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 */
29 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
30 #define CLONE_NEWNET 0x40000000 /* New network namespace */
35 #define SCHED_NORMAL 0
39 /* SCHED_ISO: reserved but not implemented yet */
48 #include <asm/param.h> /* for HZ */
50 #include <linux/capability.h>
51 #include <linux/threads.h>
52 #include <linux/kernel.h>
53 #include <linux/types.h>
54 #include <linux/timex.h>
55 #include <linux/jiffies.h>
56 #include <linux/rbtree.h>
57 #include <linux/thread_info.h>
58 #include <linux/cpumask.h>
59 #include <linux/errno.h>
60 #include <linux/nodemask.h>
62 #include <asm/system.h>
63 #include <asm/semaphore.h>
65 #include <asm/ptrace.h>
67 #include <asm/cputime.h>
69 #include <linux/smp.h>
70 #include <linux/sem.h>
71 #include <linux/signal.h>
72 #include <linux/securebits.h>
73 #include <linux/fs_struct.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/seccomp.h>
80 #include <linux/rcupdate.h>
81 #include <linux/futex.h>
82 #include <linux/rtmutex.h>
84 #include <linux/time.h>
85 #include <linux/param.h>
86 #include <linux/resource.h>
87 #include <linux/timer.h>
88 #include <linux/hrtimer.h>
89 #include <linux/task_io_accounting.h>
91 #include <asm/processor.h>
94 struct futex_pi_state;
98 * List of flags we want to share for kernel threads,
99 * if only because they are not used by them anyway.
101 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
104 * These are the constant used to fake the fixed-point load-average
105 * counting. Some notes:
106 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
107 * a load-average precision of 10 bits integer + 11 bits fractional
108 * - if you want to count load-averages more often, you need more
109 * precision, or rounding will get you. With 2-second counting freq,
110 * the EXP_n values would be 1981, 2034 and 2043 if still using only
113 extern unsigned long avenrun[]; /* Load averages */
115 #define FSHIFT 11 /* nr of bits of precision */
116 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
117 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
118 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
119 #define EXP_5 2014 /* 1/exp(5sec/5min) */
120 #define EXP_15 2037 /* 1/exp(5sec/15min) */
122 #define CALC_LOAD(load,exp,n) \
124 load += n*(FIXED_1-exp); \
127 extern unsigned long total_forks;
128 extern int nr_threads;
129 DECLARE_PER_CPU(unsigned long, process_counts);
130 extern int nr_processes(void);
131 extern unsigned long nr_running(void);
132 extern unsigned long nr_uninterruptible(void);
133 extern unsigned long nr_active(void);
134 extern unsigned long nr_iowait(void);
135 extern unsigned long weighted_cpuload(const int cpu);
140 #ifdef CONFIG_SCHED_DEBUG
141 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
142 extern void proc_sched_set_task(struct task_struct *p);
144 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
147 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
150 static inline void proc_sched_set_task(struct task_struct *p)
154 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
160 * Task state bitmask. NOTE! These bits are also
161 * encoded in fs/proc/array.c: get_task_state().
