4 #include <linux/auxvec.h> /* For AT_VECTOR_SIZE */
9 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
10 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
11 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
12 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
13 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
14 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
15 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
16 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
17 #define CLONE_THREAD 0x00010000 /* Same thread group? */
18 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
19 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
20 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
21 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
22 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
23 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
24 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
25 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
26 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
27 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
28 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
33 #define SCHED_NORMAL 0
44 #include <asm/param.h> /* for HZ */
46 #include <linux/capability.h>
47 #include <linux/threads.h>
48 #include <linux/kernel.h>
49 #include <linux/types.h>
50 #include <linux/timex.h>
51 #include <linux/jiffies.h>
52 #include <linux/rbtree.h>
53 #include <linux/thread_info.h>
54 #include <linux/cpumask.h>
55 #include <linux/errno.h>
56 #include <linux/nodemask.h>
58 #include <asm/system.h>
59 #include <asm/semaphore.h>
61 #include <asm/ptrace.h>
63 #include <asm/cputime.h>
65 #include <linux/smp.h>
66 #include <linux/sem.h>
67 #include <linux/signal.h>
68 #include <linux/securebits.h>
69 #include <linux/fs_struct.h>
70 #include <linux/compiler.h>
71 #include <linux/completion.h>
72 #include <linux/pid.h>
73 #include <linux/percpu.h>
74 #include <linux/topology.h>
75 #include <linux/seccomp.h>
76 #include <linux/rcupdate.h>
77 #include <linux/futex.h>
78 #include <linux/rtmutex.h>
80 #include <linux/time.h>
81 #include <linux/param.h>
82 #include <linux/resource.h>
83 #include <linux/timer.h>
84 #include <linux/hrtimer.h>
86 #include <asm/processor.h>
89 struct futex_pi_state;
92 * List of flags we want to share for kernel threads,
93 * if only because they are not used by them anyway.
95 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
98 * These are the constant used to fake the fixed-point load-average
99 * counting. Some notes:
100 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
101 * a load-average precision of 10 bits integer + 11 bits fractional
102 * - if you want to count load-averages more often, you need more
103 * precision, or rounding will get you. With 2-second counting freq,
104 * the EXP_n values would be 1981, 2034 and 2043 if still using only
107 extern unsigned long avenrun[]; /* Load averages */
109 #define FSHIFT 11 /* nr of bits of precision */
110 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
111 #define LOAD_FREQ (5*HZ) /* 5 sec intervals */
112 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
113 #define EXP_5 2014 /* 1/exp(5sec/5min) */
114 #define EXP_15 2037 /* 1/exp(5sec/15min) */
116 #define CALC_LOAD(load,exp,n) \
118 load += n*(FIXED_1-exp); \
121 extern unsigned long total_forks;
122 extern int nr_threads;
123 DECLARE_PER_CPU(unsigned long, process_counts);
124 extern int nr_processes(void);
125 extern unsigned long nr_running(void);
126 extern unsigned long nr_uninterruptible(void);
127 extern unsigned long nr_active(void);
128 extern unsigned long nr_iowait(void);
129 extern unsigned long weighted_cpuload(const int cpu);
133 * Task state bitmask. NOTE! These bits are also
134 * encoded in fs/proc/array.c: get_task_state().
136 * We have two separate sets of flags: task->state
137 * is about runnability, while task->exit_state are
138 * about the task exiting. Confusing, but this way
139 * modifying one set can't modify the other one by
142 #define TASK_RUNNING 0
143 #define TASK_INTERRUPTIBLE 1
144 #define TASK_UNINTERRUPTIBLE 2
145 #define TASK_STOPPED 4
146 #define TASK_TRACED 8
147 /* in tsk->exit_state */
148 #define EXIT_ZOMBIE 16
150 /* in tsk->state again */
151 #define TASK_NONINTERACTIVE 64
152 #define TASK_DEAD 128
154 #define __set_task_state(tsk, state_value) \
155 do { (tsk)->state = (state_value); } while (0)
156 #define set_task_state(tsk, state_value) \
157 set_mb((tsk)->state, (state_value))
160 * set_current_state() includes a barrier so that the write of current->state
161 * is correctly serialised wrt the caller's subsequent test of whether to
164 * set_current_state(TASK_UNINTERRUPTIBLE);
165 * if (do_i_need_to_sleep())
168 * If the caller does not need such serialisation then use __set_current_state()
170 #define __set_current_state(state_value) \
171 do { current->state = (state_value); } while (0)
172 #define set_current_state(state_value) \
173 set_mb(current->state, (state_value))
175 /* Task command name length */
176 #define TASK_COMM_LEN 16
178 #include <linux/spinlock.h>
181 * This serializes "schedule()" and also protects
182 * the run-queue from deletions/modifications (but
183 * _adding_ to the beginning of the run-queue has
186 extern rwlock_t tasklist_lock;
187 extern spinlock_t mmlist_lock;
191 extern void sched_init(void);
192 extern void sched_init_smp(void);
193 extern void init_idle(struct task_struct *idle, int cpu);
195 extern cpumask_t nohz_cpu_mask;
197 extern void show_state(void);
198 extern void show_regs(struct pt_regs *);
201 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
202 * task), SP is the stack pointer of the first frame that should be shown in the back
203 * trace (or NULL if the entire call-chain of the task should be shown).
