blob: b0ad6f30679eac4cd128f924bd8b1010b4c75be1 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
4#include <asm/param.h> /* for HZ */
5
6#include <linux/config.h>
7#include <linux/capability.h>
8#include <linux/threads.h>
9#include <linux/kernel.h>
10#include <linux/types.h>
11#include <linux/timex.h>
12#include <linux/jiffies.h>
13#include <linux/rbtree.h>
14#include <linux/thread_info.h>
15#include <linux/cpumask.h>
16#include <linux/errno.h>
17#include <linux/nodemask.h>
18
19#include <asm/system.h>
20#include <asm/semaphore.h>
21#include <asm/page.h>
22#include <asm/ptrace.h>
23#include <asm/mmu.h>
24#include <asm/cputime.h>
25
26#include <linux/smp.h>
27#include <linux/sem.h>
28#include <linux/signal.h>
29#include <linux/securebits.h>
30#include <linux/fs_struct.h>
31#include <linux/compiler.h>
32#include <linux/completion.h>
33#include <linux/pid.h>
34#include <linux/percpu.h>
35#include <linux/topology.h>
36#include <linux/seccomp.h>
37
H. J. Lu36d57ac2005-09-06 15:16:49 -070038#include <linux/auxvec.h> /* For AT_VECTOR_SIZE */
39
Linus Torvalds1da177e2005-04-16 15:20:36 -070040struct exec_domain;
41
42/*
43 * cloning flags:
44 */
45#define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
46#define CLONE_VM 0x00000100 /* set if VM shared between processes */
47#define CLONE_FS 0x00000200 /* set if fs info shared between processes */
48#define CLONE_FILES 0x00000400 /* set if open files shared between processes */
49#define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
50#define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
51#define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
52#define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
53#define CLONE_THREAD 0x00010000 /* Same thread group? */
54#define CLONE_NEWNS 0x00020000 /* New namespace group? */
55#define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
56#define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
57#define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
58#define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
59#define CLONE_DETACHED 0x00400000 /* Unused, ignored */
60#define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
61#define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
62#define CLONE_STOPPED 0x02000000 /* Start in stopped state */
63
64/*
65 * List of flags we want to share for kernel threads,
66 * if only because they are not used by them anyway.
67 */
68#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
69
70/*
71 * These are the constant used to fake the fixed-point load-average
72 * counting. Some notes:
73 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
74 * a load-average precision of 10 bits integer + 11 bits fractional
75 * - if you want to count load-averages more often, you need more
76 * precision, or rounding will get you. With 2-second counting freq,
77 * the EXP_n values would be 1981, 2034 and 2043 if still using only
78 * 11 bit fractions.
79 */
80extern unsigned long avenrun[]; /* Load averages */
81
82#define FSHIFT 11 /* nr of bits of precision */
83#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
84#define LOAD_FREQ (5*HZ) /* 5 sec intervals */
85#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
86#define EXP_5 2014 /* 1/exp(5sec/5min) */
87#define EXP_15 2037 /* 1/exp(5sec/15min) */
88
89#define CALC_LOAD(load,exp,n) \
90 load *= exp; \
91 load += n*(FIXED_1-exp); \
92 load >>= FSHIFT;
93
94extern unsigned long total_forks;
95extern int nr_threads;
96extern int last_pid;
97DECLARE_PER_CPU(unsigned long, process_counts);
98extern int nr_processes(void);
99extern unsigned long nr_running(void);
100extern unsigned long nr_uninterruptible(void);
101extern unsigned long nr_iowait(void);
102
103#include <linux/time.h>
104#include <linux/param.h>
105#include <linux/resource.h>
106#include <linux/timer.h>
107
108#include <asm/processor.h>
109
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700110/*
111 * Task state bitmask. NOTE! These bits are also
112 * encoded in fs/proc/array.c: get_task_state().
113 *
114 * We have two separate sets of flags: task->state
115 * is about runnability, while task->exit_state are
116 * about the task exiting. Confusing, but this way
117 * modifying one set can't modify the other one by
118 * mistake.
119 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700120#define TASK_RUNNING 0
121#define TASK_INTERRUPTIBLE 1
122#define TASK_UNINTERRUPTIBLE 2
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700123#define TASK_STOPPED 4
124#define TASK_TRACED 8
125/* in tsk->exit_state */
126#define EXIT_ZOMBIE 16
127#define EXIT_DEAD 32
128/* in tsk->state again */
129#define TASK_NONINTERACTIVE 64
Linus Torvalds1da177e2005-04-16 15:20:36 -0700130
131#define __set_task_state(tsk, state_value) \
132 do { (tsk)->state = (state_value); } while (0)
133#define set_task_state(tsk, state_value) \
134 set_mb((tsk)->state, (state_value))
135
Andrew Morton498d0c52005-09-13 01:25:14 -0700136/*
137 * set_current_state() includes a barrier so that the write of current->state
138 * is correctly serialised wrt the caller's subsequent test of whether to
139 * actually sleep:
140 *
141 * set_current_state(TASK_UNINTERRUPTIBLE);
142 * if (do_i_need_to_sleep())
143 * schedule();
144 *
145 * If the caller does not need such serialisation then use __set_current_state()
146 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147#define __set_current_state(state_value) \
148 do { current->state = (state_value); } while (0)
149#define set_current_state(state_value) \
150 set_mb(current->state, (state_value))
151
152/* Task command name length */
153#define TASK_COMM_LEN 16
154
155/*
156 * Scheduling policies
157 */
158#define SCHED_NORMAL 0
159#define SCHED_FIFO 1
160#define SCHED_RR 2
161
162struct sched_param {
163 int sched_priority;
164};
165
166#ifdef __KERNEL__
167
168#include <linux/spinlock.h>
169
170/*
171 * This serializes "schedule()" and also protects
172 * the run-queue from deletions/modifications (but
173 * _adding_ to the beginning of the run-queue has
174 * a separate lock).
175 */
176extern rwlock_t tasklist_lock;
177extern spinlock_t mmlist_lock;
178
179typedef struct task_struct task_t;
180
181extern void sched_init(void);
182extern void sched_init_smp(void);
183extern void init_idle(task_t *idle, int cpu);
184
185extern cpumask_t nohz_cpu_mask;
186
187extern void show_state(void);
188extern void show_regs(struct pt_regs *);
189
190/*
191 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
192 * task), SP is the stack pointer of the first frame that should be shown in the back
193 * trace (or NULL if the entire call-chain of the task should be shown).
