]> git.proxmox.com Git - mirror_ubuntu-kernels.git/blob - include/linux/sched.h
[PATCH] introduce lock_task_sighand() helper
[mirror_ubuntu-kernels.git] / include / linux / sched.h
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 #include <asm/param.h> /* for HZ */
5
6 #include <linux/config.h>
7 #include <linux/capability.h>
8 #include <linux/threads.h>
9 #include <linux/kernel.h>
10 #include <linux/types.h>
11 #include <linux/timex.h>
12 #include <linux/jiffies.h>
13 #include <linux/rbtree.h>
14 #include <linux/thread_info.h>
15 #include <linux/cpumask.h>
16 #include <linux/errno.h>
17 #include <linux/nodemask.h>
18
19 #include <asm/system.h>
20 #include <asm/semaphore.h>
21 #include <asm/page.h>
22 #include <asm/ptrace.h>
23 #include <asm/mmu.h>
24 #include <asm/cputime.h>
25
26 #include <linux/smp.h>
27 #include <linux/sem.h>
28 #include <linux/signal.h>
29 #include <linux/securebits.h>
30 #include <linux/fs_struct.h>
31 #include <linux/compiler.h>
32 #include <linux/completion.h>
33 #include <linux/pid.h>
34 #include <linux/percpu.h>
35 #include <linux/topology.h>
36 #include <linux/seccomp.h>
37 #include <linux/rcupdate.h>
38 #include <linux/futex.h>
39
40 #include <linux/auxvec.h> /* For AT_VECTOR_SIZE */
41
42 struct exec_domain;
43
44 /*
45 * cloning flags:
46 */
47 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
48 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
49 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
50 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
51 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
52 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
53 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
54 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
55 #define CLONE_THREAD 0x00010000 /* Same thread group? */
56 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
57 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
58 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
59 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
60 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
61 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
62 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
63 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
64 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
65
66 /*
67 * List of flags we want to share for kernel threads,
68 * if only because they are not used by them anyway.
69 */
70 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
71
72 /*
73 * These are the constant used to fake the fixed-point load-average
74 * counting. Some notes:
75 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
76 * a load-average precision of 10 bits integer + 11 bits fractional
77 * - if you want to count load-averages more often, you need more
78 * precision, or rounding will get you. With 2-second counting freq,
79 * the EXP_n values would be 1981, 2034 and 2043 if still using only
80 * 11 bit fractions.
81 */
82 extern unsigned long avenrun[]; /* Load averages */
83
84 #define FSHIFT 11 /* nr of bits of precision */
85 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
86 #define LOAD_FREQ (5*HZ) /* 5 sec intervals */
87 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
88 #define EXP_5 2014 /* 1/exp(5sec/5min) */
89 #define EXP_15 2037 /* 1/exp(5sec/15min) */
90
91 #define CALC_LOAD(load,exp,n) \
92 load *= exp; \
93 load += n*(FIXED_1-exp); \
94 load >>= FSHIFT;
95
96 extern unsigned long total_forks;
97 extern int nr_threads;
98 extern int last_pid;
99 DECLARE_PER_CPU(unsigned long, process_counts);
100 extern int nr_processes(void);
101 extern unsigned long nr_running(void);
102 extern unsigned long nr_uninterruptible(void);
103 extern unsigned long nr_iowait(void);
104
105 #include <linux/time.h>
106 #include <linux/param.h>
107 #include <linux/resource.h>
108 #include <linux/timer.h>
109 #include <linux/hrtimer.h>
110
111 #include <asm/processor.h>
112
113 /*
114 * Task state bitmask. NOTE! These bits are also
115 * encoded in fs/proc/array.c: get_task_state().
116 *
117 * We have two separate sets of flags: task->state
118 * is about runnability, while task->exit_state are
119 * about the task exiting. Confusing, but this way
120 * modifying one set can't modify the other one by
121 * mistake.
122 */
123 #define TASK_RUNNING 0
124 #define TASK_INTERRUPTIBLE 1
125 #define TASK_UNINTERRUPTIBLE 2
126 #define TASK_STOPPED 4
127 #define TASK_TRACED 8
128 /* in tsk->exit_state */
129 #define EXIT_ZOMBIE 16
130 #define EXIT_DEAD 32
131 /* in tsk->state again */
132 #define TASK_NONINTERACTIVE 64
133
134 #define __set_task_state(tsk, state_value) \
135 do { (tsk)->state = (state_value); } while (0)
136 #define set_task_state(tsk, state_value) \
137 set_mb((tsk)->state, (state_value))
138
139 /*
140 * set_current_state() includes a barrier so that the write of current->state
141 * is correctly serialised wrt the caller's subsequent test of whether to
142 * actually sleep:
143 *
144 * set_current_state(TASK_UNINTERRUPTIBLE);
145 * if (do_i_need_to_sleep())
146 * schedule();
147 *
148 * If the caller does not need such serialisation then use __set_current_state()
149 */
150 #define __set_current_state(state_value) \
151 do { current->state = (state_value); } while (0)
152 #define set_current_state(state_value) \
153 set_mb(current->state, (state_value))
154
155 /* Task command name length */
156 #define TASK_COMM_LEN 16
157
158 /*
159 * Scheduling policies
160 */
161 #define SCHED_NORMAL 0
162 #define SCHED_FIFO 1
163 #define SCHED_RR 2
164 #define SCHED_BATCH 3
165
166 struct sched_param {
167 int sched_priority;
168 };
169
170 #ifdef __KERNEL__
171
172 #include <linux/spinlock.h>
173
174 /*
175 * This serializes "schedule()" and also protects
176 * the run-queue from deletions/modifications (but
177 * _adding_ to the beginning of the run-queue has
178 * a separate lock).
