]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blob - include/linux/sched.h
Merge git://git.infradead.org/embedded-2.6
[mirror_ubuntu-artful-kernel.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/page.h>
65 #include <asm/ptrace.h>
66 #include <asm/cputime.h>
67
68 #include <linux/smp.h>
69 #include <linux/sem.h>
70 #include <linux/signal.h>
71 #include <linux/fs_struct.h>
72 #include <linux/compiler.h>
73 #include <linux/completion.h>
74 #include <linux/pid.h>
75 #include <linux/percpu.h>
76 #include <linux/topology.h>
77 #include <linux/proportions.h>
78 #include <linux/seccomp.h>
79 #include <linux/rcupdate.h>
80 #include <linux/rtmutex.h>
81
82 #include <linux/time.h>
83 #include <linux/param.h>
84 #include <linux/resource.h>
85 #include <linux/timer.h>
86 #include <linux/hrtimer.h>
87 #include <linux/task_io_accounting.h>
88 #include <linux/kobject.h>
89 #include <linux/latencytop.h>
90
91 #include <asm/processor.h>
92
93 struct mem_cgroup;
94 struct exec_domain;
95 struct futex_pi_state;
96 struct robust_list_head;
97 struct bio;
98
99 /*
100 * List of flags we want to share for kernel threads,
101 * if only because they are not used by them anyway.
102 */
103 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
104
105 /*
106 * These are the constant used to fake the fixed-point load-average
107 * counting. Some notes:
108 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
109 * a load-average precision of 10 bits integer + 11 bits fractional
110 * - if you want to count load-averages more often, you need more
111 * precision, or rounding will get you. With 2-second counting freq,
112 * the EXP_n values would be 1981, 2034 and 2043 if still using only
113 * 11 bit fractions.
114 */
115 extern unsigned long avenrun[]; /* Load averages */
116
117 #define FSHIFT 11 /* nr of bits of precision */
118 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
119 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
120 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
121 #define EXP_5 2014 /* 1/exp(5sec/5min) */
122 #define EXP_15 2037 /* 1/exp(5sec/15min) */
123
124 #define CALC_LOAD(load,exp,n) \
125 load *= exp; \
126 load += n*(FIXED_1-exp); \
127 load >>= FSHIFT;
128
129 extern unsigned long total_forks;
130 extern int nr_threads;
131 DECLARE_PER_CPU(unsigned long, process_counts);
132 extern int nr_processes(void);
133 extern unsigned long nr_running(void);
134 extern unsigned long nr_uninterruptible(void);
135 extern unsigned long nr_active(void);
136 extern unsigned long nr_iowait(void);
137
138 struct seq_file;
139 struct cfs_rq;
140 struct task_group;
141 #ifdef CONFIG_SCHED_DEBUG
142 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
143 extern void proc_sched_set_task(struct task_struct *p);
144 extern void
145 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
146 #else
147 static inline void
148 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
149 {
150 }
151 static inline void proc_sched_set_task(struct task_struct *p)
152 {
153 }
154 static inline void
155 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
156 {
157 }
158 #endif
159
160 extern unsigned long long time_sync_thresh;
161
162 /*
163 * Task state bitmask. NOTE! These bits are also
164 * encoded in fs/proc/array.c: get_task_state().
165 *
166 * We have two separate sets of flags: task->state
167 * is about runnability, while task->exit_state are
168 * about the task exiting. Confusing, but this way
169 * modifying one set can't modify the other one by
170 * mistake.
171 */
172 #define TASK_RUNNING 0
173 #define TASK_INTERRUPTIBLE 1
174 #define TASK_UNINTERRUPTIBLE 2
175 #define __TASK_STOPPED 4
176 #define __TASK_TRACED 8
177 /* in tsk->exit_state */
178 #define EXIT_ZOMBIE 16
179 #define EXIT_DEAD 32
180 /* in tsk->state again */
181 #define TASK_DEAD 64
182 #define TASK_WAKEKILL 128
183
184 /* Convenience macros for the sake of set_task_state */
185 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
186 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
187 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
188
189 /* Convenience macros for the sake of wake_up */
190 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
191 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
192
193 /* get_task_state() */
194 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
195 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
196 __TASK_TRACED)
197
198 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
199 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
200 #define task_is_stopped_or_traced(task) \
201 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
202 #define task_contributes_to_load(task) \
203 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
204
205 #define __set_task_state(tsk, state_value) \
206 do { (tsk)->state = (state_value); } while (0)
207 #define set_task_state(tsk, state_value) \
208 set_mb((tsk)->state, (state_value))
209
210 /*
211 * set_current_state() includes a barrier so that the write of current->state
212 * is correctly serialised wrt the caller's subsequent test of whether to
213 * actually sleep:
214 *
215 * set_current_state(TASK_UNINTERRUPTIBLE);
216 * if (do_i_need_to_sleep())
217 * schedule();
218 *
219 * If the caller does not need such serialisation then use __set_current_state()
220 */
221 #define __set_current_state(state_value) \
222 do { current->state = (state_value); } while (0)
223 #define set_current_state(state_value) \
224 set_mb(current->state, (state_value))
225
226 /* Task command name length */
227 #define TASK_COMM_LEN 16
228
229 #include <linux/spinlock.h>
230
231 /*
232 * This serializes "schedule()" and also protects
233 * the run-queue from deletions/modifications (but
234 * _adding_ to the beginning of the run-queue has
235 * a separate lock).
236 */
237 extern rwlock_t tasklist_lock;
238 extern spinlock_t mmlist_lock;
239
240 struct task_struct;
241
242 extern void sched_init(void);
243 extern void sched_init_smp(void);
244 extern asmlinkage void schedule_tail(struct task_struct *prev);
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 int runqueue_is_locked(void);
249
250 extern cpumask_t nohz_cpu_mask;
251 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
252 extern int select_nohz_load_balancer(int cpu);
253 #else
254 static inline int select_nohz_load_balancer(int cpu)
255 {
256 return 0;
257 }
258 #endif
259
260 extern unsigned long rt_needs_cpu(int cpu);
261
262 /*
263 * Only dump TASK_* tasks. (0 for all tasks)
264 */
265 extern void show_state_filter(unsigned long state_filter);
266
267 static inline void show_state(void)
268 {
269 show_state_filter(0);
270 }
271
272 extern void show_regs(struct pt_regs *);
273
274 /*
275 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
276 * task), SP is the stack pointer of the first frame that should be shown in the back
277 * trace (or NULL if the entire call-chain of the task should be shown).
278 */
279 extern void show_stack(struct task_struct *task, unsigned long *sp);
280
281 void io_schedule(void);
282 long io_schedule_timeout(long timeout);
283
284 extern void cpu_init (void);
285 extern void trap_init(void);
286 extern void account_process_tick(struct task_struct *task, int user);
287 extern void update_process_times(int user);
288 extern void scheduler_tick(void);
289 extern void hrtick_resched(void);
290
291 extern void sched_show_task(struct task_struct *p);
292
293 #ifdef CONFIG_DETECT_SOFTLOCKUP
294 extern void softlockup_tick(void);
295 extern void spawn_softlockup_task(void);
296 extern void touch_softlockup_watchdog(void);
297 extern void touch_all_softlockup_watchdogs(void);
298 extern unsigned int softlockup_panic;
299 extern unsigned long sysctl_hung_task_check_count;
300 extern unsigned long sysctl_hung_task_timeout_secs;
301 extern unsigned long sysctl_hung_task_warnings;
302 extern int softlockup_thresh;
303 #else
304 static inline void softlockup_tick(void)
305 {
306 }
307 static inline void spawn_softlockup_task(void)
308 {
309 }
310 static inline void touch_softlockup_watchdog(void)
311 {
312 }
313 static inline void touch_all_softlockup_watchdogs(void)
314 {
315 }
316 #endif
317
318
319 /* Attach to any functions which should be ignored in wchan output. */
320 #define __sched __attribute__((__section__(".sched.text")))
321
322 /* Linker adds these: start and end of __sched functions */
323 extern char __sched_text_start[], __sched_text_end[];
324
325 /* Is this address in the __sched functions? */
326 extern int in_sched_functions(unsigned long addr);
327
328 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
329 extern signed long schedule_timeout(signed long timeout);
330 extern signed long schedule_timeout_interruptible(signed long timeout);
331 extern signed long schedule_timeout_killable(signed long timeout);
332 extern signed long schedule_timeout_uninterruptible(signed long timeout);
333 asmlinkage void schedule(void);
334
335 struct nsproxy;
336 struct user_namespace;
337
338 /* Maximum number of active map areas.. This is a random (large) number */
339 #define DEFAULT_MAX_MAP_COUNT 65536
340
341 extern int sysctl_max_map_count;
342
343 #include <linux/aio.h>
344
345 extern unsigned long
346 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
347 unsigned long, unsigned long);
348 extern unsigned long
349 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
350 unsigned long len, unsigned long pgoff,
351 unsigned long flags);
352 extern void arch_unmap_area(struct mm_struct *, unsigned long);
353 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
354
355 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
356 /*
357 * The mm counters are not protected by its page_table_lock,
358 * so must be incremented atomically.
