone is a single word that can safely be read normally.
447 * getrlimit/setrlimit use task_lock(current->group_leader) to
448 * protect this instead of the siglock, because they really
449 * have no need to disable irqs.
450 */
451 struct rlimit rlim[RLIM_NLIMITS];
452
453 struct list_head cpu_timers[3];
454
455 /* keep the process-shared keyrings here so that they do the right
456 * thing in threads created with CLONE_THREAD */
457#ifdef CONFIG_KEYS
458 struct key *session_keyring; /* keyring inherited over fork */
459 struct key *process_keyring; /* keyring private to this process */
460#endif
KaiGai Kohei0e464812006-06-25 05:49:24 -0700461#ifdef CONFIG_BSD_PROCESS_ACCT
462 struct pacct_struct pacct; /* per-process accounting information */
463#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700464};
465
Nick Piggin4866cde2005-06-25 14:57:23 -0700466/* Context switch must be unlocked if interrupts are to be enabled */
467#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
468# define __ARCH_WANT_UNLOCKED_CTXSW
469#endif
470
Linus Torvalds1da177e2005-04-16 15:20:36 -0700471/*
472 * Bits in flags field of signal_struct.
473 */
474#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
475#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
476#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
477#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
478
479
480/*
481 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
Ingo Molnarb0a94992006-01-14 13:20:41 -0800482 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
483 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
484 * values are inverted: lower p->prio value means higher priority.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700485 *
486 * The MAX_USER_RT_PRIO value allows the actual maximum
487 * RT priority to be separate from the value exported to
488 * user-space. This allows kernel threads to set their
489 * priority to a value higher than any user task. Note:
490 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
491 */
492
493#define MAX_USER_RT_PRIO 100
494#define MAX_RT_PRIO MAX_USER_RT_PRIO
495
496#define MAX_PRIO (MAX_RT_PRIO + 40)
497
Ingo Molnarb29739f2006-06-27 02:54:51 -0700498#define rt_prio(prio) unlikely((prio) < MAX_RT_PRIO)
499#define rt_task(p) rt_prio((p)->prio)
Con Kolivasd425b272006-03-31 02:31:29 -0800500#define batch_task(p) (unlikely((p)->policy == SCHED_BATCH))
Ingo Molnarb29739f2006-06-27 02:54:51 -0700501#define has_rt_policy(p) \
502 unlikely((p)->policy != SCHED_NORMAL && (p)->policy != SCHED_BATCH)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700503
504/*
505 * Some day this will be a full-fledged user tracking system..
506 */
507struct user_struct {
508 atomic_t __count; /* reference count */
509 atomic_t processes; /* How many processes does this user have? */
510 atomic_t files; /* How many open files does this user have? */
511 atomic_t sigpending; /* How many pending signals does this user have? */
Amy Griffis2d9048e2006-06-01 13:10:59 -0700512#ifdef CONFIG_INOTIFY_USER
Robert Love0eeca282005-07-12 17:06:03 -0400513 atomic_t inotify_watches; /* How many inotify watches does this user have? */
514 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
515#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700516 /* protected by mq_lock */
517 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
518 unsigned long locked_shm; /* How many pages of mlocked shm ? */
519
520#ifdef CONFIG_KEYS
521 struct key *uid_keyring; /* UID specific keyring */
522 struct key *session_keyring; /* UID's default session keyring */
523#endif
524
525 /* Hash table maintenance information */
526 struct list_head uidhash_list;
527 uid_t uid;
528};
529
530extern struct user_struct *find_user(uid_t);
531
532extern struct user_struct root_user;
533#define INIT_USER (&root_user)
534
535typedef struct prio_array prio_array_t;
536struct backing_dev_info;
537struct reclaim_state;
538
539#ifdef CONFIG_SCHEDSTATS
540struct sched_info {
541 /* cumulative counters */
542 unsigned long cpu_time, /* time spent on the cpu */
543 run_delay, /* time spent waiting on a runqueue */
544 pcnt; /* # of timeslices run on this cpu */
545
546 /* timestamps */
547 unsigned long last_arrival, /* when we last ran on a cpu */
548 last_queued; /* when we were last queued to run */
549};
550
551extern struct file_operations proc_schedstat_operations;
552#endif
553
554enum idle_type
555{
556 SCHED_IDLE,
557 NOT_IDLE,
558 NEWLY_IDLE,
559 MAX_IDLE_TYPES
560};
561
562/*
563 * sched-domains (multiprocessor balancing) declarations:
564 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700565#define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
566
Peter Williams2dd73a42006-06-27 02:54:34 -0700567#ifdef CONFIG_SMP
Linus Torvalds1da177e2005-04-16 15:20:36 -0700568#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
569#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
570#define SD_BALANCE_EXEC 4 /* Balance on exec */
Nick Piggin147cbb42005-06-25 14:57:19 -0700571#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
572#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
573#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
574#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
575#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -0700576#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
577
578#define BALANCE_FOR_POWER ((sched_mc_power_savings || sched_smt_power_savings) \
579 ? SD_POWERSAVINGS_BALANCE : 0)
580
Linus Torvalds1da177e2005-04-16 15:20:36 -0700581
582struct sched_group {
583 struct sched_group *next; /* Must be a circular list */
584 cpumask_t cpumask;
585
586 /*
587 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
588 * single CPU. This is read only (except for setup, hotplug CPU).
