rt-mutex: fix chain walk early wakeup bug
[GitHub/mt8127/android_kernel_alcatel_ttab.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;
d89d8796 91struct bio;
1da177e4 92
1da177e4
LT
93/*
94 * List of flags we want to share for kernel threads,
95 * if only because they are not used by them anyway.
96 */
97#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
98
99/*
100 * These are the constant used to fake the fixed-point load-average
101 * counting. Some notes:
102 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
103 * a load-average precision of 10 bits integer + 11 bits fractional
104 * - if you want to count load-averages more often, you need more
105 * precision, or rounding will get you. With 2-second counting freq,
106 * the EXP_n values would be 1981, 2034 and 2043 if still using only
107 * 11 bit fractions.
108 */
109extern unsigned long avenrun[]; /* Load averages */
110
111#define FSHIFT 11 /* nr of bits of precision */
112#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
113#define LOAD_FREQ (5*HZ) /* 5 sec intervals */
114#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
115#define EXP_5 2014 /* 1/exp(5sec/5min) */
116#define EXP_15 2037 /* 1/exp(5sec/15min) */
117
118#define CALC_LOAD(load,exp,n) \
119 load *= exp; \
120 load += n*(FIXED_1-exp); \
121 load >>= FSHIFT;
122
123extern unsigned long total_forks;
124extern int nr_threads;
1da177e4
LT
125DECLARE_PER_CPU(unsigned long, process_counts);
126extern int nr_processes(void);
127extern unsigned long nr_running(void);
128extern unsigned long nr_uninterruptible(void);
db1b1fef 129extern unsigned long nr_active(void);
1da177e4 130extern unsigned long nr_iowait(void);
2dd73a4f 131extern unsigned long weighted_cpuload(const int cpu);
1da177e4 132
1da177e4 133
4a8342d2
LT
134/*
135 * Task state bitmask. NOTE! These bits are also
136 * encoded in fs/proc/array.c: get_task_state().
137 *
138 * We have two separate sets of flags: task->state
139 * is about runnability, while task->exit_state are
140 * about the task exiting. Confusing, but this way
141 * modifying one set can't modify the other one by
142 * mistake.
143 */
1da177e4
LT
144#define TASK_RUNNING 0
145#define TASK_INTERRUPTIBLE 1
146#define TASK_UNINTERRUPTIBLE 2
4a8342d2
LT
147#define TASK_STOPPED 4
148#define TASK_TRACED 8
149/* in tsk->exit_state */
150#define EXIT_ZOMBIE 16
151#define EXIT_DEAD 32
152/* in tsk->state again */
153#define TASK_NONINTERACTIVE 64
c394cc9f 154#define TASK_DEAD 128
1da177e4
LT
155
156#define __set_task_state(tsk, state_value) \
157 do { (tsk)->state = (state_value); } while (0)
158#define set_task_state(tsk, state_value) \
159 set_mb((tsk)->state, (state_value))
160
498d0c57
AM
161/*
162 * set_current_state() includes a barrier so that the write of current->state
163 * is correctly serialised wrt the caller's subsequent test of whether to
164 * actually sleep:
165 *
166 * set_current_state(TASK_UNINTERRUPTIBLE);
167 * if (do_i_need_to_sleep())
168 * schedule();
169 *
170 * If the caller does not need such serialisation then use __set_current_state()
171 */
1da177e4
LT
172#define __set_current_state(state_value) \
173 do { current->state = (state_value); } while (0)
174#define set_current_state(state_value) \
175 set_mb(current->state, (state_value))
176
177/* Task command name length */
178#define TASK_COMM_LEN 16
179
1da177e4
LT
180#include <linux/spinlock.h>
181
182/*
183 * This serializes "schedule()" and also protects
184 * the run-queue from deletions/modifications (but
185 * _adding_ to the beginning of the run-queue has
186 * a separate lock).
187 */
188extern rwlock_t tasklist_lock;
189extern spinlock_t mmlist_lock;
190
36c8b586 191struct task_struct;
1da177e4
LT
192
193extern void sched_init(void);
194extern void sched_init_smp(void);
36c8b586 195extern void init_idle(struct task_struct *idle, int cpu);
1da177e4
LT
196
197extern cpumask_t nohz_cpu_mask;
46cb4b7c
SS
198#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
199extern int select_nohz_load_balancer(int cpu);
200#else
201static inline int select_nohz_load_balancer(int cpu)
202{
203 return 0;
204}
205#endif
1da177e4 206
e59e2ae2 207/*
39bc89fd 208 * Only dump TASK_* tasks. (0 for all tasks)
e59e2ae2
IM
209 */
210extern void show_state_filter(unsigned long state_filter);
211
212static inline void show_state(void)
213{
39bc89fd 214 show_state_filter(0);
e59e2ae2
IM
215}
216
1da177e4
LT
217extern void show_regs(struct pt_regs *);
218
219/*
220 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
221 * task), SP is the stack pointer of the first frame that should be shown in the back
222 * trace (or NULL if the entire call-chain of the task should be shown).
223 */
224extern void show_stack(struct task_struct *task, unsigned long *sp);
225
226void io_schedule(void);
227long io_schedule_timeout(long timeout);
228
229extern void cpu_init (void);
230extern void trap_init(void);
231extern void update_process_times(int user);
232extern void scheduler_tick(void);
233
8446f1d3 234#ifdef CONFIG_DETECT_SOFTLOCKUP
6687a97d 235extern void softlockup_tick(void);
8446f1d3
IM
236extern void spawn_softlockup_task(void);
237extern void touch_softlockup_watchdog(void);
04c9167f 238extern void touch_all_softlockup_watchdogs(void);
8446f1d3 239#else
6687a97d 240static inline void softlockup_tick(void)
8446f1d3
IM
241{
242}
243static inline void spawn_softlockup_task(void)
244{
245}
246static inline void touch_softlockup_watchdog(void)
247{
248}
04c9167f
JF
249static inline void touch_all_softlockup_watchdogs(void)
250{
251}
8446f1d3
IM
252#endif
253
254
1da177e4
LT
255/* Attach to any functions which should be ignored in wchan output. */
256#define __sched __attribute__((__section__(".sched.text")))
257/* Is this address in the __sched functions? */
258extern int in_sched_functions(unsigned long addr);
259
260#define MAX_SCHEDULE_TIMEOUT LONG_MAX
261extern signed long FASTCALL(schedule_timeout(signed long timeout));
64ed93a2
NA
262extern signed long schedule_timeout_interruptible(signed long timeout);
263extern signed long schedule_timeout_uninterruptible(signed long timeout);
1da177e4
LT
264asmlinkage void schedule(void);
265
ab516013 266struct nsproxy;
1da177e4
LT
267
268/* Maximum number of active map areas.. This is a random (large) number */
269#define DEFAULT_MAX_MAP_COUNT 65536
270
271extern int sysctl_max_map_count;
272
273#include <linux/aio.h>
274
275extern unsigned long
276arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
277 unsigned long, unsigned long);
278extern unsigned long
279arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
280 unsigned long len, unsigned long pgoff,
281 unsigned long flags);
1363c3cd
WW
282extern void arch_unmap_area(struct mm_struct *, unsigned long);
283extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
1da177e4 284
f412ac08
HD
285#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
286/*
287 * The mm counters are not protected by its page_table_lock,
288 * so must be incremented atomically.