194 */
195extern void show_stack(struct task_struct *task, unsigned long *sp);
196
197void io_schedule(void);
198long io_schedule_timeout(long timeout);
199
200extern void cpu_init (void);
201extern void trap_init(void);
202extern void update_process_times(int user);
203extern void scheduler_tick(void);
204
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700205#ifdef CONFIG_DETECT_SOFTLOCKUP
206extern void softlockup_tick(struct pt_regs *regs);
207extern void spawn_softlockup_task(void);
208extern void touch_softlockup_watchdog(void);
209#else
210static inline void softlockup_tick(struct pt_regs *regs)
211{
212}
213static inline void spawn_softlockup_task(void)
214{
215}
216static inline void touch_softlockup_watchdog(void)
217{
218}
219#endif
220
221
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222/* Attach to any functions which should be ignored in wchan output. */
223#define __sched __attribute__((__section__(".sched.text")))
224/* Is this address in the __sched functions? */
225extern int in_sched_functions(unsigned long addr);
226
227#define MAX_SCHEDULE_TIMEOUT LONG_MAX
228extern signed long FASTCALL(schedule_timeout(signed long timeout));
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700229extern signed long schedule_timeout_interruptible(signed long timeout);
230extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700231asmlinkage void schedule(void);
232
233struct namespace;
234
235/* Maximum number of active map areas.. This is a random (large) number */
236#define DEFAULT_MAX_MAP_COUNT 65536
237
238extern int sysctl_max_map_count;
239
240#include <linux/aio.h>
241
242extern unsigned long
243arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
244 unsigned long, unsigned long);
245extern unsigned long
246arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
247 unsigned long len, unsigned long pgoff,
248 unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700249extern void arch_unmap_area(struct mm_struct *, unsigned long);
250extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700251
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700252#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
253/*
254 * The mm counters are not protected by its page_table_lock,
255 * so must be incremented atomically.
256 */
257#ifdef ATOMIC64_INIT
258#define set_mm_counter(mm, member, value) atomic64_set(&(mm)->_##member, value)
259#define get_mm_counter(mm, member) ((unsigned long)atomic64_read(&(mm)->_##member))
260#define add_mm_counter(mm, member, value) atomic64_add(value, &(mm)->_##member)
261#define inc_mm_counter(mm, member) atomic64_inc(&(mm)->_##member)
262#define dec_mm_counter(mm, member) atomic64_dec(&(mm)->_##member)
263typedef atomic64_t mm_counter_t;
264#else /* !ATOMIC64_INIT */
265/*
266 * The counters wrap back to 0 at 2^32 * PAGE_SIZE,
267 * that is, at 16TB if using 4kB page size.
268 */
269#define set_mm_counter(mm, member, value) atomic_set(&(mm)->_##member, value)
270#define get_mm_counter(mm, member) ((unsigned long)atomic_read(&(mm)->_##member))
271#define add_mm_counter(mm, member, value) atomic_add(value, &(mm)->_##member)
272#define inc_mm_counter(mm, member) atomic_inc(&(mm)->_##member)
273#define dec_mm_counter(mm, member) atomic_dec(&(mm)->_##member)
274typedef atomic_t mm_counter_t;
275#endif /* !ATOMIC64_INIT */
276
277#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
278/*
279 * The mm counters are protected by its page_table_lock,
280 * so can be incremented directly.
281 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700282#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
283#define get_mm_counter(mm, member) ((mm)->_##member)
284#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
285#define inc_mm_counter(mm, member) (mm)->_##member++
286#define dec_mm_counter(mm, member) (mm)->_##member--
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700287typedef unsigned long mm_counter_t;
Hugh Dickins42946212005-10-29 18:16:05 -0700288
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700289#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
290
291#define get_mm_rss(mm) \
292 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
Hugh Dickins365e9c872005-10-29 18:16:18 -0700293#define update_hiwater_rss(mm) do { \
294 unsigned long _rss = get_mm_rss(mm); \
295 if ((mm)->hiwater_rss < _rss) \
296 (mm)->hiwater_rss = _rss; \
297} while (0)
298#define update_hiwater_vm(mm) do { \
299 if ((mm)->hiwater_vm < (mm)->total_vm) \
300 (mm)->hiwater_vm = (mm)->total_vm; \
301} while (0)
302
Linus Torvalds1da177e2005-04-16 15:20:36 -0700303struct mm_struct {
304 struct vm_area_struct * mmap; /* list of VMAs */
305 struct rb_root mm_rb;
306 struct vm_area_struct * mmap_cache; /* last find_vma result */
307 unsigned long (*get_unmapped_area) (struct file *filp,
308 unsigned long addr, unsigned long len,
309 unsigned long pgoff, unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700310 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
311 unsigned long mmap_base; /* base of mmap area */
312 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
313 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700314 pgd_t * pgd;
315 atomic_t mm_users; /* How many users with user space? */
316 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
317 int map_count; /* number of VMAs */
318 struct rw_semaphore mmap_sem;
319 spinlock_t page_table_lock; /* Protects page tables and some counters */
320
321 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
322 * together off init_mm.mmlist, and are protected
323 * by mmlist_lock
324 */
325
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700326 /* Special counters, in some configurations protected by the
327 * page_table_lock, in other configurations by being atomic.