589 */
590 unsigned long cpu_power;
591};
592
593struct sched_domain {
594 /* These fields must be setup */
595 struct sched_domain *parent; /* top domain must be null terminated */
596 struct sched_group *groups; /* the balancing groups of the domain */
597 cpumask_t span; /* span of all CPUs in this domain */
598 unsigned long min_interval; /* Minimum balance interval ms */
599 unsigned long max_interval; /* Maximum balance interval ms */
600 unsigned int busy_factor; /* less balancing by factor if busy */
601 unsigned int imbalance_pct; /* No balance until over watermark */
602 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
603 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
604 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
Nick Piggin78979862005-06-25 14:57:13 -0700605 unsigned int busy_idx;
606 unsigned int idle_idx;
607 unsigned int newidle_idx;
608 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700609 unsigned int forkexec_idx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610 int flags; /* See SD_* */
611
612 /* Runtime fields. */
613 unsigned long last_balance; /* init to jiffies. units in jiffies */
614 unsigned int balance_interval; /* initialise to 1. units in ms. */
615 unsigned int nr_balance_failed; /* initialise to 0 */
616
617#ifdef CONFIG_SCHEDSTATS
618 /* load_balance() stats */
619 unsigned long lb_cnt[MAX_IDLE_TYPES];
620 unsigned long lb_failed[MAX_IDLE_TYPES];
621 unsigned long lb_balanced[MAX_IDLE_TYPES];
622 unsigned long lb_imbalance[MAX_IDLE_TYPES];
623 unsigned long lb_gained[MAX_IDLE_TYPES];
624 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
625 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
626 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
627
628 /* Active load balancing */
629 unsigned long alb_cnt;
630 unsigned long alb_failed;
631 unsigned long alb_pushed;
632
Nick Piggin68767a02005-06-25 14:57:20 -0700633 /* SD_BALANCE_EXEC stats */
634 unsigned long sbe_cnt;
635 unsigned long sbe_balanced;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700636 unsigned long sbe_pushed;
637
Nick Piggin68767a02005-06-25 14:57:20 -0700638 /* SD_BALANCE_FORK stats */
639 unsigned long sbf_cnt;
640 unsigned long sbf_balanced;
641 unsigned long sbf_pushed;
642
Linus Torvalds1da177e2005-04-16 15:20:36 -0700643 /* try_to_wake_up() stats */
644 unsigned long ttwu_wake_remote;
645 unsigned long ttwu_move_affine;
646 unsigned long ttwu_move_balance;
647#endif
648};
649
Srivatsa Vaddagiri51888ca2006-06-27 02:54:38 -0700650extern int partition_sched_domains(cpumask_t *partition1,
Dinakar Guniguntala1a20ff22005-06-25 14:57:33 -0700651 cpumask_t *partition2);
akpm@osdl.org198e2f12006-01-12 01:05:30 -0800652
653/*
654 * Maximum cache size the migration-costs auto-tuning code will
655 * search from:
656 */
657extern unsigned int max_cache_size;
658
659#endif /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660
661
662struct io_context; /* See blkdev.h */
663void exit_io_context(void);
664struct cpuset;
665
666#define NGROUPS_SMALL 32
667#define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
668struct group_info {
669 int ngroups;
670 atomic_t usage;
671 gid_t small_block[NGROUPS_SMALL];
672 int nblocks;
673 gid_t *blocks[0];
674};
675
676/*
677 * get_group_info() must be called with the owning task locked (via task_lock())
678 * when task != current. The reason being that the vast majority of callers are
679 * looking at current->group_info, which can not be changed except by the
680 * current task. Changing current->group_info requires the task lock, too.
