blob: a30e91f40da6740f791a89989e2edb44e18ed5b2 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#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
38struct exec_domain;
39
40/*
41 * cloning flags:
42 */
43#define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
44#define CLONE_VM 0x00000100 /* set if VM shared between processes */
45#define CLONE_FS 0x00000200 /* set if fs info shared between processes */
46#define CLONE_FILES 0x00000400 /* set if open files shared between processes */
47#define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
48#define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
49#define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
50#define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
51#define CLONE_THREAD 0x00010000 /* Same thread group? */
52#define CLONE_NEWNS 0x00020000 /* New namespace group? */
53#define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
54#define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
55#define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
56#define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
57#define CLONE_DETACHED 0x00400000 /* Unused, ignored */
58#define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
59#define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
60#define CLONE_STOPPED 0x02000000 /* Start in stopped state */
61
62/*
63 * List of flags we want to share for kernel threads,
64 * if only because they are not used by them anyway.
65 */
66#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
67
68/*
69 * These are the constant used to fake the fixed-point load-average
70 * counting. Some notes:
71 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
72 * a load-average precision of 10 bits integer + 11 bits fractional
73 * - if you want to count load-averages more often, you need more
74 * precision, or rounding will get you. With 2-second counting freq,
75 * the EXP_n values would be 1981, 2034 and 2043 if still using only
76 * 11 bit fractions.
77 */
78extern unsigned long avenrun[]; /* Load averages */
79
80#define FSHIFT 11 /* nr of bits of precision */
81#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
82#define LOAD_FREQ (5*HZ) /* 5 sec intervals */
83#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
84#define EXP_5 2014 /* 1/exp(5sec/5min) */
85#define EXP_15 2037 /* 1/exp(5sec/15min) */
86
87#define CALC_LOAD(load,exp,n) \
88 load *= exp; \
89 load += n*(FIXED_1-exp); \
90 load >>= FSHIFT;
91
92extern unsigned long total_forks;
93extern int nr_threads;
94extern int last_pid;
95DECLARE_PER_CPU(unsigned long, process_counts);
96extern int nr_processes(void);
97extern unsigned long nr_running(void);
98extern unsigned long nr_uninterruptible(void);
99extern unsigned long nr_iowait(void);
100
101#include <linux/time.h>
102#include <linux/param.h>
103#include <linux/resource.h>
104#include <linux/timer.h>
105
106#include <asm/processor.h>
107
108#define TASK_RUNNING 0
109#define TASK_INTERRUPTIBLE 1
110#define TASK_UNINTERRUPTIBLE 2
111#define TASK_STOPPED 4
112#define TASK_TRACED 8
113#define EXIT_ZOMBIE 16
114#define EXIT_DEAD 32
115
116#define __set_task_state(tsk, state_value) \
117 do { (tsk)->state = (state_value); } while (0)
118#define set_task_state(tsk, state_value) \
119 set_mb((tsk)->state, (state_value))
120
121#define __set_current_state(state_value) \
122 do { current->state = (state_value); } while (0)
123#define set_current_state(state_value) \
124 set_mb(current->state, (state_value))
125
126/* Task command name length */
127#define TASK_COMM_LEN 16
128
129/*
130 * Scheduling policies
131 */
132#define SCHED_NORMAL 0
133#define SCHED_FIFO 1
134#define SCHED_RR 2
135
136struct sched_param {
137 int sched_priority;
138};
139
140#ifdef __KERNEL__
141
142#include <linux/spinlock.h>
143
144/*
145 * This serializes "schedule()" and also protects
146 * the run-queue from deletions/modifications (but
147 * _adding_ to the beginning of the run-queue has
148 * a separate lock).
149 */
150extern rwlock_t tasklist_lock;
151extern spinlock_t mmlist_lock;
152
153typedef struct task_struct task_t;
154
155extern void sched_init(void);
156extern void sched_init_smp(void);
157extern void init_idle(task_t *idle, int cpu);
158
159extern cpumask_t nohz_cpu_mask;
160
161extern void show_state(void);
162extern void show_regs(struct pt_regs *);
163
164/*
165 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
166 * task), SP is the stack pointer of the first frame that should be shown in the back
167 * trace (or NULL if the entire call-chain of the task should be shown).