328 */
Hugh Dickins42946212005-10-29 18:16:05 -0700329 mm_counter_t _file_rss;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700330 mm_counter_t _anon_rss;
331
Hugh Dickinsf4499522005-10-29 18:16:19 -0700332 unsigned long hiwater_rss; /* High-watermark of RSS usage */
333 unsigned long hiwater_vm; /* High-water virtual memory usage */
334
335 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
336 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
337 unsigned long start_code, end_code, start_data, end_data;
338 unsigned long start_brk, brk, start_stack;
339 unsigned long arg_start, arg_end, env_start, env_end;
340
H. J. Lu36d57ac2005-09-06 15:16:49 -0700341 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700342
Alan Coxd6e71142005-06-23 00:09:43 -0700343 unsigned dumpable:2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700344 cpumask_t cpu_vm_mask;
345
346 /* Architecture-specific MM context */
347 mm_context_t context;
348
349 /* Token based thrashing protection. */
350 unsigned long swap_token_time;
351 char recent_pagein;
352
353 /* coredumping support */
354 int core_waiters;
355 struct completion *core_startup_done, core_done;
356
357 /* aio bits */
358 rwlock_t ioctx_list_lock;
359 struct kioctx *ioctx_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700360};
361
362struct sighand_struct {
363 atomic_t count;
364 struct k_sigaction action[_NSIG];
365 spinlock_t siglock;
366};
367
368/*
369 * NOTE! "signal_struct" does not have it's own
370 * locking, because a shared signal_struct always
371 * implies a shared sighand_struct, so locking
372 * sighand_struct is always a proper superset of
373 * the locking of signal_struct.
374 */
375struct signal_struct {
376 atomic_t count;
377 atomic_t live;
378
379 wait_queue_head_t wait_chldexit; /* for wait4() */
380
381 /* current thread group signal load-balancing target: */
382 task_t *curr_target;
383
384 /* shared signal handling: */
385 struct sigpending shared_pending;
386
387 /* thread group exit support */
388 int group_exit_code;
389 /* overloaded:
390 * - notify group_exit_task when ->count is equal to notify_count
391 * - everyone except group_exit_task is stopped during signal delivery
392 * of fatal signals, group_exit_task processes the signal.
393 */
394 struct task_struct *group_exit_task;
395 int notify_count;
396
397 /* thread group stop support, overloads group_exit_code too */
398 int group_stop_count;
399 unsigned int flags; /* see SIGNAL_* flags below */
400
401 /* POSIX.1b Interval Timers */
402 struct list_head posix_timers;
403
404 /* ITIMER_REAL timer for the process */
405 struct timer_list real_timer;
406 unsigned long it_real_value, it_real_incr;
407
408 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
409 cputime_t it_prof_expires, it_virt_expires;
410 cputime_t it_prof_incr, it_virt_incr;
411
412 /* job control IDs */
413 pid_t pgrp;
414 pid_t tty_old_pgrp;
415 pid_t session;
416 /* boolean value for session group leader */
417 int leader;
418
419 struct tty_struct *tty; /* NULL if no tty */
420
421 /*
422 * Cumulative resource counters for dead threads in the group,
423 * and for reaped dead child processes forked by this group.
424 * Live threads maintain their own counters and add to these
425 * in __exit_signal, except for the group leader.
426 */
427 cputime_t utime, stime, cutime, cstime;
428 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
429 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
430
431 /*
432 * Cumulative ns of scheduled CPU time for dead threads in the
433 * group, not including a zombie group leader. (This only differs
434 * from jiffies_to_ns(utime + stime) if sched_clock uses something
435 * other than jiffies.)
436 */
437 unsigned long long sched_time;
438
439 /*
440 * We don't bother to synchronize most readers of this at all,
441 * because there is no reader checking a limit that actually needs
442 * to get both rlim_cur and rlim_max atomically, and either one
443 * alone is a single word that can safely be read normally.
444 * getrlimit/setrlimit use task_lock(current->group_leader) to
445 * protect this instead of the siglock, because they really
446 * have no need to disable irqs.
447 */
448 struct rlimit rlim[RLIM_NLIMITS];
449
450 struct list_head cpu_timers[3];
451
452 /* keep the process-shared keyrings here so that they do the right
453 * thing in threads created with CLONE_THREAD */
454#ifdef CONFIG_KEYS
455 struct key *session_keyring; /* keyring inherited over fork */
456 struct key *process_keyring; /* keyring private to this process */
457#endif
458};
459
Nick Piggin4866cde2005-06-25 14:57:23 -0700460/* Context switch must be unlocked if interrupts are to be enabled */
461#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
462# define __ARCH_WANT_UNLOCKED_CTXSW
463#endif
464
Linus Torvalds1da177e2005-04-16 15:20:36 -0700465/*
466 * Bits in flags field of signal_struct.
467 */
468#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
469#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
470#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
471#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
472
473
474/*
475 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
476 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL tasks are
477 * in the range MAX_RT_PRIO..MAX_PRIO-1. Priority values
478 * are inverted: lower p->prio value means higher priority.
479 *
480 * The MAX_USER_RT_PRIO value allows the actual maximum
481 * RT priority to be separate from the value exported to
482 * user-space. This allows kernel threads to set their
483 * priority to a value higher than any user task. Note:
484 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
485 */
486
487#define MAX_USER_RT_PRIO 100
488#define MAX_RT_PRIO MAX_USER_RT_PRIO
489
490#define MAX_PRIO (MAX_RT_PRIO + 40)
491
492#define rt_task(p) (unlikely((p)->prio < MAX_RT_PRIO))
493
494/*
495 * Some day this will be a full-fledged user tracking system..
496 */
497struct user_struct {
498 atomic_t __count; /* reference count */
499 atomic_t processes; /* How many processes does this user have? */
500 atomic_t files; /* How many open files does this user have? */
501 atomic_t sigpending; /* How many pending signals does this user have? */
Robert Love0eeca282005-07-12 17:06:03 -0400502#ifdef CONFIG_INOTIFY
503 atomic_t inotify_watches; /* How many inotify watches does this user have? */
504 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
505#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700506 /* protected by mq_lock */
507 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
508 unsigned long locked_shm; /* How many pages of mlocked shm ? */
509
510#ifdef CONFIG_KEYS
511 struct key *uid_keyring; /* UID specific keyring */
512 struct key *session_keyring; /* UID's default session keyring */
513#endif
514
515 /* Hash table maintenance information */
516 struct list_head uidhash_list;
517 uid_t uid;
518};
519
520extern struct user_struct *find_user(uid_t);
521
522extern struct user_struct root_user;
523#define INIT_USER (&root_user)
524
525typedef struct prio_array prio_array_t;
526struct backing_dev_info;
527struct reclaim_state;
528
529#ifdef CONFIG_SCHEDSTATS
530struct sched_info {
531 /* cumulative counters */
532 unsigned long cpu_time, /* time spent on the cpu */
533 run_delay, /* time spent waiting on a runqueue */
534 pcnt; /* # of timeslices run on this cpu */
535
536 /* timestamps */
537 unsigned long last_arrival, /* when we last ran on a cpu */
538 last_queued; /* when we were last queued to run */
539};
540
541extern struct file_operations proc_schedstat_operations;
542#endif
543
544enum idle_type
545{
546 SCHED_IDLE,
547 NOT_IDLE,
548 NEWLY_IDLE,
549 MAX_IDLE_TYPES
550};
551
552/*
553 * sched-domains (multiprocessor balancing) declarations:
554 */
555#ifdef CONFIG_SMP
556#define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
557
558#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
559#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
560#define SD_BALANCE_EXEC 4 /* Balance on exec */
Nick Piggin147cbb42005-06-25 14:57:19 -0700561#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
562#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
563#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
564#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
565#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700566
567struct sched_group {
568 struct sched_group *next; /* Must be a circular list */
569 cpumask_t cpumask;
570
571 /*
572 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
573 * single CPU. This is read only (except for setup, hotplug CPU).
