]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blob - include/linux/sched.h
sched: improve affine wakeups
[mirror_ubuntu-zesty-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/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 asmlinkage void schedule_tail(struct task_struct *prev);
246 extern void init_idle(struct task_struct *idle, int cpu);
247 extern void init_idle_bootup_task(struct task_struct *idle);
248
249 extern cpumask_t nohz_cpu_mask;
250 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
251 extern int select_nohz_load_balancer(int cpu);
252 #else
253 static inline int select_nohz_load_balancer(int cpu)
254 {
255 return 0;
256 }
257 #endif
258
259 extern unsigned long rt_needs_cpu(int cpu);
260
261 /*
262 * Only dump TASK_* tasks. (0 for all tasks)
263 */
264 extern void show_state_filter(unsigned long state_filter);
265
266 static inline void show_state(void)
267 {
268 show_state_filter(0);
269 }
270
271 extern void show_regs(struct pt_regs *);
272
273 /*
274 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
275 * task), SP is the stack pointer of the first frame that should be shown in the back
276 * trace (or NULL if the entire call-chain of the task should be shown).
277 */
278 extern void show_stack(struct task_struct *task, unsigned long *sp);
279
280 void io_schedule(void);
281 long io_schedule_timeout(long timeout);
282
283 extern void cpu_init (void);
284 extern void trap_init(void);
285 extern void account_process_tick(struct task_struct *task, int user);
286 extern void update_process_times(int user);
287 extern void scheduler_tick(void);
288 extern void hrtick_resched(void);
289
290 extern void sched_show_task(struct task_struct *p);
291
292 #ifdef CONFIG_DETECT_SOFTLOCKUP
293 extern void softlockup_tick(void);
294 extern void spawn_softlockup_task(void);
295 extern void touch_softlockup_watchdog(void);
296 extern void touch_all_softlockup_watchdogs(void);
297 extern unsigned long softlockup_thresh;
298 extern unsigned long sysctl_hung_task_check_count;
299 extern unsigned long sysctl_hung_task_timeout_secs;
300 extern unsigned long sysctl_hung_task_warnings;
301 #else
302 static inline void softlockup_tick(void)
303 {
304 }
305 static inline void spawn_softlockup_task(void)
306 {
307 }
308 static inline void touch_softlockup_watchdog(void)
309 {
310 }
311 static inline void touch_all_softlockup_watchdogs(void)
312 {
313 }
314 #endif
315
316
317 /* Attach to any functions which should be ignored in wchan output. */
318 #define __sched __attribute__((__section__(".sched.text")))
319
320 /* Linker adds these: start and end of __sched functions */
321 extern char __sched_text_start[], __sched_text_end[];
322
323 /* Is this address in the __sched functions? */
324 extern int in_sched_functions(unsigned long addr);
325
326 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
327 extern signed long schedule_timeout(signed long timeout);
328 extern signed long schedule_timeout_interruptible(signed long timeout);
329 extern signed long schedule_timeout_killable(signed long timeout);
330 extern signed long schedule_timeout_uninterruptible(signed long timeout);
331 asmlinkage void schedule(void);
332
333 struct nsproxy;
334 struct user_namespace;
335
336 /* Maximum number of active map areas.. This is a random (large) number */
337 #define DEFAULT_MAX_MAP_COUNT 65536
338
339 extern int sysctl_max_map_count;
340
341 #include <linux/aio.h>
342
343 extern unsigned long
344 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
345 unsigned long, unsigned long);
346 extern unsigned long
347 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
348 unsigned long len, unsigned long pgoff,
349 unsigned long flags);
350 extern void arch_unmap_area(struct mm_struct *, unsigned long);
351 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
352
353 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
354 /*
355 * The mm counters are not protected by its page_table_lock,
356 * so must be incremented atomically.
357 */
358 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
359 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
360 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
361 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
362 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
363
364 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
365 /*
366 * The mm counters are protected by its page_table_lock,
367 * so can be incremented directly.
368 */
369 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
370 #define get_mm_counter(mm, member) ((mm)->_##member)
371 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
372 #define inc_mm_counter(mm, member) (mm)->_##member++
373 #define dec_mm_counter(mm, member) (mm)->_##member--
374
375 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
376
377 #define get_mm_rss(mm) \
378 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
379 #define update_hiwater_rss(mm) do { \
380 unsigned long _rss = get_mm_rss(mm); \
381 if ((mm)->hiwater_rss < _rss) \
382 (mm)->hiwater_rss = _rss; \
383 } while (0)
384 #define update_hiwater_vm(mm) do { \
385 if ((mm)->hiwater_vm < (mm)->total_vm) \
386 (mm)->hiwater_vm = (mm)->total_vm; \
387 } while (0)
388
389 extern void set_dumpable(struct mm_struct *mm, int value);
390 extern int get_dumpable(struct mm_struct *mm);
391
392 /* mm flags */
393 /* dumpable bits */
394 #define MMF_DUMPABLE 0 /* core dump is permitted */
395 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
396 #define MMF_DUMPABLE_BITS 2
397
398 /* coredump filter bits */
399 #define MMF_DUMP_ANON_PRIVATE 2
400 #define MMF_DUMP_ANON_SHARED 3
401 #define MMF_DUMP_MAPPED_PRIVATE 4
402 #define MMF_DUMP_MAPPED_SHARED 5
403 #define MMF_DUMP_ELF_HEADERS 6
404 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
405 #define MMF_DUMP_FILTER_BITS 5
406 #define MMF_DUMP_FILTER_MASK \
407 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
408 #define MMF_DUMP_FILTER_DEFAULT \
409 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
410
411 struct sighand_struct {
412 atomic_t count;
413 struct k_sigaction action[_NSIG];
414 spinlock_t siglock;
415 wait_queue_head_t signalfd_wqh;
416 };
417
418 struct pacct_struct {
419 int ac_flag;
420 long ac_exitcode;
421 unsigned long ac_mem;
422 cputime_t ac_utime, ac_stime;
423 unsigned long ac_minflt, ac_majflt;
424 };
425
426 /*
427 * NOTE! "signal_struct" does not have it's own
428 * locking, because a shared signal_struct always
429 * implies a shared sighand_struct, so locking
430 * sighand_struct is always a proper superset of
431 * the locking of signal_struct.
432 */
433 struct signal_struct {
434 atomic_t count;
435 atomic_t live;
436
437 wait_queue_head_t wait_chldexit; /* for wait4() */
438
439 /* current thread group signal load-balancing target: */
440 struct task_struct *curr_target;
441
442 /* shared signal handling: */
443 struct sigpending shared_pending;
444
445 /* thread group exit support */
446 int group_exit_code;
447 /* overloaded:
448 * - notify group_exit_task when ->count is equal to notify_count
449 * - everyone except group_exit_task is stopped during signal delivery
450 * of fatal signals, group_exit_task processes the signal.
451 */
452 struct task_struct *group_exit_task;
453 int notify_count;
454
455 /* thread group stop support, overloads group_exit_code too */
456 int group_stop_count;
457 unsigned int flags; /* see SIGNAL_* flags below */
458
459 /* POSIX.1b Interval Timers */
460 struct list_head posix_timers;
461
462 /* ITIMER_REAL timer for the process */
463 struct hrtimer real_timer;
464 struct pid *leader_pid;
465 ktime_t it_real_incr;
466
467 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
468 cputime_t it_prof_expires, it_virt_expires;
469 cputime_t it_prof_incr, it_virt_incr;
470
471 /* job control IDs */
472
473 /*
474 * pgrp and session fields are deprecated.