289 */
d3cb4871
CL
290#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
291#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
292#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
293#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
294#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
295typedef atomic_long_t mm_counter_t;
f412ac08
HD
296
297#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
298/*
299 * The mm counters are protected by its page_table_lock,
300 * so can be incremented directly.
301 */
1da177e4
LT
302#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
303#define get_mm_counter(mm, member) ((mm)->_##member)
304#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
305#define inc_mm_counter(mm, member) (mm)->_##member++
306#define dec_mm_counter(mm, member) (mm)->_##member--
f412ac08
HD
307typedef unsigned long mm_counter_t;
308
309#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
4294621f 310
f412ac08
HD
311#define get_mm_rss(mm) \
312 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
365e9c87
HD
313#define update_hiwater_rss(mm) do { \
314 unsigned long _rss = get_mm_rss(mm); \
315 if ((mm)->hiwater_rss < _rss) \
316 (mm)->hiwater_rss = _rss; \
317} while (0)
318#define update_hiwater_vm(mm) do { \
319 if ((mm)->hiwater_vm < (mm)->total_vm) \
320 (mm)->hiwater_vm = (mm)->total_vm; \
321} while (0)
322
1da177e4
LT
323struct mm_struct {
324 struct vm_area_struct * mmap; /* list of VMAs */
325 struct rb_root mm_rb;
326 struct vm_area_struct * mmap_cache; /* last find_vma result */
327 unsigned long (*get_unmapped_area) (struct file *filp,
328 unsigned long addr, unsigned long len,
329 unsigned long pgoff, unsigned long flags);
1363c3cd 330 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
0551fbd2
BH
331 unsigned long mmap_base; /* base of mmap area */
332 unsigned long task_size; /* size of task vm space */
333 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
1363c3cd 334 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
1da177e4
LT
335 pgd_t * pgd;
336 atomic_t mm_users; /* How many users with user space? */
337 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
338 int map_count; /* number of VMAs */
339 struct rw_semaphore mmap_sem;
340 spinlock_t page_table_lock; /* Protects page tables and some counters */
341
342 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
343 * together off init_mm.mmlist, and are protected
344 * by mmlist_lock
345 */
346
f412ac08
HD
347 /* Special counters, in some configurations protected by the
348 * page_table_lock, in other configurations by being atomic.
349 */
4294621f 350 mm_counter_t _file_rss;
1da177e4
LT
351 mm_counter_t _anon_rss;
352
f449952b
HD
353 unsigned long hiwater_rss; /* High-watermark of RSS usage */
354 unsigned long hiwater_vm; /* High-water virtual memory usage */
355
356 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
357 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
358 unsigned long start_code, end_code, start_data, end_data;
359 unsigned long start_brk, brk, start_stack;
360 unsigned long arg_start, arg_end, env_start, env_end;
361
36d57ac4 362 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
1da177e4 363
1da177e4
LT
364 cpumask_t cpu_vm_mask;
365
366 /* Architecture-specific MM context */
367 mm_context_t context;
368
7602bdf2
AC
369 /* Swap token stuff */
370 /*
371 * Last value of global fault stamp as seen by this process.
372 * In other words, this value gives an indication of how long
373 * it has been since this task got the token.
374 * Look at mm/thrash.c
375 */
376 unsigned int faultstamp;
377 unsigned int token_priority;
378 unsigned int last_interval;
1da177e4 379
36de6437
ACM
380 unsigned char dumpable:2;
381
1da177e4
LT
382 /* coredumping support */
383 int core_waiters;
384 struct completion *core_startup_done, core_done;
385
386 /* aio bits */
387 rwlock_t ioctx_list_lock;
388 struct kioctx *ioctx_list;
1da177e4
LT
389};
390
391struct sighand_struct {
392 atomic_t count;
393 struct k_sigaction action[_NSIG];
394 spinlock_t siglock;
fba2afaa 395 struct list_head signalfd_list;
1da177e4
LT
396};
397
0e464814 398struct pacct_struct {
f6ec29a4
KK
399 int ac_flag;
400 long ac_exitcode;
0e464814 401 unsigned long ac_mem;
77787bfb
KK
402 cputime_t ac_utime, ac_stime;
403 unsigned long ac_minflt, ac_majflt;
0e464814
KK
404};
405
1da177e4
LT
406/*
407 * NOTE! "signal_struct" does not have it's own
408 * locking, because a shared signal_struct always
409 * implies a shared sighand_struct, so locking
410 * sighand_struct is always a proper superset of
411 * the locking of signal_struct.
412 */
413struct signal_struct {
414 atomic_t count;
415 atomic_t live;
416
417 wait_queue_head_t wait_chldexit; /* for wait4() */
418
419 /* current thread group signal load-balancing target: */
36c8b586 420 struct task_struct *curr_target;
1da177e4
LT
421
422 /* shared signal handling: */
423 struct sigpending shared_pending;
424
425 /* thread group exit support */
426 int group_exit_code;
427 /* overloaded:
428 * - notify group_exit_task when ->count is equal to notify_count
429 * - everyone except group_exit_task is stopped during signal delivery
430 * of fatal signals, group_exit_task processes the signal.
431 */
432 struct task_struct *group_exit_task;
433 int notify_count;
434
435 /* thread group stop support, overloads group_exit_code too */
436 int group_stop_count;
437 unsigned int flags; /* see SIGNAL_* flags below */
438
439 /* POSIX.1b Interval Timers */
440 struct list_head posix_timers;
441
442 /* ITIMER_REAL timer for the process */
2ff678b8 443 struct hrtimer real_timer;
05cfb614 444 struct task_struct *tsk;
2ff678b8 445 ktime_t it_real_incr;
1da177e4
LT
446
447 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
448 cputime_t it_prof_expires, it_virt_expires;
449 cputime_t it_prof_incr, it_virt_incr;
450
451 /* job control IDs */
452 pid_t pgrp;
ab521dc0 453 struct pid *tty_old_pgrp;
1ec320af
CLG
454
455 union {
456 pid_t session __deprecated;
457 pid_t __session;
458 };
459
1da177e4
LT
460 /* boolean value for session group leader */
461 int leader;
462
463 struct tty_struct *tty; /* NULL if no tty */
464
465 /*
466 * Cumulative resource counters for dead threads in the group,
467 * and for reaped dead child processes forked by this group.
468 * Live threads maintain their own counters and add to these
469 * in __exit_signal, except for the group leader.
470 */
471 cputime_t utime, stime, cutime, cstime;
472 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
473 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
6eaeeaba 474 unsigned long inblock, oublock, cinblock, coublock;
1da177e4
LT
475
476 /*
477 * Cumulative ns of scheduled CPU time for dead threads in the
478 * group, not including a zombie group leader. (This only differs
479 * from jiffies_to_ns(utime + stime) if sched_clock uses something
480 * other than jiffies.)
481 */
482 unsigned long long sched_time;
483
484 /*
485 * We don't bother to synchronize most readers of this at all,
486 * because there is no reader checking a limit that actually needs
487 * to get both rlim_cur and rlim_max atomically, and either one
488 * alone is a single word that can safely be read normally.
489 * getrlimit/setrlimit use task_lock(current->group_leader) to
490 * protect this instead of the siglock, because they really
491 * have no need to disable irqs.