574 */
575 unsigned long cpu_power;
576};
577
578struct sched_domain {
579 /* These fields must be setup */
580 struct sched_domain *parent; /* top domain must be null terminated */
581 struct sched_group *groups; /* the balancing groups of the domain */
582 cpumask_t span; /* span of all CPUs in this domain */
583 unsigned long min_interval; /* Minimum balance interval ms */
584 unsigned long max_interval; /* Maximum balance interval ms */
585 unsigned int busy_factor; /* less balancing by factor if busy */
586 unsigned int imbalance_pct; /* No balance until over watermark */
587 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
588 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
589 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
Nick Piggin78979862005-06-25 14:57:13 -0700590 unsigned int busy_idx;
591 unsigned int idle_idx;
592 unsigned int newidle_idx;
593 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700594 unsigned int forkexec_idx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700595 int flags; /* See SD_* */
596
597 /* Runtime fields. */
598 unsigned long last_balance; /* init to jiffies. units in jiffies */
599 unsigned int balance_interval; /* initialise to 1. units in ms. */
600 unsigned int nr_balance_failed; /* initialise to 0 */
601
602#ifdef CONFIG_SCHEDSTATS
603 /* load_balance() stats */
604 unsigned long lb_cnt[MAX_IDLE_TYPES];
605 unsigned long lb_failed[MAX_IDLE_TYPES];
606 unsigned long lb_balanced[MAX_IDLE_TYPES];
607 unsigned long lb_imbalance[MAX_IDLE_TYPES];
608 unsigned long lb_gained[MAX_IDLE_TYPES];
609 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
610 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
611 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
612
613 /* Active load balancing */
614 unsigned long alb_cnt;
615 unsigned long alb_failed;
616 unsigned long alb_pushed;
617
Nick Piggin68767a02005-06-25 14:57:20 -0700618 /* SD_BALANCE_EXEC stats */
619 unsigned long sbe_cnt;
620 unsigned long sbe_balanced;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700621 unsigned long sbe_pushed;
622
Nick Piggin68767a02005-06-25 14:57:20 -0700623 /* SD_BALANCE_FORK stats */
624 unsigned long sbf_cnt;
625 unsigned long sbf_balanced;
626 unsigned long sbf_pushed;
627
Linus Torvalds1da177e2005-04-16 15:20:36 -0700628 /* try_to_wake_up() stats */
629 unsigned long ttwu_wake_remote;
630 unsigned long ttwu_move_affine;
631 unsigned long ttwu_move_balance;
632#endif
633};
634
Dinakar Guniguntala1a20ff22005-06-25 14:57:33 -0700635extern void partition_sched_domains(cpumask_t *partition1,
636 cpumask_t *partition2);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700637#endif /* CONFIG_SMP */
638
639
640struct io_context; /* See blkdev.h */
641void exit_io_context(void);
642struct cpuset;
643
644#define NGROUPS_SMALL 32
645#define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
646struct group_info {
647 int ngroups;
648 atomic_t usage;
649 gid_t small_block[NGROUPS_SMALL];
650 int nblocks;
651 gid_t *blocks[0];
652};
653
654/*
655 * get_group_info() must be called with the owning task locked (via task_lock())
656 * when task != current. The reason being that the vast majority of callers are
657 * looking at current->group_info, which can not be changed except by the
658 * current task. Changing current->group_info requires the task lock, too.
659 */
660#define get_group_info(group_info) do { \
661 atomic_inc(&(group_info)->usage); \
662} while (0)
663
664#define put_group_info(group_info) do { \
665 if (atomic_dec_and_test(&(group_info)->usage)) \
666 groups_free(group_info); \
667} while (0)
668
David Howells3e301482005-06-23 22:00:56 -0700669extern struct group_info *groups_alloc(int gidsetsize);
670extern void groups_free(struct group_info *group_info);
671extern int set_current_groups(struct group_info *group_info);
672extern int groups_search(struct group_info *group_info, gid_t grp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700673/* access the groups "array" with this macro */
674#define GROUP_AT(gi, i) \
675 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
676
Chen, Kenneth W383f2832005-09-09 13:02:02 -0700677#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
678extern void prefetch_stack(struct task_struct*);
679#else
680static inline void prefetch_stack(struct task_struct *t) { }
681#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700682
683struct audit_context; /* See audit.c */
684struct mempolicy;
685
686struct task_struct {
687 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
688 struct thread_info *thread_info;
689 atomic_t usage;
690 unsigned long flags; /* per process flags, defined below */
691 unsigned long ptrace;
692
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700693 int lock_depth; /* BKL lock depth */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700694
Nick Piggin4866cde2005-06-25 14:57:23 -0700695#if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
696 int oncpu;
697#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698 int prio, static_prio;
699 struct list_head run_list;
700 prio_array_t *array;
701
Jens Axboe22e2c502005-06-27 10:55:12 +0200702 unsigned short ioprio;
703
Linus Torvalds1da177e2005-04-16 15:20:36 -0700704 unsigned long sleep_avg;
705 unsigned long long timestamp, last_ran;
706 unsigned long long sched_time; /* sched_clock time spent running */
707 int activated;
708
709 unsigned long policy;
710 cpumask_t cpus_allowed;
711 unsigned int time_slice, first_time_slice;
712
713#ifdef CONFIG_SCHEDSTATS
714 struct sched_info sched_info;
715#endif
716
717 struct list_head tasks;
718 /*
719 * ptrace_list/ptrace_children forms the list of my children
720 * that were stolen by a ptracer.