475 * use the task_session_Xnr and task_pgrp_Xnr routines below
476 */
477
478 union {
479 pid_t pgrp __deprecated;
480 pid_t __pgrp;
481 };
482
483 struct pid *tty_old_pgrp;
484
485 union {
486 pid_t session __deprecated;
487 pid_t __session;
488 };
489
490 /* boolean value for session group leader */
491 int leader;
492
493 struct tty_struct *tty; /* NULL if no tty */
494
495 /*
496 * Cumulative resource counters for dead threads in the group,
497 * and for reaped dead child processes forked by this group.
498 * Live threads maintain their own counters and add to these
499 * in __exit_signal, except for the group leader.
500 */
501 cputime_t utime, stime, cutime, cstime;
502 cputime_t gtime;
503 cputime_t cgtime;
504 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
505 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
506 unsigned long inblock, oublock, cinblock, coublock;
507
508 /*
509 * Cumulative ns of scheduled CPU time for dead threads in the
510 * group, not including a zombie group leader. (This only differs
511 * from jiffies_to_ns(utime + stime) if sched_clock uses something
512 * other than jiffies.)
513 */
514 unsigned long long sum_sched_runtime;
515
516 /*
517 * We don't bother to synchronize most readers of this at all,
518 * because there is no reader checking a limit that actually needs
519 * to get both rlim_cur and rlim_max atomically, and either one
520 * alone is a single word that can safely be read normally.
521 * getrlimit/setrlimit use task_lock(current->group_leader) to
522 * protect this instead of the siglock, because they really
523 * have no need to disable irqs.
524 */
525 struct rlimit rlim[RLIM_NLIMITS];
526
527 struct list_head cpu_timers[3];
528
529 /* keep the process-shared keyrings here so that they do the right
530 * thing in threads created with CLONE_THREAD */
531 #ifdef CONFIG_KEYS
532 struct key *session_keyring; /* keyring inherited over fork */
533 struct key *process_keyring; /* keyring private to this process */
534 #endif
535 #ifdef CONFIG_BSD_PROCESS_ACCT
536 struct pacct_struct pacct; /* per-process accounting information */
537 #endif
538 #ifdef CONFIG_TASKSTATS
539 struct taskstats *stats;
540 #endif
541 #ifdef CONFIG_AUDIT
542 unsigned audit_tty;
543 struct tty_audit_buf *tty_audit_buf;
544 #endif
545 };
546
547 /* Context switch must be unlocked if interrupts are to be enabled */
548 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
549 # define __ARCH_WANT_UNLOCKED_CTXSW
550 #endif
551
552 /*
553 * Bits in flags field of signal_struct.
554 */
555 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
556 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
557 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
558 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
559
560 /* If true, all threads except ->group_exit_task have pending SIGKILL */
561 static inline int signal_group_exit(const struct signal_struct *sig)
562 {
563 return (sig->flags & SIGNAL_GROUP_EXIT) ||
564 (sig->group_exit_task != NULL);
565 }
566
567 /*
568 * Some day this will be a full-fledged user tracking system..
569 */
570 struct user_struct {
571 atomic_t __count; /* reference count */
572 atomic_t processes; /* How many processes does this user have? */
573 atomic_t files; /* How many open files does this user have? */
574 atomic_t sigpending; /* How many pending signals does this user have? */
575 #ifdef CONFIG_INOTIFY_USER
576 atomic_t inotify_watches; /* How many inotify watches does this user have? */
577 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
578 #endif
579 #ifdef CONFIG_POSIX_MQUEUE
580 /* protected by mq_lock */
581 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
582 #endif
583 unsigned long locked_shm; /* How many pages of mlocked shm ? */
584
585 #ifdef CONFIG_KEYS
586 struct key *uid_keyring; /* UID specific keyring */
587 struct key *session_keyring; /* UID's default session keyring */
588 #endif
589
590 /* Hash table maintenance information */
591 struct hlist_node uidhash_node;
592 uid_t uid;
593
594 #ifdef CONFIG_USER_SCHED
595 struct task_group *tg;
596 #ifdef CONFIG_SYSFS
597 struct kobject kobj;
598 struct work_struct work;
599 #endif
600 #endif
601 };
602
603 extern int uids_sysfs_init(void);
604
605 extern struct user_struct *find_user(uid_t);
606
607 extern struct user_struct root_user;
608 #define INIT_USER (&root_user)
609
610 struct backing_dev_info;
611 struct reclaim_state;
612
613 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
614 struct sched_info {
615 /* cumulative counters */
616 unsigned long pcount; /* # of times run on this cpu */
617 unsigned long long cpu_time, /* time spent on the cpu */
618 run_delay; /* time spent waiting on a runqueue */
619
620 /* timestamps */
621 unsigned long long last_arrival,/* when we last ran on a cpu */
622 last_queued; /* when we were last queued to run */
623 #ifdef CONFIG_SCHEDSTATS
624 /* BKL stats */
625 unsigned int bkl_count;
626 #endif
627 };
628 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
629
630 #ifdef CONFIG_SCHEDSTATS
631 extern const struct file_operations proc_schedstat_operations;
632 #endif /* CONFIG_SCHEDSTATS */
633
634 #ifdef CONFIG_TASK_DELAY_ACCT
635 struct task_delay_info {
636 spinlock_t lock;
637 unsigned int flags; /* Private per-task flags */
638
639 /* For each stat XXX, add following, aligned appropriately
640 *
641 * struct timespec XXX_start, XXX_end;
642 * u64 XXX_delay;
643 * u32 XXX_count;
644 *
645 * Atomicity of updates to XXX_delay, XXX_count protected by
646 * single lock above (split into XXX_lock if contention is an issue).
647 */
648
649 /*
650 * XXX_count is incremented on every XXX operation, the delay
651 * associated with the operation is added to XXX_delay.
652 * XXX_delay contains the accumulated delay time in nanoseconds.
653 */
654 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
655 u64 blkio_delay; /* wait for sync block io completion */
656 u64 swapin_delay; /* wait for swapin block io completion */
657 u32 blkio_count; /* total count of the number of sync block */
658 /* io operations performed */
659 u32 swapin_count; /* total count of the number of swapin block */
660 /* io operations performed */
661 };
662 #endif /* CONFIG_TASK_DELAY_ACCT */
663
664 static inline int sched_info_on(void)
665 {
666 #ifdef CONFIG_SCHEDSTATS
667 return 1;
668 #elif defined(CONFIG_TASK_DELAY_ACCT)
669 extern int delayacct_on;
670 return delayacct_on;
671 #else
672 return 0;
673 #endif
674 }
675
676 enum cpu_idle_type {
677 CPU_IDLE,
678 CPU_NOT_IDLE,
679 CPU_NEWLY_IDLE,
680 CPU_MAX_IDLE_TYPES
681 };
682
683 /*
684 * sched-domains (multiprocessor balancing) declarations:
685 */
686
687 /*
688 * Increase resolution of nice-level calculations:
689 */
690 #define SCHED_LOAD_SHIFT 10
691 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
692
693 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
694
695 #ifdef CONFIG_SMP
696 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
697 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
698 #define SD_BALANCE_EXEC 4 /* Balance on exec */
699 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
700 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
701 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
702 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
703 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
704 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
705 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
706 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
707
708 #define BALANCE_FOR_MC_POWER \
709 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
710
711 #define BALANCE_FOR_PKG_POWER \
712 ((sched_mc_power_savings || sched_smt_power_savings) ? \
713 SD_POWERSAVINGS_BALANCE : 0)
714
715 #define test_sd_parent(sd, flag) ((sd->parent && \
716 (sd->parent->flags & flag)) ? 1 : 0)
717
718
719 struct sched_group {
720 struct sched_group *next; /* Must be a circular list */
721 cpumask_t cpumask;
722
723 /*
724 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
725 * single CPU. This is read only (except for setup, hotplug CPU).