163 * We have two separate sets of flags: task->state
164 * is about runnability, while task->exit_state are
165 * about the task exiting. Confusing, but this way
166 * modifying one set can't modify the other one by
169 #define TASK_RUNNING 0
170 #define TASK_INTERRUPTIBLE 1
171 #define TASK_UNINTERRUPTIBLE 2
172 #define TASK_STOPPED 4
173 #define TASK_TRACED 8
174 /* in tsk->exit_state */
175 #define EXIT_ZOMBIE 16
177 /* in tsk->state again */
180 #define __set_task_state(tsk, state_value) \
181 do { (tsk)->state = (state_value); } while (0)
182 #define set_task_state(tsk, state_value) \
183 set_mb((tsk)->state, (state_value))
186 * set_current_state() includes a barrier so that the write of current->state
187 * is correctly serialised wrt the caller's subsequent test of whether to
190 * set_current_state(TASK_UNINTERRUPTIBLE);
191 * if (do_i_need_to_sleep())
194 * If the caller does not need such serialisation then use __set_current_state()
196 #define __set_current_state(state_value) \
197 do { current->state = (state_value); } while (0)
198 #define set_current_state(state_value) \
199 set_mb(current->state, (state_value))
201 /* Task command name length */
202 #define TASK_COMM_LEN 16
204 #include <linux/spinlock.h>
207 * This serializes "schedule()" and also protects
208 * the run-queue from deletions/modifications (but
209 * _adding_ to the beginning of the run-queue has
212 extern rwlock_t tasklist_lock;
213 extern spinlock_t mmlist_lock;
217 extern void sched_init(void);
218 extern void sched_init_smp(void);
219 extern void init_idle(struct task_struct *idle, int cpu);
220 extern void init_idle_bootup_task(struct task_struct *idle);
222 extern cpumask_t nohz_cpu_mask;
223 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
224 extern int select_nohz_load_balancer(int cpu);
226 static inline int select_nohz_load_balancer(int cpu)
233 * Only dump TASK_* tasks. (0 for all tasks)
235 extern void show_state_filter(unsigned long state_filter);
237 static inline void show_state(void)
239 show_state_filter(0);
242 extern void show_regs(struct pt_regs *);
245 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
246 * task), SP is the stack pointer of the first frame that should be shown in the back
247 * trace (or NULL if the entire call-chain of the task should be shown).
249 extern void show_stack(struct task_struct *task, unsigned long *sp);
251 void io_schedule(void);
252 long io_schedule_timeout(long timeout);
254 extern void cpu_init (void);
255 extern void trap_init(void);
256 extern void update_process_times(int user);
257 extern void scheduler_tick(void);
259 #ifdef CONFIG_DETECT_SOFTLOCKUP
260 extern void softlockup_tick(void);
261 extern void spawn_softlockup_task(void);
262 extern void touch_softlockup_watchdog(void);
263 extern void touch_all_softlockup_watchdogs(void);
265 static inline void softlockup_tick(void)
268 static inline void spawn_softlockup_task(void)
271 static inline void touch_softlockup_watchdog(void)
274 static inline void touch_all_softlockup_watchdogs(void)
280 /* Attach to any functions which should be ignored in wchan output. */
281 #define __sched __attribute__((__section__(".sched.text")))
282 /* Is this address in the __sched functions? */
283 extern int in_sched_functions(unsigned long addr);
285 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
286 extern signed long FASTCALL(schedule_timeout(signed long timeout));
287 extern signed long schedule_timeout_interruptible(signed long timeout);
288 extern signed long schedule_timeout_uninterruptible(signed long timeout);
289 asmlinkage void schedule(void);
292 struct user_namespace;
294 /* Maximum number of active map areas.. This is a random (large) number */
295 #define DEFAULT_MAX_MAP_COUNT 65536
297 extern int sysctl_max_map_count;
299 #include <linux/aio.h>
302 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
303 unsigned long, unsigned long);
305 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
306 unsigned long len, unsigned long pgoff,
307 unsigned long flags);
308 extern void arch_unmap_area(struct mm_struct *, unsigned long);
309 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
311 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
313 * The mm counters are not protected by its page_table_lock,
314 * so must be incremented atomically.
316 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
317 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
318 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
319 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
320 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
321 typedef atomic_long_t mm_counter_t;
323 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
325 * The mm counters are protected by its page_table_lock,
326 * so can be incremented directly.