721 */
722 struct list_head ptrace_children;
723 struct list_head ptrace_list;
724
725 struct mm_struct *mm, *active_mm;
726
727/* task state */
728 struct linux_binfmt *binfmt;
729 long exit_state;
730 int exit_code, exit_signal;
731 int pdeath_signal; /* The signal sent when the parent dies */
732 /* ??? */
733 unsigned long personality;
734 unsigned did_exec:1;
735 pid_t pid;
736 pid_t tgid;
737 /*
738 * pointers to (original) parent process, youngest child, younger sibling,
739 * older sibling, respectively. (p->father can be replaced with
740 * p->parent->pid)
741 */
742 struct task_struct *real_parent; /* real parent process (when being debugged) */
743 struct task_struct *parent; /* parent process */
744 /*
745 * children/sibling forms the list of my children plus the
746 * tasks I'm ptracing.
747 */
748 struct list_head children; /* list of my children */
749 struct list_head sibling; /* linkage in my parent's children list */
750 struct task_struct *group_leader; /* threadgroup leader */
751
752 /* PID/PID hash table linkage. */
753 struct pid pids[PIDTYPE_MAX];
754
755 struct completion *vfork_done; /* for vfork() */
756 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
757 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
758
759 unsigned long rt_priority;
760 cputime_t utime, stime;
761 unsigned long nvcsw, nivcsw; /* context switch counts */
762 struct timespec start_time;
763/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
764 unsigned long min_flt, maj_flt;
765
766 cputime_t it_prof_expires, it_virt_expires;
767 unsigned long long it_sched_expires;
768 struct list_head cpu_timers[3];
769
770/* process credentials */
771 uid_t uid,euid,suid,fsuid;
772 gid_t gid,egid,sgid,fsgid;
773 struct group_info *group_info;
774 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
775 unsigned keep_capabilities:1;
776 struct user_struct *user;
777#ifdef CONFIG_KEYS
778 struct key *thread_keyring; /* keyring private to this thread */
David Howells3e301482005-06-23 22:00:56 -0700779 unsigned char jit_keyring; /* default keyring to attach requested keys to */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700780#endif
781 int oomkilladj; /* OOM kill score adjustment (bit shift). */
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -0700782 char comm[TASK_COMM_LEN]; /* executable name excluding path
783 - access with [gs]et_task_comm (which lock
784 it with task_lock())
785 - initialized normally by flush_old_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786/* file system info */
787 int link_count, total_link_count;
788/* ipc stuff */
789 struct sysv_sem sysvsem;
790/* CPU-specific state of this task */
791 struct thread_struct thread;
792/* filesystem information */
793 struct fs_struct *fs;
794/* open file information */
795 struct files_struct *files;
796/* namespace */
797 struct namespace *namespace;
798/* signal handlers */
799 struct signal_struct *signal;
800 struct sighand_struct *sighand;
801
802 sigset_t blocked, real_blocked;
803 struct sigpending pending;
804
805 unsigned long sas_ss_sp;
806 size_t sas_ss_size;
807 int (*notifier)(void *priv);
808 void *notifier_data;
809 sigset_t *notifier_mask;
810
811 void *security;
812 struct audit_context *audit_context;
813 seccomp_t seccomp;
814
815/* Thread group tracking */
816 u32 parent_exec_id;
817 u32 self_exec_id;
818/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
819 spinlock_t alloc_lock;
820/* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
821 spinlock_t proc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700822
823/* journalling filesystem info */
824 void *journal_info;
825
826/* VM state */
827 struct reclaim_state *reclaim_state;
828
829 struct dentry *proc_dentry;
830 struct backing_dev_info *backing_dev_info;
831
832 struct io_context *io_context;
833
834 unsigned long ptrace_message;
835 siginfo_t *last_siginfo; /* For ptrace use. */
836/*
837 * current io wait handle: wait queue entry to use for io waits
838 * If this thread is processing aio, this points at the waitqueue
839 * inside the currently handled kiocb. It may be NULL (i.e. default
840 * to a stack based synchronous wait) if its doing sync IO.
841 */
842 wait_queue_t *io_wait;
843/* i/o counters(bytes read/written, #syscalls */
844 u64 rchar, wchar, syscr, syscw;
845#if defined(CONFIG_BSD_PROCESS_ACCT)
846 u64 acct_rss_mem1; /* accumulated rss usage */
847 u64 acct_vm_mem1; /* accumulated virtual memory usage */
848 clock_t acct_stimexpd; /* clock_t-converted stime since last update */
849#endif
850#ifdef CONFIG_NUMA
851 struct mempolicy *mempolicy;
852 short il_next;
853#endif
854#ifdef CONFIG_CPUSETS
855 struct cpuset *cpuset;
856 nodemask_t mems_allowed;
857 int cpuset_mems_generation;
858#endif
Jens Axboe22e2c502005-06-27 10:55:12 +0200859 atomic_t fs_excl; /* holding fs exclusive resources */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700860};
861
862static inline pid_t process_group(struct task_struct *tsk)
863{
864 return tsk->signal->pgrp;
865}
866
867/**
868 * pid_alive - check that a task structure is not stale
869 * @p: Task structure to be checked.
870 *
871 * Test if a process is not yet dead (at most zombie state)
872 * If pid_alive fails, then pointers within the task structure
873 * can be stale and must not be dereferenced.
