Memory controller: accounting setup
[deliverable/linux.git] / include / linux / sched.h
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 /*
5 * cloning flags:
6 */
7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 #define CLONE_NEWNET 0x40000000 /* New network namespace */
30 #define CLONE_IO 0x80000000 /* Clone io context */
31
32 /*
33 * Scheduling policies
34 */
35 #define SCHED_NORMAL 0
36 #define SCHED_FIFO 1
37 #define SCHED_RR 2
38 #define SCHED_BATCH 3
39 /* SCHED_ISO: reserved but not implemented yet */
40 #define SCHED_IDLE 5
41
42 #ifdef __KERNEL__
43
44 struct sched_param {
45 int sched_priority;
46 };
47
48 #include <asm/param.h> /* for HZ */
49
50 #include <linux/capability.h>
51 #include <linux/threads.h>
52 #include <linux/kernel.h>
53 #include <linux/types.h>
54 #include <linux/timex.h>
55 #include <linux/jiffies.h>
56 #include <linux/rbtree.h>
57 #include <linux/thread_info.h>
58 #include <linux/cpumask.h>
59 #include <linux/errno.h>
60 #include <linux/nodemask.h>
61 #include <linux/mm_types.h>
62
63 #include <asm/system.h>
64 #include <asm/semaphore.h>
65 #include <asm/page.h>
66 #include <asm/ptrace.h>
67 #include <asm/cputime.h>
68
69 #include <linux/smp.h>
70 #include <linux/sem.h>
71 #include <linux/signal.h>
72 #include <linux/securebits.h>
73 #include <linux/fs_struct.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rtmutex.h>
83
84 #include <linux/time.h>
85 #include <linux/param.h>
86 #include <linux/resource.h>
87 #include <linux/timer.h>
88 #include <linux/hrtimer.h>
89 #include <linux/task_io_accounting.h>
90 #include <linux/kobject.h>
91 #include <linux/latencytop.h>
92
93 #include <asm/processor.h>
94
95 struct mem_cgroup;
96 struct exec_domain;
97 struct futex_pi_state;
98 struct robust_list_head;
99 struct bio;
100
101 /*
102 * List of flags we want to share for kernel threads,
103 * if only because they are not used by them anyway.
104 */
105 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
106
107 /*
108 * These are the constant used to fake the fixed-point load-average
109 * counting. Some notes:
110 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
111 * a load-average precision of 10 bits integer + 11 bits fractional
112 * - if you want to count load-averages more often, you need more
113 * precision, or rounding will get you. With 2-second counting freq,
114 * the EXP_n values would be 1981, 2034 and 2043 if still using only
115 * 11 bit fractions.
116 */
117 extern unsigned long avenrun[]; /* Load averages */
118
119 #define FSHIFT 11 /* nr of bits of precision */
120 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
121 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
122 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
123 #define EXP_5 2014 /* 1/exp(5sec/5min) */
124 #define EXP_15 2037 /* 1/exp(5sec/15min) */
125
126 #define CALC_LOAD(load,exp,n) \
127 load *= exp; \
128 load += n*(FIXED_1-exp); \
129 load >>= FSHIFT;
130
131 extern unsigned long total_forks;
132 extern int nr_threads;
133 DECLARE_PER_CPU(unsigned long, process_counts);
134 extern int nr_processes(void);
135 extern unsigned long nr_running(void);
136 extern unsigned long nr_uninterruptible(void);
137 extern unsigned long nr_active(void);
138 extern unsigned long nr_iowait(void);
139 extern unsigned long weighted_cpuload(const int cpu);
140
141 struct seq_file;
142 struct cfs_rq;
143 struct task_group;
144 #ifdef CONFIG_SCHED_DEBUG
145 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
146 extern void proc_sched_set_task(struct task_struct *p);
147 extern void
148 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
149 #else
150 static inline void
151 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
152 {
153 }
154 static inline void proc_sched_set_task(struct task_struct *p)
155 {
156 }
157 static inline void
158 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
159 {
160 }
161 #endif
162
163 /*
164 * Task state bitmask. NOTE! These bits are also
165 * encoded in fs/proc/array.c: get_task_state().
166 *
167 * We have two separate sets of flags: task->state
168 * is about runnability, while task->exit_state are
169 * about the task exiting. Confusing, but this way
170 * modifying one set can't modify the other one by
171 * mistake.
172 */
173 #define TASK_RUNNING 0
174 #define TASK_INTERRUPTIBLE 1
175 #define TASK_UNINTERRUPTIBLE 2
176 #define __TASK_STOPPED 4
177 #define __TASK_TRACED 8
178 /* in tsk->exit_state */
179 #define EXIT_ZOMBIE 16
180 #define EXIT_DEAD 32
181 /* in tsk->state again */
182 #define TASK_DEAD 64
183 #define TASK_WAKEKILL 128
184
185 /* Convenience macros for the sake of set_task_state */
186 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
187 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
188 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
189
190 /* Convenience macros for the sake of wake_up */
191 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
192 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
193
194 /* get_task_state() */
195 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
196 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
197 __TASK_TRACED)
198
199 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
200 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
201 #define task_is_stopped_or_traced(task) \
202 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
203 #define task_contributes_to_load(task) \
204 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
205
206 #define __set_task_state(tsk, state_value) \
207 do { (tsk)->state = (state_value); } while (0)
208 #define set_task_state(tsk, state_value) \
209 set_mb((tsk)->state, (state_value))
210
211 /*
212 * set_current_state() includes a barrier so that the write of current->state
213 * is correctly serialised wrt the caller's subsequent test of whether to
214 * actually sleep:
215 *
216 * set_current_state(TASK_UNINTERRUPTIBLE);
217 * if (do_i_need_to_sleep())
218 * schedule();
219 *
220 * If the caller does not need such serialisation then use __set_current_state()
221 */
222 #define __set_current_state(state_value) \
223 do { current->state = (state_value); } while (0)
224 #define set_current_state(state_value) \
225 set_mb(current->state, (state_value))
226
227 /* Task command name length */
228 #define TASK_COMM_LEN 16
229
230 #include <linux/spinlock.h>
231
232 /*
233 * This serializes "schedule()" and also protects
234 * the run-queue from deletions/modifications (but
235 * _adding_ to the beginning of the run-queue has
236 * a separate lock).
237 */
238 extern rwlock_t tasklist_lock;
239 extern spinlock_t mmlist_lock;
240
241 struct task_struct;
242
243 extern void sched_init(void);
244 extern void sched_init_smp(void);
245 extern void init_idle(struct task_struct *idle, int cpu);
246 extern void init_idle_bootup_task(struct task_struct *idle);
247
248 extern cpumask_t nohz_cpu_mask;
249 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
250 extern int select_nohz_load_balancer(int cpu);
251 #else
252 static inline int select_nohz_load_balancer(int cpu)
253 {
254 return 0;
255 }
256 #endif
257
258 extern unsigned long rt_needs_cpu(int cpu);
259
260 /*
261 * Only dump TASK_* tasks. (0 for all tasks)
262 */
263 extern void show_state_filter(unsigned long state_filter);
264
265 static inline void show_state(void)
266 {
267 show_state_filter(0);
268 }
269
270 extern void show_regs(struct pt_regs *);
271
272 /*
273 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
274 * task), SP is the stack pointer of the first frame that should be shown in the back
275 * trace (or NULL if the entire call-chain of the task should be shown).