205 extern void show_stack(struct task_struct *task, unsigned long *sp);
207 void io_schedule(void);
208 long io_schedule_timeout(long timeout);
210 extern void cpu_init (void);
211 extern void trap_init(void);
212 extern void update_process_times(int user);
213 extern void scheduler_tick(void);
215 #ifdef CONFIG_DETECT_SOFTLOCKUP
216 extern void softlockup_tick(void);
217 extern void spawn_softlockup_task(void);
218 extern void touch_softlockup_watchdog(void);
220 static inline void softlockup_tick(void)
223 static inline void spawn_softlockup_task(void)
226 static inline void touch_softlockup_watchdog(void)
232 /* Attach to any functions which should be ignored in wchan output. */
233 #define __sched __attribute__((__section__(".sched.text")))
234 /* Is this address in the __sched functions? */
235 extern int in_sched_functions(unsigned long addr);
237 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
238 extern signed long FASTCALL(schedule_timeout(signed long timeout));
239 extern signed long schedule_timeout_interruptible(signed long timeout);
240 extern signed long schedule_timeout_uninterruptible(signed long timeout);
241 asmlinkage void schedule(void);
245 /* Maximum number of active map areas.. This is a random (large) number */
246 #define DEFAULT_MAX_MAP_COUNT 65536
248 extern int sysctl_max_map_count;
250 #include <linux/aio.h>
253 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
254 unsigned long, unsigned long);
256 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
257 unsigned long len, unsigned long pgoff,
258 unsigned long flags);
259 extern void arch_unmap_area(struct mm_struct *, unsigned long);
260 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
262 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
264 * The mm counters are not protected by its page_table_lock,
265 * so must be incremented atomically.
267 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
268 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
269 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
270 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
271 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
272 typedef atomic_long_t mm_counter_t;
274 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
276 * The mm counters are protected by its page_table_lock,
277 * so can be incremented directly.
279 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
280 #define get_mm_counter(mm, member) ((mm)->_##member)
281 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
282 #define inc_mm_counter(mm, member) (mm)->_##member++
283 #define dec_mm_counter(mm, member) (mm)->_##member--
284 typedef unsigned long mm_counter_t;
286 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
288 #define get_mm_rss(mm) \
289 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
290 #define update_hiwater_rss(mm) do { \
291 unsigned long _rss = get_mm_rss(mm); \
292 if ((mm)->hiwater_rss < _rss) \
293 (mm)->hiwater_rss = _rss; \
295 #define update_hiwater_vm(mm) do { \
296 if ((mm)->hiwater_vm < (mm)->total_vm) \
297 (mm)->hiwater_vm = (mm)->total_vm; \
301 struct vm_area_struct * mmap; /* list of VMAs */
302 struct rb_root mm_rb;
303 struct vm_area_struct * mmap_cache; /* last find_vma result */
304 unsigned long (*get_unmapped_area) (struct file *filp,
305 unsigned long addr, unsigned long len,
306 unsigned long pgoff, unsigned long flags);
307 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
308 unsigned long mmap_base; /* base of mmap area */
309 unsigned long task_size; /* size of task vm space */
310 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
311 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
313 atomic_t mm_users; /* How many users with user space? */
314 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
315 int map_count; /* number of VMAs */
316 struct rw_semaphore mmap_sem;
317 spinlock_t page_table_lock; /* Protects page tables and some counters */
319 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
320 * together off init_mm.mmlist, and are protected
324 /* Special counters, in some configurations protected by the
325 * page_table_lock, in other configurations by being atomic.
327 mm_counter_t _file_rss;
328 mm_counter_t _anon_rss;
330 unsigned long hiwater_rss; /* High-watermark of RSS usage */
331 unsigned long hiwater_vm; /* High-water virtual memory usage */
333 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
334 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
335 unsigned long start_code, end_code, start_data, end_data;
336 unsigned long start_brk, brk, start_stack;
337 unsigned long arg_start, arg_end, env_start, env_end;
339 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
342 cpumask_t cpu_vm_mask;
344 /* Architecture-specific MM context */
345 mm_context_t context;
347 /* Token based thrashing protection. */
348 unsigned long swap_token_time;
351 /* coredumping support */
353 struct completion *core_startup_done, core_done;
356 rwlock_t ioctx_list_lock;
357 struct kioctx *ioctx_list;
360 struct sighand_struct {
362 struct k_sigaction action[_NSIG];
366 struct pacct_struct {
369 unsigned long ac_mem;
370 cputime_t ac_utime, ac_stime;
371 unsigned long ac_minflt, ac_majflt;
375 * NOTE! "signal_struct" does not have it's own
376 * locking, because a shared signal_struct always
377 * implies a shared sighand_struct, so locking
378 * sighand_struct is always a proper superset of
379 * the locking of signal_struct.
381 struct signal_struct {
385 wait_queue_head_t wait_chldexit; /* for wait4() */
387 /* current thread group signal load-balancing target: */
388 struct task_struct *curr_target;
390 /* shared signal handling: */
391 struct sigpending shared_pending;
393 /* thread group exit support */
396 * - notify group_exit_task when ->count is equal to notify_count
397 * - everyone except group_exit_task is stopped during signal delivery
398 * of fatal signals, group_exit_task processes the signal.
400 struct task_struct *group_exit_task;
403 /* thread group stop support, overloads group_exit_code too */
404 int group_stop_count;
405 unsigned int flags; /* see SIGNAL_* flags below */
407 /* POSIX.1b Interval Timers */
408 struct list_head posix_timers;
410 /* ITIMER_REAL timer for the process */
411 struct hrtimer real_timer;
412 struct task_struct *tsk;
413 ktime_t it_real_incr;
415 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
416 cputime_t it_prof_expires, it_virt_expires;
417 cputime_t it_prof_incr, it_virt_incr;
419 /* job control IDs */
423 /* boolean value for session group leader */
426 struct tty_struct *tty; /* NULL if no tty */
429 * Cumulative resource counters for dead threads in the group,
430 * and for reaped dead child processes forked by this group.