168 */
169extern void show_stack(struct task_struct *task, unsigned long *sp);
170
171void io_schedule(void);
172long io_schedule_timeout(long timeout);
173
174extern void cpu_init (void);
175extern void trap_init(void);
176extern void update_process_times(int user);
177extern void scheduler_tick(void);
178
179/* Attach to any functions which should be ignored in wchan output. */
180#define __sched __attribute__((__section__(".sched.text")))
181/* Is this address in the __sched functions? */
182extern int in_sched_functions(unsigned long addr);
183
184#define MAX_SCHEDULE_TIMEOUT LONG_MAX
185extern signed long FASTCALL(schedule_timeout(signed long timeout));
186asmlinkage void schedule(void);
187
188struct namespace;
189
190/* Maximum number of active map areas.. This is a random (large) number */
191#define DEFAULT_MAX_MAP_COUNT 65536
192
193extern int sysctl_max_map_count;
194
195#include <linux/aio.h>
196
197extern unsigned long
198arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
199 unsigned long, unsigned long);
200extern unsigned long
201arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
202 unsigned long len, unsigned long pgoff,
203 unsigned long flags);
204extern void arch_unmap_area(struct vm_area_struct *area);
205extern void arch_unmap_area_topdown(struct vm_area_struct *area);
206
207#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
208#define get_mm_counter(mm, member) ((mm)->_##member)
209#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
210#define inc_mm_counter(mm, member) (mm)->_##member++
211#define dec_mm_counter(mm, member) (mm)->_##member--
212typedef unsigned long mm_counter_t;
213
214struct mm_struct {
215 struct vm_area_struct * mmap; /* list of VMAs */
216 struct rb_root mm_rb;
217 struct vm_area_struct * mmap_cache; /* last find_vma result */
218 unsigned long (*get_unmapped_area) (struct file *filp,
219 unsigned long addr, unsigned long len,
220 unsigned long pgoff, unsigned long flags);
221 void (*unmap_area) (struct vm_area_struct *area);
222 unsigned long mmap_base; /* base of mmap area */
223 unsigned long free_area_cache; /* first hole */
224 pgd_t * pgd;
225 atomic_t mm_users; /* How many users with user space? */
226 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
227 int map_count; /* number of VMAs */
228 struct rw_semaphore mmap_sem;
229 spinlock_t page_table_lock; /* Protects page tables and some counters */
230
231 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
232 * together off init_mm.mmlist, and are protected
233 * by mmlist_lock
234 */
235
236 unsigned long start_code, end_code, start_data, end_data;
237 unsigned long start_brk, brk, start_stack;
238 unsigned long arg_start, arg_end, env_start, env_end;
239 unsigned long total_vm, locked_vm, shared_vm;
240 unsigned long exec_vm, stack_vm, reserved_vm, def_flags, nr_ptes;
241
242 /* Special counters protected by the page_table_lock */
243 mm_counter_t _rss;
244 mm_counter_t _anon_rss;
245
246 unsigned long saved_auxv[42]; /* for /proc/PID/auxv */
247
248 unsigned dumpable:1;
249 cpumask_t cpu_vm_mask;
250
251 /* Architecture-specific MM context */
252 mm_context_t context;
253
254 /* Token based thrashing protection. */
255 unsigned long swap_token_time;
256 char recent_pagein;
257
258 /* coredumping support */
259 int core_waiters;
260 struct completion *core_startup_done, core_done;
261
262 /* aio bits */
263 rwlock_t ioctx_list_lock;
264 struct kioctx *ioctx_list;
265
266 struct kioctx default_kioctx;
267
268 unsigned long hiwater_rss; /* High-water RSS usage */
269 unsigned long hiwater_vm; /* High-water virtual memory usage */
270};
271
272struct sighand_struct {
273 atomic_t count;
274 struct k_sigaction action[_NSIG];
275 spinlock_t siglock;
276};
277
278/*
279 * NOTE! "signal_struct" does not have it's own
280 * locking, because a shared signal_struct always
281 * implies a shared sighand_struct, so locking
282 * sighand_struct is always a proper superset of
283 * the locking of signal_struct.
284 */
285struct signal_struct {
286 atomic_t count;
287 atomic_t live;
288
289 wait_queue_head_t wait_chldexit; /* for wait4() */
290
291 /* current thread group signal load-balancing target: */
292 task_t *curr_target;
293
294 /* shared signal handling: */
295 struct sigpending shared_pending;
296
297 /* thread group exit support */
298 int group_exit_code;
299 /* overloaded:
300 * - notify group_exit_task when ->count is equal to notify_count
301 * - everyone except group_exit_task is stopped during signal delivery
302 * of fatal signals, group_exit_task processes the signal.
303 */
304 struct task_struct *group_exit_task;
305 int notify_count;
306
307 /* thread group stop support, overloads group_exit_code too */
308 int group_stop_count;
309 unsigned int flags; /* see SIGNAL_* flags below */
310
311 /* POSIX.1b Interval Timers */
312 struct list_head posix_timers;
313
314 /* ITIMER_REAL timer for the process */
315 struct timer_list real_timer;
316 unsigned long it_real_value, it_real_incr;
317
318 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
319 cputime_t it_prof_expires, it_virt_expires;
320 cputime_t it_prof_incr, it_virt_incr;
321
322 /* job control IDs */
323 pid_t pgrp;
324 pid_t tty_old_pgrp;
325 pid_t session;
326 /* boolean value for session group leader */
327 int leader;
328
329 struct tty_struct *tty; /* NULL if no tty */
330
331 /*
332 * Cumulative resource counters for dead threads in the group,
333 * and for reaped dead child processes forked by this group.
334 * Live threads maintain their own counters and add to these
335 * in __exit_signal, except for the group leader.
336 */
337 cputime_t utime, stime, cutime, cstime;
338 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
339 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
340
341 /*
342 * Cumulative ns of scheduled CPU time for dead threads in the
343 * group, not including a zombie group leader. (This only differs
344 * from jiffies_to_ns(utime + stime) if sched_clock uses something
345 * other than jiffies.)