179 */
180 extern rwlock_t tasklist_lock;
181 extern spinlock_t mmlist_lock;
182
183 typedef struct task_struct task_t;
184
185 extern void sched_init(void);
186 extern void sched_init_smp(void);
187 extern void init_idle(task_t *idle, int cpu);
188
189 extern cpumask_t nohz_cpu_mask;
190
191 extern void show_state(void);
192 extern void show_regs(struct pt_regs *);
193
194 /*
195 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
196 * task), SP is the stack pointer of the first frame that should be shown in the back
197 * trace (or NULL if the entire call-chain of the task should be shown).
198 */
199 extern void show_stack(struct task_struct *task, unsigned long *sp);
200
201 void io_schedule(void);
202 long io_schedule_timeout(long timeout);
203
204 extern void cpu_init (void);
205 extern void trap_init(void);
206 extern void update_process_times(int user);
207 extern void scheduler_tick(void);
208
209 #ifdef CONFIG_DETECT_SOFTLOCKUP
210 extern void softlockup_tick(void);
211 extern void spawn_softlockup_task(void);
212 extern void touch_softlockup_watchdog(void);
213 #else
214 static inline void softlockup_tick(void)
215 {
216 }
217 static inline void spawn_softlockup_task(void)
218 {
219 }
220 static inline void touch_softlockup_watchdog(void)
221 {
222 }
223 #endif
224
225
226 /* Attach to any functions which should be ignored in wchan output. */
227 #define __sched __attribute__((__section__(".sched.text")))
228 /* Is this address in the __sched functions? */
229 extern int in_sched_functions(unsigned long addr);
230
231 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
232 extern signed long FASTCALL(schedule_timeout(signed long timeout));
233 extern signed long schedule_timeout_interruptible(signed long timeout);
234 extern signed long schedule_timeout_uninterruptible(signed long timeout);
235 asmlinkage void schedule(void);
236
237 struct namespace;
238
239 /* Maximum number of active map areas.. This is a random (large) number */
240 #define DEFAULT_MAX_MAP_COUNT 65536
241
242 extern int sysctl_max_map_count;
243
244 #include <linux/aio.h>
245
246 extern unsigned long
247 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
248 unsigned long, unsigned long);
249 extern unsigned long
250 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
251 unsigned long len, unsigned long pgoff,
252 unsigned long flags);
253 extern void arch_unmap_area(struct mm_struct *, unsigned long);
254 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
255
256 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
257 /*
258 * The mm counters are not protected by its page_table_lock,
259 * so must be incremented atomically.
260 */
261 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
262 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
263 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
264 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
265 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
266 typedef atomic_long_t mm_counter_t;
267
268 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
269 /*
270 * The mm counters are protected by its page_table_lock,
271 * so can be incremented directly.
272 */
273 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
274 #define get_mm_counter(mm, member) ((mm)->_##member)
275 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
276 #define inc_mm_counter(mm, member) (mm)->_##member++
277 #define dec_mm_counter(mm, member) (mm)->_##member--
278 typedef unsigned long mm_counter_t;
279
280 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
281
282 #define get_mm_rss(mm) \
283 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
284 #define update_hiwater_rss(mm) do { \
285 unsigned long _rss = get_mm_rss(mm); \
286 if ((mm)->hiwater_rss < _rss) \
287 (mm)->hiwater_rss = _rss; \
288 } while (0)
289 #define update_hiwater_vm(mm) do { \
290 if ((mm)->hiwater_vm < (mm)->total_vm) \
291 (mm)->hiwater_vm = (mm)->total_vm; \
292 } while (0)
293
294 struct mm_struct {
295 struct vm_area_struct * mmap; /* list of VMAs */
296 struct rb_root mm_rb;
297 struct vm_area_struct * mmap_cache; /* last find_vma result */
298 unsigned long (*get_unmapped_area) (struct file *filp,
299 unsigned long addr, unsigned long len,
300 unsigned long pgoff, unsigned long flags);
301 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
302 unsigned long mmap_base; /* base of mmap area */
303 unsigned long task_size; /* size of task vm space */
304 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
305 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
306 pgd_t * pgd;
307 atomic_t mm_users; /* How many users with user space? */
308 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
309 int map_count; /* number of VMAs */
310 struct rw_semaphore mmap_sem;
311 spinlock_t page_table_lock; /* Protects page tables and some counters */
312
313 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
314 * together off init_mm.mmlist, and are protected
315 * by mmlist_lock
316 */
317
318 /* Special counters, in some configurations protected by the
319 * page_table_lock, in other configurations by being atomic.
320 */
321 mm_counter_t _file_rss;
322 mm_counter_t _anon_rss;
323
324 unsigned long hiwater_rss; /* High-watermark of RSS usage */
325 unsigned long hiwater_vm; /* High-water virtual memory usage */
326
327 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
328 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
329 unsigned long start_code, end_code, start_data, end_data;
330 unsigned long start_brk, brk, start_stack;
331 unsigned long arg_start, arg_end, env_start, env_end;
332
333 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
334
335 unsigned dumpable:2;
336 cpumask_t cpu_vm_mask;
337
338 /* Architecture-specific MM context */
339 mm_context_t context;
340
341 /* Token based thrashing protection. */
342 unsigned long swap_token_time;
343 char recent_pagein;
344
345 /* coredumping support */
346 int core_waiters;
347 struct completion *core_startup_done, core_done;
348
349 /* aio bits */
350 rwlock_t ioctx_list_lock;
351 struct kioctx *ioctx_list;
352 };
353
354 struct sighand_struct {
355 atomic_t count;
356 struct k_sigaction action[_NSIG];
357 spinlock_t siglock;
358 };
359
360 /*
361 * NOTE! "signal_struct" does not have it's own
362 * locking, because a shared signal_struct always
363 * implies a shared sighand_struct, so locking
364 * sighand_struct is always a proper superset of
365 * the locking of signal_struct.