431 * Live threads maintain their own counters and add to these
432 * in __exit_signal, except for the group leader.
434 cputime_t utime, stime, cutime, cstime;
435 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
436 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
439 * Cumulative ns of scheduled CPU time for dead threads in the
440 * group, not including a zombie group leader. (This only differs
441 * from jiffies_to_ns(utime + stime) if sched_clock uses something
442 * other than jiffies.)
444 unsigned long long sched_time;
447 * We don't bother to synchronize most readers of this at all,
448 * because there is no reader checking a limit that actually needs
449 * to get both rlim_cur and rlim_max atomically, and either one
450 * alone is a single word that can safely be read normally.
451 * getrlimit/setrlimit use task_lock(current->group_leader) to
452 * protect this instead of the siglock, because they really
453 * have no need to disable irqs.
455 struct rlimit rlim[RLIM_NLIMITS];
457 struct list_head cpu_timers[3];
459 /* keep the process-shared keyrings here so that they do the right
460 * thing in threads created with CLONE_THREAD */
462 struct key *session_keyring; /* keyring inherited over fork */
463 struct key *process_keyring; /* keyring private to this process */
465 #ifdef CONFIG_BSD_PROCESS_ACCT
466 struct pacct_struct pacct; /* per-process accounting information */
468 #ifdef CONFIG_TASKSTATS
469 spinlock_t stats_lock;
470 struct taskstats *stats;
474 /* Context switch must be unlocked if interrupts are to be enabled */
475 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
476 # define __ARCH_WANT_UNLOCKED_CTXSW
480 * Bits in flags field of signal_struct.
482 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
483 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
484 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
485 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
489 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
490 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
491 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
492 * values are inverted: lower p->prio value means higher priority.
494 * The MAX_USER_RT_PRIO value allows the actual maximum
495 * RT priority to be separate from the value exported to
496 * user-space. This allows kernel threads to set their
497 * priority to a value higher than any user task. Note:
498 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
501 #define MAX_USER_RT_PRIO 100
502 #define MAX_RT_PRIO MAX_USER_RT_PRIO
504 #define MAX_PRIO (MAX_RT_PRIO + 40)
506 #define rt_prio(prio) unlikely((prio) < MAX_RT_PRIO)
507 #define rt_task(p) rt_prio((p)->prio)
508 #define batch_task(p) (unlikely((p)->policy == SCHED_BATCH))
509 #define is_rt_policy(p) ((p) != SCHED_NORMAL && (p) != SCHED_BATCH)
510 #define has_rt_policy(p) unlikely(is_rt_policy((p)->policy))
513 * Some day this will be a full-fledged user tracking system..
516 atomic_t __count; /* reference count */
517 atomic_t processes; /* How many processes does this user have? */
518 atomic_t files; /* How many open files does this user have? */
519 atomic_t sigpending; /* How many pending signals does this user have? */
520 #ifdef CONFIG_INOTIFY_USER
521 atomic_t inotify_watches; /* How many inotify watches does this user have? */
522 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
524 /* protected by mq_lock */
525 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
526 unsigned long locked_shm; /* How many pages of mlocked shm ? */
529 struct key *uid_keyring; /* UID specific keyring */
530 struct key *session_keyring; /* UID's default session keyring */
533 /* Hash table maintenance information */
534 struct list_head uidhash_list;
538 extern struct user_struct *find_user(uid_t);
540 extern struct user_struct root_user;
541 #define INIT_USER (&root_user)
543 struct backing_dev_info;
544 struct reclaim_state;
546 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
548 /* cumulative counters */
549 unsigned long cpu_time, /* time spent on the cpu */
550 run_delay, /* time spent waiting on a runqueue */
551 pcnt; /* # of timeslices run on this cpu */
554 unsigned long last_arrival, /* when we last ran on a cpu */
555 last_queued; /* when we were last queued to run */
557 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
559 #ifdef CONFIG_SCHEDSTATS
560 extern struct file_operations proc_schedstat_operations;
561 #endif /* CONFIG_SCHEDSTATS */
563 #ifdef CONFIG_TASK_DELAY_ACCT
564 struct task_delay_info {
566 unsigned int flags; /* Private per-task flags */
568 /* For each stat XXX, add following, aligned appropriately
570 * struct timespec XXX_start, XXX_end;
574 * Atomicity of updates to XXX_delay, XXX_count protected by
575 * single lock above (split into XXX_lock if contention is an issue).
579 * XXX_count is incremented on every XXX operation, the delay
580 * associated with the operation is added to XXX_delay.
581 * XXX_delay contains the accumulated delay time in nanoseconds.
583 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
584 u64 blkio_delay; /* wait for sync block io completion */
585 u64 swapin_delay; /* wait for swapin block io completion */
586 u32 blkio_count; /* total count of the number of sync block */
587 /* io operations performed */
588 u32 swapin_count; /* total count of the number of swapin block */
589 /* io operations performed */
591 #endif /* CONFIG_TASK_DELAY_ACCT */
593 static inline int sched_info_on(void)
595 #ifdef CONFIG_SCHEDSTATS
597 #elif defined(CONFIG_TASK_DELAY_ACCT)
598 extern int delayacct_on;
614 * sched-domains (multiprocessor balancing) declarations:
616 #define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
619 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
620 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
621 #define SD_BALANCE_EXEC 4 /* Balance on exec */
622 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
623 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
624 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
625 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
626 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
627 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
628 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
630 #define BALANCE_FOR_MC_POWER \
631 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
633 #define BALANCE_FOR_PKG_POWER \
634 ((sched_mc_power_savings || sched_smt_power_savings) ? \
635 SD_POWERSAVINGS_BALANCE : 0)
637 #define test_sd_parent(sd, flag) ((sd->parent && \
638 (sd->parent->flags & flag)) ? 1 : 0)
642 struct sched_group *next; /* Must be a circular list */
646 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
647 * single CPU. This is read only (except for setup, hotplug CPU).