346 */
347 unsigned long long sched_time;
348
349 /*
350 * We don't bother to synchronize most readers of this at all,
351 * because there is no reader checking a limit that actually needs
352 * to get both rlim_cur and rlim_max atomically, and either one
353 * alone is a single word that can safely be read normally.
354 * getrlimit/setrlimit use task_lock(current->group_leader) to
355 * protect this instead of the siglock, because they really
356 * have no need to disable irqs.
357 */
358 struct rlimit rlim[RLIM_NLIMITS];
359
360 struct list_head cpu_timers[3];
361
362 /* keep the process-shared keyrings here so that they do the right
363 * thing in threads created with CLONE_THREAD */
364#ifdef CONFIG_KEYS
365 struct key *session_keyring; /* keyring inherited over fork */
366 struct key *process_keyring; /* keyring private to this process */
367#endif
368};
369
370/*
371 * Bits in flags field of signal_struct.
372 */
373#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
374#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
375#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
376#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
377
378
379/*
380 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
381 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL tasks are
382 * in the range MAX_RT_PRIO..MAX_PRIO-1. Priority values
383 * are inverted: lower p->prio value means higher priority.
384 *
385 * The MAX_USER_RT_PRIO value allows the actual maximum
386 * RT priority to be separate from the value exported to
387 * user-space. This allows kernel threads to set their
388 * priority to a value higher than any user task. Note:
389 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
390 */
391
392#define MAX_USER_RT_PRIO 100
393#define MAX_RT_PRIO MAX_USER_RT_PRIO
394
395#define MAX_PRIO (MAX_RT_PRIO + 40)
396
397#define rt_task(p) (unlikely((p)->prio < MAX_RT_PRIO))
398
399/*
400 * Some day this will be a full-fledged user tracking system..
401 */
402struct user_struct {
403 atomic_t __count; /* reference count */
404 atomic_t processes; /* How many processes does this user have? */
405 atomic_t files; /* How many open files does this user have? */
406 atomic_t sigpending; /* How many pending signals does this user have? */
407 /* protected by mq_lock */
408 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
409 unsigned long locked_shm; /* How many pages of mlocked shm ? */
410
411#ifdef CONFIG_KEYS
412 struct key *uid_keyring; /* UID specific keyring */
413 struct key *session_keyring; /* UID's default session keyring */
414#endif
415
416 /* Hash table maintenance information */
417 struct list_head uidhash_list;
418 uid_t uid;
419};
420
421extern struct user_struct *find_user(uid_t);
422
423extern struct user_struct root_user;
424#define INIT_USER (&root_user)
425
426typedef struct prio_array prio_array_t;
427struct backing_dev_info;
428struct reclaim_state;
429
430#ifdef CONFIG_SCHEDSTATS
431struct sched_info {
432 /* cumulative counters */
433 unsigned long cpu_time, /* time spent on the cpu */
434 run_delay, /* time spent waiting on a runqueue */
435 pcnt; /* # of timeslices run on this cpu */
436
437 /* timestamps */
438 unsigned long last_arrival, /* when we last ran on a cpu */
439 last_queued; /* when we were last queued to run */
440};
441
442extern struct file_operations proc_schedstat_operations;
443#endif
444
445enum idle_type
446{
447 SCHED_IDLE,
448 NOT_IDLE,
449 NEWLY_IDLE,
450 MAX_IDLE_TYPES
451};
452
453/*
454 * sched-domains (multiprocessor balancing) declarations:
455 */
456#ifdef CONFIG_SMP
457#define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
458
459#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
460#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
461#define SD_BALANCE_EXEC 4 /* Balance on exec */
462#define SD_WAKE_IDLE 8 /* Wake to idle CPU on task wakeup */
463#define SD_WAKE_AFFINE 16 /* Wake task to waking CPU */
464#define SD_WAKE_BALANCE 32 /* Perform balancing at task wakeup */
465#define SD_SHARE_CPUPOWER 64 /* Domain members share cpu power */
466
467struct sched_group {
468 struct sched_group *next; /* Must be a circular list */
469 cpumask_t cpumask;
470
471 /*
472 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
473 * single CPU. This is read only (except for setup, hotplug CPU).