366 */
367 struct signal_struct {
368 atomic_t count;
369 atomic_t live;
370
371 wait_queue_head_t wait_chldexit; /* for wait4() */
372
373 /* current thread group signal load-balancing target: */
374 task_t *curr_target;
375
376 /* shared signal handling: */
377 struct sigpending shared_pending;
378
379 /* thread group exit support */
380 int group_exit_code;
381 /* overloaded:
382 * - notify group_exit_task when ->count is equal to notify_count
383 * - everyone except group_exit_task is stopped during signal delivery
384 * of fatal signals, group_exit_task processes the signal.
385 */
386 struct task_struct *group_exit_task;
387 int notify_count;
388
389 /* thread group stop support, overloads group_exit_code too */
390 int group_stop_count;
391 unsigned int flags; /* see SIGNAL_* flags below */
392
393 /* POSIX.1b Interval Timers */
394 struct list_head posix_timers;
395
396 /* ITIMER_REAL timer for the process */
397 struct hrtimer real_timer;
398 struct task_struct *tsk;
399 ktime_t it_real_incr;
400
401 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
402 cputime_t it_prof_expires, it_virt_expires;
403 cputime_t it_prof_incr, it_virt_incr;
404
405 /* job control IDs */
406 pid_t pgrp;
407 pid_t tty_old_pgrp;
408 pid_t session;
409 /* boolean value for session group leader */
410 int leader;
411
412 struct tty_struct *tty; /* NULL if no tty */
413
414 /*
415 * Cumulative resource counters for dead threads in the group,
416 * and for reaped dead child processes forked by this group.
417 * Live threads maintain their own counters and add to these
418 * in __exit_signal, except for the group leader.
419 */
420 cputime_t utime, stime, cutime, cstime;
421 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
422 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
423
424 /*
425 * Cumulative ns of scheduled CPU time for dead threads in the
426 * group, not including a zombie group leader. (This only differs
427 * from jiffies_to_ns(utime + stime) if sched_clock uses something
428 * other than jiffies.)
429 */
430 unsigned long long sched_time;
431
432 /*
433 * We don't bother to synchronize most readers of this at all,
434 * because there is no reader checking a limit that actually needs
435 * to get both rlim_cur and rlim_max atomically, and either one
436 * alone is a single word that can safely be read normally.
437 * getrlimit/setrlimit use task_lock(current->group_leader) to
438 * protect this instead of the siglock, because they really
439 * have no need to disable irqs.
440 */
441 struct rlimit rlim[RLIM_NLIMITS];
442
443 struct list_head cpu_timers[3];
444
445 /* keep the process-shared keyrings here so that they do the right
446 * thing in threads created with CLONE_THREAD */
447 #ifdef CONFIG_KEYS
448 struct key *session_keyring; /* keyring inherited over fork */
449 struct key *process_keyring; /* keyring private to this process */
450 #endif
451 };
452
453 /* Context switch must be unlocked if interrupts are to be enabled */
454 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
455 # define __ARCH_WANT_UNLOCKED_CTXSW
456 #endif
457
458 /*
459 * Bits in flags field of signal_struct.
460 */
461 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
462 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
463 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
464 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
465
466
467 /*
468 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
469 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
470 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
471 * values are inverted: lower p->prio value means higher priority.
472 *
473 * The MAX_USER_RT_PRIO value allows the actual maximum
474 * RT priority to be separate from the value exported to
475 * user-space. This allows kernel threads to set their
476 * priority to a value higher than any user task. Note:
477 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
478 */
479
480 #define MAX_USER_RT_PRIO 100
481 #define MAX_RT_PRIO MAX_USER_RT_PRIO
482
483 #define MAX_PRIO (MAX_RT_PRIO + 40)
484
485 #define rt_task(p) (unlikely((p)->prio < MAX_RT_PRIO))
486
487 /*
488 * Some day this will be a full-fledged user tracking system..
489 */
490 struct user_struct {
491 atomic_t __count; /* reference count */
492 atomic_t processes; /* How many processes does this user have? */
493 atomic_t files; /* How many open files does this user have? */
494 atomic_t sigpending; /* How many pending signals does this user have? */
495 #ifdef CONFIG_INOTIFY
496 atomic_t inotify_watches; /* How many inotify watches does this user have? */
497 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
498 #endif
499 /* protected by mq_lock */
500 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
501 unsigned long locked_shm; /* How many pages of mlocked shm ? */
502
503 #ifdef CONFIG_KEYS
504 struct key *uid_keyring; /* UID specific keyring */
505 struct key *session_keyring; /* UID's default session keyring */
506 #endif
507
508 /* Hash table maintenance information */
509 struct list_head uidhash_list;
510 uid_t uid;
511 };
512
513 extern struct user_struct *find_user(uid_t);
514
515 extern struct user_struct root_user;
516 #define INIT_USER (&root_user)
517
518 typedef struct prio_array prio_array_t;
519 struct backing_dev_info;
520 struct reclaim_state;
521
522 #ifdef CONFIG_SCHEDSTATS
523 struct sched_info {
524 /* cumulative counters */
525 unsigned long cpu_time, /* time spent on the cpu */
526 run_delay, /* time spent waiting on a runqueue */
527 pcnt; /* # of timeslices run on this cpu */
528
529 /* timestamps */
530 unsigned long last_arrival, /* when we last ran on a cpu */
531 last_queued; /* when we were last queued to run */
532 };
533
534 extern struct file_operations proc_schedstat_operations;
535 #endif
536
537 enum idle_type
538 {
539 SCHED_IDLE,
540 NOT_IDLE,
541 NEWLY_IDLE,
542 MAX_IDLE_TYPES
543 };
544
545 /*
546 * sched-domains (multiprocessor balancing) declarations:
547 */
548 #ifdef CONFIG_SMP
549 #define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
550
551 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
552 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
553 #define SD_BALANCE_EXEC 4 /* Balance on exec */
554 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
555 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
556 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
557 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
558 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
559
560 struct sched_group {
561 struct sched_group *next; /* Must be a circular list */
562 cpumask_t cpumask;
563
564 /*
565 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
566 * single CPU. This is read only (except for setup, hotplug CPU).