276 */
277 extern void show_stack(struct task_struct *task, unsigned long *sp);
278
279 void io_schedule(void);
280 long io_schedule_timeout(long timeout);
281
282 extern void cpu_init (void);
283 extern void trap_init(void);
284 extern void account_process_tick(struct task_struct *task, int user);
285 extern void update_process_times(int user);
286 extern void scheduler_tick(void);
287 extern void hrtick_resched(void);
288
289 extern void sched_show_task(struct task_struct *p);
290
291 #ifdef CONFIG_DETECT_SOFTLOCKUP
292 extern void softlockup_tick(void);
293 extern void spawn_softlockup_task(void);
294 extern void touch_softlockup_watchdog(void);
295 extern void touch_all_softlockup_watchdogs(void);
296 extern unsigned long softlockup_thresh;
297 extern unsigned long sysctl_hung_task_check_count;
298 extern unsigned long sysctl_hung_task_timeout_secs;
299 extern unsigned long sysctl_hung_task_warnings;
300 #else
301 static inline void softlockup_tick(void)
302 {
303 }
304 static inline void spawn_softlockup_task(void)
305 {
306 }
307 static inline void touch_softlockup_watchdog(void)
308 {
309 }
310 static inline void touch_all_softlockup_watchdogs(void)
311 {
312 }
313 #endif
314
315
316 /* Attach to any functions which should be ignored in wchan output. */
317 #define __sched __attribute__((__section__(".sched.text")))
318
319 /* Linker adds these: start and end of __sched functions */
320 extern char __sched_text_start[], __sched_text_end[];
321
322 /* Is this address in the __sched functions? */
323 extern int in_sched_functions(unsigned long addr);
324
325 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
326 extern signed long FASTCALL(schedule_timeout(signed long timeout));
327 extern signed long schedule_timeout_interruptible(signed long timeout);
328 extern signed long schedule_timeout_killable(signed long timeout);
329 extern signed long schedule_timeout_uninterruptible(signed long timeout);
330 asmlinkage void schedule(void);
331
332 struct nsproxy;
333 struct user_namespace;
334
335 /* Maximum number of active map areas.. This is a random (large) number */
336 #define DEFAULT_MAX_MAP_COUNT 65536
337
338 extern int sysctl_max_map_count;
339
340 #include <linux/aio.h>
341
342 extern unsigned long
343 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
344 unsigned long, unsigned long);
345 extern unsigned long
346 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
347 unsigned long len, unsigned long pgoff,
348 unsigned long flags);
349 extern void arch_unmap_area(struct mm_struct *, unsigned long);
350 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
351
352 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
353 /*
354 * The mm counters are not protected by its page_table_lock,
355 * so must be incremented atomically.
356 */
357 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
358 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
359 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
360 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
361 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
362
363 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
364 /*
365 * The mm counters are protected by its page_table_lock,
366 * so can be incremented directly.
367 */
368 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
369 #define get_mm_counter(mm, member) ((mm)->_##member)
370 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
371 #define inc_mm_counter(mm, member) (mm)->_##member++
372 #define dec_mm_counter(mm, member) (mm)->_##member--
373
374 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
375
376 #define get_mm_rss(mm) \
377 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
378 #define update_hiwater_rss(mm) do { \
379 unsigned long _rss = get_mm_rss(mm); \
380 if ((mm)->hiwater_rss < _rss) \
381 (mm)->hiwater_rss = _rss; \
382 } while (0)
383 #define update_hiwater_vm(mm) do { \
384 if ((mm)->hiwater_vm < (mm)->total_vm) \
385 (mm)->hiwater_vm = (mm)->total_vm; \
386 } while (0)
387
388 extern void set_dumpable(struct mm_struct *mm, int value);
389 extern int get_dumpable(struct mm_struct *mm);
390
391 /* mm flags */
392 /* dumpable bits */
393 #define MMF_DUMPABLE 0 /* core dump is permitted */
394 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
395 #define MMF_DUMPABLE_BITS 2
396
397 /* coredump filter bits */
398 #define MMF_DUMP_ANON_PRIVATE 2
399 #define MMF_DUMP_ANON_SHARED 3
400 #define MMF_DUMP_MAPPED_PRIVATE 4
401 #define MMF_DUMP_MAPPED_SHARED 5
402 #define MMF_DUMP_ELF_HEADERS 6
403 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
404 #define MMF_DUMP_FILTER_BITS 5
405 #define MMF_DUMP_FILTER_MASK \
406 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
407 #define MMF_DUMP_FILTER_DEFAULT \
408 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
409
410 struct sighand_struct {
411 atomic_t count;
412 struct k_sigaction action[_NSIG];
413 spinlock_t siglock;
414 wait_queue_head_t signalfd_wqh;
415 };
416
417 struct pacct_struct {
418 int ac_flag;
419 long ac_exitcode;
420 unsigned long ac_mem;
421 cputime_t ac_utime, ac_stime;
422 unsigned long ac_minflt, ac_majflt;
423 };
424
425 /*
426 * NOTE! "signal_struct" does not have it's own
427 * locking, because a shared signal_struct always
428 * implies a shared sighand_struct, so locking
429 * sighand_struct is always a proper superset of
430 * the locking of signal_struct.
431 */
432 struct signal_struct {
433 atomic_t count;
434 atomic_t live;
435
436 wait_queue_head_t wait_chldexit; /* for wait4() */
437
438 /* current thread group signal load-balancing target: */
439 struct task_struct *curr_target;
440
441 /* shared signal handling: */
442 struct sigpending shared_pending;
443
444 /* thread group exit support */
445 int group_exit_code;
446 /* overloaded:
447 * - notify group_exit_task when ->count is equal to notify_count
448 * - everyone except group_exit_task is stopped during signal delivery
449 * of fatal signals, group_exit_task processes the signal.
450 */
451 struct task_struct *group_exit_task;
452 int notify_count;
453
454 /* thread group stop support, overloads group_exit_code too */
455 int group_stop_count;
456 unsigned int flags; /* see SIGNAL_* flags below */
457
458 /* POSIX.1b Interval Timers */
459 struct list_head posix_timers;
460
461 /* ITIMER_REAL timer for the process */
462 struct hrtimer real_timer;
463 struct task_struct *tsk;
464 ktime_t it_real_incr;
465
466 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
467 cputime_t it_prof_expires, it_virt_expires;
468 cputime_t it_prof_incr, it_virt_incr;
469
470 /* job control IDs */
471
472 /*
473 * pgrp and session fields are deprecated.
474 * use the task_session_Xnr and task_pgrp_Xnr routines below
475 */
476
477 union {
478 pid_t pgrp __deprecated;
479 pid_t __pgrp;
480 };
481
482 struct pid *tty_old_pgrp;
483
484 union {
485 pid_t session __deprecated;
486 pid_t __session;
487 };
488
489 /* boolean value for session group leader */
490 int leader;
491
492 struct tty_struct *tty; /* NULL if no tty */
493
494 /*
495 * Cumulative resource counters for dead threads in the group,
496 * and for reaped dead child processes forked by this group.
497 * Live threads maintain their own counters and add to these
498 * in __exit_signal, except for the group leader.
499 */
500 cputime_t utime, stime, cutime, cstime;
501 cputime_t gtime;
502 cputime_t cgtime;
503 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
504 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
505 unsigned long inblock, oublock, cinblock, coublock;
506
507 /*
508 * Cumulative ns of scheduled CPU time for dead threads in the
509 * group, not including a zombie group leader. (This only differs
510 * from jiffies_to_ns(utime + stime) if sched_clock uses something
511 * other than jiffies.)
512 */
513 unsigned long long sum_sched_runtime;
514
515 /*
516 * We don't bother to synchronize most readers of this at all,
517 * because there is no reader checking a limit that actually needs
518 * to get both rlim_cur and rlim_max atomically, and either one
519 * alone is a single word that can safely be read normally.
520 * getrlimit/setrlimit use task_lock(current->group_leader) to
521 * protect this instead of the siglock, because they really
522 * have no need to disable irqs.
523 */
524 struct rlimit rlim[RLIM_NLIMITS];
525
526 struct list_head cpu_timers[3];
527
528 /* keep the process-shared keyrings here so that they do the right
529 * thing in threads created with CLONE_THREAD */
530 #ifdef CONFIG_KEYS
531 struct key *session_keyring; /* keyring inherited over fork */
532 struct key *process_keyring; /* keyring private to this process */
533 #endif
534 #ifdef CONFIG_BSD_PROCESS_ACCT
535 struct pacct_struct pacct; /* per-process accounting information */
536 #endif
537 #ifdef CONFIG_TASKSTATS
538 struct taskstats *stats;
539 #endif
540 #ifdef CONFIG_AUDIT
541 unsigned audit_tty;
542 struct tty_audit_buf *tty_audit_buf;
543 #endif
544 };
545
546 /* Context switch must be unlocked if interrupts are to be enabled */
547 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
548 # define __ARCH_WANT_UNLOCKED_CTXSW
549 #endif
550
551 /*
552 * Bits in flags field of signal_struct.