359 */
360 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
361 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
362 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
363 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
364 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
365
366 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
367 /*
368 * The mm counters are protected by its page_table_lock,
369 * so can be incremented directly.
370 */
371 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
372 #define get_mm_counter(mm, member) ((mm)->_##member)
373 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
374 #define inc_mm_counter(mm, member) (mm)->_##member++
375 #define dec_mm_counter(mm, member) (mm)->_##member--
376
377 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
378
379 #define get_mm_rss(mm) \
380 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
381 #define update_hiwater_rss(mm) do { \
382 unsigned long _rss = get_mm_rss(mm); \
383 if ((mm)->hiwater_rss < _rss) \
384 (mm)->hiwater_rss = _rss; \
385 } while (0)
386 #define update_hiwater_vm(mm) do { \
387 if ((mm)->hiwater_vm < (mm)->total_vm) \
388 (mm)->hiwater_vm = (mm)->total_vm; \
389 } while (0)
390
391 extern void set_dumpable(struct mm_struct *mm, int value);
392 extern int get_dumpable(struct mm_struct *mm);
393
394 /* mm flags */
395 /* dumpable bits */
396 #define MMF_DUMPABLE 0 /* core dump is permitted */
397 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
398 #define MMF_DUMPABLE_BITS 2
399
400 /* coredump filter bits */
401 #define MMF_DUMP_ANON_PRIVATE 2
402 #define MMF_DUMP_ANON_SHARED 3
403 #define MMF_DUMP_MAPPED_PRIVATE 4
404 #define MMF_DUMP_MAPPED_SHARED 5
405 #define MMF_DUMP_ELF_HEADERS 6
406 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
407 #define MMF_DUMP_FILTER_BITS 5
408 #define MMF_DUMP_FILTER_MASK \
409 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
410 #define MMF_DUMP_FILTER_DEFAULT \
411 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
412
413 struct sighand_struct {
414 atomic_t count;
415 struct k_sigaction action[_NSIG];
416 spinlock_t siglock;
417 wait_queue_head_t signalfd_wqh;
418 };
419
420 struct pacct_struct {
421 int ac_flag;
422 long ac_exitcode;
423 unsigned long ac_mem;
424 cputime_t ac_utime, ac_stime;
425 unsigned long ac_minflt, ac_majflt;
426 };
427
428 /*
429 * NOTE! "signal_struct" does not have it's own
430 * locking, because a shared signal_struct always
431 * implies a shared sighand_struct, so locking
432 * sighand_struct is always a proper superset of
433 * the locking of signal_struct.
434 */
435 struct signal_struct {
436 atomic_t count;
437 atomic_t live;
438
439 wait_queue_head_t wait_chldexit; /* for wait4() */
440
441 /* current thread group signal load-balancing target: */
442 struct task_struct *curr_target;
443
444 /* shared signal handling: */
445 struct sigpending shared_pending;
446
447 /* thread group exit support */
448 int group_exit_code;
449 /* overloaded:
450 * - notify group_exit_task when ->count is equal to notify_count
451 * - everyone except group_exit_task is stopped during signal delivery
452 * of fatal signals, group_exit_task processes the signal.
453 */
454 struct task_struct *group_exit_task;
455 int notify_count;
456
457 /* thread group stop support, overloads group_exit_code too */
458 int group_stop_count;
459 unsigned int flags; /* see SIGNAL_* flags below */
460
461 /* POSIX.1b Interval Timers */
462 struct list_head posix_timers;
463
464 /* ITIMER_REAL timer for the process */
465 struct hrtimer real_timer;
466 struct pid *leader_pid;
467 ktime_t it_real_incr;
468
469 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
470 cputime_t it_prof_expires, it_virt_expires;
471 cputime_t it_prof_incr, it_virt_incr;
472
473 /* job control IDs */
474
475 /*
476 * pgrp and session fields are deprecated.
477 * use the task_session_Xnr and task_pgrp_Xnr routines below
478 */
479
480 union {
481 pid_t pgrp __deprecated;
482 pid_t __pgrp;
483 };
484
485 struct pid *tty_old_pgrp;
486
487 union {
488 pid_t session __deprecated;
489 pid_t __session;
490 };
491
492 /* boolean value for session group leader */
493 int leader;
494
495 struct tty_struct *tty; /* NULL if no tty */
496
497 /*
498 * Cumulative resource counters for dead threads in the group,
499 * and for reaped dead child processes forked by this group.
500 * Live threads maintain their own counters and add to these
501 * in __exit_signal, except for the group leader.
502 */
503 cputime_t utime, stime, cutime, cstime;
504 cputime_t gtime;
505 cputime_t cgtime;
506 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
507 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
508 unsigned long inblock, oublock, cinblock, coublock;
509 #ifdef CONFIG_TASK_XACCT
510 u64 rchar, wchar, syscr, syscw;
511 #endif
512 struct task_io_accounting ioac;
513
514 /*
515 * Cumulative ns of scheduled CPU time for dead threads in the
516 * group, not including a zombie group leader. (This only differs
517 * from jiffies_to_ns(utime + stime) if sched_clock uses something
518 * other than jiffies.)
519 */
520 unsigned long long sum_sched_runtime;
521
522 /*
523 * We don't bother to synchronize most readers of this at all,
524 * because there is no reader checking a limit that actually needs
525 * to get both rlim_cur and rlim_max atomically, and either one
526 * alone is a single word that can safely be read normally.
527 * getrlimit/setrlimit use task_lock(current->group_leader) to
528 * protect this instead of the siglock, because they really
529 * have no need to disable irqs.
530 */
531 struct rlimit rlim[RLIM_NLIMITS];
532
533 struct list_head cpu_timers[3];
534
535 /* keep the process-shared keyrings here so that they do the right
536 * thing in threads created with CLONE_THREAD */
537 #ifdef CONFIG_KEYS
538 struct key *session_keyring; /* keyring inherited over fork */
539 struct key *process_keyring; /* keyring private to this process */
540 #endif
541 #ifdef CONFIG_BSD_PROCESS_ACCT
542 struct pacct_struct pacct; /* per-process accounting information */
543 #endif
544 #ifdef CONFIG_TASKSTATS
545 struct taskstats *stats;
546 #endif
547 #ifdef CONFIG_AUDIT
548 unsigned audit_tty;
549 struct tty_audit_buf *tty_audit_buf;
550 #endif
551 };
552
553 /* Context switch must be unlocked if interrupts are to be enabled */
554 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
555 # define __ARCH_WANT_UNLOCKED_CTXSW
556 #endif
557
558 /*
559 * Bits in flags field of signal_struct.
560 */
561 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
562 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
563 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
564 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
565 /*
566 * Pending notifications to parent.
567 */
568 #define SIGNAL_CLD_STOPPED 0x00000010
569 #define SIGNAL_CLD_CONTINUED 0x00000020
570 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
571
572 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
573
574 /* If true, all threads except ->group_exit_task have pending SIGKILL */
575 static inline int signal_group_exit(const struct signal_struct *sig)
576 {
577 return (sig->flags & SIGNAL_GROUP_EXIT) ||
578 (sig->group_exit_task != NULL);
579 }
580
581 /*
582 * Some day this will be a full-fledged user tracking system..