328 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
329 #define get_mm_counter(mm, member) ((mm)->_##member)
330 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
331 #define inc_mm_counter(mm, member) (mm)->_##member++
332 #define dec_mm_counter(mm, member) (mm)->_##member--
333 typedef unsigned long mm_counter_t;
335 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
337 #define get_mm_rss(mm) \
338 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
339 #define update_hiwater_rss(mm) do { \
340 unsigned long _rss = get_mm_rss(mm); \
341 if ((mm)->hiwater_rss < _rss) \
342 (mm)->hiwater_rss = _rss; \
344 #define update_hiwater_vm(mm) do { \
345 if ((mm)->hiwater_vm < (mm)->total_vm) \
346 (mm)->hiwater_vm = (mm)->total_vm; \
349 extern void set_dumpable(struct mm_struct *mm, int value);
350 extern int get_dumpable(struct mm_struct *mm);
354 #define MMF_DUMPABLE 0 /* core dump is permitted */
355 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
356 #define MMF_DUMPABLE_BITS 2
358 /* coredump filter bits */
359 #define MMF_DUMP_ANON_PRIVATE 2
360 #define MMF_DUMP_ANON_SHARED 3
361 #define MMF_DUMP_MAPPED_PRIVATE 4
362 #define MMF_DUMP_MAPPED_SHARED 5
363 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
364 #define MMF_DUMP_FILTER_BITS 4
365 #define MMF_DUMP_FILTER_MASK \
366 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
367 #define MMF_DUMP_FILTER_DEFAULT \
368 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
371 struct vm_area_struct * mmap; /* list of VMAs */
372 struct rb_root mm_rb;
373 struct vm_area_struct * mmap_cache; /* last find_vma result */
374 unsigned long (*get_unmapped_area) (struct file *filp,
375 unsigned long addr, unsigned long len,
376 unsigned long pgoff, unsigned long flags);
377 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
378 unsigned long mmap_base; /* base of mmap area */
379 unsigned long task_size; /* size of task vm space */
380 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
381 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
383 atomic_t mm_users; /* How many users with user space? */
384 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
385 int map_count; /* number of VMAs */
386 struct rw_semaphore mmap_sem;
387 spinlock_t page_table_lock; /* Protects page tables and some counters */
389 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
390 * together off init_mm.mmlist, and are protected
394 /* Special counters, in some configurations protected by the
395 * page_table_lock, in other configurations by being atomic.
397 mm_counter_t _file_rss;
398 mm_counter_t _anon_rss;
400 unsigned long hiwater_rss; /* High-watermark of RSS usage */
401 unsigned long hiwater_vm; /* High-water virtual memory usage */
403 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
404 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
405 unsigned long start_code, end_code, start_data, end_data;
406 unsigned long start_brk, brk, start_stack;
407 unsigned long arg_start, arg_end, env_start, env_end;
409 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
411 cpumask_t cpu_vm_mask;
413 /* Architecture-specific MM context */
414 mm_context_t context;
416 /* Swap token stuff */
418 * Last value of global fault stamp as seen by this process.
419 * In other words, this value gives an indication of how long
420 * it has been since this task got the token.
421 * Look at mm/thrash.c
423 unsigned int faultstamp;
424 unsigned int token_priority;
425 unsigned int last_interval;
427 unsigned long flags; /* Must use atomic bitops to access the bits */
429 /* coredumping support */
431 struct completion *core_startup_done, core_done;
434 rwlock_t ioctx_list_lock;
435 struct kioctx *ioctx_list;
438 struct sighand_struct {
440 struct k_sigaction action[_NSIG];
442 wait_queue_head_t signalfd_wqh;
445 struct pacct_struct {
448 unsigned long ac_mem;
449 cputime_t ac_utime, ac_stime;
450 unsigned long ac_minflt, ac_majflt;
454 * NOTE! "signal_struct" does not have it's own
455 * locking, because a shared signal_struct always
456 * implies a shared sighand_struct, so locking
457 * sighand_struct is always a proper superset of
458 * the locking of signal_struct.
460 struct signal_struct {
464 wait_queue_head_t wait_chldexit; /* for wait4() */
466 /* current thread group signal load-balancing target: */
467 struct task_struct *curr_target;
469 /* shared signal handling: */
470 struct sigpending shared_pending;
472 /* thread group exit support */
475 * - notify group_exit_task when ->count is equal to notify_count
476 * - everyone except group_exit_task is stopped during signal delivery
477 * of fatal signals, group_exit_task processes the signal.