681 */
682#define get_group_info(group_info) do { \
683 atomic_inc(&(group_info)->usage); \
684} while (0)
685
686#define put_group_info(group_info) do { \
687 if (atomic_dec_and_test(&(group_info)->usage)) \
688 groups_free(group_info); \
689} while (0)
690
David Howells3e301482005-06-23 22:00:56 -0700691extern struct group_info *groups_alloc(int gidsetsize);
692extern void groups_free(struct group_info *group_info);
693extern int set_current_groups(struct group_info *group_info);
694extern int groups_search(struct group_info *group_info, gid_t grp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700695/* access the groups "array" with this macro */
696#define GROUP_AT(gi, i) \
697 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
698
Chen, Kenneth W383f2832005-09-09 13:02:02 -0700699#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
700extern void prefetch_stack(struct task_struct*);
701#else
702static inline void prefetch_stack(struct task_struct *t) { }
703#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700704
705struct audit_context; /* See audit.c */
706struct mempolicy;
Jens Axboeb92ce552006-04-11 13:52:07 +0200707struct pipe_inode_info;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700708
Con Kolivas3dee3862006-03-31 02:31:23 -0800709enum sleep_type {
710 SLEEP_NORMAL,
711 SLEEP_NONINTERACTIVE,
712 SLEEP_INTERACTIVE,
713 SLEEP_INTERRUPTED,
714};
715
Linus Torvalds1da177e2005-04-16 15:20:36 -0700716struct task_struct {
717 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
718 struct thread_info *thread_info;
719 atomic_t usage;
720 unsigned long flags; /* per process flags, defined below */
721 unsigned long ptrace;
722
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700723 int lock_depth; /* BKL lock depth */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700724
Peter Williams2dd73a42006-06-27 02:54:34 -0700725#ifdef CONFIG_SMP
726#ifdef __ARCH_WANT_UNLOCKED_CTXSW
Nick Piggin4866cde2005-06-25 14:57:23 -0700727 int oncpu;
728#endif
Peter Williams2dd73a42006-06-27 02:54:34 -0700729#endif
730 int load_weight; /* for niceness load balancing purposes */
Ingo Molnarb29739f2006-06-27 02:54:51 -0700731 int prio, static_prio, normal_prio;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700732 struct list_head run_list;
733 prio_array_t *array;
734
Jens Axboe22e2c502005-06-27 10:55:12 +0200735 unsigned short ioprio;
Jens Axboe2056a782006-03-23 20:00:26 +0100736 unsigned int btrace_seq;
Jens Axboe22e2c502005-06-27 10:55:12 +0200737
Linus Torvalds1da177e2005-04-16 15:20:36 -0700738 unsigned long sleep_avg;
739 unsigned long long timestamp, last_ran;
740 unsigned long long sched_time; /* sched_clock time spent running */
Con Kolivas3dee3862006-03-31 02:31:23 -0800741 enum sleep_type sleep_type;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700742
743 unsigned long policy;
744 cpumask_t cpus_allowed;
745 unsigned int time_slice, first_time_slice;
746
747#ifdef CONFIG_SCHEDSTATS
748 struct sched_info sched_info;
749#endif
750
751 struct list_head tasks;
752 /*
753 * ptrace_list/ptrace_children forms the list of my children
754 * that were stolen by a ptracer.
755 */
756 struct list_head ptrace_children;
757 struct list_head ptrace_list;
758
759 struct mm_struct *mm, *active_mm;
760
761/* task state */
762 struct linux_binfmt *binfmt;
763 long exit_state;
764 int exit_code, exit_signal;
765 int pdeath_signal; /* The signal sent when the parent dies */
766 /* ??? */
767 unsigned long personality;
768 unsigned did_exec:1;
769 pid_t pid;
770 pid_t tgid;
771 /*
772 * pointers to (original) parent process, youngest child, younger sibling,
773 * older sibling, respectively. (p->father can be replaced with
774 * p->parent->pid)
775 */
776 struct task_struct *real_parent; /* real parent process (when being debugged) */
777 struct task_struct *parent; /* parent process */
778 /*
779 * children/sibling forms the list of my children plus the
780 * tasks I'm ptracing.
781 */
782 struct list_head children; /* list of my children */
783 struct list_head sibling; /* linkage in my parent's children list */
784 struct task_struct *group_leader; /* threadgroup leader */
785
786 /* PID/PID hash table linkage. */
Eric W. Biederman92476d72006-03-31 02:31:42 -0800787 struct pid_link pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -0800788 struct list_head thread_group;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700789
790 struct completion *vfork_done; /* for vfork() */
791 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
792 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
793
794 unsigned long rt_priority;
795 cputime_t utime, stime;
796 unsigned long nvcsw, nivcsw; /* context switch counts */
797 struct timespec start_time;
798/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
799 unsigned long min_flt, maj_flt;
800
801 cputime_t it_prof_expires, it_virt_expires;