583 */
584 struct user_struct {
585 atomic_t __count; /* reference count */
586 atomic_t processes; /* How many processes does this user have? */
587 atomic_t files; /* How many open files does this user have? */
588 atomic_t sigpending; /* How many pending signals does this user have? */
589 #ifdef CONFIG_INOTIFY_USER
590 atomic_t inotify_watches; /* How many inotify watches does this user have? */
591 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
592 #endif
593 #ifdef CONFIG_POSIX_MQUEUE
594 /* protected by mq_lock */
595 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
596 #endif
597 unsigned long locked_shm; /* How many pages of mlocked shm ? */
598
599 #ifdef CONFIG_KEYS
600 struct key *uid_keyring; /* UID specific keyring */
601 struct key *session_keyring; /* UID's default session keyring */
602 #endif
603
604 /* Hash table maintenance information */
605 struct hlist_node uidhash_node;
606 uid_t uid;
607
608 #ifdef CONFIG_USER_SCHED
609 struct task_group *tg;
610 #ifdef CONFIG_SYSFS
611 struct kobject kobj;
612 struct work_struct work;
613 #endif
614 #endif
615 };
616
617 extern int uids_sysfs_init(void);
618
619 extern struct user_struct *find_user(uid_t);
620
621 extern struct user_struct root_user;
622 #define INIT_USER (&root_user)
623
624 struct backing_dev_info;
625 struct reclaim_state;
626
627 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
628 struct sched_info {
629 /* cumulative counters */
630 unsigned long pcount; /* # of times run on this cpu */
631 unsigned long long cpu_time, /* time spent on the cpu */
632 run_delay; /* time spent waiting on a runqueue */
633
634 /* timestamps */
635 unsigned long long last_arrival,/* when we last ran on a cpu */
636 last_queued; /* when we were last queued to run */
637 #ifdef CONFIG_SCHEDSTATS
638 /* BKL stats */
639 unsigned int bkl_count;
640 #endif
641 };
642 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
643
644 #ifdef CONFIG_SCHEDSTATS
645 extern const struct file_operations proc_schedstat_operations;
646 #endif /* CONFIG_SCHEDSTATS */
647
648 #ifdef CONFIG_TASK_DELAY_ACCT
649 struct task_delay_info {
650 spinlock_t lock;
651 unsigned int flags; /* Private per-task flags */
652
653 /* For each stat XXX, add following, aligned appropriately
654 *
655 * struct timespec XXX_start, XXX_end;
656 * u64 XXX_delay;
657 * u32 XXX_count;
658 *
659 * Atomicity of updates to XXX_delay, XXX_count protected by
660 * single lock above (split into XXX_lock if contention is an issue).
661 */
662
663 /*
664 * XXX_count is incremented on every XXX operation, the delay
665 * associated with the operation is added to XXX_delay.
666 * XXX_delay contains the accumulated delay time in nanoseconds.
667 */
668 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
669 u64 blkio_delay; /* wait for sync block io completion */
670 u64 swapin_delay; /* wait for swapin block io completion */
671 u32 blkio_count; /* total count of the number of sync block */
672 /* io operations performed */
673 u32 swapin_count; /* total count of the number of swapin block */
674 /* io operations performed */
675
676 struct timespec freepages_start, freepages_end;
677 u64 freepages_delay; /* wait for memory reclaim */
678 u32 freepages_count; /* total count of memory reclaim */
679 };
680 #endif /* CONFIG_TASK_DELAY_ACCT */
681
682 static inline int sched_info_on(void)
683 {
684 #ifdef CONFIG_SCHEDSTATS
685 return 1;
686 #elif defined(CONFIG_TASK_DELAY_ACCT)
687 extern int delayacct_on;
688 return delayacct_on;
689 #else
690 return 0;
691 #endif
692 }
693
694 enum cpu_idle_type {
695 CPU_IDLE,
696 CPU_NOT_IDLE,
697 CPU_NEWLY_IDLE,
698 CPU_MAX_IDLE_TYPES
699 };
700
701 /*
702 * sched-domains (multiprocessor balancing) declarations:
703 */
704
705 /*
706 * Increase resolution of nice-level calculations:
707 */
708 #define SCHED_LOAD_SHIFT 10
709 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
710
711 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
712
713 #ifdef CONFIG_SMP
714 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
715 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
716 #define SD_BALANCE_EXEC 4 /* Balance on exec */
717 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
718 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
719 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
720 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
721 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
722 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
723 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
724 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
725 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
726
727 #define BALANCE_FOR_MC_POWER \
728 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
729
730 #define BALANCE_FOR_PKG_POWER \
731 ((sched_mc_power_savings || sched_smt_power_savings) ? \
732 SD_POWERSAVINGS_BALANCE : 0)
733
734 #define test_sd_parent(sd, flag) ((sd->parent && \
735 (sd->parent->flags & flag)) ? 1 : 0)
736
737
738 struct sched_group {
739 struct sched_group *next; /* Must be a circular list */
740 cpumask_t cpumask;
741
742 /*
743 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
744 * single CPU. This is read only (except for setup, hotplug CPU).
745 * Note : Never change cpu_power without recompute its reciprocal
746 */
747 unsigned int __cpu_power;
748 /*
749 * reciprocal value of cpu_power to avoid expensive divides
750 * (see include/linux/reciprocal_div.h)
751 */
752 u32 reciprocal_cpu_power;
753 };
754
755 enum sched_domain_level {
756 SD_LV_NONE = 0,
757 SD_LV_SIBLING,
758 SD_LV_MC,
759 SD_LV_CPU,
760 SD_LV_NODE,
761 SD_LV_ALLNODES,
762 SD_LV_MAX
763 };
764
765 struct sched_domain_attr {
766 int relax_domain_level;
767 };
768
769 #define SD_ATTR_INIT (struct sched_domain_attr) { \
770 .relax_domain_level = -1, \
771 }
772
773 struct sched_domain {
774 /* These fields must be setup */
775 struct sched_domain *parent; /* top domain must be null terminated */
776 struct sched_domain *child; /* bottom domain must be null terminated */
777 struct sched_group *groups; /* the balancing groups of the domain */
778 cpumask_t span; /* span of all CPUs in this domain */
779 unsigned long min_interval; /* Minimum balance interval ms */
780 unsigned long max_interval; /* Maximum balance interval ms */
781 unsigned int busy_factor; /* less balancing by factor if busy */
782 unsigned int imbalance_pct; /* No balance until over watermark */
783 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
784 unsigned int busy_idx;
785 unsigned int idle_idx;
786 unsigned int newidle_idx;
787 unsigned int wake_idx;
788 unsigned int forkexec_idx;
789 int flags; /* See SD_* */
790 enum sched_domain_level level;
791
792 /* Runtime fields. */
793 unsigned long last_balance; /* init to jiffies. units in jiffies */
794 unsigned int balance_interval; /* initialise to 1. units in ms. */
795 unsigned int nr_balance_failed; /* initialise to 0 */
796
797 u64 last_update;
798
799 #ifdef CONFIG_SCHEDSTATS
800 /* load_balance() stats */
801 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
802 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
803 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
804 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
805 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
806 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
807 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
808 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
809
810 /* Active load balancing */
811 unsigned int alb_count;
812 unsigned int alb_failed;
813 unsigned int alb_pushed;
814
815 /* SD_BALANCE_EXEC stats */
816 unsigned int sbe_count;
817 unsigned int sbe_balanced;
818 unsigned int sbe_pushed;
819
820 /* SD_BALANCE_FORK stats */
821 unsigned int sbf_count;
822 unsigned int sbf_balanced;
823 unsigned int sbf_pushed;
824
825 /* try_to_wake_up() stats */
826 unsigned int ttwu_wake_remote;
827 unsigned int ttwu_move_affine;
828 unsigned int ttwu_move_balance;
829 #endif
830 };
831
832 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new,
833 struct sched_domain_attr *dattr_new);
834 extern int arch_reinit_sched_domains(void);
835
836 #else /* CONFIG_SMP */
837
838 struct sched_domain_attr;
839
840 static inline void
841 partition_sched_domains(int ndoms_new, cpumask_t *doms_new,
842 struct sched_domain_attr *dattr_new)
843 {
844 }
845 #endif /* !CONFIG_SMP */
846
847 struct io_context; /* See blkdev.h */
848 #define NGROUPS_SMALL 32
849 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
850 struct group_info {
851 int ngroups;
852 atomic_t usage;
853 gid_t small_block[NGROUPS_SMALL];
854 int nblocks;
855 gid_t *blocks[0];
856 };
857
858 /*
859 * get_group_info() must be called with the owning task locked (via task_lock())
860 * when task != current. The reason being that the vast majority of callers are
861 * looking at current->group_info, which can not be changed except by the
862 * current task. Changing current->group_info requires the task lock, too.