567 */
568 unsigned long cpu_power;
569 };
570
571 struct sched_domain {
572 /* These fields must be setup */
573 struct sched_domain *parent; /* top domain must be null terminated */
574 struct sched_group *groups; /* the balancing groups of the domain */
575 cpumask_t span; /* span of all CPUs in this domain */
576 unsigned long min_interval; /* Minimum balance interval ms */
577 unsigned long max_interval; /* Maximum balance interval ms */
578 unsigned int busy_factor; /* less balancing by factor if busy */
579 unsigned int imbalance_pct; /* No balance until over watermark */
580 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
581 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
582 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
583 unsigned int busy_idx;
584 unsigned int idle_idx;
585 unsigned int newidle_idx;
586 unsigned int wake_idx;
587 unsigned int forkexec_idx;
588 int flags; /* See SD_* */
589
590 /* Runtime fields. */
591 unsigned long last_balance; /* init to jiffies. units in jiffies */
592 unsigned int balance_interval; /* initialise to 1. units in ms. */
593 unsigned int nr_balance_failed; /* initialise to 0 */
594
595 #ifdef CONFIG_SCHEDSTATS
596 /* load_balance() stats */
597 unsigned long lb_cnt[MAX_IDLE_TYPES];
598 unsigned long lb_failed[MAX_IDLE_TYPES];
599 unsigned long lb_balanced[MAX_IDLE_TYPES];
600 unsigned long lb_imbalance[MAX_IDLE_TYPES];
601 unsigned long lb_gained[MAX_IDLE_TYPES];
602 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
603 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
604 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
605
606 /* Active load balancing */
607 unsigned long alb_cnt;
608 unsigned long alb_failed;
609 unsigned long alb_pushed;
610
611 /* SD_BALANCE_EXEC stats */
612 unsigned long sbe_cnt;
613 unsigned long sbe_balanced;
614 unsigned long sbe_pushed;
615
616 /* SD_BALANCE_FORK stats */
617 unsigned long sbf_cnt;
618 unsigned long sbf_balanced;
619 unsigned long sbf_pushed;
620
621 /* try_to_wake_up() stats */
622 unsigned long ttwu_wake_remote;
623 unsigned long ttwu_move_affine;
624 unsigned long ttwu_move_balance;
625 #endif
626 };
627
628 extern void partition_sched_domains(cpumask_t *partition1,
629 cpumask_t *partition2);
630
631 /*
632 * Maximum cache size the migration-costs auto-tuning code will
633 * search from:
634 */
635 extern unsigned int max_cache_size;
636
637 #endif /* CONFIG_SMP */
638
639
640 struct io_context; /* See blkdev.h */
641 void exit_io_context(void);
642 struct cpuset;
643
644 #define NGROUPS_SMALL 32
645 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
646 struct group_info {
647 int ngroups;
648 atomic_t usage;
649 gid_t small_block[NGROUPS_SMALL];
650 int nblocks;
651 gid_t *blocks[0];
652 };
653
654 /*
655 * get_group_info() must be called with the owning task locked (via task_lock())
656 * when task != current. The reason being that the vast majority of callers are
657 * looking at current->group_info, which can not be changed except by the
658 * current task. Changing current->group_info requires the task lock, too.
659 */
660 #define get_group_info(group_info) do { \
661 atomic_inc(&(group_info)->usage); \
662 } while (0)
663
664 #define put_group_info(group_info) do { \
665 if (atomic_dec_and_test(&(group_info)->usage)) \
666 groups_free(group_info); \
667 } while (0)
668
669 extern struct group_info *groups_alloc(int gidsetsize);
670 extern void groups_free(struct group_info *group_info);
671 extern int set_current_groups(struct group_info *group_info);
672 extern int groups_search(struct group_info *group_info, gid_t grp);
673 /* access the groups "array" with this macro */
674 #define GROUP_AT(gi, i) \
675 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
676
677 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
678 extern void prefetch_stack(struct task_struct*);
679 #else
680 static inline void prefetch_stack(struct task_struct *t) { }
681 #endif
682
683 struct audit_context; /* See audit.c */
684 struct mempolicy;
685
686 struct task_struct {
687 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
688 struct thread_info *thread_info;
689 atomic_t usage;
690 unsigned long flags; /* per process flags, defined below */
691 unsigned long ptrace;
692
693 int lock_depth; /* BKL lock depth */
694
695 #if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
696 int oncpu;
697 #endif
698 int prio, static_prio;
699 struct list_head run_list;
700 prio_array_t *array;
701
702 unsigned short ioprio;
703 unsigned int btrace_seq;
704
705 unsigned long sleep_avg;
706 unsigned long long timestamp, last_ran;
707 unsigned long long sched_time; /* sched_clock time spent running */
708 int activated;
709
710 unsigned long policy;
711 cpumask_t cpus_allowed;
712 unsigned int time_slice, first_time_slice;
713
714 #ifdef CONFIG_SCHEDSTATS
715 struct sched_info sched_info;
716 #endif
717
718 struct list_head tasks;
719 /*
720 * ptrace_list/ptrace_children forms the list of my children
721 * that were stolen by a ptracer.