553 */
554 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
555 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
556 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
557 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
558
559 /* If true, all threads except ->group_exit_task have pending SIGKILL */
560 static inline int signal_group_exit(const struct signal_struct *sig)
561 {
562 return (sig->flags & SIGNAL_GROUP_EXIT) ||
563 (sig->group_exit_task != NULL);
564 }
565
566 /*
567 * Some day this will be a full-fledged user tracking system..
568 */
569 struct user_struct {
570 atomic_t __count; /* reference count */
571 atomic_t processes; /* How many processes does this user have? */
572 atomic_t files; /* How many open files does this user have? */
573 atomic_t sigpending; /* How many pending signals does this user have? */
574 #ifdef CONFIG_INOTIFY_USER
575 atomic_t inotify_watches; /* How many inotify watches does this user have? */
576 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
577 #endif
578 #ifdef CONFIG_POSIX_MQUEUE
579 /* protected by mq_lock */
580 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
581 #endif
582 unsigned long locked_shm; /* How many pages of mlocked shm ? */
583
584 #ifdef CONFIG_KEYS
585 struct key *uid_keyring; /* UID specific keyring */
586 struct key *session_keyring; /* UID's default session keyring */
587 #endif
588
589 /* Hash table maintenance information */
590 struct hlist_node uidhash_node;
591 uid_t uid;
592
593 #ifdef CONFIG_FAIR_USER_SCHED
594 struct task_group *tg;
595 #ifdef CONFIG_SYSFS
596 struct kobject kobj;
597 struct work_struct work;
598 #endif
599 #endif
600 };
601
602 extern int uids_sysfs_init(void);
603
604 extern struct user_struct *find_user(uid_t);
605
606 extern struct user_struct root_user;
607 #define INIT_USER (&root_user)
608
609 struct backing_dev_info;
610 struct reclaim_state;
611
612 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
613 struct sched_info {
614 /* cumulative counters */
615 unsigned long pcount; /* # of times run on this cpu */
616 unsigned long long cpu_time, /* time spent on the cpu */
617 run_delay; /* time spent waiting on a runqueue */
618
619 /* timestamps */
620 unsigned long long last_arrival,/* when we last ran on a cpu */
621 last_queued; /* when we were last queued to run */
622 #ifdef CONFIG_SCHEDSTATS
623 /* BKL stats */
624 unsigned int bkl_count;
625 #endif
626 };
627 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
628
629 #ifdef CONFIG_SCHEDSTATS
630 extern const struct file_operations proc_schedstat_operations;
631 #endif /* CONFIG_SCHEDSTATS */
632
633 #ifdef CONFIG_TASK_DELAY_ACCT
634 struct task_delay_info {
635 spinlock_t lock;
636 unsigned int flags; /* Private per-task flags */
637
638 /* For each stat XXX, add following, aligned appropriately
639 *
640 * struct timespec XXX_start, XXX_end;
641 * u64 XXX_delay;
642 * u32 XXX_count;
643 *
644 * Atomicity of updates to XXX_delay, XXX_count protected by
645 * single lock above (split into XXX_lock if contention is an issue).
646 */
647
648 /*
649 * XXX_count is incremented on every XXX operation, the delay
650 * associated with the operation is added to XXX_delay.
651 * XXX_delay contains the accumulated delay time in nanoseconds.
652 */
653 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
654 u64 blkio_delay; /* wait for sync block io completion */
655 u64 swapin_delay; /* wait for swapin block io completion */
656 u32 blkio_count; /* total count of the number of sync block */
657 /* io operations performed */
658 u32 swapin_count; /* total count of the number of swapin block */
659 /* io operations performed */
660 };
661 #endif /* CONFIG_TASK_DELAY_ACCT */
662
663 static inline int sched_info_on(void)
664 {
665 #ifdef CONFIG_SCHEDSTATS
666 return 1;
667 #elif defined(CONFIG_TASK_DELAY_ACCT)
668 extern int delayacct_on;
669 return delayacct_on;
670 #else
671 return 0;
672 #endif
673 }
674
675 enum cpu_idle_type {
676 CPU_IDLE,
677 CPU_NOT_IDLE,
678 CPU_NEWLY_IDLE,
679 CPU_MAX_IDLE_TYPES
680 };
681
682 /*
683 * sched-domains (multiprocessor balancing) declarations:
684 */
685
686 /*
687 * Increase resolution of nice-level calculations:
688 */
689 #define SCHED_LOAD_SHIFT 10
690 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
691
692 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
693
694 #ifdef CONFIG_SMP
695 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
696 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
697 #define SD_BALANCE_EXEC 4 /* Balance on exec */
698 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
699 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
700 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
701 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
702 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
703 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
704 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
705 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
706
707 #define BALANCE_FOR_MC_POWER \
708 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
709
710 #define BALANCE_FOR_PKG_POWER \
711 ((sched_mc_power_savings || sched_smt_power_savings) ? \
712 SD_POWERSAVINGS_BALANCE : 0)
713
714 #define test_sd_parent(sd, flag) ((sd->parent && \
715 (sd->parent->flags & flag)) ? 1 : 0)
716
717
718 struct sched_group {
719 struct sched_group *next; /* Must be a circular list */
720 cpumask_t cpumask;
721
722 /*
723 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
724 * single CPU. This is read only (except for setup, hotplug CPU).
725 * Note : Never change cpu_power without recompute its reciprocal
726 */
727 unsigned int __cpu_power;
728 /*
729 * reciprocal value of cpu_power to avoid expensive divides
730 * (see include/linux/reciprocal_div.h)
731 */
732 u32 reciprocal_cpu_power;
733 };
734
735 struct sched_domain {
736 /* These fields must be setup */
737 struct sched_domain *parent; /* top domain must be null terminated */
738 struct sched_domain *child; /* bottom domain must be null terminated */
739 struct sched_group *groups; /* the balancing groups of the domain */
740 cpumask_t span; /* span of all CPUs in this domain */
741 unsigned long min_interval; /* Minimum balance interval ms */
742 unsigned long max_interval; /* Maximum balance interval ms */
743 unsigned int busy_factor; /* less balancing by factor if busy */
744 unsigned int imbalance_pct; /* No balance until over watermark */
745 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
746 unsigned int busy_idx;
747 unsigned int idle_idx;
748 unsigned int newidle_idx;
749 unsigned int wake_idx;
750 unsigned int forkexec_idx;
751 int flags; /* See SD_* */
752
753 /* Runtime fields. */
754 unsigned long last_balance; /* init to jiffies. units in jiffies */
755 unsigned int balance_interval; /* initialise to 1. units in ms. */
756 unsigned int nr_balance_failed; /* initialise to 0 */
757
758 #ifdef CONFIG_SCHEDSTATS
759 /* load_balance() stats */
760 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
761 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
762 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
763 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
764 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
765 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
766 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
767 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
768
769 /* Active load balancing */
770 unsigned int alb_count;
771 unsigned int alb_failed;
772 unsigned int alb_pushed;
773
774 /* SD_BALANCE_EXEC stats */
775 unsigned int sbe_count;
776 unsigned int sbe_balanced;
777 unsigned int sbe_pushed;
778
779 /* SD_BALANCE_FORK stats */
780 unsigned int sbf_count;
781 unsigned int sbf_balanced;
782 unsigned int sbf_pushed;
783
784 /* try_to_wake_up() stats */
785 unsigned int ttwu_wake_remote;
786 unsigned int ttwu_move_affine;
787 unsigned int ttwu_move_balance;
788 #endif
789 };
790
791 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new);
792
793 #endif /* CONFIG_SMP */
794
795 /*
796 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
797 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
798 * task of nice 0 or enough lower priority tasks to bring up the
799 * weighted_cpuload
800 */
801 static inline int above_background_load(void)
802 {
803 unsigned long cpu;
804
805 for_each_online_cpu(cpu) {
806 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
807 return 1;
808 }
809 return 0;
810 }
811
812 struct io_context; /* See blkdev.h */
813 #define NGROUPS_SMALL 32
814 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
815 struct group_info {
816 int ngroups;
817 atomic_t usage;
818 gid_t small_block[NGROUPS_SMALL];
819 int nblocks;
820 gid_t *blocks[0];
821 };
822
823 /*
824 * get_group_info() must be called with the owning task locked (via task_lock())
825 * when task != current. The reason being that the vast majority of callers are
826 * looking at current->group_info, which can not be changed except by the
827 * current task. Changing current->group_info requires the task lock, too.