802 unsigned long long it_sched_expires;
803 struct list_head cpu_timers[3];
804
805/* process credentials */
806 uid_t uid,euid,suid,fsuid;
807 gid_t gid,egid,sgid,fsgid;
808 struct group_info *group_info;
809 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
810 unsigned keep_capabilities:1;
811 struct user_struct *user;
812#ifdef CONFIG_KEYS
David Howellsb5f545c2006-01-08 01:02:47 -0800813 struct key *request_key_auth; /* assumed request_key authority */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700814 struct key *thread_keyring; /* keyring private to this thread */
David Howells3e301482005-06-23 22:00:56 -0700815 unsigned char jit_keyring; /* default keyring to attach requested keys to */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700816#endif
817 int oomkilladj; /* OOM kill score adjustment (bit shift). */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700818 char comm[TASK_COMM_LEN]; /* executable name excluding path
819 - access with [gs]et_task_comm (which lock
820 it with task_lock())
821 - initialized normally by flush_old_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700822/* file system info */
823 int link_count, total_link_count;
824/* ipc stuff */
825 struct sysv_sem sysvsem;
826/* CPU-specific state of this task */
827 struct thread_struct thread;
828/* filesystem information */
829 struct fs_struct *fs;
830/* open file information */
831 struct files_struct *files;
832/* namespace */
833 struct namespace *namespace;
834/* signal handlers */
835 struct signal_struct *signal;
836 struct sighand_struct *sighand;
837
838 sigset_t blocked, real_blocked;
David Woodhouse150256d2006-01-18 17:43:57 -0800839 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700840 struct sigpending pending;
841
842 unsigned long sas_ss_sp;
843 size_t sas_ss_size;
844 int (*notifier)(void *priv);
845 void *notifier_data;
846 sigset_t *notifier_mask;
847
848 void *security;
849 struct audit_context *audit_context;
850 seccomp_t seccomp;
851
852/* Thread group tracking */
853 u32 parent_exec_id;
854 u32 self_exec_id;
855/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
856 spinlock_t alloc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700857
Ingo Molnarb29739f2006-06-27 02:54:51 -0700858 /* Protection of the PI data structures: */
859 spinlock_t pi_lock;
860
Ingo Molnar408894e2006-01-09 15:59:20 -0800861#ifdef CONFIG_DEBUG_MUTEXES
862 /* mutex deadlock detection */
863 struct mutex_waiter *blocked_on;
864#endif
865
Linus Torvalds1da177e2005-04-16 15:20:36 -0700866/* journalling filesystem info */
867 void *journal_info;
868
869/* VM state */
870 struct reclaim_state *reclaim_state;
871
Linus Torvalds1da177e2005-04-16 15:20:36 -0700872 struct backing_dev_info *backing_dev_info;
873
874 struct io_context *io_context;
875
876 unsigned long ptrace_message;
877 siginfo_t *last_siginfo; /* For ptrace use. */
878/*
879 * current io wait handle: wait queue entry to use for io waits
880 * If this thread is processing aio, this points at the waitqueue
881 * inside the currently handled kiocb. It may be NULL (i.e. default
882 * to a stack based synchronous wait) if its doing sync IO.
883 */
884 wait_queue_t *io_wait;
885/* i/o counters(bytes read/written, #syscalls */
886 u64 rchar, wchar, syscr, syscw;
887#if defined(CONFIG_BSD_PROCESS_ACCT)
888 u64 acct_rss_mem1; /* accumulated rss usage */
889 u64 acct_vm_mem1; /* accumulated virtual memory usage */
890 clock_t acct_stimexpd; /* clock_t-converted stime since last update */
891#endif
892#ifdef CONFIG_NUMA
893 struct mempolicy *mempolicy;
894 short il_next;
895#endif
896#ifdef CONFIG_CPUSETS
897 struct cpuset *cpuset;
898 nodemask_t mems_allowed;
899 int cpuset_mems_generation;
Paul Jackson825a46a2006-03-24 03:16:03 -0800900 int cpuset_mem_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700901#endif
Ingo Molnar0771dfe2006-03-27 01:16:22 -0800902 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -0800903#ifdef CONFIG_COMPAT
904 struct compat_robust_list_head __user *compat_robust_list;
905#endif
Ingo Molnar0771dfe2006-03-27 01:16:22 -0800906
Jens Axboe22e2c502005-06-27 10:55:12 +0200907 atomic_t fs_excl; /* holding fs exclusive resources */
Ingo Molnare56d0902006-01-08 01:01:37 -0800908 struct rcu_head rcu;
Jens Axboeb92ce552006-04-11 13:52:07 +0200909
910 /*
911 * cache last used pipe for splice
912 */
913 struct pipe_inode_info *splice_pipe;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700914};
915
916static inline pid_t process_group(struct task_struct *tsk)
917{
918 return tsk->signal->pgrp;
919}
920
921/**
922 * pid_alive - check that a task structure is not stale
923 * @p: Task structure to be checked.
924 *
925 * Test if a process is not yet dead (at most zombie state)
926 * If pid_alive fails, then pointers within the task structure
927 * can be stale and must not be dereferenced.