874 */
875static inline int pid_alive(struct task_struct *p)
876{
877 return p->pids[PIDTYPE_PID].nr != 0;
878}
879
880extern void free_task(struct task_struct *tsk);
881extern void __put_task_struct(struct task_struct *tsk);
882#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
883#define put_task_struct(tsk) \
884do { if (atomic_dec_and_test(&(tsk)->usage)) __put_task_struct(tsk); } while(0)
885
886/*
887 * Per process flags
888 */
889#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
890 /* Not implemented yet, only for 486*/
891#define PF_STARTING 0x00000002 /* being created */
892#define PF_EXITING 0x00000004 /* getting shut down */
893#define PF_DEAD 0x00000008 /* Dead */
894#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
895#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
896#define PF_DUMPCORE 0x00000200 /* dumped core */
897#define PF_SIGNALED 0x00000400 /* killed by a signal */
898#define PF_MEMALLOC 0x00000800 /* Allocating memory */
899#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
900#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
901#define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */
902#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
903#define PF_FROZEN 0x00010000 /* frozen for system suspend */
904#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
905#define PF_KSWAPD 0x00040000 /* I am kswapd */
906#define PF_SWAPOFF 0x00080000 /* I am in swapoff */
907#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
908#define PF_SYNCWRITE 0x00200000 /* I am doing a sync write */
909#define PF_BORROWED_MM 0x00400000 /* I am a kthread doing use_mm */
910#define PF_RANDOMIZE 0x00800000 /* randomize virtual address space */
911
912/*
913 * Only the _current_ task can read/write to tsk->flags, but other
914 * tasks can access tsk->flags in readonly mode for example
915 * with tsk_used_math (like during threaded core dumping).
916 * There is however an exception to this rule during ptrace
917 * or during fork: the ptracer task is allowed to write to the
918 * child->flags of its traced child (same goes for fork, the parent
919 * can write to the child->flags), because we're guaranteed the
920 * child is not running and in turn not changing child->flags
921 * at the same time the parent does it.
922 */
923#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
924#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
925#define clear_used_math() clear_stopped_child_used_math(current)
926#define set_used_math() set_stopped_child_used_math(current)
927#define conditional_stopped_child_used_math(condition, child) \
928 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
929#define conditional_used_math(condition) \
930 conditional_stopped_child_used_math(condition, current)
931#define copy_to_stopped_child_used_math(child) \
932 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
933/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
934#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
935#define used_math() tsk_used_math(current)
936
937#ifdef CONFIG_SMP
938extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
939#else
940static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
941{
Paul Jackson4098f992005-10-30 15:03:21 -0800942 if (!cpu_isset(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700943 return -EINVAL;
944 return 0;
945}
946#endif
947
948extern unsigned long long sched_clock(void);
949extern unsigned long long current_sched_time(const task_t *current_task);
950
951/* sched_exec is called by processes performing an exec */
952#ifdef CONFIG_SMP
953extern void sched_exec(void);
954#else
955#define sched_exec() {}
956#endif
957
958#ifdef CONFIG_HOTPLUG_CPU
959extern void idle_task_exit(void);
960#else
961static inline void idle_task_exit(void) {}
962#endif
963
964extern void sched_idle_next(void);
965extern void set_user_nice(task_t *p, long nice);
966extern int task_prio(const task_t *p);
967extern int task_nice(const task_t *p);
Matt Mackalle43379f2005-05-01 08:59:00 -0700968extern int can_nice(const task_t *p, const int nice);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700969extern int task_curr(const task_t *p);
970extern int idle_cpu(int cpu);
971extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
972extern task_t *idle_task(int cpu);
Keith Owensa2a97982005-09-11 17:19:06 +1000973extern task_t *curr_task(int cpu);
974extern void set_curr_task(int cpu, task_t *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700975
976void yield(void);
977
978/*
979 * The default (Linux) execution domain.
980 */
981extern struct exec_domain default_exec_domain;
982
983union thread_union {
984 struct thread_info thread_info;
985 unsigned long stack[THREAD_SIZE/sizeof(long)];
986};
987
988#ifndef __HAVE_ARCH_KSTACK_END
989static inline int kstack_end(void *addr)
990{
991 /* Reliable end of stack detection:
992 * Some APM bios versions misalign the stack
993 */
994 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
995}
996#endif
997
998extern union thread_union init_thread_union;
999extern struct task_struct init_task;
1000
1001extern struct mm_struct init_mm;
1002
1003#define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1004extern struct task_struct *find_task_by_pid_type(int type, int pid);
1005extern void set_special_pids(pid_t session, pid_t pgrp);
1006extern void __set_special_pids(pid_t session, pid_t pgrp);
1007
1008/* per-UID process charging. */
1009extern struct user_struct * alloc_uid(uid_t);
1010static inline struct user_struct *get_uid(struct user_struct *u)
1011{
1012 atomic_inc(&u->__count);
1013 return u;
1014}
1015extern void free_uid(struct user_struct *);
1016extern void switch_uid(struct user_struct *);
1017
1018#include <asm/current.h>
1019
1020extern void do_timer(struct pt_regs *);
1021
1022extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1023extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1024extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1025 unsigned long clone_flags));
1026#ifdef CONFIG_SMP
1027 extern void kick_process(struct task_struct *tsk);
1028#else
1029 static inline void kick_process(struct task_struct *tsk) { }
1030#endif
Nick Piggin476d1392005-06-25 14:57:29 -07001031extern void FASTCALL(sched_fork(task_t * p, int clone_flags));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001032extern void FASTCALL(sched_exit(task_t * p));
1033
1034extern int in_group_p(gid_t);
1035extern int in_egroup_p(gid_t);
1036
1037extern void proc_caches_init(void);
1038extern void flush_signals(struct task_struct *);
1039extern void flush_signal_handlers(struct task_struct *, int force_default);
1040extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1041