649 unsigned long cpu_power;
652 struct sched_domain {
653 /* These fields must be setup */
654 struct sched_domain *parent; /* top domain must be null terminated */
655 struct sched_domain *child; /* bottom domain must be null terminated */
656 struct sched_group *groups; /* the balancing groups of the domain */
657 cpumask_t span; /* span of all CPUs in this domain */
658 unsigned long min_interval; /* Minimum balance interval ms */
659 unsigned long max_interval; /* Maximum balance interval ms */
660 unsigned int busy_factor; /* less balancing by factor if busy */
661 unsigned int imbalance_pct; /* No balance until over watermark */
662 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
663 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
664 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
665 unsigned int busy_idx;
666 unsigned int idle_idx;
667 unsigned int newidle_idx;
668 unsigned int wake_idx;
669 unsigned int forkexec_idx;
670 int flags; /* See SD_* */
672 /* Runtime fields. */
673 unsigned long last_balance; /* init to jiffies. units in jiffies */
674 unsigned int balance_interval; /* initialise to 1. units in ms. */
675 unsigned int nr_balance_failed; /* initialise to 0 */
677 #ifdef CONFIG_SCHEDSTATS
678 /* load_balance() stats */
679 unsigned long lb_cnt[MAX_IDLE_TYPES];
680 unsigned long lb_failed[MAX_IDLE_TYPES];
681 unsigned long lb_balanced[MAX_IDLE_TYPES];
682 unsigned long lb_imbalance[MAX_IDLE_TYPES];
683 unsigned long lb_gained[MAX_IDLE_TYPES];
684 unsigned long lb_hot_gained[MAX_IDLE_TYPES];
685 unsigned long lb_nobusyg[MAX_IDLE_TYPES];
686 unsigned long lb_nobusyq[MAX_IDLE_TYPES];
688 /* Active load balancing */
689 unsigned long alb_cnt;
690 unsigned long alb_failed;
691 unsigned long alb_pushed;
693 /* SD_BALANCE_EXEC stats */
694 unsigned long sbe_cnt;
695 unsigned long sbe_balanced;
696 unsigned long sbe_pushed;
698 /* SD_BALANCE_FORK stats */
699 unsigned long sbf_cnt;
700 unsigned long sbf_balanced;
701 unsigned long sbf_pushed;
703 /* try_to_wake_up() stats */
704 unsigned long ttwu_wake_remote;
705 unsigned long ttwu_move_affine;
706 unsigned long ttwu_move_balance;
710 extern int partition_sched_domains(cpumask_t *partition1,
711 cpumask_t *partition2);
714 * Maximum cache size the migration-costs auto-tuning code will
717 extern unsigned int max_cache_size;
719 #endif /* CONFIG_SMP */
722 struct io_context; /* See blkdev.h */
725 #define NGROUPS_SMALL 32
726 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
730 gid_t small_block[NGROUPS_SMALL];
736 * get_group_info() must be called with the owning task locked (via task_lock())
737 * when task != current. The reason being that the vast majority of callers are
738 * looking at current->group_info, which can not be changed except by the
739 * current task. Changing current->group_info requires the task lock, too.
741 #define get_group_info(group_info) do { \
742 atomic_inc(&(group_info)->usage); \
745 #define put_group_info(group_info) do { \
746 if (atomic_dec_and_test(&(group_info)->usage)) \
747 groups_free(group_info); \
750 extern struct group_info *groups_alloc(int gidsetsize);
751 extern void groups_free(struct group_info *group_info);
752 extern int set_current_groups(struct group_info *group_info);
753 extern int groups_search(struct group_info *group_info, gid_t grp);
754 /* access the groups "array" with this macro */
755 #define GROUP_AT(gi, i) \
756 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
758 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
759 extern void prefetch_stack(struct task_struct *t);
761 static inline void prefetch_stack(struct task_struct *t) { }
764 struct audit_context; /* See audit.c */
766 struct pipe_inode_info;
767 struct uts_namespace;
771 SLEEP_NONINTERACTIVE,
779 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
780 struct thread_info *thread_info;
782 unsigned long flags; /* per process flags, defined below */
783 unsigned long ptrace;
785 int lock_depth; /* BKL lock depth */
788 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
792 int load_weight; /* for niceness load balancing purposes */
793 int prio, static_prio, normal_prio;
794 struct list_head run_list;
795 struct prio_array *array;
797 unsigned short ioprio;
798 #ifdef CONFIG_BLK_DEV_IO_TRACE
799 unsigned int btrace_seq;
801 unsigned long sleep_avg;
802 unsigned long long timestamp, last_ran;
803 unsigned long long sched_time; /* sched_clock time spent running */
804 enum sleep_type sleep_type;
806 unsigned long policy;
807 cpumask_t cpus_allowed;
808 unsigned int time_slice, first_time_slice;
810 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
811 struct sched_info sched_info;
814 struct list_head tasks;
816 * ptrace_list/ptrace_children forms the list of my children
817 * that were stolen by a ptracer.
819 struct list_head ptrace_children;
820 struct list_head ptrace_list;
822 struct mm_struct *mm, *active_mm;
825 struct linux_binfmt *binfmt;
827 int exit_code, exit_signal;
828 int pdeath_signal; /* The signal sent when the parent dies */
830 unsigned long personality;
835 #ifdef CONFIG_CC_STACKPROTECTOR
836 /* Canary value for the -fstack-protector gcc feature */
837 unsigned long stack_canary;
840 * pointers to (original) parent process, youngest child, younger sibling,
841 * older sibling, respectively. (p->father can be replaced with
844 struct task_struct *real_parent; /* real parent process (when being debugged) */
845 struct task_struct *parent; /* parent process */
847 * children/sibling forms the list of my children plus the
848 * tasks I'm ptracing.