928 */
929static inline int pid_alive(struct task_struct *p)
930{
Eric W. Biederman92476d72006-03-31 02:31:42 -0800931 return p->pids[PIDTYPE_PID].pid != NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700932}
933
934extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700935#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -0800936
Andrew Morton158d9eb2006-03-31 02:31:34 -0800937extern void __put_task_struct(struct task_struct *t);
Ingo Molnare56d0902006-01-08 01:01:37 -0800938
939static inline void put_task_struct(struct task_struct *t)
940{
941 if (atomic_dec_and_test(&t->usage))
Eric W. Biederman8c7904a2006-03-31 02:31:37 -0800942 __put_task_struct(t);
Ingo Molnare56d0902006-01-08 01:01:37 -0800943}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700944
945/*
946 * Per process flags
947 */
948#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
949 /* Not implemented yet, only for 486*/
950#define PF_STARTING 0x00000002 /* being created */
951#define PF_EXITING 0x00000004 /* getting shut down */
952#define PF_DEAD 0x00000008 /* Dead */
953#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
954#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
955#define PF_DUMPCORE 0x00000200 /* dumped core */
956#define PF_SIGNALED 0x00000400 /* killed by a signal */
957#define PF_MEMALLOC 0x00000800 /* Allocating memory */
958#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
959#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
960#define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */
961#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
962#define PF_FROZEN 0x00010000 /* frozen for system suspend */
963#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
964#define PF_KSWAPD 0x00040000 /* I am kswapd */
965#define PF_SWAPOFF 0x00080000 /* I am in swapoff */
966#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
Jens Axboeb31dc662006-06-13 08:26:10 +0200967#define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
968#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
969#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
970#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
971#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
Paul Jacksonc61afb12006-03-24 03:16:08 -0800972#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700973
974/*
975 * Only the _current_ task can read/write to tsk->flags, but other
976 * tasks can access tsk->flags in readonly mode for example
977 * with tsk_used_math (like during threaded core dumping).
978 * There is however an exception to this rule during ptrace
979 * or during fork: the ptracer task is allowed to write to the
980 * child->flags of its traced child (same goes for fork, the parent
981 * can write to the child->flags), because we're guaranteed the
982 * child is not running and in turn not changing child->flags
983 * at the same time the parent does it.
984 */
985#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
986#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
987#define clear_used_math() clear_stopped_child_used_math(current)
988#define set_used_math() set_stopped_child_used_math(current)
989#define conditional_stopped_child_used_math(condition, child) \
990 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
991#define conditional_used_math(condition) \
992 conditional_stopped_child_used_math(condition, current)
993#define copy_to_stopped_child_used_math(child) \
994 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
995/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
996#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
997#define used_math() tsk_used_math(current)
998
999#ifdef CONFIG_SMP
1000extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
1001#else
1002static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
1003{
Paul Jackson4098f992005-10-30 15:03:21 -08001004 if (!cpu_isset(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001005 return -EINVAL;
1006 return 0;
1007}
1008#endif
1009
1010extern unsigned long long sched_clock(void);
1011extern unsigned long long current_sched_time(const task_t *current_task);
1012
1013/* sched_exec is called by processes performing an exec */
1014#ifdef CONFIG_SMP
1015extern void sched_exec(void);
1016#else
1017#define sched_exec() {}
1018#endif
1019
1020#ifdef CONFIG_HOTPLUG_CPU
1021extern void idle_task_exit(void);
1022#else
1023static inline void idle_task_exit(void) {}
1024#endif
1025
1026extern void sched_idle_next(void);
Ingo Molnarb29739f2006-06-27 02:54:51 -07001027
1028#ifdef CONFIG_RT_MUTEXES
1029extern int rt_mutex_getprio(task_t *p);
1030extern void rt_mutex_setprio(task_t *p, int prio);
1031#else
1032static inline int rt_mutex_getprio(task_t *p)
1033{
1034 return p->normal_prio;
1035}
1036#endif
1037
Linus Torvalds1da177e2005-04-16 15:20:36 -07001038extern void set_user_nice(task_t *p, long nice);
1039extern int task_prio(const task_t *p);
1040extern int task_nice(const task_t *p);
Matt Mackalle43379f2005-05-01 08:59:00 -07001041extern int can_nice(const task_t *p, const int nice);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001042extern int task_curr(const task_t *p);
1043extern int idle_cpu(int cpu);
1044extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1045extern task_t *idle_task(int cpu);
Keith Owensa2a97982005-09-11 17:19:06 +10001046extern task_t *curr_task(int cpu);
1047extern void set_curr_task(int cpu, task_t *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001048
1049void yield(void);
1050
1051/*
1052 * The default (Linux) execution domain.
