]> bbs.cooldavid.org Git - net-next-2.6.git/blame - include/linux/sched.h
signal/timer/event: signalfd core
[net-next-2.6.git] / include / linux / sched.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
b7b3c76a
DW
4#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 */
071df104 27#define CLONE_NEWUTS 0x04000000 /* New utsname group? */
25b21cb2 28#define CLONE_NEWIPC 0x08000000 /* New ipcs */
b7b3c76a
DW
29
30/*
31 * Scheduling policies
32 */
33#define SCHED_NORMAL 0
34#define SCHED_FIFO 1
35#define SCHED_RR 2
36#define SCHED_BATCH 3
37
a3b6714e 38#ifdef __KERNEL__
b7b3c76a
DW
39
40struct sched_param {
41 int sched_priority;
42};
43
1da177e4
LT
44#include <asm/param.h> /* for HZ */
45
1da177e4
LT
46#include <linux/capability.h>
47#include <linux/threads.h>
48#include <linux/kernel.h>
49#include <linux/types.h>
50#include <linux/timex.h>
51#include <linux/jiffies.h>
52#include <linux/rbtree.h>
53#include <linux/thread_info.h>
54#include <linux/cpumask.h>
55#include <linux/errno.h>
56#include <linux/nodemask.h>
57
58#include <asm/system.h>
59#include <asm/semaphore.h>
60#include <asm/page.h>
61#include <asm/ptrace.h>
62#include <asm/mmu.h>
63#include <asm/cputime.h>
64
65#include <linux/smp.h>
66#include <linux/sem.h>
67#include <linux/signal.h>
68#include <linux/securebits.h>
69#include <linux/fs_struct.h>
70#include <linux/compiler.h>
71#include <linux/completion.h>
72#include <linux/pid.h>
73#include <linux/percpu.h>
74#include <linux/topology.h>
75#include <linux/seccomp.h>
e56d0903 76#include <linux/rcupdate.h>
0771dfef 77#include <linux/futex.h>
23f78d4a 78#include <linux/rtmutex.h>
1da177e4 79
a3b6714e
DW
80#include <linux/time.h>
81#include <linux/param.h>
82#include <linux/resource.h>
83#include <linux/timer.h>
84#include <linux/hrtimer.h>
7c3ab738 85#include <linux/task_io_accounting.h>
a3b6714e
DW
86
87#include <asm/processor.h>
36d57ac4 88
1da177e4 89struct exec_domain;
c87e2837 90struct futex_pi_state;
1da177e4 91
1da177e4
LT
92/*
93 * List of flags we want to share for kernel threads,
94 * if only because they are not used by them anyway.
95 */
96#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
97
98/*
99 * These are the constant used to fake the fixed-point load-average
100 * counting. Some notes:
101 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
102 * a load-average precision of 10 bits integer + 11 bits fractional
103 * - if you want to count load-averages more often, you need more
104 * precision, or rounding will get you. With 2-second counting freq,
105 * the EXP_n values would be 1981, 2034 and 2043 if still using only
106 * 11 bit fractions.
107 */
108extern unsigned long avenrun[]; /* Load averages */
109
110#define FSHIFT 11 /* nr of bits of precision */
111#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
112#define LOAD_FREQ (5*HZ) /* 5 sec intervals */
113#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
114#define EXP_5 2014 /* 1/exp(5sec/5min) */
115#define EXP_15 2037 /* 1/exp(5sec/15min) */
116
117#define CALC_LOAD(load,exp,n) \
118 load *= exp; \
119 load += n*(FIXED_1-exp); \
120 load >>= FSHIFT;
121
122extern unsigned long total_forks;
123extern int nr_threads;
1da177e4
LT
124DECLARE_PER_CPU(unsigned long, process_counts);
125extern int nr_processes(void);
126extern unsigned long nr_running(void);
127extern unsigned long nr_uninterruptible(void);
db1b1fef 128extern unsigned long nr_active(void);
1da177e4 129extern unsigned long nr_iowait(void);
2dd73a4f 130extern unsigned long weighted_cpuload(const int cpu);
1da177e4 131
1da177e4 132
4a8342d2
LT
133/*
134 * Task state bitmask. NOTE! These bits are also
135 * encoded in fs/proc/array.c: get_task_state().
136 *
137 * We have two separate sets of flags: task->state
138 * is about runnability, while task->exit_state are
139 * about the task exiting. Confusing, but this way
140 * modifying one set can't modify the other one by
141 * mistake.
142 */
1da177e4
LT
143#define TASK_RUNNING 0
144#define TASK_INTERRUPTIBLE 1
145#define TASK_UNINTERRUPTIBLE 2
4a8342d2
LT
146#define TASK_STOPPED 4
147#define TASK_TRACED 8
148/* in tsk->exit_state */
149#define EXIT_ZOMBIE 16
150#define EXIT_DEAD 32
151/* in tsk->state again */
152#define TASK_NONINTERACTIVE 64
c394cc9f 153#define TASK_DEAD 128
1da177e4
LT
154
155#define __set_task_state(tsk, state_value) \
156 do { (tsk)->state = (state_value); } while (0)
157#define set_task_state(tsk, state_value) \
158 set_mb((tsk)->state, (state_value))
159
498d0c57
AM
160/*
161 * set_current_state() includes a barrier so that the write of current->state
162 * is correctly serialised wrt the caller's subsequent test of whether to
163 * actually sleep:
164 *
165 * set_current_state(TASK_UNINTERRUPTIBLE);
166 * if (do_i_need_to_sleep())
167 * schedule();
168 *
169 * If the caller does not need such serialisation then use __set_current_state()
170 */
1da177e4
LT
171#define __set_current_state(state_value) \
172 do { current->state = (state_value); } while (0)
173#define set_current_state(state_value) \
174 set_mb(current->state, (state_value))
175
176/* Task command name length */
177#define TASK_COMM_LEN 16
178
1da177e4
LT
179#include <linux/spinlock.h>
180
181/*
182 * This serializes "schedule()" and also protects
183 * the run-queue from deletions/modifications (but
184 * _adding_ to the beginning of the run-queue has
185 * a separate lock).
186 */
187extern rwlock_t tasklist_lock;
188extern spinlock_t mmlist_lock;
189
36c8b586 190struct task_struct;
1da177e4
LT
191
192extern void sched_init(void);
193extern void sched_init_smp(void);
36c8b586 194extern void init_idle(struct task_struct *idle, int cpu);
1da177e4
LT
195
196extern cpumask_t nohz_cpu_mask;
46cb4b7c
SS
197#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
198extern int select_nohz_load_balancer(int cpu);
199#else
200static inline int select_nohz_load_balancer(int cpu)
201{
202 return 0;
203}
204#endif
1da177e4 205
e59e2ae2 206/*
39bc89fd 207 * Only dump TASK_* tasks. (0 for all tasks)
e59e2ae2
IM
208 */
209extern void show_state_filter(unsigned long state_filter);
210
211static inline void show_state(void)
212{
39bc89fd 213 show_state_filter(0);
e59e2ae2
IM
214}
215
1da177e4
LT
216extern void show_regs(struct pt_regs *);
217
218/*
219 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
220 * task), SP is the stack pointer of the first frame that should be shown in the back
221 * trace (or NULL if the entire call-chain of the task should be shown).