492 */
493 struct rlimit rlim[RLIM_NLIMITS];
494
495 struct list_head cpu_timers[3];
496
497 /* keep the process-shared keyrings here so that they do the right
498 * thing in threads created with CLONE_THREAD */
499#ifdef CONFIG_KEYS
500 struct key *session_keyring; /* keyring inherited over fork */
501 struct key *process_keyring; /* keyring private to this process */
502#endif
0e464814
KK
503#ifdef CONFIG_BSD_PROCESS_ACCT
504 struct pacct_struct pacct; /* per-process accounting information */
505#endif
ad4ecbcb 506#ifdef CONFIG_TASKSTATS
ad4ecbcb
SN
507 struct taskstats *stats;
508#endif
1da177e4
LT
509};
510
4866cde0
NP
511/* Context switch must be unlocked if interrupts are to be enabled */
512#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
513# define __ARCH_WANT_UNLOCKED_CTXSW
514#endif
515
1da177e4
LT
516/*
517 * Bits in flags field of signal_struct.
518 */
519#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
520#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
521#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
522#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
523
524
525/*
526 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
b0a9499c
IM
527 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
528 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
529 * values are inverted: lower p->prio value means higher priority.
1da177e4
LT
530 *
531 * The MAX_USER_RT_PRIO value allows the actual maximum
532 * RT priority to be separate from the value exported to
533 * user-space. This allows kernel threads to set their
534 * priority to a value higher than any user task. Note:
535 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
536 */
537
538#define MAX_USER_RT_PRIO 100
539#define MAX_RT_PRIO MAX_USER_RT_PRIO
540
541#define MAX_PRIO (MAX_RT_PRIO + 40)
542
b29739f9
IM
543#define rt_prio(prio) unlikely((prio) < MAX_RT_PRIO)
544#define rt_task(p) rt_prio((p)->prio)
d425b274 545#define batch_task(p) (unlikely((p)->policy == SCHED_BATCH))
57a6f51c
ON
546#define is_rt_policy(p) ((p) != SCHED_NORMAL && (p) != SCHED_BATCH)
547#define has_rt_policy(p) unlikely(is_rt_policy((p)->policy))
1da177e4
LT
548
549/*
550 * Some day this will be a full-fledged user tracking system..
551 */
552struct user_struct {
553 atomic_t __count; /* reference count */
554 atomic_t processes; /* How many processes does this user have? */
555 atomic_t files; /* How many open files does this user have? */
556 atomic_t sigpending; /* How many pending signals does this user have? */
2d9048e2 557#ifdef CONFIG_INOTIFY_USER
0eeca283
RL
558 atomic_t inotify_watches; /* How many inotify watches does this user have? */
559 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
560#endif
1da177e4
LT
561 /* protected by mq_lock */
562 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
563 unsigned long locked_shm; /* How many pages of mlocked shm ? */
564
565#ifdef CONFIG_KEYS
566 struct key *uid_keyring; /* UID specific keyring */
567 struct key *session_keyring; /* UID's default session keyring */
568#endif
569
570 /* Hash table maintenance information */
571 struct list_head uidhash_list;
572 uid_t uid;
573};
574
575extern struct user_struct *find_user(uid_t);
576
577extern struct user_struct root_user;
578#define INIT_USER (&root_user)
579
1da177e4
LT
580struct backing_dev_info;
581struct reclaim_state;
582
52f17b6c 583#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
584struct sched_info {
585 /* cumulative counters */
586 unsigned long cpu_time, /* time spent on the cpu */
587 run_delay, /* time spent waiting on a runqueue */
588 pcnt; /* # of timeslices run on this cpu */
589
590 /* timestamps */
591 unsigned long last_arrival, /* when we last ran on a cpu */
592 last_queued; /* when we were last queued to run */
593};
52f17b6c 594#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
1da177e4 595
52f17b6c 596#ifdef CONFIG_SCHEDSTATS
15ad7cdc 597extern const struct file_operations proc_schedstat_operations;
52f17b6c 598#endif /* CONFIG_SCHEDSTATS */
1da177e4 599
ca74e92b
SN
600#ifdef CONFIG_TASK_DELAY_ACCT
601struct task_delay_info {
602 spinlock_t lock;
603 unsigned int flags; /* Private per-task flags */
604
605 /* For each stat XXX, add following, aligned appropriately
606 *
607 * struct timespec XXX_start, XXX_end;
608 * u64 XXX_delay;
609 * u32 XXX_count;
610 *
611 * Atomicity of updates to XXX_delay, XXX_count protected by
612 * single lock above (split into XXX_lock if contention is an issue).
613 */
0ff92245
SN
614
615 /*
616 * XXX_count is incremented on every XXX operation, the delay
617 * associated with the operation is added to XXX_delay.
618 * XXX_delay contains the accumulated delay time in nanoseconds.
619 */
620 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
621 u64 blkio_delay; /* wait for sync block io completion */
622 u64 swapin_delay; /* wait for swapin block io completion */
623 u32 blkio_count; /* total count of the number of sync block */
624 /* io operations performed */
625 u32 swapin_count; /* total count of the number of swapin block */
626 /* io operations performed */
ca74e92b 627};
52f17b6c
CS
628#endif /* CONFIG_TASK_DELAY_ACCT */
629
630static inline int sched_info_on(void)
631{
632#ifdef CONFIG_SCHEDSTATS
633 return 1;
634#elif defined(CONFIG_TASK_DELAY_ACCT)
635 extern int delayacct_on;
636 return delayacct_on;
637#else
638 return 0;
ca74e92b 639#endif
52f17b6c 640}
ca74e92b 641
1da177e4
LT
642enum idle_type
643{
644 SCHED_IDLE,
645 NOT_IDLE,
646 NEWLY_IDLE,
647 MAX_IDLE_TYPES
648};
649
650/*
651 * sched-domains (multiprocessor balancing) declarations:
652 */
1da177e4
LT
653#define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
654
2dd73a4f 655#ifdef CONFIG_SMP
1da177e4
LT
656#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
657#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
658#define SD_BALANCE_EXEC 4 /* Balance on exec */
147cbb4b
NP
659#define SD_BALANCE_FORK 8 /* Balance on fork, clone */
660#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
661#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
662#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
663#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
5c45bf27 664#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
89c4710e 665#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
08c183f3 666#define SD_SERIALIZE 1024 /* Only a single load balancing instance */
5c45bf27 667
89c4710e
SS
668#define BALANCE_FOR_MC_POWER \
669 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
670
671#define BALANCE_FOR_PKG_POWER \
672 ((sched_mc_power_savings || sched_smt_power_savings) ? \
673 SD_POWERSAVINGS_BALANCE : 0)
674
675#define test_sd_parent(sd, flag) ((sd->parent && \
676 (sd->parent->flags & flag)) ? 1 : 0)
5c45bf27 677
1da177e4
LT
678
679struct sched_group {
680 struct sched_group *next; /* Must be a circular list */
681 cpumask_t cpumask;
682
683 /*
684 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
685 * single CPU. This is read only (except for setup, hotplug CPU).