474 */
475 unsigned long cpu_power;
476};
477
478struct sched_domain {
479 /* These fields must be setup */
480 struct sched_domain *parent; /* top domain must be null terminated */
481 struct sched_group *groups; /* the balancing groups of the domain */
482 cpumask_t span; /* span of all CPUs in this domain */
483 unsigned long min_interval; /* Minimum balance interval ms */
484 unsigned long max_interval; /* Maximum balance interval ms */
485 unsigned int busy_factor; /* less balancing by factor if busy */
486 unsigned int imbalance_pct; /* No balance until over watermark */
487 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
488 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
489 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
490 int flags; /* See SD_* */
491
492 /* Runtime fields. */
493 unsigned long last_balance; /* init to jiffies. units in jiffies */
494 unsigned int balance_interval; /* initialise to 1. units in ms. */
495 unsigned int nr_balance_failed; /* initialise to 0 */
496
497#ifdef CONFIG_SCHEDSTATS
498 /* load_balance() stats */
499 unsigned long lb_cnt[MAX_IDLE_TYPES];
500 unsigned long lb_failed[MAX_IDLE_TYPES];
501 unsigned long lb_balanced[MAX_IDLE_TYPES];
502 unsigned long lb_imbalance[MAX_IDLE_TYPES];
503 unsigned long lb_gained[MAX_IDLE_TYPES];
504 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
505 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
506 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
507
508 /* Active load balancing */
509 unsigned long alb_cnt;
510 unsigned long alb_failed;
511 unsigned long alb_pushed;
512
513 /* sched_balance_exec() stats */
514 unsigned long sbe_attempts;
515 unsigned long sbe_pushed;
516
517 /* try_to_wake_up() stats */
518 unsigned long ttwu_wake_remote;
519 unsigned long ttwu_move_affine;
520 unsigned long ttwu_move_balance;
521#endif
522};
523
524#ifdef ARCH_HAS_SCHED_DOMAIN
525/* Useful helpers that arch setup code may use. Defined in kernel/sched.c */
526extern cpumask_t cpu_isolated_map;
527extern void init_sched_build_groups(struct sched_group groups[],
528 cpumask_t span, int (*group_fn)(int cpu));
529extern void cpu_attach_domain(struct sched_domain *sd, int cpu);
530#endif /* ARCH_HAS_SCHED_DOMAIN */
531#endif /* CONFIG_SMP */
532
533
534struct io_context; /* See blkdev.h */
535void exit_io_context(void);
536struct cpuset;
537
538#define NGROUPS_SMALL 32
539#define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
540struct group_info {
541 int ngroups;
542 atomic_t usage;
543 gid_t small_block[NGROUPS_SMALL];
544 int nblocks;
545 gid_t *blocks[0];
546};
547
548/*
549 * get_group_info() must be called with the owning task locked (via task_lock())
550 * when task != current. The reason being that the vast majority of callers are
551 * looking at current->group_info, which can not be changed except by the
552 * current task. Changing current->group_info requires the task lock, too.
553 */
554#define get_group_info(group_info) do { \
555 atomic_inc(&(group_info)->usage); \
556} while (0)
557
558#define put_group_info(group_info) do { \
559 if (atomic_dec_and_test(&(group_info)->usage)) \
560 groups_free(group_info); \
561} while (0)
562
563struct group_info *groups_alloc(int gidsetsize);
564void groups_free(struct group_info *group_info);
565int set_current_groups(struct group_info *group_info);
566/* access the groups "array" with this macro */
567#define GROUP_AT(gi, i) \
568 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
569
570
571struct audit_context; /* See audit.c */
572struct mempolicy;
573
574struct task_struct {
575 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
576 struct thread_info *thread_info;
577 atomic_t usage;
578 unsigned long flags; /* per process flags, defined below */
579 unsigned long ptrace;
580
581 int lock_depth; /* Lock depth */
582
583 int prio, static_prio;
584 struct list_head run_list;
585 prio_array_t *array;
586
587 unsigned long sleep_avg;
588 unsigned long long timestamp, last_ran;
589 unsigned long long sched_time; /* sched_clock time spent running */
590 int activated;
591
592 unsigned long policy;
593 cpumask_t cpus_allowed;
594 unsigned int time_slice, first_time_slice;
595
596#ifdef CONFIG_SCHEDSTATS
597 struct sched_info sched_info;
598#endif
599
600 struct list_head tasks;
601 /*
602 * ptrace_list/ptrace_children forms the list of my children
603 * that were stolen by a ptracer.
604 */
605 struct list_head ptrace_children;
606 struct list_head ptrace_list;
607
608 struct mm_struct *mm, *active_mm;
609
610/* task state */
611 struct linux_binfmt *binfmt;
612 long exit_state;
613 int exit_code, exit_signal;
614 int pdeath_signal; /* The signal sent when the parent dies */
615 /* ??? */
616 unsigned long personality;
617 unsigned did_exec:1;
618 pid_t pid;
619 pid_t tgid;
620 /*
621 * pointers to (original) parent process, youngest child, younger sibling,
622 * older sibling, respectively. (p->father can be replaced with
623 * p->parent->pid)
624 */
625 struct task_struct *real_parent; /* real parent process (when being debugged) */
626 struct task_struct *parent; /* parent process */
627 /*
628 * children/sibling forms the list of my children plus the
629 * tasks I'm ptracing.