828 */
829 #define get_group_info(group_info) do { \
830 atomic_inc(&(group_info)->usage); \
831 } while (0)
832
833 #define put_group_info(group_info) do { \
834 if (atomic_dec_and_test(&(group_info)->usage)) \
835 groups_free(group_info); \
836 } while (0)
837
838 extern struct group_info *groups_alloc(int gidsetsize);
839 extern void groups_free(struct group_info *group_info);
840 extern int set_current_groups(struct group_info *group_info);
841 extern int groups_search(struct group_info *group_info, gid_t grp);
842 /* access the groups "array" with this macro */
843 #define GROUP_AT(gi, i) \
844 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
845
846 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
847 extern void prefetch_stack(struct task_struct *t);
848 #else
849 static inline void prefetch_stack(struct task_struct *t) { }
850 #endif
851
852 struct audit_context; /* See audit.c */
853 struct mempolicy;
854 struct pipe_inode_info;
855 struct uts_namespace;
856
857 struct rq;
858 struct sched_domain;
859
860 struct sched_class {
861 const struct sched_class *next;
862
863 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
864 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
865 void (*yield_task) (struct rq *rq);
866 int (*select_task_rq)(struct task_struct *p, int sync);
867
868 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
869
870 struct task_struct * (*pick_next_task) (struct rq *rq);
871 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
872
873 #ifdef CONFIG_SMP
874 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
875 struct rq *busiest, unsigned long max_load_move,
876 struct sched_domain *sd, enum cpu_idle_type idle,
877 int *all_pinned, int *this_best_prio);
878
879 int (*move_one_task) (struct rq *this_rq, int this_cpu,
880 struct rq *busiest, struct sched_domain *sd,
881 enum cpu_idle_type idle);
882 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
883 void (*post_schedule) (struct rq *this_rq);
884 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
885 #endif
886
887 void (*set_curr_task) (struct rq *rq);
888 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
889 void (*task_new) (struct rq *rq, struct task_struct *p);
890 void (*set_cpus_allowed)(struct task_struct *p, cpumask_t *newmask);
891
892 void (*join_domain)(struct rq *rq);
893 void (*leave_domain)(struct rq *rq);
894
895 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
896 int running);
897 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
898 int running);
899 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
900 int oldprio, int running);
901 };
902
903 struct load_weight {
904 unsigned long weight, inv_weight;
905 };
906
907 /*
908 * CFS stats for a schedulable entity (task, task-group etc)
909 *
910 * Current field usage histogram:
911 *
912 * 4 se->block_start
913 * 4 se->run_node
914 * 4 se->sleep_start
915 * 6 se->load.weight
916 */
917 struct sched_entity {
918 struct load_weight load; /* for load-balancing */
919 struct rb_node run_node;
920 unsigned int on_rq;
921
922 u64 exec_start;
923 u64 sum_exec_runtime;
924 u64 vruntime;
925 u64 prev_sum_exec_runtime;
926
927 #ifdef CONFIG_SCHEDSTATS
928 u64 wait_start;
929 u64 wait_max;
930 u64 wait_count;
931 u64 wait_sum;
932
933 u64 sleep_start;
934 u64 sleep_max;
935 s64 sum_sleep_runtime;
936
937 u64 block_start;
938 u64 block_max;
939 u64 exec_max;
940 u64 slice_max;
941
942 u64 nr_migrations;
943 u64 nr_migrations_cold;
944 u64 nr_failed_migrations_affine;
945 u64 nr_failed_migrations_running;
946 u64 nr_failed_migrations_hot;
947 u64 nr_forced_migrations;
948 u64 nr_forced2_migrations;
949
950 u64 nr_wakeups;
951 u64 nr_wakeups_sync;
952 u64 nr_wakeups_migrate;
953 u64 nr_wakeups_local;
954 u64 nr_wakeups_remote;
955 u64 nr_wakeups_affine;
956 u64 nr_wakeups_affine_attempts;
957 u64 nr_wakeups_passive;
958 u64 nr_wakeups_idle;
959 #endif
960
961 #ifdef CONFIG_FAIR_GROUP_SCHED
962 struct sched_entity *parent;
963 /* rq on which this entity is (to be) queued: */
964 struct cfs_rq *cfs_rq;
965 /* rq "owned" by this entity/group: */
966 struct cfs_rq *my_q;
967 #endif
968 };
969
970 struct sched_rt_entity {
971 struct list_head run_list;
972 unsigned int time_slice;
973 unsigned long timeout;
974 int nr_cpus_allowed;
975
976 #ifdef CONFIG_FAIR_GROUP_SCHED
977 struct sched_rt_entity *parent;
978 /* rq on which this entity is (to be) queued: */
979 struct rt_rq *rt_rq;
980 /* rq "owned" by this entity/group: */
981 struct rt_rq *my_q;
982 #endif
983 };
984
985 struct task_struct {
986 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
987 void *stack;
988 atomic_t usage;
989 unsigned int flags; /* per process flags, defined below */
990 unsigned int ptrace;
991
992 int lock_depth; /* BKL lock depth */
993
994 #ifdef CONFIG_SMP
995 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
996 int oncpu;
997 #endif
998 #endif
999
1000 int prio, static_prio, normal_prio;
1001 const struct sched_class *sched_class;
1002 struct sched_entity se;
1003 struct sched_rt_entity rt;
1004
1005 #ifdef CONFIG_PREEMPT_NOTIFIERS
1006 /* list of struct preempt_notifier: */
1007 struct hlist_head preempt_notifiers;
1008 #endif
1009
1010 /*
1011 * fpu_counter contains the number of consecutive context switches
1012 * that the FPU is used. If this is over a threshold, the lazy fpu
1013 * saving becomes unlazy to save the trap. This is an unsigned char
1014 * so that after 256 times the counter wraps and the behavior turns
1015 * lazy again; this to deal with bursty apps that only use FPU for
1016 * a short time
1017 */
1018 unsigned char fpu_counter;
1019 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1020 #ifdef CONFIG_BLK_DEV_IO_TRACE
1021 unsigned int btrace_seq;
1022 #endif
1023
1024 unsigned int policy;
1025 cpumask_t cpus_allowed;
1026
1027 #ifdef CONFIG_PREEMPT_RCU
1028 int rcu_read_lock_nesting;
1029 int rcu_flipctr_idx;
1030 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1031
1032 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1033 struct sched_info sched_info;
1034 #endif
1035
1036 struct list_head tasks;
1037 /*
1038 * ptrace_list/ptrace_children forms the list of my children
1039 * that were stolen by a ptracer.
1040 */
1041 struct list_head ptrace_children;
1042 struct list_head ptrace_list;
1043
1044 struct mm_struct *mm, *active_mm;
1045
1046 /* task state */
1047 struct linux_binfmt *binfmt;
1048 int exit_state;
1049 int exit_code, exit_signal;
1050 int pdeath_signal; /* The signal sent when the parent dies */
1051 /* ??? */
1052 unsigned int personality;
1053 unsigned did_exec:1;
1054 pid_t pid;
1055 pid_t tgid;
1056
1057 #ifdef CONFIG_CC_STACKPROTECTOR
1058 /* Canary value for the -fstack-protector gcc feature */
1059 unsigned long stack_canary;
1060 #endif
1061 /*
1062 * pointers to (original) parent process, youngest child, younger sibling,
1063 * older sibling, respectively. (p->father can be replaced with
1064 * p->parent->pid)
1065 */
1066 struct task_struct *real_parent; /* real parent process (when being debugged) */
1067 struct task_struct *parent; /* parent process */
1068 /*
1069 * children/sibling forms the list of my children plus the
1070 * tasks I'm ptracing.