222 */
223extern void show_stack(struct task_struct *task, unsigned long *sp);
224
225void io_schedule(void);
226long io_schedule_timeout(long timeout);
227
228extern void cpu_init (void);
229extern void trap_init(void);
230extern void update_process_times(int user);
231extern void scheduler_tick(void);
232
8446f1d3 233#ifdef CONFIG_DETECT_SOFTLOCKUP
6687a97d 234extern void softlockup_tick(void);
8446f1d3
IM
235extern void spawn_softlockup_task(void);
236extern void touch_softlockup_watchdog(void);
04c9167f 237extern void touch_all_softlockup_watchdogs(void);
8446f1d3 238#else
6687a97d 239static inline void softlockup_tick(void)
8446f1d3
IM
240{
241}
242static inline void spawn_softlockup_task(void)
243{
244}
245static inline void touch_softlockup_watchdog(void)
246{
247}
04c9167f
JF
248static inline void touch_all_softlockup_watchdogs(void)
249{
250}
8446f1d3
IM
251#endif
252
253
1da177e4
LT
254/* Attach to any functions which should be ignored in wchan output. */
255#define __sched __attribute__((__section__(".sched.text")))
256/* Is this address in the __sched functions? */
257extern int in_sched_functions(unsigned long addr);
258
259#define MAX_SCHEDULE_TIMEOUT LONG_MAX
260extern signed long FASTCALL(schedule_timeout(signed long timeout));
64ed93a2
NA
261extern signed long schedule_timeout_interruptible(signed long timeout);
262extern signed long schedule_timeout_uninterruptible(signed long timeout);
1da177e4
LT
263asmlinkage void schedule(void);
264
ab516013 265struct nsproxy;
1da177e4
LT
266
267/* Maximum number of active map areas.. This is a random (large) number */
268#define DEFAULT_MAX_MAP_COUNT 65536
269
270extern int sysctl_max_map_count;
271
272#include <linux/aio.h>
273
274extern unsigned long
275arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
276 unsigned long, unsigned long);
277extern unsigned long
278arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
279 unsigned long len, unsigned long pgoff,
280 unsigned long flags);
1363c3cd
WW
281extern void arch_unmap_area(struct mm_struct *, unsigned long);
282extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
1da177e4 283
f412ac08
HD
284#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
285/*
286 * The mm counters are not protected by its page_table_lock,
287 * so must be incremented atomically.
288 */
d3cb4871
CL
289#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
290#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
291#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
292#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
293#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
294typedef atomic_long_t mm_counter_t;
f412ac08
HD
295
296#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
297/*
298 * The mm counters are protected by its page_table_lock,
299 * so can be incremented directly.
300 */
1da177e4
LT
301#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
302#define get_mm_counter(mm, member) ((mm)->_##member)
303#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
304#define inc_mm_counter(mm, member) (mm)->_##member++
305#define dec_mm_counter(mm, member) (mm)->_##member--
f412ac08
HD
306typedef unsigned long mm_counter_t;
307
308#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
4294621f 309
f412ac08
HD
310#define get_mm_rss(mm) \
311 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
365e9c87
HD
312#define update_hiwater_rss(mm) do { \
313 unsigned long _rss = get_mm_rss(mm); \
314 if ((mm)->hiwater_rss < _rss) \
315 (mm)->hiwater_rss = _rss; \
316} while (0)
317#define update_hiwater_vm(mm) do { \
318 if ((mm)->hiwater_vm < (mm)->total_vm) \
319 (mm)->hiwater_vm = (mm)->total_vm; \
320} while (0)
321
1da177e4
LT
322struct mm_struct {
323 struct vm_area_struct * mmap; /* list of VMAs */
324 struct rb_root mm_rb;
325 struct vm_area_struct * mmap_cache; /* last find_vma result */
326 unsigned long (*get_unmapped_area) (struct file *filp,
327 unsigned long addr, unsigned long len,
328 unsigned long pgoff, unsigned long flags);
1363c3cd 329 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
0551fbd2
BH
330 unsigned long mmap_base; /* base of mmap area */
331 unsigned long task_size; /* size of task vm space */
332 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
1363c3cd 333 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
1da177e4
LT
334 pgd_t * pgd;
335 atomic_t mm_users; /* How many users with user space? */
336 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
337 int map_count; /* number of VMAs */
338 struct rw_semaphore mmap_sem;
339 spinlock_t page_table_lock; /* Protects page tables and some counters */
340
341 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
342 * together off init_mm.mmlist, and are protected
343 * by mmlist_lock
344 */
345
f412ac08
HD
346 /* Special counters, in some configurations protected by the
347 * page_table_lock, in other configurations by being atomic.
348 */
4294621f 349 mm_counter_t _file_rss;
1da177e4
LT
350 mm_counter_t _anon_rss;
351
f449952b
HD
352 unsigned long hiwater_rss; /* High-watermark of RSS usage */
353 unsigned long hiwater_vm; /* High-water virtual memory usage */
354
355 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
356 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
357 unsigned long start_code, end_code, start_data, end_data;
358 unsigned long start_brk, brk, start_stack;
359 unsigned long arg_start, arg_end, env_start, env_end;
360
36d57ac4 361 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
1da177e4 362
1da177e4
LT
363 cpumask_t cpu_vm_mask;
364
365 /* Architecture-specific MM context */
366 mm_context_t context;
367
7602bdf2
AC
368 /* Swap token stuff */
369 /*
370 * Last value of global fault stamp as seen by this process.
371 * In other words, this value gives an indication of how long
372 * it has been since this task got the token.
373 * Look at mm/thrash.c
374 */
375 unsigned int faultstamp;
376 unsigned int token_priority;
377 unsigned int last_interval;
1da177e4 378
36de6437
ACM
379 unsigned char dumpable:2;
380
1da177e4
LT
381 /* coredumping support */
382 int core_waiters;
383 struct completion *core_startup_done, core_done;
384
385 /* aio bits */
386 rwlock_t ioctx_list_lock;
387 struct kioctx *ioctx_list;
1da177e4
LT
388};
389
390struct sighand_struct {
391 atomic_t count;
392 struct k_sigaction action[_NSIG];
393 spinlock_t siglock;
fba2afaa 394 struct list_head signalfd_list;
1da177e4
LT
395};
396
0e464814 397struct pacct_struct {
f6ec29a4
KK
398 int ac_flag;
399 long ac_exitcode;
0e464814 400 unsigned long ac_mem;
77787bfb
KK
401 cputime_t ac_utime, ac_stime;
402 unsigned long ac_minflt, ac_majflt;
0e464814
KK
403};
404
1da177e4
LT
405/*
406 * NOTE! "signal_struct" does not have it's own
407 * locking, because a shared signal_struct always
408 * implies a shared sighand_struct, so locking
409 * sighand_struct is always a proper superset of
410 * the locking of signal_struct.
411 */
412struct signal_struct {
413 atomic_t count;
414 atomic_t live;
415
416 wait_queue_head_t wait_chldexit; /* for wait4() */
417
418 /* current thread group signal load-balancing target: */
36c8b586 419 struct task_struct *curr_target;
1da177e4
LT
420
421 /* shared signal handling: */
422 struct sigpending shared_pending;
423
424 /* thread group exit support */
425 int group_exit_code;
426 /* overloaded:
427 * - notify group_exit_task when ->count is equal to notify_count
428 * - everyone except group_exit_task is stopped during signal delivery
429 * of fatal signals, group_exit_task processes the signal.