726 * Note : Never change cpu_power without recompute its reciprocal
727 */
728 unsigned int __cpu_power;
729 /*
730 * reciprocal value of cpu_power to avoid expensive divides
731 * (see include/linux/reciprocal_div.h)
732 */
733 u32 reciprocal_cpu_power;
734 };
735
736 struct sched_domain {
737 /* These fields must be setup */
738 struct sched_domain *parent; /* top domain must be null terminated */
739 struct sched_domain *child; /* bottom domain must be null terminated */
740 struct sched_group *groups; /* the balancing groups of the domain */
741 cpumask_t span; /* span of all CPUs in this domain */
742 unsigned long min_interval; /* Minimum balance interval ms */
743 unsigned long max_interval; /* Maximum balance interval ms */
744 unsigned int busy_factor; /* less balancing by factor if busy */
745 unsigned int imbalance_pct; /* No balance until over watermark */
746 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
747 unsigned int busy_idx;
748 unsigned int idle_idx;
749 unsigned int newidle_idx;
750 unsigned int wake_idx;
751 unsigned int forkexec_idx;
752 int flags; /* See SD_* */
753
754 /* Runtime fields. */
755 unsigned long last_balance; /* init to jiffies. units in jiffies */
756 unsigned int balance_interval; /* initialise to 1. units in ms. */
757 unsigned int nr_balance_failed; /* initialise to 0 */
758
759 #ifdef CONFIG_SCHEDSTATS
760 /* load_balance() stats */
761 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
762 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
763 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
764 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
765 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
766 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
767 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
768 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
769
770 /* Active load balancing */
771 unsigned int alb_count;
772 unsigned int alb_failed;
773 unsigned int alb_pushed;
774
775 /* SD_BALANCE_EXEC stats */
776 unsigned int sbe_count;
777 unsigned int sbe_balanced;
778 unsigned int sbe_pushed;
779
780 /* SD_BALANCE_FORK stats */
781 unsigned int sbf_count;
782 unsigned int sbf_balanced;
783 unsigned int sbf_pushed;
784
785 /* try_to_wake_up() stats */
786 unsigned int ttwu_wake_remote;
787 unsigned int ttwu_move_affine;
788 unsigned int ttwu_move_balance;
789 #endif
790 };
791
792 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new);
793
794 #endif /* CONFIG_SMP */
795
796 /*
797 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
798 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
799 * task of nice 0 or enough lower priority tasks to bring up the
800 * weighted_cpuload
801 */
802 static inline int above_background_load(void)
803 {
804 unsigned long cpu;
805
806 for_each_online_cpu(cpu) {
807 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
808 return 1;
809 }
810 return 0;
811 }
812
813 struct io_context; /* See blkdev.h */
814 #define NGROUPS_SMALL 32
815 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
816 struct group_info {
817 int ngroups;
818 atomic_t usage;
819 gid_t small_block[NGROUPS_SMALL];
820 int nblocks;
821 gid_t *blocks[0];
822 };
823
824 /*
825 * get_group_info() must be called with the owning task locked (via task_lock())
826 * when task != current. The reason being that the vast majority of callers are
827 * looking at current->group_info, which can not be changed except by the
828 * current task. Changing current->group_info requires the task lock, too.
829 */
830 #define get_group_info(group_info) do { \
831 atomic_inc(&(group_info)->usage); \
832 } while (0)
833
834 #define put_group_info(group_info) do { \
835 if (atomic_dec_and_test(&(group_info)->usage)) \
836 groups_free(group_info); \
837 } while (0)
838
839 extern struct group_info *groups_alloc(int gidsetsize);
840 extern void groups_free(struct group_info *group_info);
841 extern int set_current_groups(struct group_info *group_info);
842 extern int groups_search(struct group_info *group_info, gid_t grp);
843 /* access the groups "array" with this macro */
844 #define GROUP_AT(gi, i) \
845 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
846
847 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
848 extern void prefetch_stack(struct task_struct *t);
849 #else
850 static inline void prefetch_stack(struct task_struct *t) { }
851 #endif
852
853 struct audit_context; /* See audit.c */
854 struct mempolicy;
855 struct pipe_inode_info;
856 struct uts_namespace;
857
858 struct rq;
859 struct sched_domain;
860
861 struct sched_class {
862 const struct sched_class *next;
863
864 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
865 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
866 void (*yield_task) (struct rq *rq);
867 int (*select_task_rq)(struct task_struct *p, int sync);
868
869 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
870
871 struct task_struct * (*pick_next_task) (struct rq *rq);
872 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
873
874 #ifdef CONFIG_SMP
875 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
876 struct rq *busiest, unsigned long max_load_move,
877 struct sched_domain *sd, enum cpu_idle_type idle,
878 int *all_pinned, int *this_best_prio);
879
880 int (*move_one_task) (struct rq *this_rq, int this_cpu,
881 struct rq *busiest, struct sched_domain *sd,
882 enum cpu_idle_type idle);
883 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
884 void (*post_schedule) (struct rq *this_rq);
885 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
886 #endif
887
888 void (*set_curr_task) (struct rq *rq);
889 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
890 void (*task_new) (struct rq *rq, struct task_struct *p);
891 void (*set_cpus_allowed)(struct task_struct *p, cpumask_t *newmask);
892
893 void (*join_domain)(struct rq *rq);
894 void (*leave_domain)(struct rq *rq);
895
896 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
897 int running);
898 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
899 int running);
900 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
901 int oldprio, int running);
902
903 #ifdef CONFIG_FAIR_GROUP_SCHED
904 void (*moved_group) (struct task_struct *p);
905 #endif
906 };
907
908 struct load_weight {
909 unsigned long weight, inv_weight;
910 };
911
912 /*
913 * CFS stats for a schedulable entity (task, task-group etc)