1071 */
1072 struct list_head children; /* list of my children */
1073 struct list_head sibling; /* linkage in my parent's children list */
1074 struct task_struct *group_leader; /* threadgroup leader */
1075
1076 /* PID/PID hash table linkage. */
1077 struct pid_link pids[PIDTYPE_MAX];
1078 struct list_head thread_group;
1079
1080 struct completion *vfork_done; /* for vfork() */
1081 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1082 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1083
1084 unsigned int rt_priority;
1085 cputime_t utime, stime, utimescaled, stimescaled;
1086 cputime_t gtime;
1087 cputime_t prev_utime, prev_stime;
1088 unsigned long nvcsw, nivcsw; /* context switch counts */
1089 struct timespec start_time; /* monotonic time */
1090 struct timespec real_start_time; /* boot based time */
1091 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1092 unsigned long min_flt, maj_flt;
1093
1094 cputime_t it_prof_expires, it_virt_expires;
1095 unsigned long long it_sched_expires;
1096 struct list_head cpu_timers[3];
1097
1098 /* process credentials */
1099 uid_t uid,euid,suid,fsuid;
1100 gid_t gid,egid,sgid,fsgid;
1101 struct group_info *group_info;
1102 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset;
1103 unsigned keep_capabilities:1;
1104 struct user_struct *user;
1105 #ifdef CONFIG_KEYS
1106 struct key *request_key_auth; /* assumed request_key authority */
1107 struct key *thread_keyring; /* keyring private to this thread */
1108 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1109 #endif
1110 char comm[TASK_COMM_LEN]; /* executable name excluding path
1111 - access with [gs]et_task_comm (which lock
1112 it with task_lock())
1113 - initialized normally by flush_old_exec */
1114 /* file system info */
1115 int link_count, total_link_count;
1116 #ifdef CONFIG_SYSVIPC
1117 /* ipc stuff */
1118 struct sysv_sem sysvsem;
1119 #endif
1120 #ifdef CONFIG_DETECT_SOFTLOCKUP
1121 /* hung task detection */
1122 unsigned long last_switch_timestamp;
1123 unsigned long last_switch_count;
1124 #endif
1125 /* CPU-specific state of this task */
1126 struct thread_struct thread;
1127 /* filesystem information */
1128 struct fs_struct *fs;
1129 /* open file information */
1130 struct files_struct *files;
1131 /* namespaces */
1132 struct nsproxy *nsproxy;
1133 /* signal handlers */
1134 struct signal_struct *signal;
1135 struct sighand_struct *sighand;
1136
1137 sigset_t blocked, real_blocked;
1138 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
1139 struct sigpending pending;
1140
1141 unsigned long sas_ss_sp;
1142 size_t sas_ss_size;
1143 int (*notifier)(void *priv);
1144 void *notifier_data;
1145 sigset_t *notifier_mask;
1146 #ifdef CONFIG_SECURITY
1147 void *security;
1148 #endif
1149 struct audit_context *audit_context;
1150 #ifdef CONFIG_AUDITSYSCALL
1151 uid_t loginuid;
1152 unsigned int sessionid;
1153 #endif
1154 seccomp_t seccomp;
1155
1156 /* Thread group tracking */
1157 u32 parent_exec_id;
1158 u32 self_exec_id;
1159 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1160 spinlock_t alloc_lock;
1161
1162 /* Protection of the PI data structures: */
1163 spinlock_t pi_lock;
1164
1165 #ifdef CONFIG_RT_MUTEXES
1166 /* PI waiters blocked on a rt_mutex held by this task */
1167 struct plist_head pi_waiters;
1168 /* Deadlock detection and priority inheritance handling */
1169 struct rt_mutex_waiter *pi_blocked_on;
1170 #endif
1171
1172 #ifdef CONFIG_DEBUG_MUTEXES
1173 /* mutex deadlock detection */
1174 struct mutex_waiter *blocked_on;
1175 #endif
1176 #ifdef CONFIG_TRACE_IRQFLAGS
1177 unsigned int irq_events;
1178 int hardirqs_enabled;
1179 unsigned long hardirq_enable_ip;
1180 unsigned int hardirq_enable_event;
1181 unsigned long hardirq_disable_ip;
1182 unsigned int hardirq_disable_event;
1183 int softirqs_enabled;
1184 unsigned long softirq_disable_ip;
1185 unsigned int softirq_disable_event;
1186 unsigned long softirq_enable_ip;
1187 unsigned int softirq_enable_event;
1188 int hardirq_context;
1189 int softirq_context;
1190 #endif
1191 #ifdef CONFIG_LOCKDEP
1192 # define MAX_LOCK_DEPTH 30UL
1193 u64 curr_chain_key;
1194 int lockdep_depth;
1195 struct held_lock held_locks[MAX_LOCK_DEPTH];
1196 unsigned int lockdep_recursion;
1197 #endif
1198
1199 /* journalling filesystem info */
1200 void *journal_info;
1201
1202 /* stacked block device info */
1203 struct bio *bio_list, **bio_tail;
1204
1205 /* VM state */
1206 struct reclaim_state *reclaim_state;
1207
1208 struct backing_dev_info *backing_dev_info;
1209
1210 struct io_context *io_context;
1211
1212 unsigned long ptrace_message;
1213 siginfo_t *last_siginfo; /* For ptrace use. */
1214 #ifdef CONFIG_TASK_XACCT
1215 /* i/o counters(bytes read/written, #syscalls */
1216 u64 rchar, wchar, syscr, syscw;
1217 #endif
1218 struct task_io_accounting ioac;
1219 #if defined(CONFIG_TASK_XACCT)
1220 u64 acct_rss_mem1; /* accumulated rss usage */
1221 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1222 cputime_t acct_stimexpd;/* stime since last update */
1223 #endif
1224 #ifdef CONFIG_NUMA
1225 struct mempolicy *mempolicy;
1226 short il_next;
1227 #endif
1228 #ifdef CONFIG_CPUSETS
1229 nodemask_t mems_allowed;
1230 int cpuset_mems_generation;
1231 int cpuset_mem_spread_rotor;
1232 #endif
1233 #ifdef CONFIG_CGROUPS
1234 /* Control Group info protected by css_set_lock */
1235 struct css_set *cgroups;
1236 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1237 struct list_head cg_list;
1238 #endif
1239 #ifdef CONFIG_FUTEX
1240 struct robust_list_head __user *robust_list;
1241 #ifdef CONFIG_COMPAT
1242 struct compat_robust_list_head __user *compat_robust_list;
1243 #endif
1244 struct list_head pi_state_list;
1245 struct futex_pi_state *pi_state_cache;
1246 #endif
1247 atomic_t fs_excl; /* holding fs exclusive resources */
1248 struct rcu_head rcu;
1249
1250 /*
1251 * cache last used pipe for splice
1252 */
1253 struct pipe_inode_info *splice_pipe;
1254 #ifdef CONFIG_TASK_DELAY_ACCT
1255 struct task_delay_info *delays;
1256 #endif
1257 #ifdef CONFIG_FAULT_INJECTION
1258 int make_it_fail;
1259 #endif
1260 struct prop_local_single dirties;
1261 #ifdef CONFIG_LATENCYTOP
1262 int latency_record_count;
1263 struct latency_record latency_record[LT_SAVECOUNT];
1264 #endif
1265 };
1266
1267 /*
1268 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1269 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1270 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1271 * values are inverted: lower p->prio value means higher priority.