5517d86b 686 * Note : Never change cpu_power without recompute its reciprocal
1da177e4 687 */
5517d86b
ED
688 unsigned int __cpu_power;
689 /*
690 * reciprocal value of cpu_power to avoid expensive divides
691 * (see include/linux/reciprocal_div.h)
692 */
693 u32 reciprocal_cpu_power;
1da177e4
LT
694};
695
696struct sched_domain {
697 /* These fields must be setup */
698 struct sched_domain *parent; /* top domain must be null terminated */
1a848870 699 struct sched_domain *child; /* bottom domain must be null terminated */
1da177e4
LT
700 struct sched_group *groups; /* the balancing groups of the domain */
701 cpumask_t span; /* span of all CPUs in this domain */
702 unsigned long min_interval; /* Minimum balance interval ms */
703 unsigned long max_interval; /* Maximum balance interval ms */
704 unsigned int busy_factor; /* less balancing by factor if busy */
705 unsigned int imbalance_pct; /* No balance until over watermark */
706 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
707 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
7897986b
NP
708 unsigned int busy_idx;
709 unsigned int idle_idx;
710 unsigned int newidle_idx;
711 unsigned int wake_idx;
147cbb4b 712 unsigned int forkexec_idx;
1da177e4
LT
713 int flags; /* See SD_* */
714
715 /* Runtime fields. */
716 unsigned long last_balance; /* init to jiffies. units in jiffies */
717 unsigned int balance_interval; /* initialise to 1. units in ms. */
718 unsigned int nr_balance_failed; /* initialise to 0 */
719
720#ifdef CONFIG_SCHEDSTATS
721 /* load_balance() stats */
722 unsigned long lb_cnt[MAX_IDLE_TYPES];
723 unsigned long lb_failed[MAX_IDLE_TYPES];
724 unsigned long lb_balanced[MAX_IDLE_TYPES];
725 unsigned long lb_imbalance[MAX_IDLE_TYPES];
726 unsigned long lb_gained[MAX_IDLE_TYPES];
727 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
728 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
729 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
730
731 /* Active load balancing */
732 unsigned long alb_cnt;
733 unsigned long alb_failed;
734 unsigned long alb_pushed;
735
68767a0a
NP
736 /* SD_BALANCE_EXEC stats */
737 unsigned long sbe_cnt;
738 unsigned long sbe_balanced;
1da177e4
LT
739 unsigned long sbe_pushed;
740
68767a0a
NP
741 /* SD_BALANCE_FORK stats */
742 unsigned long sbf_cnt;
743 unsigned long sbf_balanced;
744 unsigned long sbf_pushed;
745
1da177e4
LT
746 /* try_to_wake_up() stats */
747 unsigned long ttwu_wake_remote;
748 unsigned long ttwu_move_affine;
749 unsigned long ttwu_move_balance;
750#endif
751};
752
51888ca2 753extern int partition_sched_domains(cpumask_t *partition1,
1a20ff27 754 cpumask_t *partition2);
198e2f18
AM
755
756/*
757 * Maximum cache size the migration-costs auto-tuning code will
758 * search from:
759 */
760extern unsigned int max_cache_size;
761
762#endif /* CONFIG_SMP */
1da177e4
LT
763
764
765struct io_context; /* See blkdev.h */
1da177e4
LT
766struct cpuset;
767
768#define NGROUPS_SMALL 32
769#define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
770struct group_info {
771 int ngroups;
772 atomic_t usage;
773 gid_t small_block[NGROUPS_SMALL];
774 int nblocks;
775 gid_t *blocks[0];
776};
777
778/*
779 * get_group_info() must be called with the owning task locked (via task_lock())
780 * when task != current. The reason being that the vast majority of callers are
781 * looking at current->group_info, which can not be changed except by the
782 * current task. Changing current->group_info requires the task lock, too.
783 */
784#define get_group_info(group_info) do { \
785 atomic_inc(&(group_info)->usage); \
786} while (0)
787
788#define put_group_info(group_info) do { \
789 if (atomic_dec_and_test(&(group_info)->usage)) \
790 groups_free(group_info); \
791} while (0)
792
3e30148c
DH
793extern struct group_info *groups_alloc(int gidsetsize);
794extern void groups_free(struct group_info *group_info);
795extern int set_current_groups(struct group_info *group_info);
796extern int groups_search(struct group_info *group_info, gid_t grp);
1da177e4
LT
797/* access the groups "array" with this macro */
798#define GROUP_AT(gi, i) \
799 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
800
383f2835 801#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
36c8b586 802extern void prefetch_stack(struct task_struct *t);
383f2835
KC
803#else
804static inline void prefetch_stack(struct task_struct *t) { }
805#endif
1da177e4
LT
806
807struct audit_context; /* See audit.c */
808struct mempolicy;
b92ce558 809struct pipe_inode_info;
4865ecf1 810struct uts_namespace;
1da177e4 811
3dee386e
CK
812enum sleep_type {
813 SLEEP_NORMAL,
814 SLEEP_NONINTERACTIVE,
815 SLEEP_INTERACTIVE,
816 SLEEP_INTERRUPTED,
817};
818
70b97a7f
IM
819struct prio_array;
820
1da177e4
LT
821struct task_struct {
822 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
f7e4217b 823 void *stack;
1da177e4 824 atomic_t usage;
97dc32cd
WC
825 unsigned int flags; /* per process flags, defined below */
826 unsigned int ptrace;
1da177e4 827
36772092 828 int lock_depth; /* BKL lock depth */
1da177e4 829
2dd73a4f
PW
830#ifdef CONFIG_SMP
831#ifdef __ARCH_WANT_UNLOCKED_CTXSW
4866cde0
NP
832 int oncpu;
833#endif
2dd73a4f
PW
834#endif
835 int load_weight; /* for niceness load balancing purposes */
b29739f9 836 int prio, static_prio, normal_prio;
1da177e4 837 struct list_head run_list;
70b97a7f 838 struct prio_array *array;
1da177e4 839
22e2c507 840 unsigned short ioprio;
6c5c9341 841#ifdef CONFIG_BLK_DEV_IO_TRACE
2056a782 842 unsigned int btrace_seq;
6c5c9341 843#endif
1da177e4
LT
844 unsigned long sleep_avg;
845 unsigned long long timestamp, last_ran;
846 unsigned long long sched_time; /* sched_clock time spent running */
3dee386e 847 enum sleep_type sleep_type;
1da177e4 848
97dc32cd 849 unsigned int policy;
1da177e4
LT
850 cpumask_t cpus_allowed;
851 unsigned int time_slice, first_time_slice;
852
52f17b6c 853#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1da177e4
LT
854 struct sched_info sched_info;
855#endif
856
857 struct list_head tasks;
858 /*
859 * ptrace_list/ptrace_children forms the list of my children
860 * that were stolen by a ptracer.
861 */
862 struct list_head ptrace_children;
863 struct list_head ptrace_list;
864
865 struct mm_struct *mm, *active_mm;
866
867/* task state */
868 struct linux_binfmt *binfmt;
97dc32cd 869 int exit_state;
1da177e4
LT
870 int exit_code, exit_signal;
871 int pdeath_signal; /* The signal sent when the parent dies */
872 /* ??? */
97dc32cd 873 unsigned int personality;
1da177e4
LT
874 unsigned did_exec:1;
875 pid_t pid;
876 pid_t tgid;
0a425405
AV
877
878#ifdef CONFIG_CC_STACKPROTECTOR
879 /* Canary value for the -fstack-protector gcc feature */
880 unsigned long stack_canary;
881#endif
1da177e4
LT
882 /*
883 * pointers to (original) parent process, youngest child, younger sibling,
884 * older sibling, respectively. (p->father can be replaced with
885 * p->parent->pid)
886 */
887 struct task_struct *real_parent; /* real parent process (when being debugged) */
888 struct task_struct *parent; /* parent process */
889 /*
890 * children/sibling forms the list of my children plus the
891 * tasks I'm ptracing.