863 */
864 #define get_group_info(group_info) do { \
865 atomic_inc(&(group_info)->usage); \
866 } while (0)
867
868 #define put_group_info(group_info) do { \
869 if (atomic_dec_and_test(&(group_info)->usage)) \
870 groups_free(group_info); \
871 } while (0)
872
873 extern struct group_info *groups_alloc(int gidsetsize);
874 extern void groups_free(struct group_info *group_info);
875 extern int set_current_groups(struct group_info *group_info);
876 extern int groups_search(struct group_info *group_info, gid_t grp);
877 /* access the groups "array" with this macro */
878 #define GROUP_AT(gi, i) \
879 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
880
881 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
882 extern void prefetch_stack(struct task_struct *t);
883 #else
884 static inline void prefetch_stack(struct task_struct *t) { }
885 #endif
886
887 struct audit_context; /* See audit.c */
888 struct mempolicy;
889 struct pipe_inode_info;
890 struct uts_namespace;
891
892 struct rq;
893 struct sched_domain;
894
895 struct sched_class {
896 const struct sched_class *next;
897
898 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
899 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
900 void (*yield_task) (struct rq *rq);
901 int (*select_task_rq)(struct task_struct *p, int sync);
902
903 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
904
905 struct task_struct * (*pick_next_task) (struct rq *rq);
906 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
907
908 #ifdef CONFIG_SMP
909 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
910 struct rq *busiest, unsigned long max_load_move,
911 struct sched_domain *sd, enum cpu_idle_type idle,
912 int *all_pinned, int *this_best_prio);
913
914 int (*move_one_task) (struct rq *this_rq, int this_cpu,
915 struct rq *busiest, struct sched_domain *sd,
916 enum cpu_idle_type idle);
917 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
918 void (*post_schedule) (struct rq *this_rq);
919 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
920 #endif
921
922 void (*set_curr_task) (struct rq *rq);
923 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
924 void (*task_new) (struct rq *rq, struct task_struct *p);
925 void (*set_cpus_allowed)(struct task_struct *p,
926 const cpumask_t *newmask);
927
928 void (*rq_online)(struct rq *rq);
929 void (*rq_offline)(struct rq *rq);
930
931 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
932 int running);
933 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
934 int running);
935 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
936 int oldprio, int running);
937
938 #ifdef CONFIG_FAIR_GROUP_SCHED
939 void (*moved_group) (struct task_struct *p);
940 #endif
941 };
942
943 struct load_weight {
944 unsigned long weight, inv_weight;
945 };
946
947 /*
948 * CFS stats for a schedulable entity (task, task-group etc)
949 *
950 * Current field usage histogram:
951 *
952 * 4 se->block_start
953 * 4 se->run_node
954 * 4 se->sleep_start
955 * 6 se->load.weight
956 */
957 struct sched_entity {
958 struct load_weight load; /* for load-balancing */
959 struct rb_node run_node;
960 struct list_head group_node;
961 unsigned int on_rq;
962
963 u64 exec_start;
964 u64 sum_exec_runtime;
965 u64 vruntime;
966 u64 prev_sum_exec_runtime;
967
968 u64 last_wakeup;
969 u64 avg_overlap;
970
971 #ifdef CONFIG_SCHEDSTATS
972 u64 wait_start;
973 u64 wait_max;
974 u64 wait_count;
975 u64 wait_sum;
976
977 u64 sleep_start;
978 u64 sleep_max;
979 s64 sum_sleep_runtime;
980
981 u64 block_start;
982 u64 block_max;
983 u64 exec_max;
984 u64 slice_max;
985
986 u64 nr_migrations;
987 u64 nr_migrations_cold;
988 u64 nr_failed_migrations_affine;
989 u64 nr_failed_migrations_running;
990 u64 nr_failed_migrations_hot;
991 u64 nr_forced_migrations;
992 u64 nr_forced2_migrations;
993
994 u64 nr_wakeups;
995 u64 nr_wakeups_sync;
996 u64 nr_wakeups_migrate;
997 u64 nr_wakeups_local;
998 u64 nr_wakeups_remote;
999 u64 nr_wakeups_affine;
1000 u64 nr_wakeups_affine_attempts;
1001 u64 nr_wakeups_passive;
1002 u64 nr_wakeups_idle;
1003 #endif
1004
1005 #ifdef CONFIG_FAIR_GROUP_SCHED
1006 struct sched_entity *parent;
1007 /* rq on which this entity is (to be) queued: */
1008 struct cfs_rq *cfs_rq;
1009 /* rq "owned" by this entity/group: */
1010 struct cfs_rq *my_q;
1011 #endif
1012 };
1013
1014 struct sched_rt_entity {
1015 struct list_head run_list;
1016 unsigned int time_slice;
1017 unsigned long timeout;
1018 int nr_cpus_allowed;
1019
1020 struct sched_rt_entity *back;
1021 #ifdef CONFIG_RT_GROUP_SCHED
1022 struct sched_rt_entity *parent;
1023 /* rq on which this entity is (to be) queued: */
1024 struct rt_rq *rt_rq;
1025 /* rq "owned" by this entity/group: */
1026 struct rt_rq *my_q;
1027 #endif
1028 };
1029
1030 struct task_struct {
1031 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1032 void *stack;
1033 atomic_t usage;
1034 unsigned int flags; /* per process flags, defined below */
1035 unsigned int ptrace;
1036
1037 int lock_depth; /* BKL lock depth */
1038
1039 #ifdef CONFIG_SMP
1040 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1041 int oncpu;
1042 #endif
1043 #endif
1044
1045 int prio, static_prio, normal_prio;
1046 unsigned int rt_priority;
1047 const struct sched_class *sched_class;
1048 struct sched_entity se;
1049 struct sched_rt_entity rt;
1050
1051 #ifdef CONFIG_PREEMPT_NOTIFIERS
1052 /* list of struct preempt_notifier: */
1053 struct hlist_head preempt_notifiers;
1054 #endif
1055
1056 /*
1057 * fpu_counter contains the number of consecutive context switches
1058 * that the FPU is used. If this is over a threshold, the lazy fpu
1059 * saving becomes unlazy to save the trap. This is an unsigned char
1060 * so that after 256 times the counter wraps and the behavior turns
1061 * lazy again; this to deal with bursty apps that only use FPU for
1062 * a short time
1063 */
1064 unsigned char fpu_counter;
1065 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1066 #ifdef CONFIG_BLK_DEV_IO_TRACE
1067 unsigned int btrace_seq;
1068 #endif
1069
1070 unsigned int policy;
1071 cpumask_t cpus_allowed;
1072
1073 #ifdef CONFIG_PREEMPT_RCU
1074 int rcu_read_lock_nesting;
1075 int rcu_flipctr_idx;
1076 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1077
1078 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1079 struct sched_info sched_info;
1080 #endif
1081
1082 struct list_head tasks;
1083
1084 struct mm_struct *mm, *active_mm;
1085
1086 /* task state */
1087 struct linux_binfmt *binfmt;
1088 int exit_state;
1089 int exit_code, exit_signal;
1090 int pdeath_signal; /* The signal sent when the parent dies */
1091 /* ??? */
1092 unsigned int personality;
1093 unsigned did_exec:1;
1094 pid_t pid;
1095 pid_t tgid;
1096
1097 #ifdef CONFIG_CC_STACKPROTECTOR
1098 /* Canary value for the -fstack-protector gcc feature */
1099 unsigned long stack_canary;
1100 #endif
1101 /*
1102 * pointers to (original) parent process, youngest child, younger sibling,
1103 * older sibling, respectively. (p->father can be replaced with
1104 * p->real_parent->pid)
1105 */
1106 struct task_struct *real_parent; /* real parent process */
1107 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1108 /*
1109 * children/sibling forms the list of my natural children
1110 */
1111 struct list_head children; /* list of my children */
1112 struct list_head sibling; /* linkage in my parent's children list */
1113 struct task_struct *group_leader; /* threadgroup leader */
1114
1115 /*
1116 * ptraced is the list of tasks this task is using ptrace on.
1117 * This includes both natural children and PTRACE_ATTACH targets.