1053 */
1054extern struct exec_domain default_exec_domain;
1055
1056union thread_union {
1057 struct thread_info thread_info;
1058 unsigned long stack[THREAD_SIZE/sizeof(long)];
1059};
1060
1061#ifndef __HAVE_ARCH_KSTACK_END
1062static inline int kstack_end(void *addr)
1063{
1064 /* Reliable end of stack detection:
1065 * Some APM bios versions misalign the stack
1066 */
1067 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1068}
1069#endif
1070
1071extern union thread_union init_thread_union;
1072extern struct task_struct init_task;
1073
1074extern struct mm_struct init_mm;
1075
1076#define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1077extern struct task_struct *find_task_by_pid_type(int type, int pid);
1078extern void set_special_pids(pid_t session, pid_t pgrp);
1079extern void __set_special_pids(pid_t session, pid_t pgrp);
1080
1081/* per-UID process charging. */
1082extern struct user_struct * alloc_uid(uid_t);
1083static inline struct user_struct *get_uid(struct user_struct *u)
1084{
1085 atomic_inc(&u->__count);
1086 return u;
1087}
1088extern void free_uid(struct user_struct *);
1089extern void switch_uid(struct user_struct *);
1090
1091#include <asm/current.h>
1092
1093extern void do_timer(struct pt_regs *);
1094
1095extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1096extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1097extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1098 unsigned long clone_flags));
1099#ifdef CONFIG_SMP
1100 extern void kick_process(struct task_struct *tsk);
1101#else
1102 static inline void kick_process(struct task_struct *tsk) { }
1103#endif
Nick Piggin476d1392005-06-25 14:57:29 -07001104extern void FASTCALL(sched_fork(task_t * p, int clone_flags));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001105extern void FASTCALL(sched_exit(task_t * p));
1106
1107extern int in_group_p(gid_t);
1108extern int in_egroup_p(gid_t);
1109
1110extern void proc_caches_init(void);
1111extern void flush_signals(struct task_struct *);
1112extern void flush_signal_handlers(struct task_struct *, int force_default);
1113extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1114
1115static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1116{
1117 unsigned long flags;
1118 int ret;
1119
1120 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1121 ret = dequeue_signal(tsk, mask, info);
1122 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1123
1124 return ret;
1125}
1126
1127extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1128 sigset_t *mask);
1129extern void unblock_all_signals(void);
1130extern void release_task(struct task_struct * p);
1131extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1132extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1133extern int force_sigsegv(int, struct task_struct *);
1134extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1135extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1136extern int kill_pg_info(int, struct siginfo *, pid_t);
1137extern int kill_proc_info(int, struct siginfo *, pid_t);
Harald Welte46113832005-10-10 19:44:29 +02001138extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001139extern void do_notify_parent(struct task_struct *, int);
1140extern void force_sig(int, struct task_struct *);
1141extern void force_sig_specific(int, struct task_struct *);
1142extern int send_sig(int, struct task_struct *, int);
1143extern void zap_other_threads(struct task_struct *p);
1144extern int kill_pg(pid_t, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001145extern int kill_proc(pid_t, int, int);
1146extern struct sigqueue *sigqueue_alloc(void);
1147extern void sigqueue_free(struct sigqueue *);
1148extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1149extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03001150extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001151extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1152
1153/* These can be the second arg to send_sig_info/send_group_sig_info. */
1154#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1155#define SEND_SIG_PRIV ((struct siginfo *) 1)
1156#define SEND_SIG_FORCED ((struct siginfo *) 2)
1157
Oleg Nesterov621d3122005-10-30 15:03:45 -08001158static inline int is_si_special(const struct siginfo *info)
1159{
1160 return info <= SEND_SIG_FORCED;
1161}
1162
Linus Torvalds1da177e2005-04-16 15:20:36 -07001163/* True if we are on the alternate signal stack. */
1164
1165static inline int on_sig_stack(unsigned long sp)
1166{
1167 return (sp - current->sas_ss_sp < current->sas_ss_size);
1168}
1169
1170static inline int sas_ss_flags(unsigned long sp)
1171{
1172 return (current->sas_ss_size == 0 ? SS_DISABLE
1173 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1174}
1175
Linus Torvalds1da177e2005-04-16 15:20:36 -07001176/*
1177 * Routines for handling mm_structs
1178 */
1179extern struct mm_struct * mm_alloc(void);
1180
1181/* mmdrop drops the mm and the page tables */
1182extern void FASTCALL(__mmdrop(struct mm_struct *));
1183static inline void mmdrop(struct mm_struct * mm)
1184{
1185 if (atomic_dec_and_test(&mm->mm_count))
1186 __mmdrop(mm);
1187}
1188
1189/* mmput gets rid of the mappings and all user-space */
1190extern void mmput(struct mm_struct *);
1191/* Grab a reference to a task's mm, if it is not already going away */
1192extern struct mm_struct *get_task_mm(struct task_struct *task);
1193/* Remove the current tasks stale references to the old mm_struct */
1194extern void mm_release(struct task_struct *, struct mm_struct *);
1195
1196extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1197extern void flush_thread(void);
1198extern void exit_thread(void);
1199
Linus Torvalds1da177e2005-04-16 15:20:36 -07001200extern void exit_files(struct task_struct *);
Oleg Nesterov6b3934e2006-03-28 16:11:16 -08001201extern void __cleanup_signal(struct signal_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08001202extern void __cleanup_sighand(struct sighand_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001203extern void exit_itimers(struct signal_struct *);
1204
1205extern NORET_TYPE void do_group_exit(int);
1206
Linus Torvalds1da177e2005-04-16 15:20:36 -07001207extern void daemonize(const char *, ...);
1208extern int allow_signal(int);
1209extern int disallow_signal(int);
1210extern task_t *child_reaper;
1211
1212extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1213extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1214task_t *fork_idle(int);
1215
1216extern void set_task_comm(struct task_struct *tsk, char *from);
1217extern void get_task_comm(char *to, struct task_struct *tsk);
1218
1219#ifdef CONFIG_SMP
1220extern void wait_task_inactive(task_t * p);
1221#else
1222#define wait_task_inactive(p) do { } while (0)
1223#endif
1224
1225#define remove_parent(p) list_del_init(&(p)->sibling)
Oleg Nesterov8fafabd2006-03-28 16:11:05 -08001226#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001227
Eric W. Biederman5e85d4a2006-04-18 22:20:16 -07001228#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001229
1230#define for_each_process(p) \
1231 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1232
1233/*
1234 * Careful: do_each_thread/while_each_thread is a double loop so
1235 * 'break' will not work as expected - use goto instead.