892 */
893 struct list_head children; /* list of my children */
894 struct list_head sibling; /* linkage in my parent's children list */
895 struct task_struct *group_leader; /* threadgroup leader */
896
897 /* PID/PID hash table linkage. */
92476d7f 898 struct pid_link pids[PIDTYPE_MAX];
47e65328 899 struct list_head thread_group;
1da177e4
LT
900
901 struct completion *vfork_done; /* for vfork() */
902 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
903 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
904
97dc32cd 905 unsigned int rt_priority;
1da177e4
LT
906 cputime_t utime, stime;
907 unsigned long nvcsw, nivcsw; /* context switch counts */
908 struct timespec start_time;
909/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
910 unsigned long min_flt, maj_flt;
911
912 cputime_t it_prof_expires, it_virt_expires;
913 unsigned long long it_sched_expires;
914 struct list_head cpu_timers[3];
915
916/* process credentials */
917 uid_t uid,euid,suid,fsuid;
918 gid_t gid,egid,sgid,fsgid;
919 struct group_info *group_info;
920 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
921 unsigned keep_capabilities:1;
922 struct user_struct *user;
923#ifdef CONFIG_KEYS
b5f545c8 924 struct key *request_key_auth; /* assumed request_key authority */
1da177e4 925 struct key *thread_keyring; /* keyring private to this thread */
3e30148c 926 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1da177e4 927#endif
e07e23e1
AV
928 /*
929 * fpu_counter contains the number of consecutive context switches
930 * that the FPU is used. If this is over a threshold, the lazy fpu
931 * saving becomes unlazy to save the trap. This is an unsigned char
932 * so that after 256 times the counter wraps and the behavior turns
933 * lazy again; this to deal with bursty apps that only use FPU for
934 * a short time
935 */
936 unsigned char fpu_counter;
1da177e4 937 int oomkilladj; /* OOM kill score adjustment (bit shift). */
36772092
PBG
938 char comm[TASK_COMM_LEN]; /* executable name excluding path
939 - access with [gs]et_task_comm (which lock
940 it with task_lock())
941 - initialized normally by flush_old_exec */
1da177e4
LT
942/* file system info */
943 int link_count, total_link_count;
3d5b6fcc 944#ifdef CONFIG_SYSVIPC
1da177e4
LT
945/* ipc stuff */
946 struct sysv_sem sysvsem;
3d5b6fcc 947#endif
1da177e4
LT
948/* CPU-specific state of this task */
949 struct thread_struct thread;
950/* filesystem information */
951 struct fs_struct *fs;
952/* open file information */
953 struct files_struct *files;
1651e14e 954/* namespaces */
ab516013 955 struct nsproxy *nsproxy;
1da177e4
LT
956/* signal handlers */
957 struct signal_struct *signal;
958 struct sighand_struct *sighand;
959
960 sigset_t blocked, real_blocked;
150256d8 961 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
1da177e4
LT
962 struct sigpending pending;
963
964 unsigned long sas_ss_sp;
965 size_t sas_ss_size;
966 int (*notifier)(void *priv);
967 void *notifier_data;
968 sigset_t *notifier_mask;
969
970 void *security;
971 struct audit_context *audit_context;
972 seccomp_t seccomp;
973
974/* Thread group tracking */
975 u32 parent_exec_id;
976 u32 self_exec_id;
977/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
978 spinlock_t alloc_lock;
1da177e4 979
b29739f9
IM
980 /* Protection of the PI data structures: */
981 spinlock_t pi_lock;
982
23f78d4a
IM
983#ifdef CONFIG_RT_MUTEXES
984 /* PI waiters blocked on a rt_mutex held by this task */
985 struct plist_head pi_waiters;
986 /* Deadlock detection and priority inheritance handling */
987 struct rt_mutex_waiter *pi_blocked_on;
23f78d4a
IM
988#endif
989
408894ee
IM
990#ifdef CONFIG_DEBUG_MUTEXES
991 /* mutex deadlock detection */
992 struct mutex_waiter *blocked_on;
993#endif
de30a2b3
IM
994#ifdef CONFIG_TRACE_IRQFLAGS
995 unsigned int irq_events;
996 int hardirqs_enabled;
997 unsigned long hardirq_enable_ip;
998 unsigned int hardirq_enable_event;
999 unsigned long hardirq_disable_ip;
1000 unsigned int hardirq_disable_event;
1001 int softirqs_enabled;
1002 unsigned long softirq_disable_ip;
1003 unsigned int softirq_disable_event;
1004 unsigned long softirq_enable_ip;
1005 unsigned int softirq_enable_event;
1006 int hardirq_context;
1007 int softirq_context;
1008#endif
fbb9ce95
IM
1009#ifdef CONFIG_LOCKDEP
1010# define MAX_LOCK_DEPTH 30UL
1011 u64 curr_chain_key;
1012 int lockdep_depth;
1013 struct held_lock held_locks[MAX_LOCK_DEPTH];
1014 unsigned int lockdep_recursion;
1015#endif
408894ee 1016
1da177e4
LT
1017/* journalling filesystem info */
1018 void *journal_info;
1019
d89d8796
NB
1020/* stacked block device info */
1021 struct bio *bio_list, **bio_tail;
1022
1da177e4
LT
1023/* VM state */
1024 struct reclaim_state *reclaim_state;
1025
1da177e4
LT
1026 struct backing_dev_info *backing_dev_info;
1027
1028 struct io_context *io_context;
1029
1030 unsigned long ptrace_message;
1031 siginfo_t *last_siginfo; /* For ptrace use. */
1032/*
1033 * current io wait handle: wait queue entry to use for io waits
1034 * If this thread is processing aio, this points at the waitqueue
1035 * inside the currently handled kiocb. It may be NULL (i.e. default
1036 * to a stack based synchronous wait) if its doing sync IO.