1118 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1119 */
1120 struct list_head ptraced;
1121 struct list_head ptrace_entry;
1122
1123 /* PID/PID hash table linkage. */
1124 struct pid_link pids[PIDTYPE_MAX];
1125 struct list_head thread_group;
1126
1127 struct completion *vfork_done; /* for vfork() */
1128 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1129 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1130
1131 cputime_t utime, stime, utimescaled, stimescaled;
1132 cputime_t gtime;
1133 cputime_t prev_utime, prev_stime;
1134 unsigned long nvcsw, nivcsw; /* context switch counts */
1135 struct timespec start_time; /* monotonic time */
1136 struct timespec real_start_time; /* boot based time */
1137 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1138 unsigned long min_flt, maj_flt;
1139
1140 cputime_t it_prof_expires, it_virt_expires;
1141 unsigned long long it_sched_expires;
1142 struct list_head cpu_timers[3];
1143
1144 /* process credentials */
1145 uid_t uid,euid,suid,fsuid;
1146 gid_t gid,egid,sgid,fsgid;
1147 struct group_info *group_info;
1148 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset;
1149 struct user_struct *user;
1150 unsigned securebits;
1151 #ifdef CONFIG_KEYS
1152 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1153 struct key *request_key_auth; /* assumed request_key authority */
1154 struct key *thread_keyring; /* keyring private to this thread */
1155 #endif
1156 char comm[TASK_COMM_LEN]; /* executable name excluding path
1157 - access with [gs]et_task_comm (which lock
1158 it with task_lock())
1159 - initialized normally by flush_old_exec */
1160 /* file system info */
1161 int link_count, total_link_count;
1162 #ifdef CONFIG_SYSVIPC
1163 /* ipc stuff */
1164 struct sysv_sem sysvsem;
1165 #endif
1166 #ifdef CONFIG_DETECT_SOFTLOCKUP
1167 /* hung task detection */
1168 unsigned long last_switch_timestamp;
1169 unsigned long last_switch_count;
1170 #endif
1171 /* CPU-specific state of this task */
1172 struct thread_struct thread;
1173 /* filesystem information */
1174 struct fs_struct *fs;
1175 /* open file information */
1176 struct files_struct *files;
1177 /* namespaces */
1178 struct nsproxy *nsproxy;
1179 /* signal handlers */
1180 struct signal_struct *signal;
1181 struct sighand_struct *sighand;
1182
1183 sigset_t blocked, real_blocked;
1184 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1185 struct sigpending pending;
1186
1187 unsigned long sas_ss_sp;
1188 size_t sas_ss_size;
1189 int (*notifier)(void *priv);
1190 void *notifier_data;
1191 sigset_t *notifier_mask;
1192 #ifdef CONFIG_SECURITY
1193 void *security;
1194 #endif
1195 struct audit_context *audit_context;
1196 #ifdef CONFIG_AUDITSYSCALL
1197 uid_t loginuid;
1198 unsigned int sessionid;
1199 #endif
1200 seccomp_t seccomp;
1201
1202 /* Thread group tracking */
1203 u32 parent_exec_id;
1204 u32 self_exec_id;
1205 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1206 spinlock_t alloc_lock;
1207
1208 /* Protection of the PI data structures: */
1209 spinlock_t pi_lock;
1210
1211 #ifdef CONFIG_RT_MUTEXES
1212 /* PI waiters blocked on a rt_mutex held by this task */
1213 struct plist_head pi_waiters;
1214 /* Deadlock detection and priority inheritance handling */
1215 struct rt_mutex_waiter *pi_blocked_on;
1216 #endif
1217
1218 #ifdef CONFIG_DEBUG_MUTEXES
1219 /* mutex deadlock detection */
1220 struct mutex_waiter *blocked_on;
1221 #endif
1222 #ifdef CONFIG_TRACE_IRQFLAGS
1223 unsigned int irq_events;
1224 int hardirqs_enabled;
1225 unsigned long hardirq_enable_ip;
1226 unsigned int hardirq_enable_event;
1227 unsigned long hardirq_disable_ip;
1228 unsigned int hardirq_disable_event;
1229 int softirqs_enabled;
1230 unsigned long softirq_disable_ip;
1231 unsigned int softirq_disable_event;
1232 unsigned long softirq_enable_ip;
1233 unsigned int softirq_enable_event;
1234 int hardirq_context;
1235 int softirq_context;
1236 #endif
1237 #ifdef CONFIG_LOCKDEP
1238 # define MAX_LOCK_DEPTH 48UL
1239 u64 curr_chain_key;
1240 int lockdep_depth;
1241 unsigned int lockdep_recursion;
1242 struct held_lock held_locks[MAX_LOCK_DEPTH];
1243 #endif
1244
1245 /* journalling filesystem info */
1246 void *journal_info;
1247
1248 /* stacked block device info */
1249 struct bio *bio_list, **bio_tail;
1250
1251 /* VM state */
1252 struct reclaim_state *reclaim_state;
1253
1254 struct backing_dev_info *backing_dev_info;
1255
1256 struct io_context *io_context;
1257
1258 unsigned long ptrace_message;
1259 siginfo_t *last_siginfo; /* For ptrace use. */
1260 #ifdef CONFIG_TASK_XACCT
1261 /* i/o counters(bytes read/written, #syscalls */
1262 u64 rchar, wchar, syscr, syscw;
1263 #endif
1264 struct task_io_accounting ioac;
1265 #if defined(CONFIG_TASK_XACCT)
1266 u64 acct_rss_mem1; /* accumulated rss usage */
1267 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1268 cputime_t acct_timexpd; /* stime + utime since last update */
1269 #endif
1270 #ifdef CONFIG_CPUSETS
1271 nodemask_t mems_allowed;
1272 int cpuset_mems_generation;
1273 int cpuset_mem_spread_rotor;
1274 #endif
1275 #ifdef CONFIG_CGROUPS
1276 /* Control Group info protected by css_set_lock */
1277 struct css_set *cgroups;
1278 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1279 struct list_head cg_list;
1280 #endif
1281 #ifdef CONFIG_FUTEX
1282 struct robust_list_head __user *robust_list;
1283 #ifdef CONFIG_COMPAT
1284 struct compat_robust_list_head __user *compat_robust_list;
1285 #endif
1286 struct list_head pi_state_list;
1287 struct futex_pi_state *pi_state_cache;
1288 #endif
1289 #ifdef CONFIG_NUMA
1290 struct mempolicy *mempolicy;
1291 short il_next;
1292 #endif
1293 atomic_t fs_excl; /* holding fs exclusive resources */
1294 struct rcu_head rcu;
1295
1296 /*
1297 * cache last used pipe for splice
1298 */
1299 struct pipe_inode_info *splice_pipe;
1300 #ifdef CONFIG_TASK_DELAY_ACCT
1301 struct task_delay_info *delays;
1302 #endif
1303 #ifdef CONFIG_FAULT_INJECTION
1304 int make_it_fail;
1305 #endif
1306 struct prop_local_single dirties;
1307 #ifdef CONFIG_LATENCYTOP
1308 int latency_record_count;
1309 struct latency_record latency_record[LT_SAVECOUNT];
1310 #endif
1311 };
1312
1313 /*
1314 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1315 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1316 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1317 * values are inverted: lower p->prio value means higher priority.
1318 *
1319 * The MAX_USER_RT_PRIO value allows the actual maximum
1320 * RT priority to be separate from the value exported to
1321 * user-space. This allows kernel threads to set their
1322 * priority to a value higher than any user task. Note:
1323 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1324 */
1325
1326 #define MAX_USER_RT_PRIO 100
1327 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1328
1329 #define MAX_PRIO (MAX_RT_PRIO + 40)
1330 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1331
1332 static inline int rt_prio(int prio)
1333 {
1334 if (unlikely(prio < MAX_RT_PRIO))
1335 return 1;
1336 return 0;
1337 }
1338
1339 static inline int rt_task(struct task_struct *p)
1340 {
1341 return rt_prio(p->prio);
1342 }
1343
1344 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1345 {
1346 tsk->signal->__session = session;
1347 }
1348
1349 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1350 {
1351 tsk->signal->__pgrp = pgrp;
1352 }
1353
1354 static inline struct pid *task_pid(struct task_struct *task)
1355 {
1356 return task->pids[PIDTYPE_PID].pid;
1357 }
1358
1359 static inline struct pid *task_tgid(struct task_struct *task)
1360 {
1361 return task->group_leader->pids[PIDTYPE_PID].pid;
1362 }
1363
1364 static inline struct pid *task_pgrp(struct task_struct *task)
1365 {
1366 return task->group_leader->pids[PIDTYPE_PGID].pid;
1367 }
1368
1369 static inline struct pid *task_session(struct task_struct *task)
1370 {
1371 return task->group_leader->pids[PIDTYPE_SID].pid;
1372 }
1373
1374 struct pid_namespace;
1375
1376 /*
1377 * the helpers to get the task's different pids as they are seen
1378 * from various namespaces
1379 *
1380 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1381 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1382 * current.