1236 */
1237#define do_each_thread(g, t) \
1238 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1239
1240#define while_each_thread(g, t) \
1241 while ((t = next_thread(t)) != g)
1242
Eric W. Biedermande12a782006-04-10 17:16:49 -06001243/* de_thread depends on thread_group_leader not being a pid based check */
1244#define thread_group_leader(p) (p == p->group_leader)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001245
Keith Owensa9cdf412006-04-10 22:54:07 -07001246static inline task_t *next_thread(const task_t *p)
Oleg Nesterov47e65322006-03-28 16:11:25 -08001247{
1248 return list_entry(rcu_dereference(p->thread_group.next),
1249 task_t, thread_group);
1250}
1251
Linus Torvalds1da177e2005-04-16 15:20:36 -07001252static inline int thread_group_empty(task_t *p)
1253{
Oleg Nesterov47e65322006-03-28 16:11:25 -08001254 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001255}
1256
1257#define delay_group_leader(p) \
1258 (thread_group_leader(p) && !thread_group_empty(p))
1259
Linus Torvalds1da177e2005-04-16 15:20:36 -07001260/*
Eric W. Biederman260ea102006-06-23 02:05:18 -07001261 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02001262 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Jackson053199e2005-10-30 15:02:30 -08001263 * pins the final release of task.io_context. Also protects ->cpuset.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001264 *
1265 * Nests both inside and outside of read_lock(&tasklist_lock).
1266 * It must not be nested with write_lock_irq(&tasklist_lock),
1267 * neither inside nor outside.
1268 */
1269static inline void task_lock(struct task_struct *p)
1270{
1271 spin_lock(&p->alloc_lock);
1272}
1273
1274static inline void task_unlock(struct task_struct *p)
1275{
1276 spin_unlock(&p->alloc_lock);
1277}
1278
Oleg Nesterovf63ee722006-03-28 16:11:13 -08001279extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1280 unsigned long *flags);
1281
1282static inline void unlock_task_sighand(struct task_struct *tsk,
1283 unsigned long *flags)
1284{
1285 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1286}
1287
Al Virof0373602005-11-13 16:06:57 -08001288#ifndef __HAVE_THREAD_FUNCTIONS
1289
Al Viroa1261f52005-11-13 16:06:55 -08001290#define task_thread_info(task) (task)->thread_info
Al Viro9fc65872006-01-12 01:05:34 -08001291#define task_stack_page(task) ((void*)((task)->thread_info))
Al Viroa1261f52005-11-13 16:06:55 -08001292
Al Viro10ebffd2005-11-13 16:06:56 -08001293static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1294{
1295 *task_thread_info(p) = *task_thread_info(org);
1296 task_thread_info(p)->task = p;
1297}
1298
1299static inline unsigned long *end_of_stack(struct task_struct *p)
1300{
1301 return (unsigned long *)(p->thread_info + 1);
1302}
1303
Al Virof0373602005-11-13 16:06:57 -08001304#endif
1305
Linus Torvalds1da177e2005-04-16 15:20:36 -07001306/* set thread flags in other task's structures
1307 * - see asm/thread_info.h for TIF_xxxx flags available
1308 */
1309static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1310{
Al Viroa1261f52005-11-13 16:06:55 -08001311 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001312}
1313
1314static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1315{
Al Viroa1261f52005-11-13 16:06:55 -08001316 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001317}
1318
1319static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1320{
Al Viroa1261f52005-11-13 16:06:55 -08001321 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001322}
1323
1324static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1325{
Al Viroa1261f52005-11-13 16:06:55 -08001326 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001327}
1328
1329static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1330{
Al Viroa1261f52005-11-13 16:06:55 -08001331 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001332}
1333
1334static inline void set_tsk_need_resched(struct task_struct *tsk)
1335{
1336 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1337}
1338
1339static inline void clear_tsk_need_resched(struct task_struct *tsk)
1340{
1341 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1342}
1343
1344static inline int signal_pending(struct task_struct *p)
1345{
1346 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1347}
1348
1349static inline int need_resched(void)
1350{
1351 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1352}
1353
1354/*
1355 * cond_resched() and cond_resched_lock(): latency reduction via
1356 * explicit rescheduling in places that are safe. The return
1357 * value indicates whether a reschedule was done in fact.