722 */
723 struct list_head ptrace_children;
724 struct list_head ptrace_list;
725
726 struct mm_struct *mm, *active_mm;
727
728 /* task state */
729 struct linux_binfmt *binfmt;
730 long exit_state;
731 int exit_code, exit_signal;
732 int pdeath_signal; /* The signal sent when the parent dies */
733 /* ??? */
734 unsigned long personality;
735 unsigned did_exec:1;
736 pid_t pid;
737 pid_t tgid;
738 /*
739 * pointers to (original) parent process, youngest child, younger sibling,
740 * older sibling, respectively. (p->father can be replaced with
741 * p->parent->pid)
742 */
743 struct task_struct *real_parent; /* real parent process (when being debugged) */
744 struct task_struct *parent; /* parent process */
745 /*
746 * children/sibling forms the list of my children plus the
747 * tasks I'm ptracing.
748 */
749 struct list_head children; /* list of my children */
750 struct list_head sibling; /* linkage in my parent's children list */
751 struct task_struct *group_leader; /* threadgroup leader */
752
753 /* PID/PID hash table linkage. */
754 struct pid pids[PIDTYPE_MAX];
755
756 struct completion *vfork_done; /* for vfork() */
757 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
758 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
759
760 unsigned long rt_priority;
761 cputime_t utime, stime;
762 unsigned long nvcsw, nivcsw; /* context switch counts */
763 struct timespec start_time;
764 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
765 unsigned long min_flt, maj_flt;
766
767 cputime_t it_prof_expires, it_virt_expires;
768 unsigned long long it_sched_expires;
769 struct list_head cpu_timers[3];
770
771 /* process credentials */
772 uid_t uid,euid,suid,fsuid;
773 gid_t gid,egid,sgid,fsgid;
774 struct group_info *group_info;
775 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
776 unsigned keep_capabilities:1;
777 struct user_struct *user;
778 #ifdef CONFIG_KEYS
779 struct key *request_key_auth; /* assumed request_key authority */
780 struct key *thread_keyring; /* keyring private to this thread */
781 unsigned char jit_keyring; /* default keyring to attach requested keys to */
782 #endif
783 int oomkilladj; /* OOM kill score adjustment (bit shift). */
784 char comm[TASK_COMM_LEN]; /* executable name excluding path
785 - access with [gs]et_task_comm (which lock
786 it with task_lock())
787 - initialized normally by flush_old_exec */
788 /* file system info */
789 int link_count, total_link_count;
790 /* ipc stuff */
791 struct sysv_sem sysvsem;
792 /* CPU-specific state of this task */
793 struct thread_struct thread;
794 /* filesystem information */
795 struct fs_struct *fs;
796 /* open file information */
797 struct files_struct *files;
798 /* namespace */
799 struct namespace *namespace;
800 /* signal handlers */
801 struct signal_struct *signal;
802 struct sighand_struct *sighand;
803
804 sigset_t blocked, real_blocked;
805 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
806 struct sigpending pending;
807
808 unsigned long sas_ss_sp;
809 size_t sas_ss_size;
810 int (*notifier)(void *priv);
811 void *notifier_data;
812 sigset_t *notifier_mask;
813
814 void *security;
815 struct audit_context *audit_context;
816 seccomp_t seccomp;
817
818 /* Thread group tracking */
819 u32 parent_exec_id;
820 u32 self_exec_id;
821 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
822 spinlock_t alloc_lock;
823 /* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
824 spinlock_t proc_lock;
825
826 #ifdef CONFIG_DEBUG_MUTEXES
827 /* mutex deadlock detection */
828 struct mutex_waiter *blocked_on;
829 #endif
830
831 /* journalling filesystem info */
832 void *journal_info;
833
834 /* VM state */
835 struct reclaim_state *reclaim_state;
836
837 struct dentry *proc_dentry;
838 struct backing_dev_info *backing_dev_info;
839
840 struct io_context *io_context;
841
842 unsigned long ptrace_message;
843 siginfo_t *last_siginfo; /* For ptrace use. */
844 /*
845 * current io wait handle: wait queue entry to use for io waits
846 * If this thread is processing aio, this points at the waitqueue
847 * inside the currently handled kiocb. It may be NULL (i.e. default
848 * to a stack based synchronous wait) if its doing sync IO.
849 */
850 wait_queue_t *io_wait;
851 /* i/o counters(bytes read/written, #syscalls */
852 u64 rchar, wchar, syscr, syscw;
853 #if defined(CONFIG_BSD_PROCESS_ACCT)
854 u64 acct_rss_mem1; /* accumulated rss usage */
855 u64 acct_vm_mem1; /* accumulated virtual memory usage */
856 clock_t acct_stimexpd; /* clock_t-converted stime since last update */
857 #endif
858 #ifdef CONFIG_NUMA
859 struct mempolicy *mempolicy;
860 short il_next;
861 #endif
862 #ifdef CONFIG_CPUSETS
863 struct cpuset *cpuset;
864 nodemask_t mems_allowed;
865 int cpuset_mems_generation;
866 int cpuset_mem_spread_rotor;
867 #endif
868 struct robust_list_head __user *robust_list;
869 #ifdef CONFIG_COMPAT
870 struct compat_robust_list_head __user *compat_robust_list;
871 #endif
872
873 atomic_t fs_excl; /* holding fs exclusive resources */
874 struct rcu_head rcu;
875 };
876
877 static inline pid_t process_group(struct task_struct *tsk)
878 {
879 return tsk->signal->pgrp;
880 }
881
882 /**
883 * pid_alive - check that a task structure is not stale
884 * @p: Task structure to be checked.
885 *
886 * Test if a process is not yet dead (at most zombie state)
887 * If pid_alive fails, then pointers within the task structure
888 * can be stale and must not be dereferenced.