1383 * task_xid_nr_ns() : id seen from the ns specified;
1384 *
1385 * set_task_vxid() : assigns a virtual id to a task;
1386 *
1387 * see also pid_nr() etc in include/linux/pid.h
1388 */
1389
1390 static inline pid_t task_pid_nr(struct task_struct *tsk)
1391 {
1392 return tsk->pid;
1393 }
1394
1395 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1396
1397 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1398 {
1399 return pid_vnr(task_pid(tsk));
1400 }
1401
1402
1403 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1404 {
1405 return tsk->tgid;
1406 }
1407
1408 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1409
1410 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1411 {
1412 return pid_vnr(task_tgid(tsk));
1413 }
1414
1415
1416 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1417 {
1418 return tsk->signal->__pgrp;
1419 }
1420
1421 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1422
1423 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1424 {
1425 return pid_vnr(task_pgrp(tsk));
1426 }
1427
1428
1429 static inline pid_t task_session_nr(struct task_struct *tsk)
1430 {
1431 return tsk->signal->__session;
1432 }
1433
1434 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1435
1436 static inline pid_t task_session_vnr(struct task_struct *tsk)
1437 {
1438 return pid_vnr(task_session(tsk));
1439 }
1440
1441
1442 /**
1443 * pid_alive - check that a task structure is not stale
1444 * @p: Task structure to be checked.
1445 *
1446 * Test if a process is not yet dead (at most zombie state)
1447 * If pid_alive fails, then pointers within the task structure
1448 * can be stale and must not be dereferenced.
1449 */
1450 static inline int pid_alive(struct task_struct *p)
1451 {
1452 return p->pids[PIDTYPE_PID].pid != NULL;
1453 }
1454
1455 /**
1456 * is_global_init - check if a task structure is init
1457 * @tsk: Task structure to be checked.
1458 *
1459 * Check if a task structure is the first user space task the kernel created.
1460 */
1461 static inline int is_global_init(struct task_struct *tsk)
1462 {
1463 return tsk->pid == 1;
1464 }
1465
1466 /*
1467 * is_container_init:
1468 * check whether in the task is init in its own pid namespace.
1469 */
1470 extern int is_container_init(struct task_struct *tsk);
1471
1472 extern struct pid *cad_pid;
1473
1474 extern void free_task(struct task_struct *tsk);
1475 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1476
1477 extern void __put_task_struct(struct task_struct *t);
1478
1479 static inline void put_task_struct(struct task_struct *t)
1480 {
1481 if (atomic_dec_and_test(&t->usage))
1482 __put_task_struct(t);
1483 }
1484
1485 /*
1486 * Per process flags
1487 */
1488 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1489 /* Not implemented yet, only for 486*/
1490 #define PF_STARTING 0x00000002 /* being created */
1491 #define PF_EXITING 0x00000004 /* getting shut down */
1492 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1493 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1494 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1495 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1496 #define PF_DUMPCORE 0x00000200 /* dumped core */
1497 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1498 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1499 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1500 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1501 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1502 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1503 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1504 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1505 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1506 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1507 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1508 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1509 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1510 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1511 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1512 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1513 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1514 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1515 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1516 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1517
1518 /*
1519 * Only the _current_ task can read/write to tsk->flags, but other
1520 * tasks can access tsk->flags in readonly mode for example
1521 * with tsk_used_math (like during threaded core dumping).
1522 * There is however an exception to this rule during ptrace
1523 * or during fork: the ptracer task is allowed to write to the
1524 * child->flags of its traced child (same goes for fork, the parent
1525 * can write to the child->flags), because we're guaranteed the
1526 * child is not running and in turn not changing child->flags
1527 * at the same time the parent does it.
1528 */
1529 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1530 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1531 #define clear_used_math() clear_stopped_child_used_math(current)
1532 #define set_used_math() set_stopped_child_used_math(current)
1533 #define conditional_stopped_child_used_math(condition, child) \
1534 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1535 #define conditional_used_math(condition) \
1536 conditional_stopped_child_used_math(condition, current)
1537 #define copy_to_stopped_child_used_math(child) \
1538 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1539 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1540 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1541 #define used_math() tsk_used_math(current)
1542
1543 #ifdef CONFIG_SMP
1544 extern int set_cpus_allowed_ptr(struct task_struct *p,
1545 const cpumask_t *new_mask);
1546 #else
1547 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1548 const cpumask_t *new_mask)
1549 {
1550 if (!cpu_isset(0, *new_mask))
1551 return -EINVAL;
1552 return 0;
1553 }
1554 #endif
1555 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1556 {
1557 return set_cpus_allowed_ptr(p, &new_mask);
1558 }
1559
1560 extern unsigned long long sched_clock(void);
1561
1562 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1563 static inline void sched_clock_init(void)
1564 {
1565 }
1566
1567 static inline u64 sched_clock_cpu(int cpu)
1568 {
1569 return sched_clock();
1570 }
1571
1572 static inline void sched_clock_tick(void)
1573 {
1574 }
1575
1576 static inline void sched_clock_idle_sleep_event(void)
1577 {
1578 }
1579
1580 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1581 {
1582 }
1583
1584 #ifdef CONFIG_NO_HZ
1585 static inline void sched_clock_tick_stop(int cpu)
1586 {
1587 }
1588
1589 static inline void sched_clock_tick_start(int cpu)
1590 {
1591 }
1592 #endif
1593
1594 #else /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */
1595 extern void sched_clock_init(void);
1596 extern u64 sched_clock_cpu(int cpu);
1597 extern void sched_clock_tick(void);
1598 extern void sched_clock_idle_sleep_event(void);
1599 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1600 #ifdef CONFIG_NO_HZ
1601 extern void sched_clock_tick_stop(int cpu);
1602 extern void sched_clock_tick_start(int cpu);
1603 #endif
1604 #endif /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */
1605
1606 /*
1607 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1608 * clock constructed from sched_clock():
1609 */
1610 extern unsigned long long cpu_clock(int cpu);
1611
1612 extern unsigned long long
1613 task_sched_runtime(struct task_struct *task);
1614
1615 /* sched_exec is called by processes performing an exec */
1616 #ifdef CONFIG_SMP
1617 extern void sched_exec(void);
1618 #else
1619 #define sched_exec() {}
1620 #endif
1621
1622 extern void sched_clock_idle_sleep_event(void);
1623 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1624
1625 #ifdef CONFIG_HOTPLUG_CPU
1626 extern void idle_task_exit(void);
1627 #else
1628 static inline void idle_task_exit(void) {}
1629 #endif
1630
1631 extern void sched_idle_next(void);
1632
1633 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1634 extern void wake_up_idle_cpu(int cpu);
1635 #else
1636 static inline void wake_up_idle_cpu(int cpu) { }
1637 #endif
1638
1639 #ifdef CONFIG_SCHED_DEBUG
1640 extern unsigned int sysctl_sched_latency;
1641 extern unsigned int sysctl_sched_min_granularity;
1642 extern unsigned int sysctl_sched_wakeup_granularity;
1643 extern unsigned int sysctl_sched_child_runs_first;
1644 extern unsigned int sysctl_sched_features;
1645 extern unsigned int sysctl_sched_migration_cost;
1646 extern unsigned int sysctl_sched_nr_migrate;
1647 extern unsigned int sysctl_sched_shares_ratelimit;
1648
1649 int sched_nr_latency_handler(struct ctl_table *table, int write,
1650 struct file *file, void __user *buffer, size_t *length,
1651 loff_t *ppos);
1652 #endif
1653 extern unsigned int sysctl_sched_rt_period;
1654 extern int sysctl_sched_rt_runtime;
1655
1656 int sched_rt_handler(struct ctl_table *table, int write,
1657 struct file *filp, void __user *buffer, size_t *lenp,
1658 loff_t *ppos);
1659
1660 extern unsigned int sysctl_sched_compat_yield;
1661
1662 #ifdef CONFIG_RT_MUTEXES
1663 extern int rt_mutex_getprio(struct task_struct *p);
1664 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1665 extern void rt_mutex_adjust_pi(struct task_struct *p);
1666 #else
1667 static inline int rt_mutex_getprio(struct task_struct *p)
1668 {
1669 return p->normal_prio;
1670 }
1671 # define rt_mutex_adjust_pi(p) do { } while (0)
1672 #endif
1673
1674 extern void set_user_nice(struct task_struct *p, long nice);
1675 extern int task_prio(const struct task_struct *p);
1676 extern int task_nice(const struct task_struct *p);
1677 extern int can_nice(const struct task_struct *p, const int nice);
1678 extern int task_curr(const struct task_struct *p);
1679 extern int idle_cpu(int cpu);
1680 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1681 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1682 struct sched_param *);
1683 extern struct task_struct *idle_task(int cpu);
1684 extern struct task_struct *curr_task(int cpu);
1685 extern void set_curr_task(int cpu, struct task_struct *p);
1686
1687 void yield(void);
1688
1689 /*
1690 * The default (Linux) execution domain.