914 *
915 * Current field usage histogram:
916 *
917 * 4 se->block_start
918 * 4 se->run_node
919 * 4 se->sleep_start
920 * 6 se->load.weight
921 */
922 struct sched_entity {
923 struct load_weight load; /* for load-balancing */
924 struct rb_node run_node;
925 unsigned int on_rq;
926
927 u64 exec_start;
928 u64 sum_exec_runtime;
929 u64 vruntime;
930 u64 prev_sum_exec_runtime;
931
932 u64 last_wakeup;
933 u64 avg_overlap;
934
935 #ifdef CONFIG_SCHEDSTATS
936 u64 wait_start;
937 u64 wait_max;
938 u64 wait_count;
939 u64 wait_sum;
940
941 u64 sleep_start;
942 u64 sleep_max;
943 s64 sum_sleep_runtime;
944
945 u64 block_start;
946 u64 block_max;
947 u64 exec_max;
948 u64 slice_max;
949
950 u64 nr_migrations;
951 u64 nr_migrations_cold;
952 u64 nr_failed_migrations_affine;
953 u64 nr_failed_migrations_running;
954 u64 nr_failed_migrations_hot;
955 u64 nr_forced_migrations;
956 u64 nr_forced2_migrations;
957
958 u64 nr_wakeups;
959 u64 nr_wakeups_sync;
960 u64 nr_wakeups_migrate;
961 u64 nr_wakeups_local;
962 u64 nr_wakeups_remote;
963 u64 nr_wakeups_affine;
964 u64 nr_wakeups_affine_attempts;
965 u64 nr_wakeups_passive;
966 u64 nr_wakeups_idle;
967 #endif
968
969 #ifdef CONFIG_FAIR_GROUP_SCHED
970 struct sched_entity *parent;
971 /* rq on which this entity is (to be) queued: */
972 struct cfs_rq *cfs_rq;
973 /* rq "owned" by this entity/group: */
974 struct cfs_rq *my_q;
975 #endif
976 };
977
978 struct sched_rt_entity {
979 struct list_head run_list;
980 unsigned int time_slice;
981 unsigned long timeout;
982 int nr_cpus_allowed;
983
984 #ifdef CONFIG_RT_GROUP_SCHED
985 struct sched_rt_entity *parent;
986 /* rq on which this entity is (to be) queued: */
987 struct rt_rq *rt_rq;
988 /* rq "owned" by this entity/group: */
989 struct rt_rq *my_q;
990 #endif
991 };
992
993 struct task_struct {
994 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
995 void *stack;
996 atomic_t usage;
997 unsigned int flags; /* per process flags, defined below */
998 unsigned int ptrace;
999
1000 int lock_depth; /* BKL lock depth */
1001
1002 #ifdef CONFIG_SMP
1003 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1004 int oncpu;
1005 #endif
1006 #endif
1007
1008 int prio, static_prio, normal_prio;
1009 const struct sched_class *sched_class;
1010 struct sched_entity se;
1011 struct sched_rt_entity rt;
1012
1013 #ifdef CONFIG_PREEMPT_NOTIFIERS
1014 /* list of struct preempt_notifier: */
1015 struct hlist_head preempt_notifiers;
1016 #endif
1017
1018 /*
1019 * fpu_counter contains the number of consecutive context switches
1020 * that the FPU is used. If this is over a threshold, the lazy fpu
1021 * saving becomes unlazy to save the trap. This is an unsigned char
1022 * so that after 256 times the counter wraps and the behavior turns
1023 * lazy again; this to deal with bursty apps that only use FPU for
1024 * a short time
1025 */
1026 unsigned char fpu_counter;
1027 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1028 #ifdef CONFIG_BLK_DEV_IO_TRACE
1029 unsigned int btrace_seq;
1030 #endif
1031
1032 unsigned int policy;
1033 cpumask_t cpus_allowed;
1034
1035 #ifdef CONFIG_PREEMPT_RCU
1036 int rcu_read_lock_nesting;
1037 int rcu_flipctr_idx;
1038 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1039
1040 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1041 struct sched_info sched_info;
1042 #endif
1043
1044 struct list_head tasks;
1045 /*
1046 * ptrace_list/ptrace_children forms the list of my children
1047 * that were stolen by a ptracer.
1048 */
1049 struct list_head ptrace_children;
1050 struct list_head ptrace_list;
1051
1052 struct mm_struct *mm, *active_mm;
1053
1054 /* task state */
1055 struct linux_binfmt *binfmt;
1056 int exit_state;
1057 int exit_code, exit_signal;
1058 int pdeath_signal; /* The signal sent when the parent dies */
1059 /* ??? */
1060 unsigned int personality;
1061 unsigned did_exec:1;
1062 pid_t pid;
1063 pid_t tgid;
1064
1065 #ifdef CONFIG_CC_STACKPROTECTOR
1066 /* Canary value for the -fstack-protector gcc feature */
1067 unsigned long stack_canary;
1068 #endif
1069 /*
1070 * pointers to (original) parent process, youngest child, younger sibling,
1071 * older sibling, respectively. (p->father can be replaced with
1072 * p->parent->pid)
1073 */
1074 struct task_struct *real_parent; /* real parent process (when being debugged) */
1075 struct task_struct *parent; /* parent process */
1076 /*
1077 * children/sibling forms the list of my children plus the
1078 * tasks I'm ptracing.
1079 */
1080 struct list_head children; /* list of my children */
1081 struct list_head sibling; /* linkage in my parent's children list */
1082 struct task_struct *group_leader; /* threadgroup leader */
1083
1084 /* PID/PID hash table linkage. */
1085 struct pid_link pids[PIDTYPE_MAX];
1086 struct list_head thread_group;
1087
1088 struct completion *vfork_done; /* for vfork() */
1089 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1090 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1091
1092 unsigned int rt_priority;
1093 cputime_t utime, stime, utimescaled, stimescaled;
1094 cputime_t gtime;
1095 cputime_t prev_utime, prev_stime;
1096 unsigned long nvcsw, nivcsw; /* context switch counts */
1097 struct timespec start_time; /* monotonic time */
1098 struct timespec real_start_time; /* boot based time */
1099 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1100 unsigned long min_flt, maj_flt;
1101
1102 cputime_t it_prof_expires, it_virt_expires;
1103 unsigned long long it_sched_expires;
1104 struct list_head cpu_timers[3];
1105
1106 /* process credentials */
1107 uid_t uid,euid,suid,fsuid;
1108 gid_t gid,egid,sgid,fsgid;
1109 struct group_info *group_info;
1110 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset;
1111 unsigned keep_capabilities:1;
1112 struct user_struct *user;
1113 #ifdef CONFIG_KEYS
1114 struct key *request_key_auth; /* assumed request_key authority */
1115 struct key *thread_keyring; /* keyring private to this thread */
1116 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1117 #endif
1118 char comm[TASK_COMM_LEN]; /* executable name excluding path
1119 - access with [gs]et_task_comm (which lock
1120 it with task_lock())
1121 - initialized normally by flush_old_exec */
1122 /* file system info */
1123 int link_count, total_link_count;
1124 #ifdef CONFIG_SYSVIPC
1125 /* ipc stuff */
1126 struct sysv_sem sysvsem;
1127 #endif
1128 #ifdef CONFIG_DETECT_SOFTLOCKUP