1037 */
1038 wait_queue_t *io_wait;
4b98d11b 1039#ifdef CONFIG_TASK_XACCT
1da177e4
LT
1040/* i/o counters(bytes read/written, #syscalls */
1041 u64 rchar, wchar, syscr, syscw;
4b98d11b 1042#endif
7c3ab738 1043 struct task_io_accounting ioac;
8f0ab514 1044#if defined(CONFIG_TASK_XACCT)
1da177e4
LT
1045 u64 acct_rss_mem1; /* accumulated rss usage */
1046 u64 acct_vm_mem1; /* accumulated virtual memory usage */
db5fed26 1047 cputime_t acct_stimexpd;/* stime since last update */
1da177e4
LT
1048#endif
1049#ifdef CONFIG_NUMA
1050 struct mempolicy *mempolicy;
1051 short il_next;
1052#endif
1053#ifdef CONFIG_CPUSETS
1054 struct cpuset *cpuset;
1055 nodemask_t mems_allowed;
1056 int cpuset_mems_generation;
825a46af 1057 int cpuset_mem_spread_rotor;
1da177e4 1058#endif
0771dfef 1059 struct robust_list_head __user *robust_list;
34f192c6
IM
1060#ifdef CONFIG_COMPAT
1061 struct compat_robust_list_head __user *compat_robust_list;
1062#endif
c87e2837
IM
1063 struct list_head pi_state_list;
1064 struct futex_pi_state *pi_state_cache;
0771dfef 1065
22e2c507 1066 atomic_t fs_excl; /* holding fs exclusive resources */
e56d0903 1067 struct rcu_head rcu;
b92ce558
JA
1068
1069 /*
1070 * cache last used pipe for splice
1071 */
1072 struct pipe_inode_info *splice_pipe;
ca74e92b
SN
1073#ifdef CONFIG_TASK_DELAY_ACCT
1074 struct task_delay_info *delays;
f4f154fd
AM
1075#endif
1076#ifdef CONFIG_FAULT_INJECTION
1077 int make_it_fail;
ca74e92b 1078#endif
1da177e4
LT
1079};
1080
1081static inline pid_t process_group(struct task_struct *tsk)
1082{
1083 return tsk->signal->pgrp;
1084}
1085
1ec320af
CLG
1086static inline pid_t signal_session(struct signal_struct *sig)
1087{
1088 return sig->__session;
1089}
1090
937949d9
CLG
1091static inline pid_t process_session(struct task_struct *tsk)
1092{
1ec320af
CLG
1093 return signal_session(tsk->signal);
1094}
1095
1096static inline void set_signal_session(struct signal_struct *sig, pid_t session)
1097{
1098 sig->__session = session;
937949d9
CLG
1099}
1100
22c935f4
EB
1101static inline struct pid *task_pid(struct task_struct *task)
1102{
1103 return task->pids[PIDTYPE_PID].pid;
1104}
1105
1106static inline struct pid *task_tgid(struct task_struct *task)
1107{
1108 return task->group_leader->pids[PIDTYPE_PID].pid;
1109}
1110
1111static inline struct pid *task_pgrp(struct task_struct *task)
1112{
1113 return task->group_leader->pids[PIDTYPE_PGID].pid;
1114}
1115
1116static inline struct pid *task_session(struct task_struct *task)
1117{
1118 return task->group_leader->pids[PIDTYPE_SID].pid;
1119}
1120
1da177e4
LT
1121/**
1122 * pid_alive - check that a task structure is not stale
1123 * @p: Task structure to be checked.
1124 *
1125 * Test if a process is not yet dead (at most zombie state)
1126 * If pid_alive fails, then pointers within the task structure
1127 * can be stale and must not be dereferenced.
1128 */
1129static inline int pid_alive(struct task_struct *p)
1130{
92476d7f 1131 return p->pids[PIDTYPE_PID].pid != NULL;
1da177e4
LT
1132}
1133
f400e198 1134/**
3260259f
HK
1135 * is_init - check if a task structure is init
1136 * @tsk: Task structure to be checked.
1137 *
1138 * Check if a task structure is the first user space task the kernel created.
f400e198
SB
1139 */
1140static inline int is_init(struct task_struct *tsk)
1141{
1142 return tsk->pid == 1;
1143}
1144
9ec52099
CLG
1145extern struct pid *cad_pid;
1146
1da177e4 1147extern void free_task(struct task_struct *tsk);
1da177e4 1148#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
e56d0903 1149
158d9ebd 1150extern void __put_task_struct(struct task_struct *t);
e56d0903
IM
1151
1152static inline void put_task_struct(struct task_struct *t)
1153{
1154 if (atomic_dec_and_test(&t->usage))
8c7904a0 1155 __put_task_struct(t);
e56d0903 1156}
1da177e4
LT
1157
1158/*
1159 * Per process flags
1160 */
1161#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1162 /* Not implemented yet, only for 486*/
1163#define PF_STARTING 0x00000002 /* being created */
1164#define PF_EXITING 0x00000004 /* getting shut down */
1da177e4
LT
1165#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1166#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1167#define PF_DUMPCORE 0x00000200 /* dumped core */
1168#define PF_SIGNALED 0x00000400 /* killed by a signal */
1169#define PF_MEMALLOC 0x00000800 /* Allocating memory */
1170#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1171#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1da177e4
LT
1172#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1173#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1174#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1175#define PF_KSWAPD 0x00040000 /* I am kswapd */
1176#define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1177#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
b31dc66a
JA
1178#define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1179#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1180#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1181#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1182#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
c61afb18 1183#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
61a87122 1184#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
ba96a0c8 1185#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1da177e4
LT
1186
1187/*
1188 * Only the _current_ task can read/write to tsk->flags, but other
1189 * tasks can access tsk->flags in readonly mode for example
1190 * with tsk_used_math (like during threaded core dumping).
1191 * There is however an exception to this rule during ptrace
1192 * or during fork: the ptracer task is allowed to write to the
1193 * child->flags of its traced child (same goes for fork, the parent
1194 * can write to the child->flags), because we're guaranteed the
1195 * child is not running and in turn not changing child->flags
1196 * at the same time the parent does it.
1197 */
1198#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1199#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1200#define clear_used_math() clear_stopped_child_used_math(current)
1201#define set_used_math() set_stopped_child_used_math(current)
1202#define conditional_stopped_child_used_math(condition, child) \
1203 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1204#define conditional_used_math(condition) \
1205 conditional_stopped_child_used_math(condition, current)
1206#define copy_to_stopped_child_used_math(child) \
1207 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1208/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1209#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1210#define used_math() tsk_used_math(current)
1211
1212#ifdef CONFIG_SMP
36c8b586 1213extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1da177e4 1214#else
36c8b586 1215static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1da177e4 1216{
4098f991 1217 if (!cpu_isset(0, new_mask))
1da177e4
LT
1218 return -EINVAL;
1219 return 0;
1220}
1221#endif
1222
1223extern unsigned long long sched_clock(void);
36c8b586
IM
1224extern unsigned long long
1225current_sched_time(const struct task_struct *current_task);
1da177e4
LT
1226
1227/* sched_exec is called by processes performing an exec */
1228#ifdef CONFIG_SMP
1229extern void sched_exec(void);
1230#else
1231#define sched_exec() {}
1232#endif
1233
1234#ifdef CONFIG_HOTPLUG_CPU
1235extern void idle_task_exit(void);
1236#else
1237static inline void idle_task_exit(void) {}
1238#endif
1239
1240extern void sched_idle_next(void);
b29739f9
IM
1241
1242#ifdef CONFIG_RT_MUTEXES
36c8b586
IM
1243extern int rt_mutex_getprio(struct task_struct *p);
1244extern void rt_mutex_setprio(struct task_struct *p, int prio);
1245extern void rt_mutex_adjust_pi(struct task_struct *p);
b29739f9 1246#else
36c8b586 1247static inline int rt_mutex_getprio(struct task_struct *p)
b29739f9
IM
1248{
1249 return p->normal_prio;
1250}
95e02ca9 1251# define rt_mutex_adjust_pi(p) do { } while (0)
b29739f9
IM
1252#endif
1253
36c8b586
IM
1254extern void set_user_nice(struct task_struct *p, long nice);
1255extern int task_prio(const struct task_struct *p);
1256extern int task_nice(const struct task_struct *p);
1257extern int can_nice(const struct task_struct *p, const int nice);
1258extern int task_curr(const struct task_struct *p);
1da177e4
LT
1259extern int idle_cpu(int cpu);
1260extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
36c8b586
IM
1261extern struct task_struct *idle_task(int cpu);
1262extern struct task_struct *curr_task(int cpu);
1263extern void set_curr_task(int cpu, struct task_struct *p);
1da177e4
LT
1264
1265void yield(void);
1266
1267/*
1268 * The default (Linux) execution domain.