1358 * cond_resched_lock() will drop the spinlock before scheduling,
1359 * cond_resched_softirq() will enable bhs before scheduling.
1360 */
1361extern int cond_resched(void);
1362extern int cond_resched_lock(spinlock_t * lock);
1363extern int cond_resched_softirq(void);
1364
1365/*
1366 * Does a critical section need to be broken due to another
1367 * task waiting?:
1368 */
1369#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1370# define need_lockbreak(lock) ((lock)->break_lock)
1371#else
1372# define need_lockbreak(lock) 0
1373#endif
1374
1375/*
1376 * Does a critical section need to be broken due to another
1377 * task waiting or preemption being signalled:
1378 */
1379static inline int lock_need_resched(spinlock_t *lock)
1380{
1381 if (need_lockbreak(lock) || need_resched())
1382 return 1;
1383 return 0;
1384}
1385
1386/* Reevaluate whether the task has signals pending delivery.
1387 This is required every time the blocked sigset_t changes.
1388 callers must hold sighand->siglock. */
1389
1390extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1391extern void recalc_sigpending(void);
1392
1393extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1394
1395/*
1396 * Wrappers for p->thread_info->cpu access. No-op on UP.
1397 */
1398#ifdef CONFIG_SMP
1399
1400static inline unsigned int task_cpu(const struct task_struct *p)
1401{
Al Viroa1261f52005-11-13 16:06:55 -08001402 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001403}
1404
1405static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1406{
Al Viroa1261f52005-11-13 16:06:55 -08001407 task_thread_info(p)->cpu = cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001408}
1409
1410#else
1411
1412static inline unsigned int task_cpu(const struct task_struct *p)
1413{
1414 return 0;
1415}
1416
1417static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1418{
1419}
1420
1421#endif /* CONFIG_SMP */
1422
1423#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1424extern void arch_pick_mmap_layout(struct mm_struct *mm);
1425#else
1426static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1427{
1428 mm->mmap_base = TASK_UNMAPPED_BASE;
1429 mm->get_unmapped_area = arch_get_unmapped_area;
1430 mm->unmap_area = arch_unmap_area;
1431}
1432#endif
1433
1434extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1435extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1436
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -07001437#include <linux/sysdev.h>
1438extern int sched_mc_power_savings, sched_smt_power_savings;
1439extern struct sysdev_attribute attr_sched_mc_power_savings, attr_sched_smt_power_savings;
1440extern int sched_create_sysfs_power_savings_entries(struct sysdev_class *cls);
1441
Linus Torvalds1da177e2005-04-16 15:20:36 -07001442extern void normalize_rt_tasks(void);
1443
Linus Torvalds1da177e2005-04-16 15:20:36 -07001444#ifdef CONFIG_PM
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001445/*
1446 * Check if a process has been frozen
1447 */
1448static inline int frozen(struct task_struct *p)
1449{
1450 return p->flags & PF_FROZEN;
1451}
1452
1453/*
1454 * Check if there is a request to freeze a process
1455 */
1456static inline int freezing(struct task_struct *p)
1457{
1458 return p->flags & PF_FREEZE;
1459}
1460
1461/*
1462 * Request that a process be frozen
1463 * FIXME: SMP problem. We may not modify other process' flags!
1464 */
1465static inline void freeze(struct task_struct *p)
1466{
1467 p->flags |= PF_FREEZE;
1468}
1469
1470/*
1471 * Wake up a frozen process
1472 */
1473static inline int thaw_process(struct task_struct *p)
1474{
1475 if (frozen(p)) {
1476 p->flags &= ~PF_FROZEN;
1477 wake_up_process(p);
1478 return 1;
1479 }
1480 return 0;
1481}
1482
1483/*
1484 * freezing is complete, mark process as frozen
1485 */
1486static inline void frozen_process(struct task_struct *p)
1487{
1488 p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1489}
1490
1491extern void refrigerator(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001492extern int freeze_processes(void);
1493extern void thaw_processes(void);
1494
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001495static inline int try_to_freeze(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001496{
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001497 if (freezing(current)) {
1498 refrigerator();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001499 return 1;
1500 } else
1501 return 0;
1502}
1503#else
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001504static inline int frozen(struct task_struct *p) { return 0; }
1505static inline int freezing(struct task_struct *p) { return 0; }
1506static inline void freeze(struct task_struct *p) { BUG(); }
1507static inline int thaw_process(struct task_struct *p) { return 1; }
1508static inline void frozen_process(struct task_struct *p) { BUG(); }
1509
1510static inline void refrigerator(void) {}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001511static inline int freeze_processes(void) { BUG(); return 0; }
1512static inline void thaw_processes(void) {}
1513
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001514static inline int try_to_freeze(void) { return 0; }
1515
Linus Torvalds1da177e2005-04-16 15:20:36 -07001516#endif /* CONFIG_PM */
1517#endif /* __KERNEL__ */
1518
1519#endif