1129 /* hung task detection */
1130 unsigned long last_switch_timestamp;
1131 unsigned long last_switch_count;
1132 #endif
1133 /* CPU-specific state of this task */
1134 struct thread_struct thread;
1135 /* filesystem information */
1136 struct fs_struct *fs;
1137 /* open file information */
1138 struct files_struct *files;
1139 /* namespaces */
1140 struct nsproxy *nsproxy;
1141 /* signal handlers */
1142 struct signal_struct *signal;
1143 struct sighand_struct *sighand;
1144
1145 sigset_t blocked, real_blocked;
1146 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
1147 struct sigpending pending;
1148
1149 unsigned long sas_ss_sp;
1150 size_t sas_ss_size;
1151 int (*notifier)(void *priv);
1152 void *notifier_data;
1153 sigset_t *notifier_mask;
1154 #ifdef CONFIG_SECURITY
1155 void *security;
1156 #endif
1157 struct audit_context *audit_context;
1158 #ifdef CONFIG_AUDITSYSCALL
1159 uid_t loginuid;
1160 unsigned int sessionid;
1161 #endif
1162 seccomp_t seccomp;
1163
1164 /* Thread group tracking */
1165 u32 parent_exec_id;
1166 u32 self_exec_id;
1167 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1168 spinlock_t alloc_lock;
1169
1170 /* Protection of the PI data structures: */
1171 spinlock_t pi_lock;
1172
1173 #ifdef CONFIG_RT_MUTEXES
1174 /* PI waiters blocked on a rt_mutex held by this task */
1175 struct plist_head pi_waiters;
1176 /* Deadlock detection and priority inheritance handling */
1177 struct rt_mutex_waiter *pi_blocked_on;
1178 #endif
1179
1180 #ifdef CONFIG_DEBUG_MUTEXES
1181 /* mutex deadlock detection */
1182 struct mutex_waiter *blocked_on;
1183 #endif
1184 #ifdef CONFIG_TRACE_IRQFLAGS
1185 unsigned int irq_events;
1186 int hardirqs_enabled;
1187 unsigned long hardirq_enable_ip;
1188 unsigned int hardirq_enable_event;
1189 unsigned long hardirq_disable_ip;
1190 unsigned int hardirq_disable_event;
1191 int softirqs_enabled;
1192 unsigned long softirq_disable_ip;
1193 unsigned int softirq_disable_event;
1194 unsigned long softirq_enable_ip;
1195 unsigned int softirq_enable_event;
1196 int hardirq_context;
1197 int softirq_context;
1198 #endif
1199 #ifdef CONFIG_LOCKDEP
1200 # define MAX_LOCK_DEPTH 48UL
1201 u64 curr_chain_key;
1202 int lockdep_depth;
1203 struct held_lock held_locks[MAX_LOCK_DEPTH];
1204 unsigned int lockdep_recursion;
1205 #endif
1206
1207 /* journalling filesystem info */
1208 void *journal_info;
1209
1210 /* stacked block device info */
1211 struct bio *bio_list, **bio_tail;
1212
1213 /* VM state */
1214 struct reclaim_state *reclaim_state;
1215
1216 struct backing_dev_info *backing_dev_info;
1217
1218 struct io_context *io_context;
1219
1220 unsigned long ptrace_message;
1221 siginfo_t *last_siginfo; /* For ptrace use. */
1222 #ifdef CONFIG_TASK_XACCT
1223 /* i/o counters(bytes read/written, #syscalls */
1224 u64 rchar, wchar, syscr, syscw;
1225 #endif
1226 struct task_io_accounting ioac;
1227 #if defined(CONFIG_TASK_XACCT)
1228 u64 acct_rss_mem1; /* accumulated rss usage */
1229 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1230 cputime_t acct_stimexpd;/* stime since last update */
1231 #endif
1232 #ifdef CONFIG_NUMA
1233 struct mempolicy *mempolicy;
1234 short il_next;
1235 #endif
1236 #ifdef CONFIG_CPUSETS
1237 nodemask_t mems_allowed;
1238 int cpuset_mems_generation;
1239 int cpuset_mem_spread_rotor;
1240 #endif
1241 #ifdef CONFIG_CGROUPS
1242 /* Control Group info protected by css_set_lock */
1243 struct css_set *cgroups;
1244 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1245 struct list_head cg_list;
1246 #endif
1247 #ifdef CONFIG_FUTEX
1248 struct robust_list_head __user *robust_list;
1249 #ifdef CONFIG_COMPAT
1250 struct compat_robust_list_head __user *compat_robust_list;
1251 #endif
1252 struct list_head pi_state_list;
1253 struct futex_pi_state *pi_state_cache;
1254 #endif
1255 atomic_t fs_excl; /* holding fs exclusive resources */
1256 struct rcu_head rcu;
1257
1258 /*
1259 * cache last used pipe for splice
1260 */
1261 struct pipe_inode_info *splice_pipe;
1262 #ifdef CONFIG_TASK_DELAY_ACCT
1263 struct task_delay_info *delays;
1264 #endif
1265 #ifdef CONFIG_FAULT_INJECTION
1266 int make_it_fail;
1267 #endif
1268 struct prop_local_single dirties;
1269 #ifdef CONFIG_LATENCYTOP
1270 int latency_record_count;
1271 struct latency_record latency_record[LT_SAVECOUNT];
1272 #endif
1273 };
1274
1275 /*
1276 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1277 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1278 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1279 * values are inverted: lower p->prio value means higher priority.
1280 *
1281 * The MAX_USER_RT_PRIO value allows the actual maximum
1282 * RT priority to be separate from the value exported to
1283 * user-space. This allows kernel threads to set their
1284 * priority to a value higher than any user task. Note:
1285 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1286 */
1287
1288 #define MAX_USER_RT_PRIO 100
1289 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1290
1291 #define MAX_PRIO (MAX_RT_PRIO + 40)
1292 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1293
1294 static inline int rt_prio(int prio)
1295 {
1296 if (unlikely(prio < MAX_RT_PRIO))
1297 return 1;
1298 return 0;
1299 }
1300
1301 static inline int rt_task(struct task_struct *p)
1302 {
1303 return rt_prio(p->prio);
1304 }
1305
1306 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1307 {
1308 tsk->signal->__session = session;
1309 }
1310
1311 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1312 {
1313 tsk->signal->__pgrp = pgrp;
1314 }
1315
1316 static inline struct pid *task_pid(struct task_struct *task)
1317 {
1318 return task->pids[PIDTYPE_PID].pid;
1319 }
1320
1321 static inline struct pid *task_tgid(struct task_struct *task)
1322 {
1323 return task->group_leader->pids[PIDTYPE_PID].pid;
1324 }
1325
1326 static inline struct pid *task_pgrp(struct task_struct *task)
1327 {
1328 return task->group_leader->pids[PIDTYPE_PGID].pid;
1329 }
1330
1331 static inline struct pid *task_session(struct task_struct *task)
1332 {
1333 return task->group_leader->pids[PIDTYPE_SID].pid;
1334 }
1335
1336 struct pid_namespace;
1337
1338 /*
1339 * the helpers to get the task's different pids as they are seen
1340 * from various namespaces
1341 *
1342 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1343 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1344 * current.