430 */
431 struct task_struct *group_exit_task;
432 int notify_count;
433
434 /* thread group stop support, overloads group_exit_code too */
435 int group_stop_count;
436 unsigned int flags; /* see SIGNAL_* flags below */
437
438 /* POSIX.1b Interval Timers */
439 struct list_head posix_timers;
440
441 /* ITIMER_REAL timer for the process */
2ff678b8 442 struct hrtimer real_timer;
05cfb614 443 struct task_struct *tsk;
2ff678b8 444 ktime_t it_real_incr;
1da177e4
LT
445
446 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
447 cputime_t it_prof_expires, it_virt_expires;
448 cputime_t it_prof_incr, it_virt_incr;
449
450 /* job control IDs */
451 pid_t pgrp;
ab521dc0 452 struct pid *tty_old_pgrp;
1ec320af
CLG
453
454 union {
455 pid_t session __deprecated;
456 pid_t __session;
457 };
458
1da177e4
LT
459 /* boolean value for session group leader */
460 int leader;
461
462 struct tty_struct *tty; /* NULL if no tty */
463
464 /*
465 * Cumulative resource counters for dead threads in the group,
466 * and for reaped dead child processes forked by this group.
467 * Live threads maintain their own counters and add to these
468 * in __exit_signal, except for the group leader.
469 */
470 cputime_t utime, stime, cutime, cstime;
471 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
472 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
6eaeeaba 473 unsigned long inblock, oublock, cinblock, coublock;
1da177e4
LT
474
475 /*
476 * Cumulative ns of scheduled CPU time for dead threads in the
477 * group, not including a zombie group leader. (This only differs
478 * from jiffies_to_ns(utime + stime) if sched_clock uses something
479 * other than jiffies.)
480 */
481 unsigned long long sched_time;
482
483 /*
484 * We don't bother to synchronize most readers of this at all,
485 * because there is no reader checking a limit that actually needs
486 * to get both rlim_cur and rlim_max atomically, and either one
487 * alone is a single word that can safely be read normally.
488 * getrlimit/setrlimit use task_lock(current->group_leader) to
489 * protect this instead of the siglock, because they really
490 * have no need to disable irqs.
491 */
492 struct rlimit rlim[RLIM_NLIMITS];
493
494 struct list_head cpu_timers[3];
495
496 /* keep the process-shared keyrings here so that they do the right
497 * thing in threads created with CLONE_THREAD */
498#ifdef CONFIG_KEYS
499 struct key *session_keyring; /* keyring inherited over fork */
500 struct key *process_keyring; /* keyring private to this process */
501#endif
0e464814
KK
502#ifdef CONFIG_BSD_PROCESS_ACCT
503 struct pacct_struct pacct; /* per-process accounting information */
504#endif
ad4ecbcb 505#ifdef CONFIG_TASKSTATS
ad4ecbcb
SN
506 struct taskstats *stats;
507#endif
1da177e4
LT
508};
509
4866cde0
NP
510/* Context switch must be unlocked if interrupts are to be enabled */
511#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
512# define __ARCH_WANT_UNLOCKED_CTXSW
513#endif
514
1da177e4
LT
515/*
516 * Bits in flags field of signal_struct.
517 */
518#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
519#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
520#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
521#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
522
523
524/*
525 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
b0a9499c
IM
526 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
527 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
528 * values are inverted: lower p->prio value means higher priority.
1da177e4
LT
529 *
530 * The MAX_USER_RT_PRIO value allows the actual maximum
531 * RT priority to be separate from the value exported to
532 * user-space. This allows kernel threads to set their
533 * priority to a value higher than any user task. Note:
534 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
535 */
536
537#define MAX_USER_RT_PRIO 100
538#define MAX_RT_PRIO MAX_USER_RT_PRIO
539
540#define MAX_PRIO (MAX_RT_PRIO + 40)
541
b29739f9
IM
542#define rt_prio(prio) unlikely((prio) < MAX_RT_PRIO)
543#define rt_task(p) rt_prio((p)->prio)
d425b274 544#define batch_task(p) (unlikely((p)->policy == SCHED_BATCH))
57a6f51c
ON
545#define is_rt_policy(p) ((p) != SCHED_NORMAL && (p) != SCHED_BATCH)
546#define has_rt_policy(p) unlikely(is_rt_policy((p)->policy))
1da177e4
LT
547
548/*
549 * Some day this will be a full-fledged user tracking system..
550 */
551struct user_struct {
552 atomic_t __count; /* reference count */
553 atomic_t processes; /* How many processes does this user have? */
554 atomic_t files; /* How many open files does this user have? */
555 atomic_t sigpending; /* How many pending signals does this user have? */
2d9048e2 556#ifdef CONFIG_INOTIFY_USER
0eeca283
RL
557 atomic_t inotify_watches; /* How many inotify watches does this user have? */
558 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
559#endif
1da177e4
LT
560 /* protected by mq_lock */
561 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
562 unsigned long locked_shm; /* How many pages of mlocked shm ? */
563
564#ifdef CONFIG_KEYS
565 struct key *uid_keyring; /* UID specific keyring */
566 struct key *session_keyring; /* UID's default session keyring */
567#endif
568
569 /* Hash table maintenance information */
570 struct list_head uidhash_list;
571 uid_t uid;
572};
573
574extern struct user_struct *find_user(uid_t);
575
576extern struct user_struct root_user;
577#define INIT_USER (&root_user)
578
1da177e4
LT
579struct backing_dev_info;
580struct reclaim_state;
581
52f17b6c 582#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
583struct sched_info {
584 /* cumulative counters */
585 unsigned long cpu_time, /* time spent on the cpu */
586 run_delay, /* time spent waiting on a runqueue */
587 pcnt; /* # of timeslices run on this cpu */
588
589 /* timestamps */
590 unsigned long last_arrival, /* when we last ran on a cpu */
591 last_queued; /* when we were last queued to run */
592};
52f17b6c 593#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
1da177e4 594
52f17b6c 595#ifdef CONFIG_SCHEDSTATS
15ad7cdc 596extern const struct file_operations proc_schedstat_operations;
52f17b6c 597#endif /* CONFIG_SCHEDSTATS */
1da177e4 598
ca74e92b
SN
599#ifdef CONFIG_TASK_DELAY_ACCT
600struct task_delay_info {
601 spinlock_t lock;
602 unsigned int flags; /* Private per-task flags */
603
604 /* For each stat XXX, add following, aligned appropriately
605 *
606 * struct timespec XXX_start, XXX_end;
607 * u64 XXX_delay;
608 * u32 XXX_count;
609 *
610 * Atomicity of updates to XXX_delay, XXX_count protected by
611 * single lock above (split into XXX_lock if contention is an issue).
612 */
0ff92245
SN
613
614 /*
615 * XXX_count is incremented on every XXX operation, the delay
616 * associated with the operation is added to XXX_delay.
617 * XXX_delay contains the accumulated delay time in nanoseconds.