1691 */
1692 extern struct exec_domain default_exec_domain;
1693
1694 union thread_union {
1695 struct thread_info thread_info;
1696 unsigned long stack[THREAD_SIZE/sizeof(long)];
1697 };
1698
1699 #ifndef __HAVE_ARCH_KSTACK_END
1700 static inline int kstack_end(void *addr)
1701 {
1702 /* Reliable end of stack detection:
1703 * Some APM bios versions misalign the stack
1704 */
1705 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1706 }
1707 #endif
1708
1709 extern union thread_union init_thread_union;
1710 extern struct task_struct init_task;
1711
1712 extern struct mm_struct init_mm;
1713
1714 extern struct pid_namespace init_pid_ns;
1715
1716 /*
1717 * find a task by one of its numerical ids
1718 *
1719 * find_task_by_pid_type_ns():
1720 * it is the most generic call - it finds a task by all id,
1721 * type and namespace specified
1722 * find_task_by_pid_ns():
1723 * finds a task by its pid in the specified namespace
1724 * find_task_by_vpid():
1725 * finds a task by its virtual pid
1726 *
1727 * see also find_vpid() etc in include/linux/pid.h
1728 */
1729
1730 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1731 struct pid_namespace *ns);
1732
1733 extern struct task_struct *find_task_by_vpid(pid_t nr);
1734 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1735 struct pid_namespace *ns);
1736
1737 extern void __set_special_pids(struct pid *pid);
1738
1739 /* per-UID process charging. */
1740 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1741 static inline struct user_struct *get_uid(struct user_struct *u)
1742 {
1743 atomic_inc(&u->__count);
1744 return u;
1745 }
1746 extern void free_uid(struct user_struct *);
1747 extern void switch_uid(struct user_struct *);
1748 extern void release_uids(struct user_namespace *ns);
1749
1750 #include <asm/current.h>
1751
1752 extern void do_timer(unsigned long ticks);
1753
1754 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1755 extern int wake_up_process(struct task_struct *tsk);
1756 extern void wake_up_new_task(struct task_struct *tsk,
1757 unsigned long clone_flags);
1758 #ifdef CONFIG_SMP
1759 extern void kick_process(struct task_struct *tsk);
1760 #else
1761 static inline void kick_process(struct task_struct *tsk) { }
1762 #endif
1763 extern void sched_fork(struct task_struct *p, int clone_flags);
1764 extern void sched_dead(struct task_struct *p);
1765
1766 extern int in_group_p(gid_t);
1767 extern int in_egroup_p(gid_t);
1768
1769 extern void proc_caches_init(void);
1770 extern void flush_signals(struct task_struct *);
1771 extern void ignore_signals(struct task_struct *);
1772 extern void flush_signal_handlers(struct task_struct *, int force_default);
1773 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1774
1775 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1776 {
1777 unsigned long flags;
1778 int ret;
1779
1780 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1781 ret = dequeue_signal(tsk, mask, info);
1782 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1783
1784 return ret;
1785 }
1786
1787 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1788 sigset_t *mask);
1789 extern void unblock_all_signals(void);
1790 extern void release_task(struct task_struct * p);
1791 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1792 extern int force_sigsegv(int, struct task_struct *);
1793 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1794 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1795 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1796 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1797 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1798 extern int kill_pid(struct pid *pid, int sig, int priv);
1799 extern int kill_proc_info(int, struct siginfo *, pid_t);
1800 extern void do_notify_parent(struct task_struct *, int);
1801 extern void force_sig(int, struct task_struct *);
1802 extern void force_sig_specific(int, struct task_struct *);
1803 extern int send_sig(int, struct task_struct *, int);
1804 extern void zap_other_threads(struct task_struct *p);
1805 extern struct sigqueue *sigqueue_alloc(void);
1806 extern void sigqueue_free(struct sigqueue *);
1807 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1808 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1809 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1810
1811 static inline int kill_cad_pid(int sig, int priv)
1812 {
1813 return kill_pid(cad_pid, sig, priv);
1814 }
1815
1816 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1817 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1818 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1819 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1820
1821 static inline int is_si_special(const struct siginfo *info)
1822 {
1823 return info <= SEND_SIG_FORCED;
1824 }
1825
1826 /* True if we are on the alternate signal stack. */
1827
1828 static inline int on_sig_stack(unsigned long sp)
1829 {
1830 return (sp - current->sas_ss_sp < current->sas_ss_size);
1831 }
1832
1833 static inline int sas_ss_flags(unsigned long sp)
1834 {
1835 return (current->sas_ss_size == 0 ? SS_DISABLE
1836 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1837 }
1838
1839 /*
1840 * Routines for handling mm_structs
1841 */
1842 extern struct mm_struct * mm_alloc(void);
1843
1844 /* mmdrop drops the mm and the page tables */
1845 extern void __mmdrop(struct mm_struct *);
1846 static inline void mmdrop(struct mm_struct * mm)
1847 {
1848 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1849 __mmdrop(mm);
1850 }
1851
1852 /* mmput gets rid of the mappings and all user-space */
1853 extern void mmput(struct mm_struct *);
1854 /* Grab a reference to a task's mm, if it is not already going away */
1855 extern struct mm_struct *get_task_mm(struct task_struct *task);
1856 /* Remove the current tasks stale references to the old mm_struct */
1857 extern void mm_release(struct task_struct *, struct mm_struct *);
1858 /* Allocate a new mm structure and copy contents from tsk->mm */
1859 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1860
1861 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1862 extern void flush_thread(void);
1863 extern void exit_thread(void);
1864
1865 extern void exit_files(struct task_struct *);
1866 extern void __cleanup_signal(struct signal_struct *);
1867 extern void __cleanup_sighand(struct sighand_struct *);
1868
1869 extern void exit_itimers(struct signal_struct *);
1870 extern void flush_itimer_signals(void);
1871
1872 extern NORET_TYPE void do_group_exit(int);
1873
1874 extern void daemonize(const char *, ...);
1875 extern int allow_signal(int);
1876 extern int disallow_signal(int);
1877
1878 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1879 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1880 struct task_struct *fork_idle(int);
1881
1882 extern void set_task_comm(struct task_struct *tsk, char *from);
1883 extern char *get_task_comm(char *to, struct task_struct *tsk);
1884
1885 #ifdef CONFIG_SMP
1886 extern void wait_task_inactive(struct task_struct * p);
1887 #else
1888 #define wait_task_inactive(p) do { } while (0)
1889 #endif
1890
1891 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1892
1893 #define for_each_process(p) \
1894 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1895
1896 /*
1897 * Careful: do_each_thread/while_each_thread is a double loop so
1898 * 'break' will not work as expected - use goto instead.
1899 */
1900 #define do_each_thread(g, t) \
1901 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1902
1903 #define while_each_thread(g, t) \
1904 while ((t = next_thread(t)) != g)
1905
1906 /* de_thread depends on thread_group_leader not being a pid based check */
1907 #define thread_group_leader(p) (p == p->group_leader)
1908
1909 /* Do to the insanities of de_thread it is possible for a process
1910 * to have the pid of the thread group leader without actually being
1911 * the thread group leader. For iteration through the pids in proc
1912 * all we care about is that we have a task with the appropriate
1913 * pid, we don't actually care if we have the right task.
1914 */
1915 static inline int has_group_leader_pid(struct task_struct *p)
1916 {
1917 return p->pid == p->tgid;
1918 }
1919
1920 static inline
1921 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1922 {
1923 return p1->tgid == p2->tgid;
1924 }
1925
1926 static inline struct task_struct *next_thread(const struct task_struct *p)
1927 {
1928 return list_entry(rcu_dereference(p->thread_group.next),
1929 struct task_struct, thread_group);
1930 }
1931
1932 static inline int thread_group_empty(struct task_struct *p)
1933 {
1934 return list_empty(&p->thread_group);
1935 }
1936
1937 #define delay_group_leader(p) \
1938 (thread_group_leader(p) && !thread_group_empty(p))
1939
1940 /*
1941 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1942 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1943 * pins the final release of task.io_context. Also protects ->cpuset and
1944 * ->cgroup.subsys[].