479 struct task_struct *group_exit_task;
482 /* thread group stop support, overloads group_exit_code too */
483 int group_stop_count;
484 unsigned int flags; /* see SIGNAL_* flags below */
486 /* POSIX.1b Interval Timers */
487 struct list_head posix_timers;
489 /* ITIMER_REAL timer for the process */
490 struct hrtimer real_timer;
491 struct task_struct *tsk;
492 ktime_t it_real_incr;
494 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
495 cputime_t it_prof_expires, it_virt_expires;
496 cputime_t it_prof_incr, it_virt_incr;
498 /* job control IDs */
500 struct pid *tty_old_pgrp;
503 pid_t session __deprecated;
507 /* boolean value for session group leader */
510 struct tty_struct *tty; /* NULL if no tty */
513 * Cumulative resource counters for dead threads in the group,
514 * and for reaped dead child processes forked by this group.
515 * Live threads maintain their own counters and add to these
516 * in __exit_signal, except for the group leader.
518 cputime_t utime, stime, cutime, cstime;
519 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
520 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
521 unsigned long inblock, oublock, cinblock, coublock;
524 * Cumulative ns of scheduled CPU time for dead threads in the
525 * group, not including a zombie group leader. (This only differs
526 * from jiffies_to_ns(utime + stime) if sched_clock uses something
527 * other than jiffies.)
529 unsigned long long sum_sched_runtime;
532 * We don't bother to synchronize most readers of this at all,
533 * because there is no reader checking a limit that actually needs
534 * to get both rlim_cur and rlim_max atomically, and either one
535 * alone is a single word that can safely be read normally.
536 * getrlimit/setrlimit use task_lock(current->group_leader) to
537 * protect this instead of the siglock, because they really
538 * have no need to disable irqs.
540 struct rlimit rlim[RLIM_NLIMITS];
542 struct list_head cpu_timers[3];
544 /* keep the process-shared keyrings here so that they do the right
545 * thing in threads created with CLONE_THREAD */
547 struct key *session_keyring; /* keyring inherited over fork */
548 struct key *process_keyring; /* keyring private to this process */
550 #ifdef CONFIG_BSD_PROCESS_ACCT
551 struct pacct_struct pacct; /* per-process accounting information */
553 #ifdef CONFIG_TASKSTATS
554 struct taskstats *stats;
558 struct tty_audit_buf *tty_audit_buf;
562 /* Context switch must be unlocked if interrupts are to be enabled */
563 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
564 # define __ARCH_WANT_UNLOCKED_CTXSW
568 * Bits in flags field of signal_struct.
570 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
571 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
572 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
573 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
576 * Some day this will be a full-fledged user tracking system..
579 atomic_t __count; /* reference count */
580 atomic_t processes; /* How many processes does this user have? */
581 atomic_t files; /* How many open files does this user have? */
582 atomic_t sigpending; /* How many pending signals does this user have? */
583 #ifdef CONFIG_INOTIFY_USER
584 atomic_t inotify_watches; /* How many inotify watches does this user have? */
585 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
587 /* protected by mq_lock */
588 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
589 unsigned long locked_shm; /* How many pages of mlocked shm ? */
592 struct key *uid_keyring; /* UID specific keyring */
593 struct key *session_keyring; /* UID's default session keyring */
596 /* Hash table maintenance information */
597 struct hlist_node uidhash_node;
600 #ifdef CONFIG_FAIR_USER_SCHED
601 struct task_group *tg;
605 extern struct user_struct *find_user(uid_t);
607 extern struct user_struct root_user;
608 #define INIT_USER (&root_user)
610 struct backing_dev_info;
611 struct reclaim_state;
613 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
615 /* cumulative counters */
616 unsigned long pcount; /* # of times run on this cpu */
617 unsigned long long cpu_time, /* time spent on the cpu */
618 run_delay; /* time spent waiting on a runqueue */
621 unsigned long long last_arrival,/* when we last ran on a cpu */
622 last_queued; /* when we were last queued to run */
623 #ifdef CONFIG_SCHEDSTATS
625 unsigned long bkl_count;
628 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
630 #ifdef CONFIG_SCHEDSTATS
631 extern const struct file_operations proc_schedstat_operations;
632 #endif /* CONFIG_SCHEDSTATS */
634 #ifdef CONFIG_TASK_DELAY_ACCT
635 struct task_delay_info {
637 unsigned int flags; /* Private per-task flags */
639 /* For each stat XXX, add following, aligned appropriately
641 * struct timespec XXX_start, XXX_end;
645 * Atomicity of updates to XXX_delay, XXX_count protected by
646 * single lock above (split into XXX_lock if contention is an issue).