618 */
619 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
620 u64 blkio_delay; /* wait for sync block io completion */
621 u64 swapin_delay; /* wait for swapin block io completion */
622 u32 blkio_count; /* total count of the number of sync block */
623 /* io operations performed */
624 u32 swapin_count; /* total count of the number of swapin block */
625 /* io operations performed */
ca74e92b 626};
52f17b6c
CS
627#endif /* CONFIG_TASK_DELAY_ACCT */
628
629static inline int sched_info_on(void)
630{
631#ifdef CONFIG_SCHEDSTATS
632 return 1;
633#elif defined(CONFIG_TASK_DELAY_ACCT)
634 extern int delayacct_on;
635 return delayacct_on;
636#else
637 return 0;
ca74e92b 638#endif
52f17b6c 639}
ca74e92b 640
1da177e4
LT
641enum idle_type
642{
643 SCHED_IDLE,
644 NOT_IDLE,
645 NEWLY_IDLE,
646 MAX_IDLE_TYPES
647};
648
649/*
650 * sched-domains (multiprocessor balancing) declarations:
651 */
1da177e4
LT
652#define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
653
2dd73a4f 654#ifdef CONFIG_SMP
1da177e4
LT
655#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
656#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
657#define SD_BALANCE_EXEC 4 /* Balance on exec */
147cbb4b
NP
658#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
659#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
660#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
661#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
662#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
5c45bf27 663#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
89c4710e 664#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
08c183f3 665#define SD_SERIALIZE 1024 /* Only a single load balancing instance */
5c45bf27 666
89c4710e
SS
667#define BALANCE_FOR_MC_POWER \
668 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
669
670#define BALANCE_FOR_PKG_POWER \
671 ((sched_mc_power_savings || sched_smt_power_savings) ? \
672 SD_POWERSAVINGS_BALANCE : 0)
673
674#define test_sd_parent(sd, flag) ((sd->parent && \
675 (sd->parent->flags & flag)) ? 1 : 0)
5c45bf27 676
1da177e4
LT
677
678struct sched_group {
679 struct sched_group *next; /* Must be a circular list */
680 cpumask_t cpumask;
681
682 /*
683 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
684 * single CPU. This is read only (except for setup, hotplug CPU).
5517d86b 685 * Note : Never change cpu_power without recompute its reciprocal
1da177e4 686 */
5517d86b
ED
687 unsigned int __cpu_power;
688 /*
689 * reciprocal value of cpu_power to avoid expensive divides
690 * (see include/linux/reciprocal_div.h)
691 */
692 u32 reciprocal_cpu_power;
1da177e4
LT
693};
694
695struct sched_domain {
696 /* These fields must be setup */
697 struct sched_domain *parent; /* top domain must be null terminated */
1a848870 698 struct sched_domain *child; /* bottom domain must be null terminated */
1da177e4
LT
699 struct sched_group *groups; /* the balancing groups of the domain */
700 cpumask_t span; /* span of all CPUs in this domain */
701 unsigned long min_interval; /* Minimum balance interval ms */
702 unsigned long max_interval; /* Maximum balance interval ms */
703 unsigned int busy_factor; /* less balancing by factor if busy */
704 unsigned int imbalance_pct; /* No balance until over watermark */
705 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
706 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
7897986b
NP
707 unsigned int busy_idx;
708 unsigned int idle_idx;
709 unsigned int newidle_idx;
710 unsigned int wake_idx;
147cbb4b 711 unsigned int forkexec_idx;
1da177e4
LT
712 int flags; /* See SD_* */
713
714 /* Runtime fields. */
715 unsigned long last_balance; /* init to jiffies. units in jiffies */
716 unsigned int balance_interval; /* initialise to 1. units in ms. */
717 unsigned int nr_balance_failed; /* initialise to 0 */
718
719#ifdef CONFIG_SCHEDSTATS
720 /* load_balance() stats */
721 unsigned long lb_cnt[MAX_IDLE_TYPES];
722 unsigned long lb_failed[MAX_IDLE_TYPES];
723 unsigned long lb_balanced[MAX_IDLE_TYPES];
724 unsigned long lb_imbalance[MAX_IDLE_TYPES];
725 unsigned long lb_gained[MAX_IDLE_TYPES];
726 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
727 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
728 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
729
730 /* Active load balancing */
731 unsigned long alb_cnt;
732 unsigned long alb_failed;
733 unsigned long alb_pushed;
734
68767a0a
NP
735 /* SD_BALANCE_EXEC stats */
736 unsigned long sbe_cnt;
737 unsigned long sbe_balanced;
1da177e4
LT
738 unsigned long sbe_pushed;
739
68767a0a
NP
740 /* SD_BALANCE_FORK stats */
741 unsigned long sbf_cnt;
742 unsigned long sbf_balanced;
743 unsigned long sbf_pushed;
744
1da177e4
LT
745 /* try_to_wake_up() stats */
746 unsigned long ttwu_wake_remote;
747 unsigned long ttwu_move_affine;
748 unsigned long ttwu_move_balance;
749#endif
750};
751
51888ca2 752extern int partition_sched_domains(cpumask_t *partition1,
1a20ff27 753 cpumask_t *partition2);
198e2f18
AM
754
755/*
756 * Maximum cache size the migration-costs auto-tuning code will
757 * search from:
758 */
759extern unsigned int max_cache_size;
760
761#endif /* CONFIG_SMP */
1da177e4
LT
762
763
764struct io_context; /* See blkdev.h */
1da177e4
LT
765struct cpuset;
766
767#define NGROUPS_SMALL 32
768#define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
769struct group_info {
770 int ngroups;
771 atomic_t usage;
772 gid_t small_block[NGROUPS_SMALL];
773 int nblocks;
774 gid_t *blocks[0];
775};
776
777/*
778 * get_group_info() must be called with the owning task locked (via task_lock())
779 * when task != current. The reason being that the vast majority of callers are
780 * looking at current->group_info, which can not be changed except by the
781 * current task. Changing current->group_info requires the task lock, too.
782 */
783#define get_group_info(group_info) do { \
784 atomic_inc(&(group_info)->usage); \
785} while (0)
786
787#define put_group_info(group_info) do { \
788 if (atomic_dec_and_test(&(group_info)->usage)) \
789 groups_free(group_info); \
790} while (0)
791
3e30148c
DH
792extern struct group_info *groups_alloc(int gidsetsize);
793extern void groups_free(struct group_info *group_info);
794extern int set_current_groups(struct group_info *group_info);
795extern int groups_search(struct group_info *group_info, gid_t grp);
1da177e4
LT
796/* access the groups "array" with this macro */
797#define GROUP_AT(gi, i) \
798 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
799
383f2835 800#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
36c8b586 801extern void prefetch_stack(struct task_struct *t);
383f2835
KC
802#else
803static inline void prefetch_stack(struct task_struct *t) { }
804#endif
1da177e4
LT
805
806struct audit_context; /* See audit.c */
807struct mempolicy;
b92ce558 808struct pipe_inode_info;
4865ecf1 809struct uts_namespace;
1da177e4 810
3dee386e
CK
811enum sleep_type {
812 SLEEP_NORMAL,
813 SLEEP_NONINTERACTIVE,
814 SLEEP_INTERACTIVE,
815 SLEEP_INTERRUPTED,
816};
817
70b97a7f
IM
818struct prio_array;
819
1da177e4
LT
820struct task_struct {
821 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
f7e4217b 822 void *stack;
1da177e4 823 atomic_t usage;
97dc32cd
WC
824 unsigned int flags; /* per process flags, defined below */
825 unsigned int ptrace;
1da177e4 826
36772092 827 int lock_depth; /* BKL lock depth */
1da177e4 828
2dd73a4f
PW
829#ifdef CONFIG_SMP
830#ifdef __ARCH_WANT_UNLOCKED_CTXSW
4866cde0
NP
831 int oncpu;
832#endif
2dd73a4f
PW
833#endif
834 int load_weight; /* for niceness load balancing purposes */
b29739f9 835 int prio, static_prio, normal_prio;
1da177e4 836 struct list_head run_list;
70b97a7f 837 struct prio_array *array;
1da177e4 838
22e2c507 839 unsigned short ioprio;
6c5c9341 840#ifdef CONFIG_BLK_DEV_IO_TRACE
2056a782 841 unsigned int btrace_seq;
6c5c9341 842#endif
1da177e4
LT
843 unsigned long sleep_avg;
844 unsigned long long timestamp, last_ran;
845 unsigned long long sched_time; /* sched_clock time spent running */
3dee386e 846 enum sleep_type sleep_type;
1da177e4 847
97dc32cd 848 unsigned int policy;
1da177e4
LT
849 cpumask_t cpus_allowed;
850 unsigned int time_slice, first_time_slice;
851
52f17b6c 852#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
853 struct sched_info sched_info;
854#endif
855
856 struct list_head tasks;
857 /*
858 * ptrace_list/ptrace_children forms the list of my children
859 * that were stolen by a ptracer.