1345 * task_xid_nr_ns() : id seen from the ns specified;
1346 *
1347 * set_task_vxid() : assigns a virtual id to a task;
1348 *
1349 * see also pid_nr() etc in include/linux/pid.h
1350 */
1351
1352 static inline pid_t task_pid_nr(struct task_struct *tsk)
1353 {
1354 return tsk->pid;
1355 }
1356
1357 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1358
1359 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1360 {
1361 return pid_vnr(task_pid(tsk));
1362 }
1363
1364
1365 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1366 {
1367 return tsk->tgid;
1368 }
1369
1370 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1371
1372 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1373 {
1374 return pid_vnr(task_tgid(tsk));
1375 }
1376
1377
1378 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1379 {
1380 return tsk->signal->__pgrp;
1381 }
1382
1383 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1384
1385 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1386 {
1387 return pid_vnr(task_pgrp(tsk));
1388 }
1389
1390
1391 static inline pid_t task_session_nr(struct task_struct *tsk)
1392 {
1393 return tsk->signal->__session;
1394 }
1395
1396 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1397
1398 static inline pid_t task_session_vnr(struct task_struct *tsk)
1399 {
1400 return pid_vnr(task_session(tsk));
1401 }
1402
1403
1404 /**
1405 * pid_alive - check that a task structure is not stale
1406 * @p: Task structure to be checked.
1407 *
1408 * Test if a process is not yet dead (at most zombie state)
1409 * If pid_alive fails, then pointers within the task structure
1410 * can be stale and must not be dereferenced.
1411 */
1412 static inline int pid_alive(struct task_struct *p)
1413 {
1414 return p->pids[PIDTYPE_PID].pid != NULL;
1415 }
1416
1417 /**
1418 * is_global_init - check if a task structure is init
1419 * @tsk: Task structure to be checked.
1420 *
1421 * Check if a task structure is the first user space task the kernel created.
1422 */
1423 static inline int is_global_init(struct task_struct *tsk)
1424 {
1425 return tsk->pid == 1;
1426 }
1427
1428 /*
1429 * is_container_init:
1430 * check whether in the task is init in its own pid namespace.
1431 */
1432 extern int is_container_init(struct task_struct *tsk);
1433
1434 extern struct pid *cad_pid;
1435
1436 extern void free_task(struct task_struct *tsk);
1437 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1438
1439 extern void __put_task_struct(struct task_struct *t);
1440
1441 static inline void put_task_struct(struct task_struct *t)
1442 {
1443 if (atomic_dec_and_test(&t->usage))
1444 __put_task_struct(t);
1445 }
1446
1447 /*
1448 * Per process flags
1449 */
1450 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1451 /* Not implemented yet, only for 486*/
1452 #define PF_STARTING 0x00000002 /* being created */
1453 #define PF_EXITING 0x00000004 /* getting shut down */
1454 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1455 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1456 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1457 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1458 #define PF_DUMPCORE 0x00000200 /* dumped core */
1459 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1460 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1461 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1462 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1463 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1464 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1465 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1466 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1467 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1468 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1469 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1470 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1471 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1472 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1473 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1474 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1475 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1476 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1477
1478 /*
1479 * Only the _current_ task can read/write to tsk->flags, but other
1480 * tasks can access tsk->flags in readonly mode for example
1481 * with tsk_used_math (like during threaded core dumping).
1482 * There is however an exception to this rule during ptrace
1483 * or during fork: the ptracer task is allowed to write to the
1484 * child->flags of its traced child (same goes for fork, the parent
1485 * can write to the child->flags), because we're guaranteed the
1486 * child is not running and in turn not changing child->flags
1487 * at the same time the parent does it.
1488 */
1489 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1490 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1491 #define clear_used_math() clear_stopped_child_used_math(current)
1492 #define set_used_math() set_stopped_child_used_math(current)
1493 #define conditional_stopped_child_used_math(condition, child) \
1494 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1495 #define conditional_used_math(condition) \
1496 conditional_stopped_child_used_math(condition, current)
1497 #define copy_to_stopped_child_used_math(child) \
1498 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1499 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1500 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1501 #define used_math() tsk_used_math(current)
1502
1503 #ifdef CONFIG_SMP
1504 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1505 #else
1506 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1507 {
1508 if (!cpu_isset(0, new_mask))
1509 return -EINVAL;
1510 return 0;
1511 }
1512 #endif
1513
1514 extern unsigned long long sched_clock(void);
1515
1516 /*
1517 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1518 * clock constructed from sched_clock():
1519 */
1520 extern unsigned long long cpu_clock(int cpu);
1521
1522 extern unsigned long long
1523 task_sched_runtime(struct task_struct *task);
1524
1525 /* sched_exec is called by processes performing an exec */
1526 #ifdef CONFIG_SMP
1527 extern void sched_exec(void);
1528 #else
1529 #define sched_exec() {}
1530 #endif
1531
1532 extern void sched_clock_idle_sleep_event(void);
1533 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1534
1535 #ifdef CONFIG_HOTPLUG_CPU
1536 extern void idle_task_exit(void);
1537 #else
1538 static inline void idle_task_exit(void) {}
1539 #endif
1540
1541 extern void sched_idle_next(void);
1542
1543 #ifdef CONFIG_SCHED_DEBUG
1544 extern unsigned int sysctl_sched_latency;
1545 extern unsigned int sysctl_sched_min_granularity;
1546 extern unsigned int sysctl_sched_wakeup_granularity;
1547 extern unsigned int sysctl_sched_batch_wakeup_granularity;
1548 extern unsigned int sysctl_sched_child_runs_first;
1549 extern unsigned int sysctl_sched_features;
1550 extern unsigned int sysctl_sched_migration_cost;
1551 extern unsigned int sysctl_sched_nr_migrate;
1552
1553 int sched_nr_latency_handler(struct ctl_table *table, int write,
1554 struct file *file, void __user *buffer, size_t *length,
1555 loff_t *ppos);
1556 #endif
1557 extern unsigned int sysctl_sched_rt_period;
1558 extern int sysctl_sched_rt_runtime;
1559
1560 extern unsigned int sysctl_sched_compat_yield;
1561
1562 #ifdef CONFIG_RT_MUTEXES
1563 extern int rt_mutex_getprio(struct task_struct *p);
1564 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1565 extern void rt_mutex_adjust_pi(struct task_struct *p);
1566 #else
1567 static inline int rt_mutex_getprio(struct task_struct *p)
1568 {
1569 return p->normal_prio;
1570 }
1571 # define rt_mutex_adjust_pi(p) do { } while (0)
1572 #endif
1573
1574 extern void set_user_nice(struct task_struct *p, long nice);
1575 extern int task_prio(const struct task_struct *p);
1576 extern int task_nice(const struct task_struct *p);
1577 extern int can_nice(const struct task_struct *p, const int nice);
1578 extern int task_curr(const struct task_struct *p);
1579 extern int idle_cpu(int cpu);
1580 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1581 extern struct task_struct *idle_task(int cpu);
1582 extern struct task_struct *curr_task(int cpu);
1583 extern void set_curr_task(int cpu, struct task_struct *p);
1584
1585 void yield(void);
1586
1587 /*
1588 * The default (Linux) execution domain.