630 */
631 struct list_head children; /* list of my children */
632 struct list_head sibling; /* linkage in my parent's children list */
633 struct task_struct *group_leader; /* threadgroup leader */
634
635 /* PID/PID hash table linkage. */
636 struct pid pids[PIDTYPE_MAX];
637
638 struct completion *vfork_done; /* for vfork() */
639 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
640 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
641
642 unsigned long rt_priority;
643 cputime_t utime, stime;
644 unsigned long nvcsw, nivcsw; /* context switch counts */
645 struct timespec start_time;
646/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
647 unsigned long min_flt, maj_flt;
648
649 cputime_t it_prof_expires, it_virt_expires;
650 unsigned long long it_sched_expires;
651 struct list_head cpu_timers[3];
652
653/* process credentials */
654 uid_t uid,euid,suid,fsuid;
655 gid_t gid,egid,sgid,fsgid;
656 struct group_info *group_info;
657 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
658 unsigned keep_capabilities:1;
659 struct user_struct *user;
660#ifdef CONFIG_KEYS
661 struct key *thread_keyring; /* keyring private to this thread */
662#endif
663 int oomkilladj; /* OOM kill score adjustment (bit shift). */
664 char comm[TASK_COMM_LEN];
665/* file system info */
666 int link_count, total_link_count;
667/* ipc stuff */
668 struct sysv_sem sysvsem;
669/* CPU-specific state of this task */
670 struct thread_struct thread;
671/* filesystem information */
672 struct fs_struct *fs;
673/* open file information */
674 struct files_struct *files;
675/* namespace */
676 struct namespace *namespace;
677/* signal handlers */
678 struct signal_struct *signal;
679 struct sighand_struct *sighand;
680
681 sigset_t blocked, real_blocked;
682 struct sigpending pending;
683
684 unsigned long sas_ss_sp;
685 size_t sas_ss_size;
686 int (*notifier)(void *priv);
687 void *notifier_data;
688 sigset_t *notifier_mask;
689
690 void *security;
691 struct audit_context *audit_context;
692 seccomp_t seccomp;
693
694/* Thread group tracking */
695 u32 parent_exec_id;
696 u32 self_exec_id;
697/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
698 spinlock_t alloc_lock;
699/* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
700 spinlock_t proc_lock;
701/* context-switch lock */
702 spinlock_t switch_lock;
703
704/* journalling filesystem info */
705 void *journal_info;
706
707/* VM state */
708 struct reclaim_state *reclaim_state;
709
710 struct dentry *proc_dentry;
711 struct backing_dev_info *backing_dev_info;
712
713 struct io_context *io_context;
714
715 unsigned long ptrace_message;
716 siginfo_t *last_siginfo; /* For ptrace use. */
717/*
718 * current io wait handle: wait queue entry to use for io waits
719 * If this thread is processing aio, this points at the waitqueue
720 * inside the currently handled kiocb. It may be NULL (i.e. default
721 * to a stack based synchronous wait) if its doing sync IO.
722 */
723 wait_queue_t *io_wait;
724/* i/o counters(bytes read/written, #syscalls */
725 u64 rchar, wchar, syscr, syscw;
726#if defined(CONFIG_BSD_PROCESS_ACCT)
727 u64 acct_rss_mem1; /* accumulated rss usage */
728 u64 acct_vm_mem1; /* accumulated virtual memory usage */
729 clock_t acct_stimexpd; /* clock_t-converted stime since last update */
730#endif
731#ifdef CONFIG_NUMA
732 struct mempolicy *mempolicy;
733 short il_next;
734#endif
735#ifdef CONFIG_CPUSETS
736 struct cpuset *cpuset;
737 nodemask_t mems_allowed;
738 int cpuset_mems_generation;
739#endif
740};
741
742static inline pid_t process_group(struct task_struct *tsk)
743{
744 return tsk->signal->pgrp;
745}
746
747/**
748 * pid_alive - check that a task structure is not stale
749 * @p: Task structure to be checked.
750 *
751 * Test if a process is not yet dead (at most zombie state)
752 * If pid_alive fails, then pointers within the task structure
753 * can be stale and must not be dereferenced.
754 */
755static inline int pid_alive(struct task_struct *p)
756{
757 return p->pids[PIDTYPE_PID].nr != 0;
758}
759
760extern void free_task(struct task_struct *tsk);
761extern void __put_task_struct(struct task_struct *tsk);
762#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
763#define put_task_struct(tsk) \
764do { if (atomic_dec_and_test(&(tsk)->usage)) __put_task_struct(tsk); } while(0)
765
766/*
767 * Per process flags
768 */
769#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
770 /* Not implemented yet, only for 486*/
771#define PF_STARTING 0x00000002 /* being created */
772#define PF_EXITING 0x00000004 /* getting shut down */
773#define PF_DEAD 0x00000008 /* Dead */
774#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
775#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
776#define PF_DUMPCORE 0x00000200 /* dumped core */
777#define PF_SIGNALED 0x00000400 /* killed by a signal */
778#define PF_MEMALLOC 0x00000800 /* Allocating memory */
779#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
780#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
781#define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */
782#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
783#define PF_FROZEN 0x00010000 /* frozen for system suspend */
784#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
785#define PF_KSWAPD 0x00040000 /* I am kswapd */
786#define PF_SWAPOFF 0x00080000 /* I am in swapoff */
787#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
788#define PF_SYNCWRITE 0x00200000 /* I am doing a sync write */
789#define PF_BORROWED_MM 0x00400000 /* I am a kthread doing use_mm */
790#define PF_RANDOMIZE 0x00800000 /* randomize virtual address space */
791
792/*
793 * Only the _current_ task can read/write to tsk->flags, but other
794 * tasks can access tsk->flags in readonly mode for example
795 * with tsk_used_math (like during threaded core dumping).