850 struct list_head children; /* list of my children */
851 struct list_head sibling; /* linkage in my parent's children list */
852 struct task_struct *group_leader; /* threadgroup leader */
854 /* PID/PID hash table linkage. */
855 struct pid_link pids[PIDTYPE_MAX];
856 struct list_head thread_group;
858 struct completion *vfork_done; /* for vfork() */
859 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
860 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
862 unsigned long rt_priority;
863 cputime_t utime, stime;
864 unsigned long nvcsw, nivcsw; /* context switch counts */
865 struct timespec start_time;
866 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
867 unsigned long min_flt, maj_flt;
869 cputime_t it_prof_expires, it_virt_expires;
870 unsigned long long it_sched_expires;
871 struct list_head cpu_timers[3];
873 /* process credentials */
874 uid_t uid,euid,suid,fsuid;
875 gid_t gid,egid,sgid,fsgid;
876 struct group_info *group_info;
877 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
878 unsigned keep_capabilities:1;
879 struct user_struct *user;
881 struct key *request_key_auth; /* assumed request_key authority */
882 struct key *thread_keyring; /* keyring private to this thread */
883 unsigned char jit_keyring; /* default keyring to attach requested keys to */
886 * fpu_counter contains the number of consecutive context switches
887 * that the FPU is used. If this is over a threshold, the lazy fpu
888 * saving becomes unlazy to save the trap. This is an unsigned char
889 * so that after 256 times the counter wraps and the behavior turns
890 * lazy again; this to deal with bursty apps that only use FPU for
893 unsigned char fpu_counter;
894 int oomkilladj; /* OOM kill score adjustment (bit shift). */
895 char comm[TASK_COMM_LEN]; /* executable name excluding path
896 - access with [gs]et_task_comm (which lock
898 - initialized normally by flush_old_exec */
899 /* file system info */
900 int link_count, total_link_count;
901 #ifdef CONFIG_SYSVIPC
903 struct sysv_sem sysvsem;
905 /* CPU-specific state of this task */
906 struct thread_struct thread;
907 /* filesystem information */
908 struct fs_struct *fs;
909 /* open file information */
910 struct files_struct *files;
912 struct nsproxy *nsproxy;
913 /* signal handlers */
914 struct signal_struct *signal;
915 struct sighand_struct *sighand;
917 sigset_t blocked, real_blocked;
918 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
919 struct sigpending pending;
921 unsigned long sas_ss_sp;
923 int (*notifier)(void *priv);
925 sigset_t *notifier_mask;
928 struct audit_context *audit_context;
931 /* Thread group tracking */
934 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
935 spinlock_t alloc_lock;
937 /* Protection of the PI data structures: */
940 #ifdef CONFIG_RT_MUTEXES
941 /* PI waiters blocked on a rt_mutex held by this task */
942 struct plist_head pi_waiters;
943 /* Deadlock detection and priority inheritance handling */
944 struct rt_mutex_waiter *pi_blocked_on;
947 #ifdef CONFIG_DEBUG_MUTEXES
948 /* mutex deadlock detection */
949 struct mutex_waiter *blocked_on;
951 #ifdef CONFIG_TRACE_IRQFLAGS
952 unsigned int irq_events;
953 int hardirqs_enabled;
954 unsigned long hardirq_enable_ip;
955 unsigned int hardirq_enable_event;
956 unsigned long hardirq_disable_ip;
957 unsigned int hardirq_disable_event;
958 int softirqs_enabled;
959 unsigned long softirq_disable_ip;
960 unsigned int softirq_disable_event;
961 unsigned long softirq_enable_ip;
962 unsigned int softirq_enable_event;
966 #ifdef CONFIG_LOCKDEP
967 # define MAX_LOCK_DEPTH 30UL
970 struct held_lock held_locks[MAX_LOCK_DEPTH];
971 unsigned int lockdep_recursion;
974 /* journalling filesystem info */
978 struct reclaim_state *reclaim_state;
980 struct backing_dev_info *backing_dev_info;
982 struct io_context *io_context;
984 unsigned long ptrace_message;
985 siginfo_t *last_siginfo; /* For ptrace use. */
987 * current io wait handle: wait queue entry to use for io waits
988 * If this thread is processing aio, this points at the waitqueue
989 * inside the currently handled kiocb. It may be NULL (i.e. default
990 * to a stack based synchronous wait) if its doing sync IO.