1269 */
1270extern struct exec_domain default_exec_domain;
1271
1272union thread_union {
1273 struct thread_info thread_info;
1274 unsigned long stack[THREAD_SIZE/sizeof(long)];
1275};
1276
1277#ifndef __HAVE_ARCH_KSTACK_END
1278static inline int kstack_end(void *addr)
1279{
1280 /* Reliable end of stack detection:
1281 * Some APM bios versions misalign the stack
1282 */
1283 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1284}
1285#endif
1286
1287extern union thread_union init_thread_union;
1288extern struct task_struct init_task;
1289
1290extern struct mm_struct init_mm;
1291
1292#define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1293extern struct task_struct *find_task_by_pid_type(int type, int pid);
1da177e4
LT
1294extern void __set_special_pids(pid_t session, pid_t pgrp);
1295
1296/* per-UID process charging. */
1297extern struct user_struct * alloc_uid(uid_t);
1298static inline struct user_struct *get_uid(struct user_struct *u)
1299{
1300 atomic_inc(&u->__count);
1301 return u;
1302}
1303extern void free_uid(struct user_struct *);
1304extern void switch_uid(struct user_struct *);
1305
1306#include <asm/current.h>
1307
3171a030 1308extern void do_timer(unsigned long ticks);
1da177e4
LT
1309
1310extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1311extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1312extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1313 unsigned long clone_flags));
1314#ifdef CONFIG_SMP
1315 extern void kick_process(struct task_struct *tsk);
1316#else
1317 static inline void kick_process(struct task_struct *tsk) { }
1318#endif
36c8b586
IM
1319extern void FASTCALL(sched_fork(struct task_struct * p, int clone_flags));
1320extern void FASTCALL(sched_exit(struct task_struct * p));
1da177e4
LT
1321
1322extern int in_group_p(gid_t);
1323extern int in_egroup_p(gid_t);
1324
1325extern void proc_caches_init(void);
1326extern void flush_signals(struct task_struct *);
10ab825b 1327extern void ignore_signals(struct task_struct *);
1da177e4
LT
1328extern void flush_signal_handlers(struct task_struct *, int force_default);
1329extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1330
1331static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1332{
1333 unsigned long flags;
1334 int ret;
1335
1336 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1337 ret = dequeue_signal(tsk, mask, info);
1338 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1339
1340 return ret;
1341}
1342
1343extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1344 sigset_t *mask);
1345extern void unblock_all_signals(void);
1346extern void release_task(struct task_struct * p);
1347extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1348extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1349extern int force_sigsegv(int, struct task_struct *);
1350extern int force_sig_info(int, struct siginfo *, struct task_struct *);
c4b92fc1
EB
1351extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1352extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1353extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2425c08b 1354extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
c4b92fc1
EB
1355extern int kill_pgrp(struct pid *pid, int sig, int priv);
1356extern int kill_pid(struct pid *pid, int sig, int priv);
c3de4b38 1357extern int kill_proc_info(int, struct siginfo *, pid_t);
1da177e4
LT
1358extern void do_notify_parent(struct task_struct *, int);
1359extern void force_sig(int, struct task_struct *);
1360extern void force_sig_specific(int, struct task_struct *);
1361extern int send_sig(int, struct task_struct *, int);
1362extern void zap_other_threads(struct task_struct *p);
1da177e4
LT
1363extern int kill_proc(pid_t, int, int);
1364extern struct sigqueue *sigqueue_alloc(void);
1365extern void sigqueue_free(struct sigqueue *);
1366extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1367extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
9ac95f2f 1368extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1da177e4
LT
1369extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1370
9ec52099
CLG
1371static inline int kill_cad_pid(int sig, int priv)
1372{
1373 return kill_pid(cad_pid, sig, priv);
1374}
1375
1da177e4
LT
1376/* These can be the second arg to send_sig_info/send_group_sig_info. */
1377#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1378#define SEND_SIG_PRIV ((struct siginfo *) 1)
1379#define SEND_SIG_FORCED ((struct siginfo *) 2)
1380
621d3121
ON
1381static inline int is_si_special(const struct siginfo *info)
1382{
1383 return info <= SEND_SIG_FORCED;
1384}
1385
1da177e4
LT
1386/* True if we are on the alternate signal stack. */
1387
1388static inline int on_sig_stack(unsigned long sp)
1389{
1390 return (sp - current->sas_ss_sp < current->sas_ss_size);
1391}
1392
1393static inline int sas_ss_flags(unsigned long sp)
1394{
1395 return (current->sas_ss_size == 0 ? SS_DISABLE
1396 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1397}
1398
1da177e4
LT
1399/*
1400 * Routines for handling mm_structs
1401 */
1402extern struct mm_struct * mm_alloc(void);
1403
1404/* mmdrop drops the mm and the page tables */
1405extern void FASTCALL(__mmdrop(struct mm_struct *));
1406static inline void mmdrop(struct mm_struct * mm)
1407{
1408 if (atomic_dec_and_test(&mm->mm_count))
1409 __mmdrop(mm);
1410}
1411
1412/* mmput gets rid of the mappings and all user-space */
1413extern void mmput(struct mm_struct *);
1414/* Grab a reference to a task's mm, if it is not already going away */
1415extern struct mm_struct *get_task_mm(struct task_struct *task);
1416/* Remove the current tasks stale references to the old mm_struct */
1417extern void mm_release(struct task_struct *, struct mm_struct *);
1418
1419extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1420extern void flush_thread(void);
1421extern void exit_thread(void);
1422
1da177e4 1423extern void exit_files(struct task_struct *);
6b3934ef 1424extern void __cleanup_signal(struct signal_struct *);
a7e5328a 1425extern void __cleanup_sighand(struct sighand_struct *);
1da177e4
LT
1426extern void exit_itimers(struct signal_struct *);
1427
1428extern NORET_TYPE void do_group_exit(int);
1429
1da177e4
LT
1430extern void daemonize(const char *, ...);
1431extern int allow_signal(int);
1432extern int disallow_signal(int);
1da177e4
LT
1433
1434extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1435extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
36c8b586 1436struct task_struct *fork_idle(int);
1da177e4
LT
1437
1438extern void set_task_comm(struct task_struct *tsk, char *from);
1439extern void get_task_comm(char *to, struct task_struct *tsk);
1440
1441#ifdef CONFIG_SMP
36c8b586 1442extern void wait_task_inactive(struct task_struct * p);
1da177e4
LT
1443#else
1444#define wait_task_inactive(p) do { } while (0)
1445#endif
1446
1447#define remove_parent(p) list_del_init(&(p)->sibling)
8fafabd8 1448#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1da177e4 1449
5e85d4ab 1450#define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1da177e4
LT
1451
1452#define for_each_process(p) \
1453 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1454
1455/*
1456 * Careful: do_each_thread/while_each_thread is a double loop so
1457 * 'break' will not work as expected - use goto instead.