1042static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1043{
1044 unsigned long flags;
1045 int ret;
1046
1047 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1048 ret = dequeue_signal(tsk, mask, info);
1049 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1050
1051 return ret;
1052}
1053
1054extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1055 sigset_t *mask);
1056extern void unblock_all_signals(void);
1057extern void release_task(struct task_struct * p);
1058extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1059extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1060extern int force_sigsegv(int, struct task_struct *);
1061extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1062extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1063extern int kill_pg_info(int, struct siginfo *, pid_t);
1064extern int kill_proc_info(int, struct siginfo *, pid_t);
Harald Welte46113832005-10-10 19:44:29 +02001065extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001066extern void do_notify_parent(struct task_struct *, int);
1067extern void force_sig(int, struct task_struct *);
1068extern void force_sig_specific(int, struct task_struct *);
1069extern int send_sig(int, struct task_struct *, int);
1070extern void zap_other_threads(struct task_struct *p);
1071extern int kill_pg(pid_t, int, int);
1072extern int kill_sl(pid_t, int, int);
1073extern int kill_proc(pid_t, int, int);
1074extern struct sigqueue *sigqueue_alloc(void);
1075extern void sigqueue_free(struct sigqueue *);
1076extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1077extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
1078extern int do_sigaction(int, const struct k_sigaction *, struct k_sigaction *);
1079extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1080
1081/* These can be the second arg to send_sig_info/send_group_sig_info. */
1082#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1083#define SEND_SIG_PRIV ((struct siginfo *) 1)
1084#define SEND_SIG_FORCED ((struct siginfo *) 2)
1085
Oleg Nesterov621d3122005-10-30 15:03:45 -08001086static inline int is_si_special(const struct siginfo *info)
1087{
1088 return info <= SEND_SIG_FORCED;
1089}
1090
Linus Torvalds1da177e2005-04-16 15:20:36 -07001091/* True if we are on the alternate signal stack. */
1092
1093static inline int on_sig_stack(unsigned long sp)
1094{
1095 return (sp - current->sas_ss_sp < current->sas_ss_size);
1096}
1097
1098static inline int sas_ss_flags(unsigned long sp)
1099{
1100 return (current->sas_ss_size == 0 ? SS_DISABLE
1101 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1102}
1103
1104
1105#ifdef CONFIG_SECURITY
1106/* code is in security.c */
1107extern int capable(int cap);
1108#else
1109static inline int capable(int cap)
1110{
1111 if (cap_raised(current->cap_effective, cap)) {
1112 current->flags |= PF_SUPERPRIV;
1113 return 1;
1114 }
1115 return 0;
1116}
1117#endif
1118
1119/*
1120 * Routines for handling mm_structs
1121 */
1122extern struct mm_struct * mm_alloc(void);
1123
1124/* mmdrop drops the mm and the page tables */
1125extern void FASTCALL(__mmdrop(struct mm_struct *));
1126static inline void mmdrop(struct mm_struct * mm)
1127{
1128 if (atomic_dec_and_test(&mm->mm_count))
1129 __mmdrop(mm);
1130}
1131
1132/* mmput gets rid of the mappings and all user-space */
1133extern void mmput(struct mm_struct *);
1134/* Grab a reference to a task's mm, if it is not already going away */
1135extern struct mm_struct *get_task_mm(struct task_struct *task);
1136/* Remove the current tasks stale references to the old mm_struct */
1137extern void mm_release(struct task_struct *, struct mm_struct *);
1138
1139extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1140extern void flush_thread(void);
1141extern void exit_thread(void);
1142
Linus Torvalds1da177e2005-04-16 15:20:36 -07001143extern void exit_files(struct task_struct *);
1144extern void exit_signal(struct task_struct *);
1145extern void __exit_signal(struct task_struct *);
1146extern void exit_sighand(struct task_struct *);
1147extern void __exit_sighand(struct task_struct *);
1148extern void exit_itimers(struct signal_struct *);
1149
1150extern NORET_TYPE void do_group_exit(int);
1151
Linus Torvalds1da177e2005-04-16 15:20:36 -07001152extern void daemonize(const char *, ...);
1153extern int allow_signal(int);
1154extern int disallow_signal(int);
1155extern task_t *child_reaper;
1156
1157extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1158extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1159task_t *fork_idle(int);
1160
1161extern void set_task_comm(struct task_struct *tsk, char *from);
1162extern void get_task_comm(char *to, struct task_struct *tsk);
1163
1164#ifdef CONFIG_SMP
1165extern void wait_task_inactive(task_t * p);
1166#else
1167#define wait_task_inactive(p) do { } while (0)
1168#endif
1169
1170#define remove_parent(p) list_del_init(&(p)->sibling)
1171#define add_parent(p, parent) list_add_tail(&(p)->sibling,&(parent)->children)
1172
1173#define REMOVE_LINKS(p) do { \
1174 if (thread_group_leader(p)) \
1175 list_del_init(&(p)->tasks); \
1176 remove_parent(p); \
1177 } while (0)
1178
1179#define SET_LINKS(p) do { \
1180 if (thread_group_leader(p)) \
1181 list_add_tail(&(p)->tasks,&init_task.tasks); \
1182 add_parent(p, (p)->parent); \
1183 } while (0)
1184
1185#define next_task(p) list_entry((p)->tasks.next, struct task_struct, tasks)
1186#define prev_task(p) list_entry((p)->tasks.prev, struct task_struct, tasks)
1187
1188#define for_each_process(p) \
1189 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1190
1191/*
1192 * Careful: do_each_thread/while_each_thread is a double loop so
1193 * 'break' will not work as expected - use goto instead.
1194 */
1195#define do_each_thread(g, t) \
1196 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1197
1198#define while_each_thread(g, t) \
1199 while ((t = next_thread(t)) != g)
1200
1201extern task_t * FASTCALL(next_thread(const task_t *p));
1202
1203#define thread_group_leader(p) (p->pid == p->tgid)
1204
1205static inline int thread_group_empty(task_t *p)
1206{
1207 return list_empty(&p->pids[PIDTYPE_TGID].pid_list);
1208}
1209
1210#define delay_group_leader(p) \
1211 (thread_group_leader(p) && !thread_group_empty(p))
1212
1213extern void unhash_process(struct task_struct *p);
1214
1215/*
1216 * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02001217 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Jackson053199e2005-10-30 15:02:30 -08001218 * pins the final release of task.io_context. Also protects ->cpuset.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001219 *
1220 * Nests both inside and outside of read_lock(&tasklist_lock).
1221 * It must not be nested with write_lock_irq(&tasklist_lock),
1222 * neither inside nor outside.