992 wait_queue_t *io_wait;
993 /* i/o counters(bytes read/written, #syscalls */
994 u64 rchar, wchar, syscr, syscw;
995 #if defined(CONFIG_TASK_XACCT)
996 u64 acct_rss_mem1; /* accumulated rss usage */
997 u64 acct_vm_mem1; /* accumulated virtual memory usage */
998 cputime_t acct_stimexpd;/* stime since last update */
1001 struct mempolicy *mempolicy;
1004 #ifdef CONFIG_CPUSETS
1005 struct cpuset *cpuset;
1006 nodemask_t mems_allowed;
1007 int cpuset_mems_generation;
1008 int cpuset_mem_spread_rotor;
1010 struct robust_list_head __user *robust_list;
1011 #ifdef CONFIG_COMPAT
1012 struct compat_robust_list_head __user *compat_robust_list;
1014 struct list_head pi_state_list;
1015 struct futex_pi_state *pi_state_cache;
1017 atomic_t fs_excl; /* holding fs exclusive resources */
1018 struct rcu_head rcu;
1021 * cache last used pipe for splice
1023 struct pipe_inode_info *splice_pipe;
1024 #ifdef CONFIG_TASK_DELAY_ACCT
1025 struct task_delay_info *delays;
1029 static inline pid_t process_group(struct task_struct *tsk)
1031 return tsk->signal->pgrp;
1034 static inline struct pid *task_pid(struct task_struct *task)
1036 return task->pids[PIDTYPE_PID].pid;
1039 static inline struct pid *task_tgid(struct task_struct *task)
1041 return task->group_leader->pids[PIDTYPE_PID].pid;
1044 static inline struct pid *task_pgrp(struct task_struct *task)
1046 return task->group_leader->pids[PIDTYPE_PGID].pid;
1049 static inline struct pid *task_session(struct task_struct *task)
1051 return task->group_leader->pids[PIDTYPE_SID].pid;
1055 * pid_alive - check that a task structure is not stale
1056 * @p: Task structure to be checked.
1058 * Test if a process is not yet dead (at most zombie state)
1059 * If pid_alive fails, then pointers within the task structure
1060 * can be stale and must not be dereferenced.
1062 static inline int pid_alive(struct task_struct *p)
1064 return p->pids[PIDTYPE_PID].pid != NULL;
1068 * is_init - check if a task structure is the first user space
1069 * task the kernel created.
1070 * @p: Task structure to be checked.
1072 static inline int is_init(struct task_struct *tsk)
1074 return tsk->pid == 1;
1077 extern struct pid *cad_pid;
1079 extern void free_task(struct task_struct *tsk);
1080 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1082 extern void __put_task_struct(struct task_struct *t);
1084 static inline void put_task_struct(struct task_struct *t)
1086 if (atomic_dec_and_test(&t->usage))
1087 __put_task_struct(t);
1093 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1094 /* Not implemented yet, only for 486*/
1095 #define PF_STARTING 0x00000002 /* being created */
1096 #define PF_EXITING 0x00000004 /* getting shut down */
1097 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1098 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1099 #define PF_DUMPCORE 0x00000200 /* dumped core */
1100 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1101 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1102 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1103 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1104 #define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */
1105 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1106 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1107 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1108 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1109 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1110 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1111 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1112 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1113 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1114 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1115 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1116 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1117 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1120 * Only the _current_ task can read/write to tsk->flags, but other
1121 * tasks can access tsk->flags in readonly mode for example
1122 * with tsk_used_math (like during threaded core dumping).
1123 * There is however an exception to this rule during ptrace
1124 * or during fork: the ptracer task is allowed to write to the
1125 * child->flags of its traced child (same goes for fork, the parent
1126 * can write to the child->flags), because we're guaranteed the
1127 * child is not running and in turn not changing child->flags
1128 * at the same time the parent does it.
1130 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1131 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1132 #define clear_used_math() clear_stopped_child_used_math(current)
1133 #define set_used_math() set_stopped_child_used_math(current)
1134 #define conditional_stopped_child_used_math(condition, child) \
1135 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1136 #define conditional_used_math(condition) \
1137 conditional_stopped_child_used_math(condition, current)
1138 #define copy_to_stopped_child_used_math(child) \
1139 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1140 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1141 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1142 #define used_math() tsk_used_math(current)
1145 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1147 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1149 if (!cpu_isset(0, new_mask))
1155 extern unsigned long long sched_clock(void);
1156 extern unsigned long long
1157 current_sched_time(const struct task_struct *current_task);
1159 /* sched_exec is called by processes performing an exec */
1161 extern void sched_exec(void);
1163 #define sched_exec() {}
1166 #ifdef CONFIG_HOTPLUG_CPU
1167 extern void idle_task_exit(void);
1169 static inline void idle_task_exit(void) {}
1172 extern void sched_idle_next(void);
1174 #ifdef CONFIG_RT_MUTEXES
1175 extern int rt_mutex_getprio(struct task_struct *p);
1176 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1177 extern void rt_mutex_adjust_pi(struct task_struct *p);
1179 static inline int rt_mutex_getprio(struct task_struct *p)
1181 return p->normal_prio;
1183 # define rt_mutex_adjust_pi(p) do { } while (0)
1186 extern void set_user_nice(struct task_struct *p, long nice);
1187 extern int task_prio(const struct task_struct *p);
1188 extern int task_nice(const struct task_struct *p);
1189 extern int can_nice(const struct task_struct *p, const int nice);
1190 extern int task_curr(const struct task_struct *p);
1191 extern int idle_cpu(int cpu);
1192 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1193 extern struct task_struct *idle_task(int cpu);
1194 extern struct task_struct *curr_task(int cpu);
1195 extern void set_curr_task(int cpu, struct task_struct *p);
1200 * The default (Linux) execution domain.