796 * There is however an exception to this rule during ptrace
797 * or during fork: the ptracer task is allowed to write to the
798 * child->flags of its traced child (same goes for fork, the parent
799 * can write to the child->flags), because we're guaranteed the
800 * child is not running and in turn not changing child->flags
801 * at the same time the parent does it.
802 */
803#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
804#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
805#define clear_used_math() clear_stopped_child_used_math(current)
806#define set_used_math() set_stopped_child_used_math(current)
807#define conditional_stopped_child_used_math(condition, child) \
808 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
809#define conditional_used_math(condition) \
810 conditional_stopped_child_used_math(condition, current)
811#define copy_to_stopped_child_used_math(child) \
812 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
813/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
814#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
815#define used_math() tsk_used_math(current)
816
817#ifdef CONFIG_SMP
818extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
819#else
820static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
821{
822 if (!cpus_intersects(new_mask, cpu_online_map))
823 return -EINVAL;
824 return 0;
825}
826#endif
827
828extern unsigned long long sched_clock(void);
829extern unsigned long long current_sched_time(const task_t *current_task);
830
831/* sched_exec is called by processes performing an exec */
832#ifdef CONFIG_SMP
833extern void sched_exec(void);
834#else
835#define sched_exec() {}
836#endif
837
838#ifdef CONFIG_HOTPLUG_CPU
839extern void idle_task_exit(void);
840#else
841static inline void idle_task_exit(void) {}
842#endif
843
844extern void sched_idle_next(void);
845extern void set_user_nice(task_t *p, long nice);
846extern int task_prio(const task_t *p);
847extern int task_nice(const task_t *p);
848extern int task_curr(const task_t *p);
849extern int idle_cpu(int cpu);
850extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
851extern task_t *idle_task(int cpu);
852
853void yield(void);
854
855/*
856 * The default (Linux) execution domain.
857 */
858extern struct exec_domain default_exec_domain;
859
860union thread_union {
861 struct thread_info thread_info;
862 unsigned long stack[THREAD_SIZE/sizeof(long)];
863};
864
865#ifndef __HAVE_ARCH_KSTACK_END
866static inline int kstack_end(void *addr)
867{
868 /* Reliable end of stack detection:
869 * Some APM bios versions misalign the stack
870 */
871 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
872}
873#endif
874
875extern union thread_union init_thread_union;
876extern struct task_struct init_task;
877
878extern struct mm_struct init_mm;
879
880#define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
881extern struct task_struct *find_task_by_pid_type(int type, int pid);
882extern void set_special_pids(pid_t session, pid_t pgrp);
883extern void __set_special_pids(pid_t session, pid_t pgrp);
884
885/* per-UID process charging. */
886extern struct user_struct * alloc_uid(uid_t);
887static inline struct user_struct *get_uid(struct user_struct *u)
888{
889 atomic_inc(&u->__count);
890 return u;
891}
892extern void free_uid(struct user_struct *);
893extern void switch_uid(struct user_struct *);
894
895#include <asm/current.h>
896
897extern void do_timer(struct pt_regs *);
898
899extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
900extern int FASTCALL(wake_up_process(struct task_struct * tsk));
901extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
902 unsigned long clone_flags));
903#ifdef CONFIG_SMP
904 extern void kick_process(struct task_struct *tsk);
905#else
906 static inline void kick_process(struct task_struct *tsk) { }
907#endif
908extern void FASTCALL(sched_fork(task_t * p));
909extern void FASTCALL(sched_exit(task_t * p));
910
911extern int in_group_p(gid_t);
912extern int in_egroup_p(gid_t);
913
914extern void proc_caches_init(void);
915extern void flush_signals(struct task_struct *);
916extern void flush_signal_handlers(struct task_struct *, int force_default);
917extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
918
919static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
920{
921 unsigned long flags;
922 int ret;
923
924 spin_lock_irqsave(&tsk->sighand->siglock, flags);
925 ret = dequeue_signal(tsk, mask, info);
926 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
927
928 return ret;
929}
930
931extern void block_all_signals(int (*notifier)(void *priv), void *priv,
932 sigset_t *mask);
933extern void unblock_all_signals(void);
934extern void release_task(struct task_struct * p);
935extern int send_sig_info(int, struct siginfo *, struct task_struct *);
936extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
937extern int force_sigsegv(int, struct task_struct *);
938extern int force_sig_info(int, struct siginfo *, struct task_struct *);
939extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
940extern int kill_pg_info(int, struct siginfo *, pid_t);
941extern int kill_proc_info(int, struct siginfo *, pid_t);