889 */
890 static inline int pid_alive(struct task_struct *p)
891 {
892 return p->pids[PIDTYPE_PID].nr != 0;
893 }
894
895 extern void free_task(struct task_struct *tsk);
896 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
897
898 extern void __put_task_struct_cb(struct rcu_head *rhp);
899
900 static inline void put_task_struct(struct task_struct *t)
901 {
902 if (atomic_dec_and_test(&t->usage))
903 call_rcu(&t->rcu, __put_task_struct_cb);
904 }
905
906 /*
907 * Per process flags
908 */
909 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
910 /* Not implemented yet, only for 486*/
911 #define PF_STARTING 0x00000002 /* being created */
912 #define PF_EXITING 0x00000004 /* getting shut down */
913 #define PF_DEAD 0x00000008 /* Dead */
914 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
915 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
916 #define PF_DUMPCORE 0x00000200 /* dumped core */
917 #define PF_SIGNALED 0x00000400 /* killed by a signal */
918 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
919 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
920 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
921 #define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */
922 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
923 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
924 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
925 #define PF_KSWAPD 0x00040000 /* I am kswapd */
926 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
927 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
928 #define PF_SYNCWRITE 0x00200000 /* I am doing a sync write */
929 #define PF_BORROWED_MM 0x00400000 /* I am a kthread doing use_mm */
930 #define PF_RANDOMIZE 0x00800000 /* randomize virtual address space */
931 #define PF_SWAPWRITE 0x01000000 /* Allowed to write to swap */
932 #define PF_SPREAD_PAGE 0x04000000 /* Spread page cache over cpuset */
933 #define PF_SPREAD_SLAB 0x08000000 /* Spread some slab caches over cpuset */
934 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
935
936 /*
937 * Only the _current_ task can read/write to tsk->flags, but other
938 * tasks can access tsk->flags in readonly mode for example
939 * with tsk_used_math (like during threaded core dumping).
940 * There is however an exception to this rule during ptrace
941 * or during fork: the ptracer task is allowed to write to the
942 * child->flags of its traced child (same goes for fork, the parent
943 * can write to the child->flags), because we're guaranteed the
944 * child is not running and in turn not changing child->flags
945 * at the same time the parent does it.
946 */
947 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
948 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
949 #define clear_used_math() clear_stopped_child_used_math(current)
950 #define set_used_math() set_stopped_child_used_math(current)
951 #define conditional_stopped_child_used_math(condition, child) \
952 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
953 #define conditional_used_math(condition) \
954 conditional_stopped_child_used_math(condition, current)
955 #define copy_to_stopped_child_used_math(child) \
956 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
957 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
958 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
959 #define used_math() tsk_used_math(current)
960
961 #ifdef CONFIG_SMP
962 extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
963 #else
964 static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
965 {
966 if (!cpu_isset(0, new_mask))
967 return -EINVAL;
968 return 0;
969 }
970 #endif
971
972 extern unsigned long long sched_clock(void);
973 extern unsigned long long current_sched_time(const task_t *current_task);
974
975 /* sched_exec is called by processes performing an exec */
976 #ifdef CONFIG_SMP
977 extern void sched_exec(void);
978 #else
979 #define sched_exec() {}
980 #endif
981
982 #ifdef CONFIG_HOTPLUG_CPU
983 extern void idle_task_exit(void);
984 #else
985 static inline void idle_task_exit(void) {}
986 #endif
987
988 extern void sched_idle_next(void);
989 extern void set_user_nice(task_t *p, long nice);
990 extern int task_prio(const task_t *p);
991 extern int task_nice(const task_t *p);
992 extern int can_nice(const task_t *p, const int nice);
993 extern int task_curr(const task_t *p);
994 extern int idle_cpu(int cpu);
995 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
996 extern task_t *idle_task(int cpu);
997 extern task_t *curr_task(int cpu);
998 extern void set_curr_task(int cpu, task_t *p);
999
1000 void yield(void);
1001
1002 /*
1003 * The default (Linux) execution domain.
1004 */
1005 extern struct exec_domain default_exec_domain;
1006
1007 union thread_union {
1008 struct thread_info thread_info;
1009 unsigned long stack[THREAD_SIZE/sizeof(long)];
1010 };
1011
1012 #ifndef __HAVE_ARCH_KSTACK_END
1013 static inline int kstack_end(void *addr)
1014 {
1015 /* Reliable end of stack detection:
1016 * Some APM bios versions misalign the stack
1017 */
1018 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1019 }
1020 #endif
1021
1022 extern union thread_union init_thread_union;
1023 extern struct task_struct init_task;
1024
1025 extern struct mm_struct init_mm;
1026
1027 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1028 extern struct task_struct *find_task_by_pid_type(int type, int pid);
1029 extern void set_special_pids(pid_t session, pid_t pgrp);
1030 extern void __set_special_pids(pid_t session, pid_t pgrp);
1031
1032 /* per-UID process charging. */
1033 extern struct user_struct * alloc_uid(uid_t);
1034 static inline struct user_struct *get_uid(struct user_struct *u)
1035 {
1036 atomic_inc(&u->__count);
1037 return u;
1038 }
1039 extern void free_uid(struct user_struct *);
1040 extern void switch_uid(struct user_struct *);
1041
1042 #include <asm/current.h>
1043
1044 extern void do_timer(struct pt_regs *);
1045
1046 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1047 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1048 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1049 unsigned long clone_flags));
1050 #ifdef CONFIG_SMP
1051 extern void kick_process(struct task_struct *tsk);
1052 #else