860 */
861 struct list_head ptrace_children;
862 struct list_head ptrace_list;
863
864 struct mm_struct *mm, *active_mm;
865
866/* task state */
867 struct linux_binfmt *binfmt;
97dc32cd 868 int exit_state;
1da177e4
LT
869 int exit_code, exit_signal;
870 int pdeath_signal; /* The signal sent when the parent dies */
871 /* ??? */
97dc32cd 872 unsigned int personality;
1da177e4
LT
873 unsigned did_exec:1;
874 pid_t pid;
875 pid_t tgid;
0a425405
AV
876
877#ifdef CONFIG_CC_STACKPROTECTOR
878 /* Canary value for the -fstack-protector gcc feature */
879 unsigned long stack_canary;
880#endif
1da177e4
LT
881 /*
882 * pointers to (original) parent process, youngest child, younger sibling,
883 * older sibling, respectively. (p->father can be replaced with
884 * p->parent->pid)
885 */
886 struct task_struct *real_parent; /* real parent process (when being debugged) */
887 struct task_struct *parent; /* parent process */
888 /*
889 * children/sibling forms the list of my children plus the
890 * tasks I'm ptracing.
891 */
892 struct list_head children; /* list of my children */
893 struct list_head sibling; /* linkage in my parent's children list */
894 struct task_struct *group_leader; /* threadgroup leader */
895
896 /* PID/PID hash table linkage. */
92476d7f 897 struct pid_link pids[PIDTYPE_MAX];
47e65328 898 struct list_head thread_group;
1da177e4
LT
899
900 struct completion *vfork_done; /* for vfork() */
901 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
902 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
903
97dc32cd 904 unsigned int rt_priority;
1da177e4
LT
905 cputime_t utime, stime;
906 unsigned long nvcsw, nivcsw; /* context switch counts */
907 struct timespec start_time;
908/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
909 unsigned long min_flt, maj_flt;
910
911 cputime_t it_prof_expires, it_virt_expires;
912 unsigned long long it_sched_expires;
913 struct list_head cpu_timers[3];
914
915/* process credentials */
916 uid_t uid,euid,suid,fsuid;
917 gid_t gid,egid,sgid,fsgid;
918 struct group_info *group_info;
919 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
920 unsigned keep_capabilities:1;
921 struct user_struct *user;
922#ifdef CONFIG_KEYS
b5f545c8 923 struct key *request_key_auth; /* assumed request_key authority */
1da177e4 924 struct key *thread_keyring; /* keyring private to this thread */
3e30148c 925 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1da177e4 926#endif
e07e23e1
AV
927 /*
928 * fpu_counter contains the number of consecutive context switches
929 * that the FPU is used. If this is over a threshold, the lazy fpu
930 * saving becomes unlazy to save the trap. This is an unsigned char
931 * so that after 256 times the counter wraps and the behavior turns
932 * lazy again; this to deal with bursty apps that only use FPU for
933 * a short time
934 */
935 unsigned char fpu_counter;
1da177e4 936 int oomkilladj; /* OOM kill score adjustment (bit shift). */
36772092
PBG
937 char comm[TASK_COMM_LEN]; /* executable name excluding path
938 - access with [gs]et_task_comm (which lock
939 it with task_lock())
940 - initialized normally by flush_old_exec */
1da177e4
LT
941/* file system info */
942 int link_count, total_link_count;
3d5b6fcc 943#ifdef CONFIG_SYSVIPC
1da177e4
LT
944/* ipc stuff */
945 struct sysv_sem sysvsem;
3d5b6fcc 946#endif
1da177e4
LT
947/* CPU-specific state of this task */
948 struct thread_struct thread;
949/* filesystem information */
950 struct fs_struct *fs;
951/* open file information */
952 struct files_struct *files;
1651e14e 953/* namespaces */
ab516013 954 struct nsproxy *nsproxy;
1da177e4
LT
955/* signal handlers */
956 struct signal_struct *signal;
957 struct sighand_struct *sighand;
958
959 sigset_t blocked, real_blocked;
150256d8 960 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
1da177e4
LT
961 struct sigpending pending;
962
963 unsigned long sas_ss_sp;
964 size_t sas_ss_size;
965 int (*notifier)(void *priv);
966 void *notifier_data;
967 sigset_t *notifier_mask;
968
969 void *security;
970 struct audit_context *audit_context;
971 seccomp_t seccomp;
972
973/* Thread group tracking */
974 u32 parent_exec_id;
975 u32 self_exec_id;
976/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
977 spinlock_t alloc_lock;
1da177e4 978
b29739f9
IM
979 /* Protection of the PI data structures: */
980 spinlock_t pi_lock;
981
23f78d4a
IM
982#ifdef CONFIG_RT_MUTEXES
983 /* PI waiters blocked on a rt_mutex held by this task */
984 struct plist_head pi_waiters;
985 /* Deadlock detection and priority inheritance handling */
986 struct rt_mutex_waiter *pi_blocked_on;
23f78d4a
IM
987#endif
988
408894ee
IM
989#ifdef CONFIG_DEBUG_MUTEXES
990 /* mutex deadlock detection */
991 struct mutex_waiter *blocked_on;
992#endif
de30a2b3
IM
993#ifdef CONFIG_TRACE_IRQFLAGS
994 unsigned int irq_events;
995 int hardirqs_enabled;
996 unsigned long hardirq_enable_ip;
997 unsigned int hardirq_enable_event;
998 unsigned long hardirq_disable_ip;
999 unsigned int hardirq_disable_event;
1000 int softirqs_enabled;
1001 unsigned long softirq_disable_ip;
1002 unsigned int softirq_disable_event;
1003 unsigned long softirq_enable_ip;
1004 unsigned int softirq_enable_event;
1005 int hardirq_context;
1006 int softirq_context;
1007#endif
fbb9ce95
IM
1008#ifdef CONFIG_LOCKDEP
1009# define MAX_LOCK_DEPTH 30UL
1010 u64 curr_chain_key;
1011 int lockdep_depth;
1012 struct held_lock held_locks[MAX_LOCK_DEPTH];
1013 unsigned int lockdep_recursion;
1014#endif
408894ee 1015
1da177e4
LT
1016/* journalling filesystem info */
1017 void *journal_info;
1018
1019/* VM state */
1020 struct reclaim_state *reclaim_state;
1021
1da177e4
LT
1022 struct backing_dev_info *backing_dev_info;
1023
1024 struct io_context *io_context;
1025
1026 unsigned long ptrace_message;
1027 siginfo_t *last_siginfo; /* For ptrace use. */
1028/*
1029 * current io wait handle: wait queue entry to use for io waits
1030 * If this thread is processing aio, this points at the waitqueue
1031 * inside the currently handled kiocb. It may be NULL (i.e. default
1032 * to a stack based synchronous wait) if its doing sync IO.