942extern void do_notify_parent(struct task_struct *, int);
943extern void force_sig(int, struct task_struct *);
944extern void force_sig_specific(int, struct task_struct *);
945extern int send_sig(int, struct task_struct *, int);
946extern void zap_other_threads(struct task_struct *p);
947extern int kill_pg(pid_t, int, int);
948extern int kill_sl(pid_t, int, int);
949extern int kill_proc(pid_t, int, int);
950extern struct sigqueue *sigqueue_alloc(void);
951extern void sigqueue_free(struct sigqueue *);
952extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
953extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
954extern int do_sigaction(int, const struct k_sigaction *, struct k_sigaction *);
955extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
956
957/* These can be the second arg to send_sig_info/send_group_sig_info. */
958#define SEND_SIG_NOINFO ((struct siginfo *) 0)
959#define SEND_SIG_PRIV ((struct siginfo *) 1)
960#define SEND_SIG_FORCED ((struct siginfo *) 2)
961
962/* True if we are on the alternate signal stack. */
963
964static inline int on_sig_stack(unsigned long sp)
965{
966 return (sp - current->sas_ss_sp < current->sas_ss_size);
967}
968
969static inline int sas_ss_flags(unsigned long sp)
970{
971 return (current->sas_ss_size == 0 ? SS_DISABLE
972 : on_sig_stack(sp) ? SS_ONSTACK : 0);
973}
974
975
976#ifdef CONFIG_SECURITY
977/* code is in security.c */
978extern int capable(int cap);
979#else
980static inline int capable(int cap)
981{
982 if (cap_raised(current->cap_effective, cap)) {
983 current->flags |= PF_SUPERPRIV;
984 return 1;
985 }
986 return 0;
987}
988#endif
989
990/*
991 * Routines for handling mm_structs
992 */
993extern struct mm_struct * mm_alloc(void);
994
995/* mmdrop drops the mm and the page tables */
996extern void FASTCALL(__mmdrop(struct mm_struct *));
997static inline void mmdrop(struct mm_struct * mm)
998{
999 if (atomic_dec_and_test(&mm->mm_count))
1000 __mmdrop(mm);
1001}
1002
1003/* mmput gets rid of the mappings and all user-space */
1004extern void mmput(struct mm_struct *);
1005/* Grab a reference to a task's mm, if it is not already going away */
1006extern struct mm_struct *get_task_mm(struct task_struct *task);
1007/* Remove the current tasks stale references to the old mm_struct */
1008extern void mm_release(struct task_struct *, struct mm_struct *);
1009
1010extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1011extern void flush_thread(void);
1012extern void exit_thread(void);
1013
1014extern void exit_mm(struct task_struct *);
1015extern void exit_files(struct task_struct *);
1016extern void exit_signal(struct task_struct *);
1017extern void __exit_signal(struct task_struct *);
1018extern void exit_sighand(struct task_struct *);
1019extern void __exit_sighand(struct task_struct *);
1020extern void exit_itimers(struct signal_struct *);
1021
1022extern NORET_TYPE void do_group_exit(int);
1023
1024extern void reparent_to_init(void);
1025extern void daemonize(const char *, ...);
1026extern int allow_signal(int);
1027extern int disallow_signal(int);
1028extern task_t *child_reaper;
1029
1030extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1031extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1032task_t *fork_idle(int);
1033
1034extern void set_task_comm(struct task_struct *tsk, char *from);
1035extern void get_task_comm(char *to, struct task_struct *tsk);
1036
1037#ifdef CONFIG_SMP
1038extern void wait_task_inactive(task_t * p);
1039#else
1040#define wait_task_inactive(p) do { } while (0)
1041#endif
1042
1043#define remove_parent(p) list_del_init(&(p)->sibling)
1044#define add_parent(p, parent) list_add_tail(&(p)->sibling,&(parent)->children)
1045
1046#define REMOVE_LINKS(p) do { \
1047 if (thread_group_leader(p)) \
1048 list_del_init(&(p)->tasks); \
1049 remove_parent(p); \
1050 } while (0)
1051
1052#define SET_LINKS(p) do { \
1053 if (thread_group_leader(p)) \
1054 list_add_tail(&(p)->tasks,&init_task.tasks); \
1055 add_parent(p, (p)->parent); \
1056 } while (0)
1057
1058#define next_task(p) list_entry((p)->tasks.next, struct task_struct, tasks)
1059#define prev_task(p) list_entry((p)->tasks.prev, struct task_struct, tasks)
1060
1061#define for_each_process(p) \
1062 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1063
1064/*
1065 * Careful: do_each_thread/while_each_thread is a double loop so
1066 * 'break' will not work as expected - use goto instead.
1067 */
1068#define do_each_thread(g, t) \
1069 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1070
1071#define while_each_thread(g, t) \
1072 while ((t = next_thread(t)) != g)
1073
1074extern task_t * FASTCALL(next_thread(const task_t *p));
1075
1076#define thread_group_leader(p) (p->pid == p->tgid)
1077
1078static inline int thread_group_empty(task_t *p)
1079{
1080 return list_empty(&p->pids[PIDTYPE_TGID].pid_list);
1081}
1082
1083#define delay_group_leader(p) \
1084 (thread_group_leader(p) && !thread_group_empty(p))
1085
1086extern void unhash_process(struct task_struct *p);
1087
1088/*
1089 * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring
1090 * subscriptions and synchronises with wait4(). Also used in procfs.