1053 static inline void kick_process(struct task_struct *tsk) { }
1054 #endif
1055 extern void FASTCALL(sched_fork(task_t * p, int clone_flags));
1056 extern void FASTCALL(sched_exit(task_t * p));
1057
1058 extern int in_group_p(gid_t);
1059 extern int in_egroup_p(gid_t);
1060
1061 extern void proc_caches_init(void);
1062 extern void flush_signals(struct task_struct *);
1063 extern void flush_signal_handlers(struct task_struct *, int force_default);
1064 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1065
1066 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1067 {
1068 unsigned long flags;
1069 int ret;
1070
1071 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1072 ret = dequeue_signal(tsk, mask, info);
1073 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1074
1075 return ret;
1076 }
1077
1078 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1079 sigset_t *mask);
1080 extern void unblock_all_signals(void);
1081 extern void release_task(struct task_struct * p);
1082 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1083 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1084 extern int force_sigsegv(int, struct task_struct *);
1085 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1086 extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1087 extern int kill_pg_info(int, struct siginfo *, pid_t);
1088 extern int kill_proc_info(int, struct siginfo *, pid_t);
1089 extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t);
1090 extern void do_notify_parent(struct task_struct *, int);
1091 extern void force_sig(int, struct task_struct *);
1092 extern void force_sig_specific(int, struct task_struct *);
1093 extern int send_sig(int, struct task_struct *, int);
1094 extern void zap_other_threads(struct task_struct *p);
1095 extern int kill_pg(pid_t, int, int);
1096 extern int kill_proc(pid_t, int, int);
1097 extern struct sigqueue *sigqueue_alloc(void);
1098 extern void sigqueue_free(struct sigqueue *);
1099 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1100 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
1101 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1102 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1103
1104 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1105 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1106 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1107 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1108
1109 static inline int is_si_special(const struct siginfo *info)
1110 {
1111 return info <= SEND_SIG_FORCED;
1112 }
1113
1114 /* True if we are on the alternate signal stack. */
1115
1116 static inline int on_sig_stack(unsigned long sp)
1117 {
1118 return (sp - current->sas_ss_sp < current->sas_ss_size);
1119 }
1120
1121 static inline int sas_ss_flags(unsigned long sp)
1122 {
1123 return (current->sas_ss_size == 0 ? SS_DISABLE
1124 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1125 }
1126
1127 /*
1128 * Routines for handling mm_structs
1129 */
1130 extern struct mm_struct * mm_alloc(void);
1131
1132 /* mmdrop drops the mm and the page tables */
1133 extern void FASTCALL(__mmdrop(struct mm_struct *));
1134 static inline void mmdrop(struct mm_struct * mm)
1135 {
1136 if (atomic_dec_and_test(&mm->mm_count))
1137 __mmdrop(mm);
1138 }
1139
1140 /* mmput gets rid of the mappings and all user-space */
1141 extern void mmput(struct mm_struct *);
1142 /* Grab a reference to a task's mm, if it is not already going away */
1143 extern struct mm_struct *get_task_mm(struct task_struct *task);
1144 /* Remove the current tasks stale references to the old mm_struct */
1145 extern void mm_release(struct task_struct *, struct mm_struct *);
1146
1147 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1148 extern void flush_thread(void);
1149 extern void exit_thread(void);
1150
1151 extern void exit_files(struct task_struct *);
1152 extern void exit_signal(struct task_struct *);
1153 extern void __exit_signal(struct task_struct *);
1154 extern void exit_sighand(struct task_struct *);
1155 extern void __exit_sighand(struct task_struct *);
1156 extern void exit_itimers(struct signal_struct *);
1157
1158 extern NORET_TYPE void do_group_exit(int);
1159
1160 extern void daemonize(const char *, ...);
1161 extern int allow_signal(int);
1162 extern int disallow_signal(int);
1163 extern task_t *child_reaper;
1164
1165 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1166 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1167 task_t *fork_idle(int);
1168
1169 extern void set_task_comm(struct task_struct *tsk, char *from);
1170 extern void get_task_comm(char *to, struct task_struct *tsk);
1171
1172 #ifdef CONFIG_SMP
1173 extern void wait_task_inactive(task_t * p);
1174 #else
1175 #define wait_task_inactive(p) do { } while (0)
1176 #endif
1177
1178 #define remove_parent(p) list_del_init(&(p)->sibling)
1179 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1180
1181 #define next_task(p) list_entry((p)->tasks.next, struct task_struct, tasks)
1182 #define prev_task(p) list_entry((p)->tasks.prev, struct task_struct, tasks)
1183
1184 #define for_each_process(p) \
1185 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1186
1187 /*
1188 * Careful: do_each_thread/while_each_thread is a double loop so
1189 * 'break' will not work as expected - use goto instead.
1190 */
1191 #define do_each_thread(g, t) \
1192 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1193
1194 #define while_each_thread(g, t) \
1195 while ((t = next_thread(t)) != g)
1196
1197 extern task_t * FASTCALL(next_thread(const task_t *p));
1198
1199 #define thread_group_leader(p) (p->pid == p->tgid)
1200
1201 static inline int thread_group_empty(task_t *p)
1202 {
1203 return list_empty(&p->pids[PIDTYPE_TGID].pid_list);
1204 }
1205
1206 #define delay_group_leader(p) \
1207 (thread_group_leader(p) && !thread_group_empty(p))
1208
1209 /*
1210 * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring
1211 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1212 * pins the final release of task.io_context. Also protects ->cpuset.
1213 *
1214 * Nests both inside and outside of read_lock(&tasklist_lock).
1215 * It must not be nested with write_lock_irq(&tasklist_lock),
1216 * neither inside nor outside.