1033 */
1034 wait_queue_t *io_wait;
4b98d11b 1035#ifdef CONFIG_TASK_XACCT
1da177e4
LT
1036/* i/o counters(bytes read/written, #syscalls */
1037 u64 rchar, wchar, syscr, syscw;
4b98d11b 1038#endif
7c3ab738 1039 struct task_io_accounting ioac;
8f0ab514 1040#if defined(CONFIG_TASK_XACCT)
1da177e4
LT
1041 u64 acct_rss_mem1; /* accumulated rss usage */
1042 u64 acct_vm_mem1; /* accumulated virtual memory usage */
db5fed26 1043 cputime_t acct_stimexpd;/* stime since last update */
1da177e4
LT
1044#endif
1045#ifdef CONFIG_NUMA
1046 struct mempolicy *mempolicy;
1047 short il_next;
1048#endif
1049#ifdef CONFIG_CPUSETS
1050 struct cpuset *cpuset;
1051 nodemask_t mems_allowed;
1052 int cpuset_mems_generation;
825a46af 1053 int cpuset_mem_spread_rotor;
1da177e4 1054#endif
0771dfef 1055 struct robust_list_head __user *robust_list;
34f192c6
IM
1056#ifdef CONFIG_COMPAT
1057 struct compat_robust_list_head __user *compat_robust_list;
1058#endif
c87e2837
IM
1059 struct list_head pi_state_list;
1060 struct futex_pi_state *pi_state_cache;
0771dfef 1061
22e2c507 1062 atomic_t fs_excl; /* holding fs exclusive resources */
e56d0903 1063 struct rcu_head rcu;
b92ce558
JA
1064
1065 /*
1066 * cache last used pipe for splice
1067 */
1068 struct pipe_inode_info *splice_pipe;
ca74e92b
SN
1069#ifdef CONFIG_TASK_DELAY_ACCT
1070 struct task_delay_info *delays;
f4f154fd
AM
1071#endif
1072#ifdef CONFIG_FAULT_INJECTION
1073 int make_it_fail;
ca74e92b 1074#endif
1da177e4
LT
1075};
1076
1077static inline pid_t process_group(struct task_struct *tsk)
1078{
1079 return tsk->signal->pgrp;
1080}
1081
1ec320af
CLG
1082static inline pid_t signal_session(struct signal_struct *sig)
1083{
1084 return sig->__session;
1085}
1086
937949d9
CLG
1087static inline pid_t process_session(struct task_struct *tsk)
1088{
1ec320af
CLG
1089 return signal_session(tsk->signal);
1090}
1091
1092static inline void set_signal_session(struct signal_struct *sig, pid_t session)
1093{
1094 sig->__session = session;
937949d9
CLG
1095}
1096
22c935f4
EB
1097static inline struct pid *task_pid(struct task_struct *task)
1098{
1099 return task->pids[PIDTYPE_PID].pid;
1100}
1101
1102static inline struct pid *task_tgid(struct task_struct *task)
1103{
1104 return task->group_leader->pids[PIDTYPE_PID].pid;
1105}
1106
1107static inline struct pid *task_pgrp(struct task_struct *task)
1108{
1109 return task->group_leader->pids[PIDTYPE_PGID].pid;
1110}
1111
1112static inline struct pid *task_session(struct task_struct *task)
1113{
1114 return task->group_leader->pids[PIDTYPE_SID].pid;
1115}
1116
1da177e4
LT
1117/**
1118 * pid_alive - check that a task structure is not stale
1119 * @p: Task structure to be checked.
1120 *
1121 * Test if a process is not yet dead (at most zombie state)
1122 * If pid_alive fails, then pointers within the task structure
1123 * can be stale and must not be dereferenced.
1124 */
1125static inline int pid_alive(struct task_struct *p)
1126{
92476d7f 1127 return p->pids[PIDTYPE_PID].pid != NULL;
1da177e4
LT
1128}
1129
f400e198 1130/**
3260259f
HK
1131 * is_init - check if a task structure is init
1132 * @tsk: Task structure to be checked.
1133 *
1134 * Check if a task structure is the first user space task the kernel created.
f400e198
SB
1135 */
1136static inline int is_init(struct task_struct *tsk)
1137{
1138 return tsk->pid == 1;
1139}
1140
9ec52099
CLG
1141extern struct pid *cad_pid;
1142
1da177e4 1143extern void free_task(struct task_struct *tsk);
1da177e4 1144#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
e56d0903 1145
158d9ebd 1146extern void __put_task_struct(struct task_struct *t);
e56d0903
IM
1147
1148static inline void put_task_struct(struct task_struct *t)
1149{
1150 if (atomic_dec_and_test(&t->usage))
8c7904a0 1151 __put_task_struct(t);
e56d0903 1152}
1da177e4
LT
1153
1154/*
1155 * Per process flags
1156 */
1157#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1158 /* Not implemented yet, only for 486*/
1159#define PF_STARTING 0x00000002 /* being created */
1160#define PF_EXITING 0x00000004 /* getting shut down */
1da177e4
LT
1161#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1162#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1163#define PF_DUMPCORE 0x00000200 /* dumped core */
1164#define PF_SIGNALED 0x00000400 /* killed by a signal */
1165#define PF_MEMALLOC 0x00000800 /* Allocating memory */
1166#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1167#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1da177e4
LT
1168#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1169#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1170#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1171#define PF_KSWAPD 0x00040000 /* I am kswapd */
1172#define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1173#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
b31dc66a
JA
1174#define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1175#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1176#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1177#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1178#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
c61afb18 1179#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
61a87122 1180#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1da177e4
LT
1181
1182/*
1183 * Only the _current_ task can read/write to tsk->flags, but other
1184 * tasks can access tsk->flags in readonly mode for example
1185 * with tsk_used_math (like during threaded core dumping).
1186 * There is however an exception to this rule during ptrace
1187 * or during fork: the ptracer task is allowed to write to the
1188 * child->flags of its traced child (same goes for fork, the parent
1189 * can write to the child->flags), because we're guaranteed the
1190 * child is not running and in turn not changing child->flags
1191 * at the same time the parent does it.
1192 */
1193#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1194#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1195#define clear_used_math() clear_stopped_child_used_math(current)
1196#define set_used_math() set_stopped_child_used_math(current)
1197#define conditional_stopped_child_used_math(condition, child) \
1198 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1199#define conditional_used_math(condition) \
1200 conditional_stopped_child_used_math(condition, current)
1201#define copy_to_stopped_child_used_math(child) \
1202 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1203/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1204#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1205#define used_math() tsk_used_math(current)
1206
1207#ifdef CONFIG_SMP
36c8b586 1208extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1da177e4 1209#else
36c8b586 1210static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1da177e4 1211{
4098f991 1212 if (!cpu_isset(0, new_mask))
1da177e4
LT
1213 return -EINVAL;
1214 return 0;
1215}
1216#endif
1217
1218extern unsigned long long sched_clock(void);
36c8b586
IM
1219extern unsigned long long
1220current_sched_time(const struct task_struct *current_task);
1da177e4
LT
1221
1222/* sched_exec is called by processes performing an exec */
1223#ifdef CONFIG_SMP
1224extern void sched_exec(void);
1225#else
1226#define sched_exec() {}
1227#endif
1228
1229#ifdef CONFIG_HOTPLUG_CPU
1230extern void idle_task_exit(void);
1231#else
1232static inline void idle_task_exit(void) {}
1233#endif
1234
1235extern void sched_idle_next(void);
b29739f9
IM
1236
1237#ifdef CONFIG_RT_MUTEXES
36c8b586
IM
1238extern int rt_mutex_getprio(struct task_struct *p);
1239extern void rt_mutex_setprio(struct task_struct *p, int prio);
1240extern void rt_mutex_adjust_pi(struct task_struct *p);
b29739f9 1241#else
36c8b586 1242static inline int rt_mutex_getprio(struct task_struct *p)
b29739f9
IM
1243{
1244 return p->normal_prio;
1245}
95e02ca9 1246# define rt_mutex_adjust_pi(p) do { } while (0)
b29739f9
IM
1247#endif
1248
36c8b586
IM
1249extern void set_user_nice(struct task_struct *p, long nice);
1250extern int task_prio(const struct task_struct *p);
1251extern int task_nice(const struct task_struct *p);
1252extern int can_nice(const struct task_struct *p, const int nice);
1253extern int task_curr(const struct task_struct *p);
1da177e4
LT
1254extern int idle_cpu(int cpu);
1255extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
36c8b586
IM
1256extern struct task_struct *idle_task(int cpu);
1257extern struct task_struct *curr_task(int cpu);
1258extern void set_curr_task(int cpu, struct task_struct *p);
1da177e4
LT
1259
1260void yield(void);
1261
1262/*
1263 * The default (Linux) execution domain.