650 * XXX_count is incremented on every XXX operation, the delay
651 * associated with the operation is added to XXX_delay.
652 * XXX_delay contains the accumulated delay time in nanoseconds.
654 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
655 u64 blkio_delay; /* wait for sync block io completion */
656 u64 swapin_delay; /* wait for swapin block io completion */
657 u32 blkio_count; /* total count of the number of sync block */
658 /* io operations performed */
659 u32 swapin_count; /* total count of the number of swapin block */
660 /* io operations performed */
662 #endif /* CONFIG_TASK_DELAY_ACCT */
664 static inline int sched_info_on(void)
666 #ifdef CONFIG_SCHEDSTATS
668 #elif defined(CONFIG_TASK_DELAY_ACCT)
669 extern int delayacct_on;
684 * sched-domains (multiprocessor balancing) declarations:
688 * Increase resolution of nice-level calculations:
690 #define SCHED_LOAD_SHIFT 10
691 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
693 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
696 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
697 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
698 #define SD_BALANCE_EXEC 4 /* Balance on exec */
699 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
700 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
701 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
702 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
703 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
704 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
705 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
706 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
708 #define BALANCE_FOR_MC_POWER \
709 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
711 #define BALANCE_FOR_PKG_POWER \
712 ((sched_mc_power_savings || sched_smt_power_savings) ? \
713 SD_POWERSAVINGS_BALANCE : 0)
715 #define test_sd_parent(sd, flag) ((sd->parent && \
716 (sd->parent->flags & flag)) ? 1 : 0)
720 struct sched_group *next; /* Must be a circular list */
724 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
725 * single CPU. This is read only (except for setup, hotplug CPU).
726 * Note : Never change cpu_power without recompute its reciprocal
728 unsigned int __cpu_power;
730 * reciprocal value of cpu_power to avoid expensive divides
731 * (see include/linux/reciprocal_div.h)
733 u32 reciprocal_cpu_power;
736 struct sched_domain {
737 /* These fields must be setup */
738 struct sched_domain *parent; /* top domain must be null terminated */
739 struct sched_domain *child; /* bottom domain must be null terminated */
740 struct sched_group *groups; /* the balancing groups of the domain */
741 cpumask_t span; /* span of all CPUs in this domain */
742 unsigned long min_interval; /* Minimum balance interval ms */
743 unsigned long max_interval; /* Maximum balance interval ms */
744 unsigned int busy_factor; /* less balancing by factor if busy */
745 unsigned int imbalance_pct; /* No balance until over watermark */
746 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
747 unsigned int busy_idx;
748 unsigned int idle_idx;
749 unsigned int newidle_idx;
750 unsigned int wake_idx;
751 unsigned int forkexec_idx;
752 int flags; /* See SD_* */
754 /* Runtime fields. */
755 unsigned long last_balance; /* init to jiffies. units in jiffies */
756 unsigned int balance_interval; /* initialise to 1. units in ms. */
757 unsigned int nr_balance_failed; /* initialise to 0 */
759 #ifdef CONFIG_SCHEDSTATS
760 /* load_balance() stats */
761 unsigned long lb_count[CPU_MAX_IDLE_TYPES];
762 unsigned long lb_failed[CPU_MAX_IDLE_TYPES];
763 unsigned long lb_balanced[CPU_MAX_IDLE_TYPES];