1458 */
1459#define do_each_thread(g, t) \
1460 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1461
1462#define while_each_thread(g, t) \
1463 while ((t = next_thread(t)) != g)
1464
de12a787
EB
1465/* de_thread depends on thread_group_leader not being a pid based check */
1466#define thread_group_leader(p) (p == p->group_leader)
1da177e4 1467
0804ef4b
EB
1468/* Do to the insanities of de_thread it is possible for a process
1469 * to have the pid of the thread group leader without actually being
1470 * the thread group leader. For iteration through the pids in proc
1471 * all we care about is that we have a task with the appropriate
1472 * pid, we don't actually care if we have the right task.
1473 */
1474static inline int has_group_leader_pid(struct task_struct *p)
1475{
1476 return p->pid == p->tgid;
1477}
1478
36c8b586 1479static inline struct task_struct *next_thread(const struct task_struct *p)
47e65328
ON
1480{
1481 return list_entry(rcu_dereference(p->thread_group.next),
36c8b586 1482 struct task_struct, thread_group);
47e65328
ON
1483}
1484
36c8b586 1485static inline int thread_group_empty(struct task_struct *p)
1da177e4 1486{
47e65328 1487 return list_empty(&p->thread_group);
1da177e4
LT
1488}
1489
1490#define delay_group_leader(p) \
1491 (thread_group_leader(p) && !thread_group_empty(p))
1492
1da177e4 1493/*
260ea101 1494 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
22e2c507 1495 * subscriptions and synchronises with wait4(). Also used in procfs. Also
053199ed 1496 * pins the final release of task.io_context. Also protects ->cpuset.
1da177e4
LT
1497 *
1498 * Nests both inside and outside of read_lock(&tasklist_lock).
1499 * It must not be nested with write_lock_irq(&tasklist_lock),
1500 * neither inside nor outside.
1501 */
1502static inline void task_lock(struct task_struct *p)
1503{
1504 spin_lock(&p->alloc_lock);
1505}
1506
1507static inline void task_unlock(struct task_struct *p)
1508{
1509 spin_unlock(&p->alloc_lock);
1510}
1511
f63ee72e
ON
1512extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1513 unsigned long *flags);
1514
1515static inline void unlock_task_sighand(struct task_struct *tsk,
1516 unsigned long *flags)
1517{
1518 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1519}
1520
f037360f
AV
1521#ifndef __HAVE_THREAD_FUNCTIONS
1522
f7e4217b
RZ
1523#define task_thread_info(task) ((struct thread_info *)(task)->stack)
1524#define task_stack_page(task) ((task)->stack)
a1261f54 1525
10ebffde
AV
1526static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1527{
1528 *task_thread_info(p) = *task_thread_info(org);
1529 task_thread_info(p)->task = p;
1530}
1531
1532static inline unsigned long *end_of_stack(struct task_struct *p)
1533{
f7e4217b 1534 return (unsigned long *)(task_thread_info(p) + 1);
10ebffde
AV
1535}
1536
f037360f
AV
1537#endif
1538
1da177e4
LT
1539/* set thread flags in other task's structures
1540 * - see asm/thread_info.h for TIF_xxxx flags available
1541 */
1542static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1543{
a1261f54 1544 set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1545}
1546
1547static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1548{
a1261f54 1549 clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1550}
1551
1552static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1553{
a1261f54 1554 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1555}
1556
1557static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1558{
a1261f54 1559 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1560}
1561
1562static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1563{
a1261f54 1564 return test_ti_thread_flag(task_thread_info(tsk), flag);
1da177e4
LT
1565}
1566
1567static inline void set_tsk_need_resched(struct task_struct *tsk)
1568{
1569 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1570}
1571
1572static inline void clear_tsk_need_resched(struct task_struct *tsk)
1573{
1574 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1575}
1576
1577static inline int signal_pending(struct task_struct *p)
1578{
1579 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1580}
1581
1582static inline int need_resched(void)
1583{
1584 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1585}
1586
1587/*
1588 * cond_resched() and cond_resched_lock(): latency reduction via
1589 * explicit rescheduling in places that are safe. The return
1590 * value indicates whether a reschedule was done in fact.
1591 * cond_resched_lock() will drop the spinlock before scheduling,
1592 * cond_resched_softirq() will enable bhs before scheduling.
1593 */
1594extern int cond_resched(void);
1595extern int cond_resched_lock(spinlock_t * lock);
1596extern int cond_resched_softirq(void);
1597
1598/*
1599 * Does a critical section need to be broken due to another
1600 * task waiting?:
1601 */
1602#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1603# define need_lockbreak(lock) ((lock)->break_lock)
1604#else
1605# define need_lockbreak(lock) 0
1606#endif
1607
1608/*
1609 * Does a critical section need to be broken due to another
1610 * task waiting or preemption being signalled:
1611 */
1612static inline int lock_need_resched(spinlock_t *lock)
1613{
1614 if (need_lockbreak(lock) || need_resched())
1615 return 1;
1616 return 0;
1617}
1618
7bb44ade
RM
1619/*
1620 * Reevaluate whether the task has signals pending delivery.
1621 * Wake the task if so.
1622 * This is required every time the blocked sigset_t changes.
1623 * callers must hold sighand->siglock.
1624 */
1625extern void recalc_sigpending_and_wake(struct task_struct *t);
1da177e4
LT
1626extern void recalc_sigpending(void);
1627
1628extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1629
1630/*
1631 * Wrappers for p->thread_info->cpu access. No-op on UP.
1632 */
1633#ifdef CONFIG_SMP
1634
1635static inline unsigned int task_cpu(const struct task_struct *p)
1636{
a1261f54 1637 return task_thread_info(p)->cpu;
1da177e4
LT
1638}
1639
1640static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1641{
a1261f54 1642 task_thread_info(p)->cpu = cpu;
1da177e4
LT
1643}
1644
1645#else
1646
1647static inline unsigned int task_cpu(const struct task_struct *p)
1648{
1649 return 0;
1650}
1651
1652static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1653{
1654}
1655
1656#endif /* CONFIG_SMP */
1657
1658#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1659extern void arch_pick_mmap_layout(struct mm_struct *mm);
1660#else
1661static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1662{
1663 mm->mmap_base = TASK_UNMAPPED_BASE;
1664 mm->get_unmapped_area = arch_get_unmapped_area;
1665 mm->unmap_area = arch_unmap_area;
1666}
1667#endif
1668
1669extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1670extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1671
5c45bf27 1672extern int sched_mc_power_savings, sched_smt_power_savings;
5c45bf27 1673
1da177e4
LT
1674extern void normalize_rt_tasks(void);
1675
4b98d11b
AD
1676#ifdef CONFIG_TASK_XACCT
1677static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1678{
1679 tsk->rchar += amt;
1680}
1681
1682static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1683{
1684 tsk->wchar += amt;
1685}
1686
1687static inline void inc_syscr(struct task_struct *tsk)
1688{
1689 tsk->syscr++;
1690}
1691
1692static inline void inc_syscw(struct task_struct *tsk)
1693{
1694 tsk->syscw++;
1695}
1696#else
1697static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1698{
1699}
1700
1701static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1702{
1703}
1704
1705static inline void inc_syscr(struct task_struct *tsk)
1706{
1707}
1708
1709static inline void inc_syscw(struct task_struct *tsk)
1710{
1711}
1712#endif
1713
1da177e4
LT
1714#endif /* __KERNEL__ */
1715
1716#endif