1223 */
1224static inline void task_lock(struct task_struct *p)
1225{
1226 spin_lock(&p->alloc_lock);
1227}
1228
1229static inline void task_unlock(struct task_struct *p)
1230{
1231 spin_unlock(&p->alloc_lock);
1232}
1233
Al Virof0373602005-11-13 16:06:57 -08001234#ifndef __HAVE_THREAD_FUNCTIONS
1235
Al Viroa1261f52005-11-13 16:06:55 -08001236#define task_thread_info(task) (task)->thread_info
1237
Al Viro10ebffd2005-11-13 16:06:56 -08001238static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1239{
1240 *task_thread_info(p) = *task_thread_info(org);
1241 task_thread_info(p)->task = p;
1242}
1243
1244static inline unsigned long *end_of_stack(struct task_struct *p)
1245{
1246 return (unsigned long *)(p->thread_info + 1);
1247}
1248
Al Virof0373602005-11-13 16:06:57 -08001249#endif
1250
Linus Torvalds1da177e2005-04-16 15:20:36 -07001251/* set thread flags in other task's structures
1252 * - see asm/thread_info.h for TIF_xxxx flags available
1253 */
1254static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1255{
Al Viroa1261f52005-11-13 16:06:55 -08001256 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001257}
1258
1259static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1260{
Al Viroa1261f52005-11-13 16:06:55 -08001261 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001262}
1263
1264static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1265{
Al Viroa1261f52005-11-13 16:06:55 -08001266 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001267}
1268
1269static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1270{
Al Viroa1261f52005-11-13 16:06:55 -08001271 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001272}
1273
1274static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1275{
Al Viroa1261f52005-11-13 16:06:55 -08001276 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001277}
1278
1279static inline void set_tsk_need_resched(struct task_struct *tsk)
1280{
1281 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1282}
1283
1284static inline void clear_tsk_need_resched(struct task_struct *tsk)
1285{
1286 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1287}
1288
1289static inline int signal_pending(struct task_struct *p)
1290{
1291 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1292}
1293
1294static inline int need_resched(void)
1295{
1296 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1297}
1298
1299/*
1300 * cond_resched() and cond_resched_lock(): latency reduction via
1301 * explicit rescheduling in places that are safe. The return
1302 * value indicates whether a reschedule was done in fact.
1303 * cond_resched_lock() will drop the spinlock before scheduling,
1304 * cond_resched_softirq() will enable bhs before scheduling.
1305 */
1306extern int cond_resched(void);
1307extern int cond_resched_lock(spinlock_t * lock);
1308extern int cond_resched_softirq(void);
1309
1310/*
1311 * Does a critical section need to be broken due to another
1312 * task waiting?:
1313 */
1314#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1315# define need_lockbreak(lock) ((lock)->break_lock)
1316#else
1317# define need_lockbreak(lock) 0
1318#endif
1319
1320/*
1321 * Does a critical section need to be broken due to another
1322 * task waiting or preemption being signalled:
1323 */
1324static inline int lock_need_resched(spinlock_t *lock)
1325{
1326 if (need_lockbreak(lock) || need_resched())
1327 return 1;
1328 return 0;
1329}
1330
1331/* Reevaluate whether the task has signals pending delivery.
1332 This is required every time the blocked sigset_t changes.
1333 callers must hold sighand->siglock. */
1334
1335extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1336extern void recalc_sigpending(void);
1337
1338extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1339
1340/*
1341 * Wrappers for p->thread_info->cpu access. No-op on UP.
1342 */
1343#ifdef CONFIG_SMP
1344
1345static inline unsigned int task_cpu(const struct task_struct *p)
1346{
Al Viroa1261f52005-11-13 16:06:55 -08001347 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001348}
1349
1350static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1351{
Al Viroa1261f52005-11-13 16:06:55 -08001352 task_thread_info(p)->cpu = cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001353}
1354
1355#else
1356
1357static inline unsigned int task_cpu(const struct task_struct *p)
1358{
1359 return 0;
1360}
1361
1362static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1363{
1364}
1365
1366#endif /* CONFIG_SMP */
1367
1368#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1369extern void arch_pick_mmap_layout(struct mm_struct *mm);
1370#else
1371static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1372{
1373 mm->mmap_base = TASK_UNMAPPED_BASE;
1374 mm->get_unmapped_area = arch_get_unmapped_area;
1375 mm->unmap_area = arch_unmap_area;
1376}
1377#endif
1378
1379extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1380extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1381
1382#ifdef CONFIG_MAGIC_SYSRQ
1383
1384extern void normalize_rt_tasks(void);
1385
1386#endif
1387
Linus Torvalds1da177e2005-04-16 15:20:36 -07001388#ifdef CONFIG_PM
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001389/*
1390 * Check if a process has been frozen
1391 */
1392static inline int frozen(struct task_struct *p)
1393{
1394 return p->flags & PF_FROZEN;
1395}
1396
1397/*
1398 * Check if there is a request to freeze a process
1399 */
1400static inline int freezing(struct task_struct *p)
1401{
1402 return p->flags & PF_FREEZE;
1403}
1404
1405/*
1406 * Request that a process be frozen
1407 * FIXME: SMP problem. We may not modify other process' flags!
1408 */
1409static inline void freeze(struct task_struct *p)
1410{
1411 p->flags |= PF_FREEZE;
1412}
1413
1414/*
1415 * Wake up a frozen process
1416 */
1417static inline int thaw_process(struct task_struct *p)
1418{
1419 if (frozen(p)) {
1420 p->flags &= ~PF_FROZEN;
1421 wake_up_process(p);
1422 return 1;
1423 }
1424 return 0;
1425}
1426
1427/*
1428 * freezing is complete, mark process as frozen
1429 */
1430static inline void frozen_process(struct task_struct *p)
1431{
1432 p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1433}
1434
1435extern void refrigerator(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001436extern int freeze_processes(void);
1437extern void thaw_processes(void);
1438
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001439static inline int try_to_freeze(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001440{
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001441 if (freezing(current)) {
1442 refrigerator();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001443 return 1;
1444 } else
1445 return 0;
1446}
1447#else
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001448static inline int frozen(struct task_struct *p) { return 0; }
1449static inline int freezing(struct task_struct *p) { return 0; }
1450static inline void freeze(struct task_struct *p) { BUG(); }
1451static inline int thaw_process(struct task_struct *p) { return 1; }
1452static inline void frozen_process(struct task_struct *p) { BUG(); }
1453
1454static inline void refrigerator(void) {}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001455static inline int freeze_processes(void) { BUG(); return 0; }
1456static inline void thaw_processes(void) {}
1457
Christoph Lameter3e1d1d22005-06-24 23:13:50 -07001458static inline int try_to_freeze(void) { return 0; }
1459
Linus Torvalds1da177e2005-04-16 15:20:36 -07001460#endif /* CONFIG_PM */
1461#endif /* __KERNEL__ */
1462
1463#endif