1202 extern struct exec_domain default_exec_domain;
1204 union thread_union {
1205 struct thread_info thread_info;
1206 unsigned long stack[THREAD_SIZE/sizeof(long)];
1209 #ifndef __HAVE_ARCH_KSTACK_END
1210 static inline int kstack_end(void *addr)
1212 /* Reliable end of stack detection:
1213 * Some APM bios versions misalign the stack
1215 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1219 extern union thread_union init_thread_union;
1220 extern struct task_struct init_task;
1222 extern struct mm_struct init_mm;
1224 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1225 extern struct task_struct *find_task_by_pid_type(int type, int pid);
1226 extern void set_special_pids(pid_t session, pid_t pgrp);
1227 extern void __set_special_pids(pid_t session, pid_t pgrp);
1229 /* per-UID process charging. */
1230 extern struct user_struct * alloc_uid(uid_t);
1231 static inline struct user_struct *get_uid(struct user_struct *u)
1233 atomic_inc(&u->__count);
1236 extern void free_uid(struct user_struct *);
1237 extern void switch_uid(struct user_struct *);
1239 #include <asm/current.h>
1241 extern void do_timer(unsigned long ticks);
1243 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1244 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1245 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1246 unsigned long clone_flags));
1248 extern void kick_process(struct task_struct *tsk);
1250 static inline void kick_process(struct task_struct *tsk) { }
1252 extern void FASTCALL(sched_fork(struct task_struct * p, int clone_flags));
1253 extern void FASTCALL(sched_exit(struct task_struct * p));
1255 extern int in_group_p(gid_t);
1256 extern int in_egroup_p(gid_t);
1258 extern void proc_caches_init(void);
1259 extern void flush_signals(struct task_struct *);
1260 extern void flush_signal_handlers(struct task_struct *, int force_default);
1261 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1263 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1265 unsigned long flags;
1268 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1269 ret = dequeue_signal(tsk, mask, info);
1270 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1275 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1277 extern void unblock_all_signals(void);
1278 extern void release_task(struct task_struct * p);
1279 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1280 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1281 extern int force_sigsegv(int, struct task_struct *);
1282 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1283 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1284 extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1285 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1286 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1287 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1288 extern int kill_pid(struct pid *pid, int sig, int priv);
1289 extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1290 extern int kill_pg_info(int, struct siginfo *, pid_t);
1291 extern int kill_proc_info(int, struct siginfo *, pid_t);
1292 extern void do_notify_parent(struct task_struct *, int);
1293 extern void force_sig(int, struct task_struct *);
1294 extern void force_sig_specific(int, struct task_struct *);
1295 extern int send_sig(int, struct task_struct *, int);
1296 extern void zap_other_threads(struct task_struct *p);
1297 extern int kill_pg(pid_t, int, int);
1298 extern int kill_proc(pid_t, int, int);
1299 extern struct sigqueue *sigqueue_alloc(void);
1300 extern void sigqueue_free(struct sigqueue *);
1301 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1302 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
1303 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1304 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1306 static inline int kill_cad_pid(int sig, int priv)
1308 return kill_pid(cad_pid, sig, priv);
1311 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1312 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1313 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1314 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1316 static inline int is_si_special(const struct siginfo *info)
1318 return info <= SEND_SIG_FORCED;
1321 /* True if we are on the alternate signal stack. */
1323 static inline int on_sig_stack(unsigned long sp)
1325 return (sp - current->sas_ss_sp < current->sas_ss_size);
1328 static inline int sas_ss_flags(unsigned long sp)
1330 return (current->sas_ss_size == 0 ? SS_DISABLE
1331 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1335 * Routines for handling mm_structs
1337 extern struct mm_struct * mm_alloc(void);
1339 /* mmdrop drops the mm and the page tables */
1340 extern void FASTCALL(__mmdrop(struct mm_struct *));
1341 static inline void mmdrop(struct mm_struct * mm)
1343 if (atomic_dec_and_test(&mm->mm_count))
1347 /* mmput gets rid of the mappings and all user-space */
1348 extern void mmput(struct mm_struct *);
1349 /* Grab a reference to a task's mm, if it is not already going away */
1350 extern struct mm_struct *get_task_mm(struct task_struct *task);
1351 /* Remove the current tasks stale references to the old mm_struct */
1352 extern void mm_release(struct task_struct *, struct mm_struct *);
1354 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1355 extern void flush_thread(void);
1356 extern void exit_thread(void);
1358 extern void exit_files(struct task_struct *);
1359 extern void __cleanup_signal(struct signal_struct *);
1360 extern void __cleanup_sighand(struct sighand_struct *);
1361 extern void exit_itimers(struct signal_struct *);
1363 extern NORET_TYPE void do_group_exit(int);
1365 extern void daemonize(const char *, ...);
1366 extern int allow_signal(int);
1367 extern int disallow_signal(int);
1368 extern struct task_struct *child_reaper;
1370 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1371 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1372 struct task_struct *fork_idle(int);
1374 extern void set_task_comm(struct task_struct *tsk, char *from);
1375 extern void get_task_comm(char *to, struct task_struct *tsk);
1378 extern void wait_task_inactive(struct task_struct * p);
1380 #define wait_task_inactive(p) do { } while (0)
1383 #define remove_parent(p) list_del_init(&(p)->sibling)
1384 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1386 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1388 #define for_each_process(p) \
1389 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1392 * Careful: do_each_thread/while_each_thread is a double loop so
1393 * 'break' will not work as expected - use goto instead.
1395 #define do_each_thread(g, t) \
1396 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1398 #define while_each_thread(g, t) \
1399 while ((t = next_thread(t)) != g)
1401 /* de_thread depends on thread_group_leader not being a pid based check */
1402 #define thread_group_leader(p) (p == p->group_leader)
1404 /* Do to the insanities of de_thread it is possible for a process
1405 * to have the pid of the thread group leader without actually being
1406 * the thread group leader. For iteration through the pids in proc
1407 * all we care about is that we have a task with the appropriate
1408 * pid, we don't actually care if we have the right task.
1410 static inline int has_group_leader_pid(struct task_struct *p)
1412 return p->pid == p->tgid;
1415 static inline struct task_struct *next_thread(const struct task_struct *p)
1417 return list_entry(rcu_dereference(p->thread_group.next),
1418 struct task_struct, thread_group);
1421 static inline int thread_group_empty(struct task_struct *p)
1423 return list_empty(&p->thread_group);
1426 #define delay_group_leader(p) \
1427 (thread_group_leader(p) && !thread_group_empty(p))
1430 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1431 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1432 * pins the final release of task.io_context. Also protects ->cpuset.