1272 *
1273 * The MAX_USER_RT_PRIO value allows the actual maximum
1274 * RT priority to be separate from the value exported to
1275 * user-space. This allows kernel threads to set their
1276 * priority to a value higher than any user task. Note:
1277 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1278 */
1279
1280 #define MAX_USER_RT_PRIO 100
1281 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1282
1283 #define MAX_PRIO (MAX_RT_PRIO + 40)
1284 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1285
1286 static inline int rt_prio(int prio)
1287 {
1288 if (unlikely(prio < MAX_RT_PRIO))
1289 return 1;
1290 return 0;
1291 }
1292
1293 static inline int rt_task(struct task_struct *p)
1294 {
1295 return rt_prio(p->prio);
1296 }
1297
1298 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1299 {
1300 tsk->signal->__session = session;
1301 }
1302
1303 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1304 {
1305 tsk->signal->__pgrp = pgrp;
1306 }
1307
1308 static inline struct pid *task_pid(struct task_struct *task)
1309 {
1310 return task->pids[PIDTYPE_PID].pid;
1311 }
1312
1313 static inline struct pid *task_tgid(struct task_struct *task)
1314 {
1315 return task->group_leader->pids[PIDTYPE_PID].pid;
1316 }
1317
1318 static inline struct pid *task_pgrp(struct task_struct *task)
1319 {
1320 return task->group_leader->pids[PIDTYPE_PGID].pid;
1321 }
1322
1323 static inline struct pid *task_session(struct task_struct *task)
1324 {
1325 return task->group_leader->pids[PIDTYPE_SID].pid;
1326 }
1327
1328 struct pid_namespace;
1329
1330 /*
1331 * the helpers to get the task's different pids as they are seen
1332 * from various namespaces
1333 *
1334 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1335 * task_xid_vnr() : virtual id, i.e. the id seen from the namespace the task
1336 * belongs to. this only makes sence when called in the
1337 * context of the task that belongs to the same namespace;
1338 * task_xid_nr_ns() : id seen from the ns specified;
1339 *
1340 * set_task_vxid() : assigns a virtual id to a task;
1341 *
1342 * see also pid_nr() etc in include/linux/pid.h
1343 */
1344
1345 static inline pid_t task_pid_nr(struct task_struct *tsk)
1346 {
1347 return tsk->pid;
1348 }
1349
1350 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1351
1352 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1353 {
1354 return pid_vnr(task_pid(tsk));
1355 }
1356
1357
1358 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1359 {
1360 return tsk->tgid;
1361 }
1362
1363 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1364
1365 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1366 {
1367 return pid_vnr(task_tgid(tsk));
1368 }
1369
1370
1371 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1372 {
1373 return tsk->signal->__pgrp;
1374 }
1375
1376 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1377
1378 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1379 {
1380 return pid_vnr(task_pgrp(tsk));
1381 }
1382
1383
1384 static inline pid_t task_session_nr(struct task_struct *tsk)
1385 {
1386 return tsk->signal->__session;
1387 }
1388
1389 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1390
1391 static inline pid_t task_session_vnr(struct task_struct *tsk)
1392 {
1393 return pid_vnr(task_session(tsk));
1394 }
1395
1396
1397 /**
1398 * pid_alive - check that a task structure is not stale
1399 * @p: Task structure to be checked.
1400 *
1401 * Test if a process is not yet dead (at most zombie state)
1402 * If pid_alive fails, then pointers within the task structure
1403 * can be stale and must not be dereferenced.
1404 */
1405 static inline int pid_alive(struct task_struct *p)
1406 {
1407 return p->pids[PIDTYPE_PID].pid != NULL;
1408 }
1409
1410 /**
1411 * is_global_init - check if a task structure is init
1412 * @tsk: Task structure to be checked.
1413 *
1414 * Check if a task structure is the first user space task the kernel created.
1415 */
1416 static inline int is_global_init(struct task_struct *tsk)
1417 {
1418 return tsk->pid == 1;
1419 }
1420
1421 /*
1422 * is_container_init:
1423 * check whether in the task is init in its own pid namespace.
1424 */
1425 extern int is_container_init(struct task_struct *tsk);
1426
1427 extern struct pid *cad_pid;
1428
1429 extern void free_task(struct task_struct *tsk);
1430 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1431
1432 extern void __put_task_struct(struct task_struct *t);
1433
1434 static inline void put_task_struct(struct task_struct *t)
1435 {
1436 if (atomic_dec_and_test(&t->usage))
1437 __put_task_struct(t);
1438 }
1439
1440 /*
1441 * Per process flags
1442 */
1443 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1444 /* Not implemented yet, only for 486*/
1445 #define PF_STARTING 0x00000002 /* being created */
1446 #define PF_EXITING 0x00000004 /* getting shut down */
1447 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1448 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1449 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1450 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1451 #define PF_DUMPCORE 0x00000200 /* dumped core */
1452 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1453 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1454 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1455 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1456 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1457 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1458 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1459 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1460 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1461 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1462 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1463 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1464 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1465 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1466 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1467 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1468 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1469 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1470
1471 /*
1472 * Only the _current_ task can read/write to tsk->flags, but other
1473 * tasks can access tsk->flags in readonly mode for example
1474 * with tsk_used_math (like during threaded core dumping).
1475 * There is however an exception to this rule during ptrace
1476 * or during fork: the ptracer task is allowed to write to the
1477 * child->flags of its traced child (same goes for fork, the parent
1478 * can write to the child->flags), because we're guaranteed the
1479 * child is not running and in turn not changing child->flags
1480 * at the same time the parent does it.
1481 */
1482 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1483 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1484 #define clear_used_math() clear_stopped_child_used_math(current)
1485 #define set_used_math() set_stopped_child_used_math(current)
1486 #define conditional_stopped_child_used_math(condition, child) \
1487 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1488 #define conditional_used_math(condition) \
1489 conditional_stopped_child_used_math(condition, current)
1490 #define copy_to_stopped_child_used_math(child) \
1491 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1492 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1493 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1494 #define used_math() tsk_used_math(current)
1495
1496 #ifdef CONFIG_SMP
1497 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1498 #else
1499 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1500 {
1501 if (!cpu_isset(0, new_mask))
1502 return -EINVAL;
1503 return 0;
1504 }
1505 #endif
1506
1507 extern unsigned long long sched_clock(void);
1508
1509 /*
1510 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1511 * clock constructed from sched_clock():
1512 */
1513 extern unsigned long long cpu_clock(int cpu);
1514
1515 extern unsigned long long
1516 task_sched_runtime(struct task_struct *task);
1517
1518 /* sched_exec is called by processes performing an exec */
1519 #ifdef CONFIG_SMP
1520 extern void sched_exec(void);
1521 #else
1522 #define sched_exec() {}
1523 #endif
1524
1525 extern void sched_clock_idle_sleep_event(void);
1526 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1527
1528 #ifdef CONFIG_HOTPLUG_CPU
1529 extern void idle_task_exit(void);
1530 #else
1531 static inline void idle_task_exit(void) {}
1532 #endif
1533
1534 extern void sched_idle_next(void);
1535
1536 #ifdef CONFIG_SCHED_DEBUG
1537 extern unsigned int sysctl_sched_latency;
1538 extern unsigned int sysctl_sched_min_granularity;
1539 extern unsigned int sysctl_sched_wakeup_granularity;
1540 extern unsigned int sysctl_sched_batch_wakeup_granularity;
1541 extern unsigned int sysctl_sched_child_runs_first;
1542 extern unsigned int sysctl_sched_features;
1543 extern unsigned int sysctl_sched_migration_cost;
1544 extern unsigned int sysctl_sched_nr_migrate;
1545 extern unsigned int sysctl_sched_rt_period;
1546 extern unsigned int sysctl_sched_rt_ratio;
1547 #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP)
1548 extern unsigned int sysctl_sched_min_bal_int_shares;
1549 extern unsigned int sysctl_sched_max_bal_int_shares;
1550 #endif
1551
1552 int sched_nr_latency_handler(struct ctl_table *table, int write,
1553 struct file *file, void __user *buffer, size_t *length,
1554 loff_t *ppos);
1555 #endif
1556
1557 extern unsigned int sysctl_sched_compat_yield;
1558
1559 #ifdef CONFIG_RT_MUTEXES
1560 extern int rt_mutex_getprio(struct task_struct *p);
1561 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1562 extern void rt_mutex_adjust_pi(struct task_struct *p);
1563 #else
1564 static inline int rt_mutex_getprio(struct task_struct *p)
1565 {
1566 return p->normal_prio;
1567 }
1568 # define rt_mutex_adjust_pi(p) do { } while (0)
1569 #endif
1570
1571 extern void set_user_nice(struct task_struct *p, long nice);
1572 extern int task_prio(const struct task_struct *p);
1573 extern int task_nice(const struct task_struct *p);
1574 extern int can_nice(const struct task_struct *p, const int nice);
1575 extern int task_curr(const struct task_struct *p);
1576 extern int idle_cpu(int cpu);
1577 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1578 extern struct task_struct *idle_task(int cpu);
1579 extern struct task_struct *curr_task(int cpu);
1580 extern void set_curr_task(int cpu, struct task_struct *p);
1581
1582 void yield(void);
1583
1584 /*
1585 * The default (Linux) execution domain.