1945 *
1946 * Nests both inside and outside of read_lock(&tasklist_lock).
1947 * It must not be nested with write_lock_irq(&tasklist_lock),
1948 * neither inside nor outside.
1949 */
1950 static inline void task_lock(struct task_struct *p)
1951 {
1952 spin_lock(&p->alloc_lock);
1953 }
1954
1955 static inline void task_unlock(struct task_struct *p)
1956 {
1957 spin_unlock(&p->alloc_lock);
1958 }
1959
1960 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1961 unsigned long *flags);
1962
1963 static inline void unlock_task_sighand(struct task_struct *tsk,
1964 unsigned long *flags)
1965 {
1966 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1967 }
1968
1969 #ifndef __HAVE_THREAD_FUNCTIONS
1970
1971 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1972 #define task_stack_page(task) ((task)->stack)
1973
1974 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1975 {
1976 *task_thread_info(p) = *task_thread_info(org);
1977 task_thread_info(p)->task = p;
1978 }
1979
1980 static inline unsigned long *end_of_stack(struct task_struct *p)
1981 {
1982 return (unsigned long *)(task_thread_info(p) + 1);
1983 }
1984
1985 #endif
1986
1987 static inline int object_is_on_stack(void *obj)
1988 {
1989 void *stack = task_stack_page(current);
1990
1991 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
1992 }
1993
1994 extern void thread_info_cache_init(void);
1995
1996 /* set thread flags in other task's structures
1997 * - see asm/thread_info.h for TIF_xxxx flags available
1998 */
1999 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2000 {
2001 set_ti_thread_flag(task_thread_info(tsk), flag);
2002 }
2003
2004 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2005 {
2006 clear_ti_thread_flag(task_thread_info(tsk), flag);
2007 }
2008
2009 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2010 {
2011 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2012 }
2013
2014 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2015 {
2016 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2017 }
2018
2019 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2020 {
2021 return test_ti_thread_flag(task_thread_info(tsk), flag);
2022 }
2023
2024 static inline void set_tsk_need_resched(struct task_struct *tsk)
2025 {
2026 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2027 }
2028
2029 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2030 {
2031 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2032 }
2033
2034 static inline int test_tsk_need_resched(struct task_struct *tsk)
2035 {
2036 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2037 }
2038
2039 static inline int signal_pending(struct task_struct *p)
2040 {
2041 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2042 }
2043
2044 extern int __fatal_signal_pending(struct task_struct *p);
2045
2046 static inline int fatal_signal_pending(struct task_struct *p)
2047 {
2048 return signal_pending(p) && __fatal_signal_pending(p);
2049 }
2050
2051 static inline int signal_pending_state(long state, struct task_struct *p)
2052 {
2053 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2054 return 0;
2055 if (!signal_pending(p))
2056 return 0;
2057
2058 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2059 }
2060
2061 static inline int need_resched(void)
2062 {
2063 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2064 }
2065
2066 /*
2067 * cond_resched() and cond_resched_lock(): latency reduction via
2068 * explicit rescheduling in places that are safe. The return
2069 * value indicates whether a reschedule was done in fact.
2070 * cond_resched_lock() will drop the spinlock before scheduling,
2071 * cond_resched_softirq() will enable bhs before scheduling.
2072 */
2073 extern int _cond_resched(void);
2074 #ifdef CONFIG_PREEMPT_BKL
2075 static inline int cond_resched(void)
2076 {
2077 return 0;
2078 }
2079 #else
2080 static inline int cond_resched(void)
2081 {
2082 return _cond_resched();
2083 }
2084 #endif
2085 extern int cond_resched_lock(spinlock_t * lock);
2086 extern int cond_resched_softirq(void);
2087 static inline int cond_resched_bkl(void)
2088 {
2089 return _cond_resched();
2090 }
2091
2092 /*
2093 * Does a critical section need to be broken due to another
2094 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2095 * but a general need for low latency)
2096 */
2097 static inline int spin_needbreak(spinlock_t *lock)
2098 {
2099 #ifdef CONFIG_PREEMPT
2100 return spin_is_contended(lock);
2101 #else
2102 return 0;
2103 #endif
2104 }
2105
2106 /*
2107 * Reevaluate whether the task has signals pending delivery.
2108 * Wake the task if so.
2109 * This is required every time the blocked sigset_t changes.
2110 * callers must hold sighand->siglock.
2111 */
2112 extern void recalc_sigpending_and_wake(struct task_struct *t);
2113 extern void recalc_sigpending(void);
2114
2115 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2116
2117 /*
2118 * Wrappers for p->thread_info->cpu access. No-op on UP.
2119 */
2120 #ifdef CONFIG_SMP
2121
2122 static inline unsigned int task_cpu(const struct task_struct *p)
2123 {
2124 return task_thread_info(p)->cpu;
2125 }
2126
2127 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2128
2129 #else
2130
2131 static inline unsigned int task_cpu(const struct task_struct *p)
2132 {
2133 return 0;
2134 }
2135
2136 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2137 {
2138 }
2139
2140 #endif /* CONFIG_SMP */
2141
2142 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2143 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2144 #else
2145 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
2146 {
2147 mm->mmap_base = TASK_UNMAPPED_BASE;
2148 mm->get_unmapped_area = arch_get_unmapped_area;
2149 mm->unmap_area = arch_unmap_area;
2150 }
2151 #endif
2152
2153 #ifdef CONFIG_TRACING
2154 extern void
2155 __trace_special(void *__tr, void *__data,
2156 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2157 #else
2158 static inline void
2159 __trace_special(void *__tr, void *__data,
2160 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2161 {
2162 }
2163 #endif
2164
2165 extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask);
2166 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2167
2168 extern int sched_mc_power_savings, sched_smt_power_savings;
2169
2170 extern void normalize_rt_tasks(void);
2171
2172 #ifdef CONFIG_GROUP_SCHED
2173
2174 extern struct task_group init_task_group;
2175 #ifdef CONFIG_USER_SCHED
2176 extern struct task_group root_task_group;
2177 #endif
2178
2179 extern struct task_group *sched_create_group(struct task_group *parent);
2180 extern void sched_destroy_group(struct task_group *tg);
2181 extern void sched_move_task(struct task_struct *tsk);
2182 #ifdef CONFIG_FAIR_GROUP_SCHED
2183 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2184 extern unsigned long sched_group_shares(struct task_group *tg);
2185 #endif
2186 #ifdef CONFIG_RT_GROUP_SCHED
2187 extern int sched_group_set_rt_runtime(struct task_group *tg,
2188 long rt_runtime_us);
2189 extern long sched_group_rt_runtime(struct task_group *tg);
2190 extern int sched_group_set_rt_period(struct task_group *tg,
2191 long rt_period_us);
2192 extern long sched_group_rt_period(struct task_group *tg);
2193 #endif
2194 #endif
2195
2196 #ifdef CONFIG_TASK_XACCT
2197 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2198 {
2199 tsk->rchar += amt;
2200 }
2201
2202 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2203 {
2204 tsk->wchar += amt;
2205 }
2206
2207 static inline void inc_syscr(struct task_struct *tsk)
2208 {
2209 tsk->syscr++;
2210 }
2211
2212 static inline void inc_syscw(struct task_struct *tsk)
2213 {
2214 tsk->syscw++;
2215 }
2216 #else
2217 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2218 {
2219 }
2220
2221 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2222 {
2223 }
2224
2225 static inline void inc_syscr(struct task_struct *tsk)
2226 {
2227 }
2228
2229 static inline void inc_syscw(struct task_struct *tsk)
2230 {
2231 }
2232 #endif
2233
2234 #ifdef CONFIG_SMP
2235 void migration_init(void);
2236 #else
2237 static inline void migration_init(void)
2238 {
2239 }
2240 #endif
2241
2242 #ifndef TASK_SIZE_OF
2243 #define TASK_SIZE_OF(tsk) TASK_SIZE
2244 #endif
2245
2246 #ifdef CONFIG_MM_OWNER
2247 extern void mm_update_next_owner(struct mm_struct *mm);
2248 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2249 #else
2250 static inline void mm_update_next_owner(struct mm_struct *mm)
2251 {
2252 }
2253
2254 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2255 {
2256 }
2257 #endif /* CONFIG_MM_OWNER */
2258
2259 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2260
2261 #endif /* __KERNEL__ */
2262
2263 #endif