1264 */
1265extern struct exec_domain default_exec_domain;
1266
1267union thread_union {
1268 struct thread_info thread_info;
1269 unsigned long stack[THREAD_SIZE/sizeof(long)];
1270};
1271
1272#ifndef __HAVE_ARCH_KSTACK_END
1273static inline int kstack_end(void *addr)
1274{
1275 /* Reliable end of stack detection:
1276 * Some APM bios versions misalign the stack
1277 */
1278 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1279}
1280#endif
1281
1282extern union thread_union init_thread_union;
1283extern struct task_struct init_task;
1284
1285extern struct mm_struct init_mm;
1286
1287#define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1288extern struct task_struct *find_task_by_pid_type(int type, int pid);
1da177e4
LT
1289extern void __set_special_pids(pid_t session, pid_t pgrp);
1290
1291/* per-UID process charging. */
1292extern struct user_struct * alloc_uid(uid_t);
1293static inline struct user_struct *get_uid(struct user_struct *u)
1294{
1295 atomic_inc(&u->__count);
1296 return u;
1297}
1298extern void free_uid(struct user_struct *);
1299extern void switch_uid(struct user_struct *);
1300
1301#include <asm/current.h>
1302
3171a030 1303extern void do_timer(unsigned long ticks);
1da177e4
LT
1304
1305extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1306extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1307extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1308 unsigned long clone_flags));
1309#ifdef CONFIG_SMP
1310 extern void kick_process(struct task_struct *tsk);
1311#else
1312 static inline void kick_process(struct task_struct *tsk) { }
1313#endif
36c8b586
IM
1314extern void FASTCALL(sched_fork(struct task_struct * p, int clone_flags));
1315extern void FASTCALL(sched_exit(struct task_struct * p));
1da177e4
LT
1316
1317extern int in_group_p(gid_t);
1318extern int in_egroup_p(gid_t);
1319
1320extern void proc_caches_init(void);
1321extern void flush_signals(struct task_struct *);
10ab825b 1322extern void ignore_signals(struct task_struct *);
1da177e4
LT
1323extern void flush_signal_handlers(struct task_struct *, int force_default);
1324extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1325
1326static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1327{
1328 unsigned long flags;
1329 int ret;
1330
1331 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1332 ret = dequeue_signal(tsk, mask, info);
1333 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1334
1335 return ret;
1336}
1337
1338extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1339 sigset_t *mask);
1340extern void unblock_all_signals(void);
1341extern void release_task(struct task_struct * p);
1342extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1343extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1344extern int force_sigsegv(int, struct task_struct *);
1345extern int force_sig_info(int, struct siginfo *, struct task_struct *);
c4b92fc1
EB
1346extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1347extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1348extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2425c08b 1349extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
c4b92fc1
EB
1350extern int kill_pgrp(struct pid *pid, int sig, int priv);
1351extern int kill_pid(struct pid *pid, int sig, int priv);
c3de4b38 1352extern int kill_proc_info(int, struct siginfo *, pid_t);
1da177e4
LT
1353extern void do_notify_parent(struct task_struct *, int);
1354extern void force_sig(int, struct task_struct *);
1355extern void force_sig_specific(int, struct task_struct *);
1356extern int send_sig(int, struct task_struct *, int);
1357extern void zap_other_threads(struct task_struct *p);
1da177e4
LT
1358extern int kill_proc(pid_t, int, int);
1359extern struct sigqueue *sigqueue_alloc(void);
1360extern void sigqueue_free(struct sigqueue *);
1361extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1362extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
9ac95f2f 1363extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1da177e4
LT
1364extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1365
9ec52099
CLG
1366static inline int kill_cad_pid(int sig, int priv)
1367{
1368 return kill_pid(cad_pid, sig, priv);
1369}
1370
1da177e4
LT
1371/* These can be the second arg to send_sig_info/send_group_sig_info. */
1372#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1373#define SEND_SIG_PRIV ((struct siginfo *) 1)
1374#define SEND_SIG_FORCED ((struct siginfo *) 2)
1375
621d3121
ON
1376static inline int is_si_special(const struct siginfo *info)
1377{
1378 return info <= SEND_SIG_FORCED;
1379}
1380
1da177e4
LT
1381/* True if we are on the alternate signal stack. */
1382
1383static inline int on_sig_stack(unsigned long sp)
1384{
1385 return (sp - current->sas_ss_sp < current->sas_ss_size);
1386}
1387
1388static inline int sas_ss_flags(unsigned long sp)
1389{
1390 return (current->sas_ss_size == 0 ? SS_DISABLE
1391 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1392}
1393
1da177e4
LT
1394/*
1395 * Routines for handling mm_structs
1396 */
1397extern struct mm_struct * mm_alloc(void);
1398
1399/* mmdrop drops the mm and the page tables */
1400extern void FASTCALL(__mmdrop(struct mm_struct *));
1401static inline void mmdrop(struct mm_struct * mm)
1402{
1403 if (atomic_dec_and_test(&mm->mm_count))
1404 __mmdrop(mm);
1405}
1406
1407/* mmput gets rid of the mappings and all user-space */
1408extern void mmput(struct mm_struct *);
1409/* Grab a reference to a task's mm, if it is not already going away */
1410extern struct mm_struct *get_task_mm(struct task_struct *task);
1411/* Remove the current tasks stale references to the old mm_struct */
1412extern void mm_release(struct task_struct *, struct mm_struct *);
1413
1414extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1415extern void flush_thread(void);
1416extern void exit_thread(void);
1417
1da177e4 1418extern void exit_files(struct task_struct *);
6b3934ef 1419extern void __cleanup_signal(struct signal_struct *);
a7e5328a 1420extern void __cleanup_sighand(struct sighand_struct *);
1da177e4
LT
1421extern void exit_itimers(struct signal_struct *);
1422
1423extern NORET_TYPE void do_group_exit(int);
1424
1da177e4
LT
1425extern void daemonize(const char *, ...);
1426extern int allow_signal(int);
1427extern int disallow_signal(int);
1da177e4
LT
1428
1429extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1430extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
36c8b586 1431struct task_struct *fork_idle(int);
1da177e4
LT
1432
1433extern void set_task_comm(struct task_struct *tsk, char *from);
1434extern void get_task_comm(char *to, struct task_struct *tsk);
1435
1436#ifdef CONFIG_SMP
36c8b586 1437extern void wait_task_inactive(struct task_struct * p);
1da177e4
LT
1438#else
1439#define wait_task_inactive(p) do { } while (0)
1440#endif
1441
1442#define remove_parent(p) list_del_init(&(p)->sibling)
8fafabd8 1443#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1da177e4 1444
5e85d4ab 1445#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1da177e4
LT
1446
1447#define for_each_process(p) \
1448 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1449
1450/*
1451 * Careful: do_each_thread/while_each_thread is a double loop so
1452 * 'break' will not work as expected - use goto instead.