1586 */
1587 extern struct exec_domain default_exec_domain;
1588
1589 union thread_union {
1590 struct thread_info thread_info;
1591 unsigned long stack[THREAD_SIZE/sizeof(long)];
1592 };
1593
1594 #ifndef __HAVE_ARCH_KSTACK_END
1595 static inline int kstack_end(void *addr)
1596 {
1597 /* Reliable end of stack detection:
1598 * Some APM bios versions misalign the stack
1599 */
1600 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1601 }
1602 #endif
1603
1604 extern union thread_union init_thread_union;
1605 extern struct task_struct init_task;
1606
1607 extern struct mm_struct init_mm;
1608
1609 extern struct pid_namespace init_pid_ns;
1610
1611 /*
1612 * find a task by one of its numerical ids
1613 *
1614 * find_task_by_pid_type_ns():
1615 * it is the most generic call - it finds a task by all id,
1616 * type and namespace specified
1617 * find_task_by_pid_ns():
1618 * finds a task by its pid in the specified namespace
1619 * find_task_by_vpid():
1620 * finds a task by its virtual pid
1621 * find_task_by_pid():
1622 * finds a task by its global pid
1623 *
1624 * see also find_pid() etc in include/linux/pid.h
1625 */
1626
1627 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1628 struct pid_namespace *ns);
1629
1630 extern struct task_struct *find_task_by_pid(pid_t nr);
1631 extern struct task_struct *find_task_by_vpid(pid_t nr);
1632 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1633 struct pid_namespace *ns);
1634
1635 extern void __set_special_pids(pid_t session, pid_t pgrp);
1636
1637 /* per-UID process charging. */
1638 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1639 static inline struct user_struct *get_uid(struct user_struct *u)
1640 {
1641 atomic_inc(&u->__count);
1642 return u;
1643 }
1644 extern void free_uid(struct user_struct *);
1645 extern void switch_uid(struct user_struct *);
1646 extern void release_uids(struct user_namespace *ns);
1647
1648 #include <asm/current.h>
1649
1650 extern void do_timer(unsigned long ticks);
1651
1652 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1653 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1654 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1655 unsigned long clone_flags));
1656 #ifdef CONFIG_SMP
1657 extern void kick_process(struct task_struct *tsk);
1658 #else
1659 static inline void kick_process(struct task_struct *tsk) { }
1660 #endif
1661 extern void sched_fork(struct task_struct *p, int clone_flags);
1662 extern void sched_dead(struct task_struct *p);
1663
1664 extern int in_group_p(gid_t);
1665 extern int in_egroup_p(gid_t);
1666
1667 extern void proc_caches_init(void);
1668 extern void flush_signals(struct task_struct *);
1669 extern void ignore_signals(struct task_struct *);
1670 extern void flush_signal_handlers(struct task_struct *, int force_default);
1671 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1672
1673 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1674 {
1675 unsigned long flags;
1676 int ret;
1677
1678 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1679 ret = dequeue_signal(tsk, mask, info);
1680 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1681
1682 return ret;
1683 }
1684
1685 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1686 sigset_t *mask);
1687 extern void unblock_all_signals(void);
1688 extern void release_task(struct task_struct * p);
1689 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1690 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1691 extern int force_sigsegv(int, struct task_struct *);
1692 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1693 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1694 extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1695 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1696 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1697 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1698 extern int kill_pid(struct pid *pid, int sig, int priv);
1699 extern int kill_proc_info(int, struct siginfo *, pid_t);
1700 extern void do_notify_parent(struct task_struct *, int);
1701 extern void force_sig(int, struct task_struct *);
1702 extern void force_sig_specific(int, struct task_struct *);
1703 extern int send_sig(int, struct task_struct *, int);
1704 extern void zap_other_threads(struct task_struct *p);
1705 extern int kill_proc(pid_t, int, int);
1706 extern struct sigqueue *sigqueue_alloc(void);
1707 extern void sigqueue_free(struct sigqueue *);
1708 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1709 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
1710 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1711 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1712
1713 static inline int kill_cad_pid(int sig, int priv)
1714 {
1715 return kill_pid(cad_pid, sig, priv);
1716 }
1717
1718 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1719 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1720 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1721 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1722
1723 static inline int is_si_special(const struct siginfo *info)
1724 {
1725 return info <= SEND_SIG_FORCED;
1726 }
1727
1728 /* True if we are on the alternate signal stack. */
1729
1730 static inline int on_sig_stack(unsigned long sp)
1731 {
1732 return (sp - current->sas_ss_sp < current->sas_ss_size);
1733 }
1734
1735 static inline int sas_ss_flags(unsigned long sp)
1736 {
1737 return (current->sas_ss_size == 0 ? SS_DISABLE
1738 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1739 }
1740
1741 /*
1742 * Routines for handling mm_structs
1743 */
1744 extern struct mm_struct * mm_alloc(void);
1745
1746 /* mmdrop drops the mm and the page tables */
1747 extern void FASTCALL(__mmdrop(struct mm_struct *));
1748 static inline void mmdrop(struct mm_struct * mm)
1749 {
1750 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1751 __mmdrop(mm);
1752 }
1753
1754 /* mmput gets rid of the mappings and all user-space */
1755 extern void mmput(struct mm_struct *);
1756 /* Grab a reference to a task's mm, if it is not already going away */
1757 extern struct mm_struct *get_task_mm(struct task_struct *task);
1758 /* Remove the current tasks stale references to the old mm_struct */
1759 extern void mm_release(struct task_struct *, struct mm_struct *);
1760
1761 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1762 extern void flush_thread(void);
1763 extern void exit_thread(void);
1764
1765 extern void exit_files(struct task_struct *);
1766 extern void __cleanup_signal(struct signal_struct *);
1767 extern void __cleanup_sighand(struct sighand_struct *);
1768 extern void exit_itimers(struct signal_struct *);
1769
1770 extern NORET_TYPE void do_group_exit(int);
1771
1772 extern void daemonize(const char *, ...);
1773 extern int allow_signal(int);
1774 extern int disallow_signal(int);
1775
1776 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1777 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1778 struct task_struct *fork_idle(int);
1779
1780 extern void set_task_comm(struct task_struct *tsk, char *from);
1781 extern char *get_task_comm(char *to, struct task_struct *tsk);
1782
1783 #ifdef CONFIG_SMP
1784 extern void wait_task_inactive(struct task_struct * p);
1785 #else
1786 #define wait_task_inactive(p) do { } while (0)
1787 #endif
1788
1789 #define remove_parent(p) list_del_init(&(p)->sibling)
1790 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1791
1792 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1793
1794 #define for_each_process(p) \
1795 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1796
1797 /*
1798 * Careful: do_each_thread/while_each_thread is a double loop so
1799 * 'break' will not work as expected - use goto instead.