1589 */
1590 extern struct exec_domain default_exec_domain;
1591
1592 union thread_union {
1593 struct thread_info thread_info;
1594 unsigned long stack[THREAD_SIZE/sizeof(long)];
1595 };
1596
1597 #ifndef __HAVE_ARCH_KSTACK_END
1598 static inline int kstack_end(void *addr)
1599 {
1600 /* Reliable end of stack detection:
1601 * Some APM bios versions misalign the stack
1602 */
1603 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1604 }
1605 #endif
1606
1607 extern union thread_union init_thread_union;
1608 extern struct task_struct init_task;
1609
1610 extern struct mm_struct init_mm;
1611
1612 extern struct pid_namespace init_pid_ns;
1613
1614 /*
1615 * find a task by one of its numerical ids
1616 *
1617 * find_task_by_pid_type_ns():
1618 * it is the most generic call - it finds a task by all id,
1619 * type and namespace specified
1620 * find_task_by_pid_ns():
1621 * finds a task by its pid in the specified namespace
1622 * find_task_by_vpid():
1623 * finds a task by its virtual pid
1624 * find_task_by_pid():
1625 * finds a task by its global pid
1626 *
1627 * see also find_pid() etc in include/linux/pid.h
1628 */
1629
1630 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1631 struct pid_namespace *ns);
1632
1633 extern struct task_struct *find_task_by_pid(pid_t nr);
1634 extern struct task_struct *find_task_by_vpid(pid_t nr);
1635 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1636 struct pid_namespace *ns);
1637
1638 extern void __set_special_pids(struct pid *pid);
1639
1640 /* per-UID process charging. */
1641 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1642 static inline struct user_struct *get_uid(struct user_struct *u)
1643 {
1644 atomic_inc(&u->__count);
1645 return u;
1646 }
1647 extern void free_uid(struct user_struct *);
1648 extern void switch_uid(struct user_struct *);
1649 extern void release_uids(struct user_namespace *ns);
1650
1651 #include <asm/current.h>
1652
1653 extern void do_timer(unsigned long ticks);
1654
1655 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1656 extern int wake_up_process(struct task_struct *tsk);
1657 extern void wake_up_new_task(struct task_struct *tsk,
1658 unsigned long clone_flags);
1659 #ifdef CONFIG_SMP
1660 extern void kick_process(struct task_struct *tsk);
1661 #else
1662 static inline void kick_process(struct task_struct *tsk) { }
1663 #endif
1664 extern void sched_fork(struct task_struct *p, int clone_flags);
1665 extern void sched_dead(struct task_struct *p);
1666
1667 extern int in_group_p(gid_t);
1668 extern int in_egroup_p(gid_t);
1669
1670 extern void proc_caches_init(void);
1671 extern void flush_signals(struct task_struct *);
1672 extern void ignore_signals(struct task_struct *);
1673 extern void flush_signal_handlers(struct task_struct *, int force_default);
1674 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1675
1676 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1677 {
1678 unsigned long flags;
1679 int ret;
1680
1681 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1682 ret = dequeue_signal(tsk, mask, info);
1683 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1684
1685 return ret;
1686 }
1687
1688 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1689 sigset_t *mask);
1690 extern void unblock_all_signals(void);
1691 extern void release_task(struct task_struct * p);
1692 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1693 extern int force_sigsegv(int, struct task_struct *);
1694 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1695 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1696 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1697 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1698 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1699 extern int kill_pid(struct pid *pid, int sig, int priv);
1700 extern int kill_proc_info(int, struct siginfo *, pid_t);
1701 extern void do_notify_parent(struct task_struct *, int);
1702 extern void force_sig(int, struct task_struct *);
1703 extern void force_sig_specific(int, struct task_struct *);
1704 extern int send_sig(int, struct task_struct *, int);
1705 extern void zap_other_threads(struct task_struct *p);
1706 extern int kill_proc(pid_t, int, int);
1707 extern struct sigqueue *sigqueue_alloc(void);
1708 extern void sigqueue_free(struct sigqueue *);
1709 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1710 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
1711 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1712 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1713
1714 static inline int kill_cad_pid(int sig, int priv)
1715 {
1716 return kill_pid(cad_pid, sig, priv);
1717 }
1718
1719 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1720 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1721 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1722 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1723
1724 static inline int is_si_special(const struct siginfo *info)
1725 {
1726 return info <= SEND_SIG_FORCED;
1727 }
1728
1729 /* True if we are on the alternate signal stack. */
1730
1731 static inline int on_sig_stack(unsigned long sp)
1732 {
1733 return (sp - current->sas_ss_sp < current->sas_ss_size);
1734 }
1735
1736 static inline int sas_ss_flags(unsigned long sp)
1737 {
1738 return (current->sas_ss_size == 0 ? SS_DISABLE
1739 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1740 }
1741
1742 /*
1743 * Routines for handling mm_structs
1744 */
1745 extern struct mm_struct * mm_alloc(void);
1746
1747 /* mmdrop drops the mm and the page tables */
1748 extern void __mmdrop(struct mm_struct *);
1749 static inline void mmdrop(struct mm_struct * mm)
1750 {
1751 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1752 __mmdrop(mm);
1753 }
1754
1755 /* mmput gets rid of the mappings and all user-space */
1756 extern void mmput(struct mm_struct *);
1757 /* Grab a reference to a task's mm, if it is not already going away */
1758 extern struct mm_struct *get_task_mm(struct task_struct *task);
1759 /* Remove the current tasks stale references to the old mm_struct */
1760 extern void mm_release(struct task_struct *, struct mm_struct *);
1761
1762 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1763 extern void flush_thread(void);
1764 extern void exit_thread(void);
1765
1766 extern void exit_files(struct task_struct *);
1767 extern void __cleanup_signal(struct signal_struct *);
1768 extern void __cleanup_sighand(struct sighand_struct *);
1769 extern void exit_itimers(struct signal_struct *);
1770
1771 extern NORET_TYPE void do_group_exit(int);
1772
1773 extern void daemonize(const char *, ...);
1774 extern int allow_signal(int);
1775 extern int disallow_signal(int);
1776
1777 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1778 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1779 struct task_struct *fork_idle(int);
1780
1781 extern void set_task_comm(struct task_struct *tsk, char *from);
1782 extern char *get_task_comm(char *to, struct task_struct *tsk);
1783
1784 #ifdef CONFIG_SMP
1785 extern void wait_task_inactive(struct task_struct * p);
1786 #else
1787 #define wait_task_inactive(p) do { } while (0)
1788 #endif
1789
1790 #define remove_parent(p) list_del_init(&(p)->sibling)
1791 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1792
1793 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1794
1795 #define for_each_process(p) \
1796 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1797
1798 /*
1799 * Careful: do_each_thread/while_each_thread is a double loop so
1800 * 'break' will not work as expected - use goto instead.
1801 */
1802 #define do_each_thread(g, t) \
1803 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1804
1805 #define while_each_thread(g, t) \
1806 while ((t = next_thread(t)) != g)
1807
1808 /* de_thread depends on thread_group_leader not being a pid based check */
1809 #define thread_group_leader(p) (p == p->group_leader)
1810
1811 /* Do to the insanities of de_thread it is possible for a process
1812 * to have the pid of the thread group leader without actually being
1813 * the thread group leader. For iteration through the pids in proc
1814 * all we care about is that we have a task with the appropriate
1815 * pid, we don't actually care if we have the right task.