1434 * Nests both inside and outside of read_lock(&tasklist_lock).
1435 * It must not be nested with write_lock_irq(&tasklist_lock),
1436 * neither inside nor outside.
1438 static inline void task_lock(struct task_struct *p)
1440 spin_lock(&p->alloc_lock);
1443 static inline void task_unlock(struct task_struct *p)
1445 spin_unlock(&p->alloc_lock);
1448 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1449 unsigned long *flags);
1451 static inline void unlock_task_sighand(struct task_struct *tsk,
1452 unsigned long *flags)
1454 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1457 #ifndef __HAVE_THREAD_FUNCTIONS
1459 #define task_thread_info(task) (task)->thread_info
1460 #define task_stack_page(task) ((void*)((task)->thread_info))
1462 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1464 *task_thread_info(p) = *task_thread_info(org);
1465 task_thread_info(p)->task = p;
1468 static inline unsigned long *end_of_stack(struct task_struct *p)
1470 return (unsigned long *)(p->thread_info + 1);
1475 /* set thread flags in other task's structures
1476 * - see asm/thread_info.h for TIF_xxxx flags available
1478 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1480 set_ti_thread_flag(task_thread_info(tsk), flag);
1483 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1485 clear_ti_thread_flag(task_thread_info(tsk), flag);
1488 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1490 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1493 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1495 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1498 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1500 return test_ti_thread_flag(task_thread_info(tsk), flag);
1503 static inline void set_tsk_need_resched(struct task_struct *tsk)
1505 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1508 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1510 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1513 static inline int signal_pending(struct task_struct *p)
1515 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1518 static inline int need_resched(void)
1520 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1524 * cond_resched() and cond_resched_lock(): latency reduction via
1525 * explicit rescheduling in places that are safe. The return
1526 * value indicates whether a reschedule was done in fact.
1527 * cond_resched_lock() will drop the spinlock before scheduling,
1528 * cond_resched_softirq() will enable bhs before scheduling.
1530 extern int cond_resched(void);
1531 extern int cond_resched_lock(spinlock_t * lock);
1532 extern int cond_resched_softirq(void);
1535 * Does a critical section need to be broken due to another
1538 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1539 # define need_lockbreak(lock) ((lock)->break_lock)
1541 # define need_lockbreak(lock) 0
1545 * Does a critical section need to be broken due to another
1546 * task waiting or preemption being signalled:
1548 static inline int lock_need_resched(spinlock_t *lock)
1550 if (need_lockbreak(lock) || need_resched())
1555 /* Reevaluate whether the task has signals pending delivery.
1556 This is required every time the blocked sigset_t changes.
1557 callers must hold sighand->siglock. */
1559 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1560 extern void recalc_sigpending(void);
1562 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1565 * Wrappers for p->thread_info->cpu access. No-op on UP.
1569 static inline unsigned int task_cpu(const struct task_struct *p)
1571 return task_thread_info(p)->cpu;
1574 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1576 task_thread_info(p)->cpu = cpu;
1581 static inline unsigned int task_cpu(const struct task_struct *p)
1586 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1590 #endif /* CONFIG_SMP */
1592 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1593 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1595 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1597 mm->mmap_base = TASK_UNMAPPED_BASE;
1598 mm->get_unmapped_area = arch_get_unmapped_area;
1599 mm->unmap_area = arch_unmap_area;
1603 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1604 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1606 #include <linux/sysdev.h>
1607 extern int sched_mc_power_savings, sched_smt_power_savings;
1608 extern struct sysdev_attribute attr_sched_mc_power_savings, attr_sched_smt_power_savings;
1609 extern int sched_create_sysfs_power_savings_entries(struct sysdev_class *cls);
1611 extern void normalize_rt_tasks(void);
1615 * Check if a process has been frozen
1617 static inline int frozen(struct task_struct *p)
1619 return p->flags & PF_FROZEN;
1623 * Check if there is a request to freeze a process
1625 static inline int freezing(struct task_struct *p)
1627 return p->flags & PF_FREEZE;
1631 * Request that a process be frozen
1632 * FIXME: SMP problem. We may not modify other process' flags!
1634 static inline void freeze(struct task_struct *p)
1636 p->flags |= PF_FREEZE;
1640 * Sometimes we may need to cancel the previous 'freeze' request
1642 static inline void do_not_freeze(struct task_struct *p)
1644 p->flags &= ~PF_FREEZE;
1648 * Wake up a frozen process
1650 static inline int thaw_process(struct task_struct *p)
1653 p->flags &= ~PF_FROZEN;
1661 * freezing is complete, mark process as frozen
1663 static inline void frozen_process(struct task_struct *p)
1665 p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1668 extern void refrigerator(void);
1669 extern int freeze_processes(void);
1670 extern void thaw_processes(void);
1672 static inline int try_to_freeze(void)
1674 if (freezing(current)) {
1681 static inline int frozen(struct task_struct *p) { return 0; }
1682 static inline int freezing(struct task_struct *p) { return 0; }
1683 static inline void freeze(struct task_struct *p) { BUG(); }
1684 static inline int thaw_process(struct task_struct *p) { return 1; }
1685 static inline void frozen_process(struct task_struct *p) { BUG(); }
1687 static inline void refrigerator(void) {}
1688 static inline int freeze_processes(void) { BUG(); return 0; }
1689 static inline void thaw_processes(void) {}
1691 static inline int try_to_freeze(void) { return 0; }
1693 #endif /* CONFIG_PM */
1694 #endif /* __KERNEL__ */