1800 */
1801 #define do_each_thread(g, t) \
1802 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1803
1804 #define while_each_thread(g, t) \
1805 while ((t = next_thread(t)) != g)
1806
1807 /* de_thread depends on thread_group_leader not being a pid based check */
1808 #define thread_group_leader(p) (p == p->group_leader)
1809
1810 /* Do to the insanities of de_thread it is possible for a process
1811 * to have the pid of the thread group leader without actually being
1812 * the thread group leader. For iteration through the pids in proc
1813 * all we care about is that we have a task with the appropriate
1814 * pid, we don't actually care if we have the right task.
1815 */
1816 static inline int has_group_leader_pid(struct task_struct *p)
1817 {
1818 return p->pid == p->tgid;
1819 }
1820
1821 static inline
1822 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1823 {
1824 return p1->tgid == p2->tgid;
1825 }
1826
1827 static inline struct task_struct *next_thread(const struct task_struct *p)
1828 {
1829 return list_entry(rcu_dereference(p->thread_group.next),
1830 struct task_struct, thread_group);
1831 }
1832
1833 static inline int thread_group_empty(struct task_struct *p)
1834 {
1835 return list_empty(&p->thread_group);
1836 }
1837
1838 #define delay_group_leader(p) \
1839 (thread_group_leader(p) && !thread_group_empty(p))
1840
1841 /*
1842 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1843 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1844 * pins the final release of task.io_context. Also protects ->cpuset and
1845 * ->cgroup.subsys[].
1846 *
1847 * Nests both inside and outside of read_lock(&tasklist_lock).
1848 * It must not be nested with write_lock_irq(&tasklist_lock),
1849 * neither inside nor outside.
1850 */
1851 static inline void task_lock(struct task_struct *p)
1852 {
1853 spin_lock(&p->alloc_lock);
1854 }
1855
1856 static inline void task_unlock(struct task_struct *p)
1857 {
1858 spin_unlock(&p->alloc_lock);
1859 }
1860
1861 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1862 unsigned long *flags);
1863
1864 static inline void unlock_task_sighand(struct task_struct *tsk,
1865 unsigned long *flags)
1866 {
1867 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1868 }
1869
1870 #ifndef __HAVE_THREAD_FUNCTIONS
1871
1872 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1873 #define task_stack_page(task) ((task)->stack)
1874
1875 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1876 {
1877 *task_thread_info(p) = *task_thread_info(org);
1878 task_thread_info(p)->task = p;
1879 }
1880
1881 static inline unsigned long *end_of_stack(struct task_struct *p)
1882 {
1883 return (unsigned long *)(task_thread_info(p) + 1);
1884 }
1885
1886 #endif
1887
1888 /* set thread flags in other task's structures
1889 * - see asm/thread_info.h for TIF_xxxx flags available
1890 */
1891 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1892 {
1893 set_ti_thread_flag(task_thread_info(tsk), flag);
1894 }
1895
1896 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1897 {
1898 clear_ti_thread_flag(task_thread_info(tsk), flag);
1899 }
1900
1901 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1902 {
1903 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1904 }
1905
1906 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1907 {
1908 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1909 }
1910
1911 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1912 {
1913 return test_ti_thread_flag(task_thread_info(tsk), flag);
1914 }
1915
1916 static inline void set_tsk_need_resched(struct task_struct *tsk)
1917 {
1918 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1919 }
1920
1921 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1922 {
1923 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1924 }
1925
1926 static inline int signal_pending(struct task_struct *p)
1927 {
1928 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1929 }
1930
1931 extern int FASTCALL(__fatal_signal_pending(struct task_struct *p));
1932
1933 static inline int fatal_signal_pending(struct task_struct *p)
1934 {
1935 return signal_pending(p) && __fatal_signal_pending(p);
1936 }
1937
1938 static inline int need_resched(void)
1939 {
1940 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1941 }
1942
1943 /*
1944 * cond_resched() and cond_resched_lock(): latency reduction via
1945 * explicit rescheduling in places that are safe. The return
1946 * value indicates whether a reschedule was done in fact.
1947 * cond_resched_lock() will drop the spinlock before scheduling,
1948 * cond_resched_softirq() will enable bhs before scheduling.
1949 */
1950 #ifdef CONFIG_PREEMPT
1951 static inline int cond_resched(void)
1952 {
1953 return 0;
1954 }
1955 #else
1956 extern int _cond_resched(void);
1957 static inline int cond_resched(void)
1958 {
1959 return _cond_resched();
1960 }
1961 #endif
1962 extern int cond_resched_lock(spinlock_t * lock);
1963 extern int cond_resched_softirq(void);
1964
1965 /*
1966 * Does a critical section need to be broken due to another
1967 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1968 * but a general need for low latency)
1969 */
1970 static inline int spin_needbreak(spinlock_t *lock)
1971 {
1972 #ifdef CONFIG_PREEMPT
1973 return spin_is_contended(lock);
1974 #else
1975 return 0;
1976 #endif
1977 }
1978
1979 /*
1980 * Reevaluate whether the task has signals pending delivery.
1981 * Wake the task if so.
1982 * This is required every time the blocked sigset_t changes.
1983 * callers must hold sighand->siglock.
1984 */
1985 extern void recalc_sigpending_and_wake(struct task_struct *t);
1986 extern void recalc_sigpending(void);
1987
1988 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1989
1990 /*
1991 * Wrappers for p->thread_info->cpu access. No-op on UP.
1992 */
1993 #ifdef CONFIG_SMP
1994
1995 static inline unsigned int task_cpu(const struct task_struct *p)
1996 {
1997 return task_thread_info(p)->cpu;
1998 }
1999
2000 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2001
2002 #else
2003
2004 static inline unsigned int task_cpu(const struct task_struct *p)
2005 {
2006 return 0;
2007 }
2008
2009 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2010 {
2011 }
2012
2013 #endif /* CONFIG_SMP */
2014
2015 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2016 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2017 #else
2018 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
2019 {
2020 mm->mmap_base = TASK_UNMAPPED_BASE;
2021 mm->get_unmapped_area = arch_get_unmapped_area;
2022 mm->unmap_area = arch_unmap_area;
2023 }
2024 #endif
2025
2026 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
2027 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2028
2029 extern int sched_mc_power_savings, sched_smt_power_savings;
2030
2031 extern void normalize_rt_tasks(void);
2032
2033 #ifdef CONFIG_FAIR_GROUP_SCHED
2034
2035 extern struct task_group init_task_group;
2036
2037 extern struct task_group *sched_create_group(void);
2038 extern void sched_destroy_group(struct task_group *tg);
2039 extern void sched_move_task(struct task_struct *tsk);
2040 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2041 extern unsigned long sched_group_shares(struct task_group *tg);
2042
2043 #endif
2044
2045 #ifdef CONFIG_TASK_XACCT
2046 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2047 {
2048 tsk->rchar += amt;
2049 }
2050
2051 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2052 {
2053 tsk->wchar += amt;
2054 }
2055
2056 static inline void inc_syscr(struct task_struct *tsk)
2057 {
2058 tsk->syscr++;
2059 }
2060
2061 static inline void inc_syscw(struct task_struct *tsk)
2062 {
2063 tsk->syscw++;
2064 }
2065 #else
2066 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2067 {
2068 }
2069
2070 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2071 {
2072 }
2073
2074 static inline void inc_syscr(struct task_struct *tsk)
2075 {
2076 }
2077
2078 static inline void inc_syscw(struct task_struct *tsk)
2079 {
2080 }
2081 #endif
2082
2083 #ifdef CONFIG_SMP
2084 void migration_init(void);
2085 #else
2086 static inline void migration_init(void)
2087 {
2088 }
2089 #endif
2090
2091 #ifndef TASK_SIZE_OF
2092 #define TASK_SIZE_OF(tsk) TASK_SIZE
2093 #endif
2094
2095 #endif /* __KERNEL__ */
2096
2097 #endif
This page took 0.140643 seconds and 5 git commands to generate.