1816 */
1817 static inline int has_group_leader_pid(struct task_struct *p)
1818 {
1819 return p->pid == p->tgid;
1820 }
1821
1822 static inline
1823 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1824 {
1825 return p1->tgid == p2->tgid;
1826 }
1827
1828 static inline struct task_struct *next_thread(const struct task_struct *p)
1829 {
1830 return list_entry(rcu_dereference(p->thread_group.next),
1831 struct task_struct, thread_group);
1832 }
1833
1834 static inline int thread_group_empty(struct task_struct *p)
1835 {
1836 return list_empty(&p->thread_group);
1837 }
1838
1839 #define delay_group_leader(p) \
1840 (thread_group_leader(p) && !thread_group_empty(p))
1841
1842 /*
1843 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1844 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1845 * pins the final release of task.io_context. Also protects ->cpuset and
1846 * ->cgroup.subsys[].
1847 *
1848 * Nests both inside and outside of read_lock(&tasklist_lock).
1849 * It must not be nested with write_lock_irq(&tasklist_lock),
1850 * neither inside nor outside.
1851 */
1852 static inline void task_lock(struct task_struct *p)
1853 {
1854 spin_lock(&p->alloc_lock);
1855 }
1856
1857 static inline void task_unlock(struct task_struct *p)
1858 {
1859 spin_unlock(&p->alloc_lock);
1860 }
1861
1862 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1863 unsigned long *flags);
1864
1865 static inline void unlock_task_sighand(struct task_struct *tsk,
1866 unsigned long *flags)
1867 {
1868 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1869 }
1870
1871 #ifndef __HAVE_THREAD_FUNCTIONS
1872
1873 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1874 #define task_stack_page(task) ((task)->stack)
1875
1876 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1877 {
1878 *task_thread_info(p) = *task_thread_info(org);
1879 task_thread_info(p)->task = p;
1880 }
1881
1882 static inline unsigned long *end_of_stack(struct task_struct *p)
1883 {
1884 return (unsigned long *)(task_thread_info(p) + 1);
1885 }
1886
1887 #endif
1888
1889 /* set thread flags in other task's structures
1890 * - see asm/thread_info.h for TIF_xxxx flags available
1891 */
1892 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1893 {
1894 set_ti_thread_flag(task_thread_info(tsk), flag);
1895 }
1896
1897 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1898 {
1899 clear_ti_thread_flag(task_thread_info(tsk), flag);
1900 }
1901
1902 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1903 {
1904 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1905 }
1906
1907 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1908 {
1909 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1910 }
1911
1912 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1913 {
1914 return test_ti_thread_flag(task_thread_info(tsk), flag);
1915 }
1916
1917 static inline void set_tsk_need_resched(struct task_struct *tsk)
1918 {
1919 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1920 }
1921
1922 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1923 {
1924 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1925 }
1926
1927 static inline int signal_pending(struct task_struct *p)
1928 {
1929 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1930 }
1931
1932 extern int __fatal_signal_pending(struct task_struct *p);
1933
1934 static inline int fatal_signal_pending(struct task_struct *p)
1935 {
1936 return signal_pending(p) && __fatal_signal_pending(p);
1937 }
1938
1939 static inline int need_resched(void)
1940 {
1941 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1942 }
1943
1944 /*
1945 * cond_resched() and cond_resched_lock(): latency reduction via
1946 * explicit rescheduling in places that are safe. The return
1947 * value indicates whether a reschedule was done in fact.
1948 * cond_resched_lock() will drop the spinlock before scheduling,
1949 * cond_resched_softirq() will enable bhs before scheduling.
1950 */
1951 #ifdef CONFIG_PREEMPT
1952 static inline int cond_resched(void)
1953 {
1954 return 0;
1955 }
1956 #else
1957 extern int _cond_resched(void);
1958 static inline int cond_resched(void)
1959 {
1960 return _cond_resched();
1961 }
1962 #endif
1963 extern int cond_resched_lock(spinlock_t * lock);
1964 extern int cond_resched_softirq(void);
1965
1966 /*
1967 * Does a critical section need to be broken due to another
1968 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1969 * but a general need for low latency)
1970 */
1971 static inline int spin_needbreak(spinlock_t *lock)
1972 {
1973 #ifdef CONFIG_PREEMPT
1974 return spin_is_contended(lock);
1975 #else
1976 return 0;
1977 #endif
1978 }
1979
1980 /*
1981 * Reevaluate whether the task has signals pending delivery.
1982 * Wake the task if so.
1983 * This is required every time the blocked sigset_t changes.
1984 * callers must hold sighand->siglock.
1985 */
1986 extern void recalc_sigpending_and_wake(struct task_struct *t);
1987 extern void recalc_sigpending(void);
1988
1989 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1990
1991 /*
1992 * Wrappers for p->thread_info->cpu access. No-op on UP.
1993 */
1994 #ifdef CONFIG_SMP
1995
1996 static inline unsigned int task_cpu(const struct task_struct *p)
1997 {
1998 return task_thread_info(p)->cpu;
1999 }
2000
2001 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2002
2003 #else
2004
2005 static inline unsigned int task_cpu(const struct task_struct *p)
2006 {
2007 return 0;
2008 }
2009
2010 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2011 {
2012 }
2013
2014 #endif /* CONFIG_SMP */
2015
2016 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2017 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2018 #else
2019 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
2020 {
2021 mm->mmap_base = TASK_UNMAPPED_BASE;
2022 mm->get_unmapped_area = arch_get_unmapped_area;
2023 mm->unmap_area = arch_unmap_area;
2024 }
2025 #endif
2026
2027 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
2028 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2029
2030 extern int sched_mc_power_savings, sched_smt_power_savings;
2031
2032 extern void normalize_rt_tasks(void);
2033
2034 #ifdef CONFIG_GROUP_SCHED
2035
2036 extern struct task_group init_task_group;
2037
2038 extern struct task_group *sched_create_group(void);
2039 extern void sched_destroy_group(struct task_group *tg);
2040 extern void sched_move_task(struct task_struct *tsk);
2041 #ifdef CONFIG_FAIR_GROUP_SCHED
2042 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2043 extern unsigned long sched_group_shares(struct task_group *tg);
2044 #endif
2045 #ifdef CONFIG_RT_GROUP_SCHED
2046 extern int sched_group_set_rt_runtime(struct task_group *tg,
2047 long rt_runtime_us);
2048 extern long sched_group_rt_runtime(struct task_group *tg);
2049 #endif
2050 #endif
2051
2052 #ifdef CONFIG_TASK_XACCT
2053 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2054 {
2055 tsk->rchar += amt;
2056 }
2057
2058 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2059 {
2060 tsk->wchar += amt;
2061 }
2062
2063 static inline void inc_syscr(struct task_struct *tsk)
2064 {
2065 tsk->syscr++;
2066 }
2067
2068 static inline void inc_syscw(struct task_struct *tsk)
2069 {
2070 tsk->syscw++;
2071 }
2072 #else
2073 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2074 {
2075 }
2076
2077 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2078 {
2079 }
2080
2081 static inline void inc_syscr(struct task_struct *tsk)
2082 {
2083 }
2084
2085 static inline void inc_syscw(struct task_struct *tsk)
2086 {
2087 }
2088 #endif
2089
2090 #ifdef CONFIG_SMP
2091 void migration_init(void);
2092 #else
2093 static inline void migration_init(void)
2094 {
2095 }
2096 #endif
2097
2098 #ifndef TASK_SIZE_OF
2099 #define TASK_SIZE_OF(tsk) TASK_SIZE
2100 #endif
2101
2102 #endif /* __KERNEL__ */
2103
2104 #endif