1217 */
1218 static inline void task_lock(struct task_struct *p)
1219 {
1220 spin_lock(&p->alloc_lock);
1221 }
1222
1223 static inline void task_unlock(struct task_struct *p)
1224 {
1225 spin_unlock(&p->alloc_lock);
1226 }
1227
1228 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1229 unsigned long *flags);
1230
1231 static inline void unlock_task_sighand(struct task_struct *tsk,
1232 unsigned long *flags)
1233 {
1234 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1235 }
1236
1237 #ifndef __HAVE_THREAD_FUNCTIONS
1238
1239 #define task_thread_info(task) (task)->thread_info
1240 #define task_stack_page(task) ((void*)((task)->thread_info))
1241
1242 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1243 {
1244 *task_thread_info(p) = *task_thread_info(org);
1245 task_thread_info(p)->task = p;
1246 }
1247
1248 static inline unsigned long *end_of_stack(struct task_struct *p)
1249 {
1250 return (unsigned long *)(p->thread_info + 1);
1251 }
1252
1253 #endif
1254
1255 /* set thread flags in other task's structures
1256 * - see asm/thread_info.h for TIF_xxxx flags available
1257 */
1258 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1259 {
1260 set_ti_thread_flag(task_thread_info(tsk), flag);
1261 }
1262
1263 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1264 {
1265 clear_ti_thread_flag(task_thread_info(tsk), flag);
1266 }
1267
1268 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1269 {
1270 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1271 }
1272
1273 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1274 {
1275 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1276 }
1277
1278 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1279 {
1280 return test_ti_thread_flag(task_thread_info(tsk), flag);
1281 }
1282
1283 static inline void set_tsk_need_resched(struct task_struct *tsk)
1284 {
1285 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1286 }
1287
1288 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1289 {
1290 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1291 }
1292
1293 static inline int signal_pending(struct task_struct *p)
1294 {
1295 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1296 }
1297
1298 static inline int need_resched(void)
1299 {
1300 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1301 }
1302
1303 /*
1304 * cond_resched() and cond_resched_lock(): latency reduction via
1305 * explicit rescheduling in places that are safe. The return
1306 * value indicates whether a reschedule was done in fact.
1307 * cond_resched_lock() will drop the spinlock before scheduling,
1308 * cond_resched_softirq() will enable bhs before scheduling.
1309 */
1310 extern int cond_resched(void);
1311 extern int cond_resched_lock(spinlock_t * lock);
1312 extern int cond_resched_softirq(void);
1313
1314 /*
1315 * Does a critical section need to be broken due to another
1316 * task waiting?:
1317 */
1318 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1319 # define need_lockbreak(lock) ((lock)->break_lock)
1320 #else
1321 # define need_lockbreak(lock) 0
1322 #endif
1323
1324 /*
1325 * Does a critical section need to be broken due to another
1326 * task waiting or preemption being signalled:
1327 */
1328 static inline int lock_need_resched(spinlock_t *lock)
1329 {
1330 if (need_lockbreak(lock) || need_resched())
1331 return 1;
1332 return 0;
1333 }
1334
1335 /* Reevaluate whether the task has signals pending delivery.
1336 This is required every time the blocked sigset_t changes.
1337 callers must hold sighand->siglock. */
1338
1339 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1340 extern void recalc_sigpending(void);
1341
1342 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1343
1344 /*
1345 * Wrappers for p->thread_info->cpu access. No-op on UP.
1346 */
1347 #ifdef CONFIG_SMP
1348
1349 static inline unsigned int task_cpu(const struct task_struct *p)
1350 {
1351 return task_thread_info(p)->cpu;
1352 }
1353
1354 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1355 {
1356 task_thread_info(p)->cpu = cpu;
1357 }
1358
1359 #else
1360
1361 static inline unsigned int task_cpu(const struct task_struct *p)
1362 {
1363 return 0;
1364 }
1365
1366 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1367 {
1368 }
1369
1370 #endif /* CONFIG_SMP */
1371
1372 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1373 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1374 #else
1375 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1376 {
1377 mm->mmap_base = TASK_UNMAPPED_BASE;
1378 mm->get_unmapped_area = arch_get_unmapped_area;
1379 mm->unmap_area = arch_unmap_area;
1380 }
1381 #endif
1382
1383 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1384 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1385
1386 extern void normalize_rt_tasks(void);
1387
1388 #ifdef CONFIG_PM
1389 /*
1390 * Check if a process has been frozen
1391 */
1392 static inline int frozen(struct task_struct *p)
1393 {
1394 return p->flags & PF_FROZEN;
1395 }
1396
1397 /*
1398 * Check if there is a request to freeze a process
1399 */
1400 static inline int freezing(struct task_struct *p)
1401 {
1402 return p->flags & PF_FREEZE;
1403 }
1404
1405 /*
1406 * Request that a process be frozen
1407 * FIXME: SMP problem. We may not modify other process' flags!
1408 */
1409 static inline void freeze(struct task_struct *p)
1410 {
1411 p->flags |= PF_FREEZE;
1412 }
1413
1414 /*
1415 * Wake up a frozen process
1416 */
1417 static inline int thaw_process(struct task_struct *p)
1418 {
1419 if (frozen(p)) {
1420 p->flags &= ~PF_FROZEN;
1421 wake_up_process(p);
1422 return 1;
1423 }
1424 return 0;
1425 }
1426
1427 /*
1428 * freezing is complete, mark process as frozen
1429 */
1430 static inline void frozen_process(struct task_struct *p)
1431 {
1432 p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1433 }
1434
1435 extern void refrigerator(void);
1436 extern int freeze_processes(void);
1437 extern void thaw_processes(void);
1438
1439 static inline int try_to_freeze(void)
1440 {
1441 if (freezing(current)) {
1442 refrigerator();
1443 return 1;
1444 } else
1445 return 0;
1446 }
1447 #else
1448 static inline int frozen(struct task_struct *p) { return 0; }
1449 static inline int freezing(struct task_struct *p) { return 0; }
1450 static inline void freeze(struct task_struct *p) { BUG(); }
1451 static inline int thaw_process(struct task_struct *p) { return 1; }
1452 static inline void frozen_process(struct task_struct *p) { BUG(); }
1453
1454 static inline void refrigerator(void) {}
1455 static inline int freeze_processes(void) { BUG(); return 0; }
1456 static inline void thaw_processes(void) {}
1457
1458 static inline int try_to_freeze(void) { return 0; }
1459
1460 #endif /* CONFIG_PM */
1461 #endif /* __KERNEL__ */
1462
1463 #endif