1091 *
1092 * Nests both inside and outside of read_lock(&tasklist_lock).
1093 * It must not be nested with write_lock_irq(&tasklist_lock),
1094 * neither inside nor outside.
1095 */
1096static inline void task_lock(struct task_struct *p)
1097{
1098 spin_lock(&p->alloc_lock);
1099}
1100
1101static inline void task_unlock(struct task_struct *p)
1102{
1103 spin_unlock(&p->alloc_lock);
1104}
1105
1106/* set thread flags in other task's structures
1107 * - see asm/thread_info.h for TIF_xxxx flags available
1108 */
1109static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1110{
1111 set_ti_thread_flag(tsk->thread_info,flag);
1112}
1113
1114static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1115{
1116 clear_ti_thread_flag(tsk->thread_info,flag);
1117}
1118
1119static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1120{
1121 return test_and_set_ti_thread_flag(tsk->thread_info,flag);
1122}
1123
1124static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1125{
1126 return test_and_clear_ti_thread_flag(tsk->thread_info,flag);
1127}
1128
1129static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1130{
1131 return test_ti_thread_flag(tsk->thread_info,flag);
1132}
1133
1134static inline void set_tsk_need_resched(struct task_struct *tsk)
1135{
1136 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1137}
1138
1139static inline void clear_tsk_need_resched(struct task_struct *tsk)
1140{
1141 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1142}
1143
1144static inline int signal_pending(struct task_struct *p)
1145{
1146 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1147}
1148
1149static inline int need_resched(void)
1150{
1151 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1152}
1153
1154/*
1155 * cond_resched() and cond_resched_lock(): latency reduction via
1156 * explicit rescheduling in places that are safe. The return
1157 * value indicates whether a reschedule was done in fact.
1158 * cond_resched_lock() will drop the spinlock before scheduling,
1159 * cond_resched_softirq() will enable bhs before scheduling.
1160 */
1161extern int cond_resched(void);
1162extern int cond_resched_lock(spinlock_t * lock);
1163extern int cond_resched_softirq(void);
1164
1165/*
1166 * Does a critical section need to be broken due to another
1167 * task waiting?:
1168 */
1169#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1170# define need_lockbreak(lock) ((lock)->break_lock)
1171#else
1172# define need_lockbreak(lock) 0
1173#endif
1174
1175/*
1176 * Does a critical section need to be broken due to another
1177 * task waiting or preemption being signalled:
1178 */
1179static inline int lock_need_resched(spinlock_t *lock)
1180{
1181 if (need_lockbreak(lock) || need_resched())
1182 return 1;
1183 return 0;
1184}
1185
1186/* Reevaluate whether the task has signals pending delivery.
1187 This is required every time the blocked sigset_t changes.
1188 callers must hold sighand->siglock. */
1189
1190extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1191extern void recalc_sigpending(void);
1192
1193extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1194
1195/*
1196 * Wrappers for p->thread_info->cpu access. No-op on UP.
1197 */
1198#ifdef CONFIG_SMP
1199
1200static inline unsigned int task_cpu(const struct task_struct *p)
1201{
1202 return p->thread_info->cpu;
1203}
1204
1205static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1206{
1207 p->thread_info->cpu = cpu;
1208}
1209
1210#else
1211
1212static inline unsigned int task_cpu(const struct task_struct *p)
1213{
1214 return 0;
1215}
1216
1217static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1218{
1219}
1220
1221#endif /* CONFIG_SMP */
1222
1223#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1224extern void arch_pick_mmap_layout(struct mm_struct *mm);
1225#else
1226static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1227{
1228 mm->mmap_base = TASK_UNMAPPED_BASE;
1229 mm->get_unmapped_area = arch_get_unmapped_area;
1230 mm->unmap_area = arch_unmap_area;
1231}
1232#endif
1233
1234extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1235extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1236
1237#ifdef CONFIG_MAGIC_SYSRQ
1238
1239extern void normalize_rt_tasks(void);
1240
1241#endif
1242
1243/* try_to_freeze
1244 *
1245 * Checks whether we need to enter the refrigerator
1246 * and returns 1 if we did so.
1247 */
1248#ifdef CONFIG_PM
1249extern void refrigerator(unsigned long);
1250extern int freeze_processes(void);
1251extern void thaw_processes(void);
1252
1253static inline int try_to_freeze(unsigned long refrigerator_flags)
1254{
1255 if (unlikely(current->flags & PF_FREEZE)) {
1256 refrigerator(refrigerator_flags);
1257 return 1;
1258 } else
1259 return 0;
1260}
1261#else
1262static inline void refrigerator(unsigned long flag) {}
1263static inline int freeze_processes(void) { BUG(); return 0; }
1264static inline void thaw_processes(void) {}
1265
1266static inline int try_to_freeze(unsigned long refrigerator_flags)
1267{
1268 return 0;
1269}
1270#endif /* CONFIG_PM */
1271#endif /* __KERNEL__ */
1272
1273#endif