1453 */
1454#define do_each_thread(g, t) \
1455 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1456
1457#define while_each_thread(g, t) \
1458 while ((t = next_thread(t)) != g)
1459
de12a787
EB
1460/* de_thread depends on thread_group_leader not being a pid based check */
1461#define thread_group_leader(p) (p == p->group_leader)
1da177e4 1462
0804ef4b
EB
1463/* Do to the insanities of de_thread it is possible for a process
1464 * to have the pid of the thread group leader without actually being
1465 * the thread group leader. For iteration through the pids in proc
1466 * all we care about is that we have a task with the appropriate
1467 * pid, we don't actually care if we have the right task.
1468 */
1469static inline int has_group_leader_pid(struct task_struct *p)
1470{
1471 return p->pid == p->tgid;
1472}
1473
36c8b586 1474static inline struct task_struct *next_thread(const struct task_struct *p)
47e65328
ON
1475{
1476 return list_entry(rcu_dereference(p->thread_group.next),
36c8b586 1477 struct task_struct, thread_group);
47e65328
ON
1478}
1479
36c8b586 1480static inline int thread_group_empty(struct task_struct *p)
1da177e4 1481{
47e65328 1482 return list_empty(&p->thread_group);
1da177e4
LT
1483}
1484
1485#define delay_group_leader(p) \
1486 (thread_group_leader(p) && !thread_group_empty(p))
1487
1da177e4 1488/*
260ea101 1489 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
22e2c507 1490 * subscriptions and synchronises with wait4(). Also used in procfs. Also
053199ed 1491 * pins the final release of task.io_context. Also protects ->cpuset.
1da177e4
LT
1492 *
1493 * Nests both inside and outside of read_lock(&tasklist_lock).
1494 * It must not be nested with write_lock_irq(&tasklist_lock),
1495 * neither inside nor outside.
1496 */
1497static inline void task_lock(struct task_struct *p)
1498{
1499 spin_lock(&p->alloc_lock);
1500}
1501
1502static inline void task_unlock(struct task_struct *p)
1503{
1504 spin_unlock(&p->alloc_lock);
1505}
1506
f63ee72e
ON
1507extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1508 unsigned long *flags);
1509
1510static inline void unlock_task_sighand(struct task_struct *tsk,
1511 unsigned long *flags)
1512{
1513 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1514}
1515
f037360f
AV
1516#ifndef __HAVE_THREAD_FUNCTIONS
1517
f7e4217b
RZ
1518#define task_thread_info(task) ((struct thread_info *)(task)->stack)
1519#define task_stack_page(task) ((task)->stack)
a1261f54 1520
10ebffde
AV
1521static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1522{
1523 *task_thread_info(p) = *task_thread_info(org);
1524 task_thread_info(p)->task = p;
1525}
1526
1527static inline unsigned long *end_of_stack(struct task_struct *p)
1528{
f7e4217b 1529 return (unsigned long *)(task_thread_info(p) + 1);
10ebffde
AV
1530}
1531
f037360f
AV
1532#endif
1533
1da177e4
LT
1534/* set thread flags in other task's structures
1535 * - see asm/thread_info.h for TIF_xxxx flags available
1536 */
1537static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1538{
a1261f54 1539 set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1540}
1541
1542static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1543{
a1261f54 1544 clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1545}
1546
1547static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1548{
a1261f54 1549 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1550}
1551
1552static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1553{
a1261f54 1554 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1555}
1556
1557static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1558{
a1261f54 1559 return test_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1560}
1561
1562static inline void set_tsk_need_resched(struct task_struct *tsk)
1563{
1564 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1565}
1566
1567static inline void clear_tsk_need_resched(struct task_struct *tsk)
1568{
1569 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1570}
1571
1572static inline int signal_pending(struct task_struct *p)
1573{
1574 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1575}
1576
1577static inline int need_resched(void)
1578{
1579 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1580}
1581
1582/*
1583 * cond_resched() and cond_resched_lock(): latency reduction via
1584 * explicit rescheduling in places that are safe. The return
1585 * value indicates whether a reschedule was done in fact.
1586 * cond_resched_lock() will drop the spinlock before scheduling,
1587 * cond_resched_softirq() will enable bhs before scheduling.
1588 */
1589extern int cond_resched(void);
1590extern int cond_resched_lock(spinlock_t * lock);
1591extern int cond_resched_softirq(void);
1592
1593/*
1594 * Does a critical section need to be broken due to another
1595 * task waiting?:
1596 */
1597#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1598# define need_lockbreak(lock) ((lock)->break_lock)
1599#else
1600# define need_lockbreak(lock) 0
1601#endif
1602
1603/*
1604 * Does a critical section need to be broken due to another
1605 * task waiting or preemption being signalled:
1606 */
1607static inline int lock_need_resched(spinlock_t *lock)
1608{
1609 if (need_lockbreak(lock) || need_resched())
1610 return 1;
1611 return 0;
1612}
1613
1614/* Reevaluate whether the task has signals pending delivery.
1615 This is required every time the blocked sigset_t changes.
1616 callers must hold sighand->siglock. */
1617
1618extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1619extern void recalc_sigpending(void);
1620
1621extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1622
1623/*
1624 * Wrappers for p->thread_info->cpu access. No-op on UP.
1625 */
1626#ifdef CONFIG_SMP
1627
1628static inline unsigned int task_cpu(const struct task_struct *p)
1629{
a1261f54 1630 return task_thread_info(p)->cpu;
1da177e4
LT
1631}
1632
1633static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1634{
a1261f54 1635 task_thread_info(p)->cpu = cpu;
1da177e4
LT
1636}
1637
1638#else
1639
1640static inline unsigned int task_cpu(const struct task_struct *p)
1641{
1642 return 0;
1643}
1644
1645static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1646{
1647}
1648
1649#endif /* CONFIG_SMP */
1650
1651#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1652extern void arch_pick_mmap_layout(struct mm_struct *mm);
1653#else
1654static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1655{
1656 mm->mmap_base = TASK_UNMAPPED_BASE;
1657 mm->get_unmapped_area = arch_get_unmapped_area;
1658 mm->unmap_area = arch_unmap_area;
1659}
1660#endif
1661
1662extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1663extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1664
5c45bf27 1665extern int sched_mc_power_savings, sched_smt_power_savings;
5c45bf27 1666
1da177e4
LT
1667extern void normalize_rt_tasks(void);
1668
4b98d11b
AD
1669#ifdef CONFIG_TASK_XACCT
1670static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1671{
1672 tsk->rchar += amt;
1673}
1674
1675static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1676{
1677 tsk->wchar += amt;
1678}
1679
1680static inline void inc_syscr(struct task_struct *tsk)
1681{
1682 tsk->syscr++;
1683}
1684
1685static inline void inc_syscw(struct task_struct *tsk)
1686{
1687 tsk->syscw++;
1688}
1689#else
1690static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1691{
1692}
1693
1694static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1695{
1696}
1697
1698static inline void inc_syscr(struct task_struct *tsk)
1699{
1700}
1701
1702static inline void inc_syscw(struct task_struct *tsk)
1703{
1704}
1705#endif
1706
1da177e4
LT
1707#endif /* __KERNEL__ */
1708
1709#endif