764 unsigned long lb_imbalance[CPU_MAX_IDLE_TYPES];
765 unsigned long lb_gained[CPU_MAX_IDLE_TYPES];
766 unsigned long lb_hot_gained[CPU_MAX_IDLE_TYPES];
767 unsigned long lb_nobusyg[CPU_MAX_IDLE_TYPES];
768 unsigned long lb_nobusyq[CPU_MAX_IDLE_TYPES];
770 /* Active load balancing */
771 unsigned long alb_count;
772 unsigned long alb_failed;
773 unsigned long alb_pushed;
775 /* SD_BALANCE_EXEC stats */
776 unsigned long sbe_count;
777 unsigned long sbe_balanced;
778 unsigned long sbe_pushed;
780 /* SD_BALANCE_FORK stats */
781 unsigned long sbf_count;
782 unsigned long sbf_balanced;
783 unsigned long sbf_pushed;
785 /* try_to_wake_up() stats */
786 unsigned long ttwu_wake_remote;
787 unsigned long ttwu_move_affine;
788 unsigned long ttwu_move_balance;
792 extern int partition_sched_domains(cpumask_t *partition1,
793 cpumask_t *partition2);
795 #endif /* CONFIG_SMP */
798 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
799 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
800 * task of nice 0 or enough lower priority tasks to bring up the
803 static inline int above_background_load(void)
807 for_each_online_cpu(cpu) {
808 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
814 struct io_context; /* See blkdev.h */
817 #define NGROUPS_SMALL 32
818 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
822 gid_t small_block[NGROUPS_SMALL];
828 * get_group_info() must be called with the owning task locked (via task_lock())
829 * when task != current. The reason being that the vast majority of callers are
830 * looking at current->group_info, which can not be changed except by the
831 * current task. Changing current->group_info requires the task lock, too.
833 #define get_group_info(group_info) do { \
834 atomic_inc(&(group_info)->usage); \
837 #define put_group_info(group_info) do { \
838 if (atomic_dec_and_test(&(group_info)->usage)) \
839 groups_free(group_info); \
842 extern struct group_info *groups_alloc(int gidsetsize);
843 extern void groups_free(struct group_info *group_info);
844 extern int set_current_groups(struct group_info *group_info);
845 extern int groups_search(struct group_info *group_info, gid_t grp);
846 /* access the groups "array" with this macro */
847 #define GROUP_AT(gi, i) \
848 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
850 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
851 extern void prefetch_stack(struct task_struct *t);
853 static inline void prefetch_stack(struct task_struct *t) { }
856 struct audit_context; /* See audit.c */
858 struct pipe_inode_info;
859 struct uts_namespace;
865 const struct sched_class *next;
867 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
868 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
869 void (*yield_task) (struct rq *rq);
871 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
873 struct task_struct * (*pick_next_task) (struct rq *rq);
874 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
876 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
878 unsigned long max_nr_move, unsigned long max_load_move,
879 struct sched_domain *sd, enum cpu_idle_type idle,
880 int *all_pinned, int *this_best_prio);
882 void (*set_curr_task) (struct rq *rq);
883 void (*task_tick) (struct rq *rq, struct task_struct *p);
884 void (*task_new) (struct rq *rq, struct task_struct *p);
888 unsigned long weight, inv_weight;
892 * CFS stats for a schedulable entity (task, task-group etc)
894 * Current field usage histogram:
901 struct sched_entity {
902 struct load_weight load; /* for load-balancing */
903 struct rb_node run_node;
907 u64 sum_exec_runtime;
909 u64 prev_sum_exec_runtime;
911 #ifdef CONFIG_SCHEDSTATS
917 s64 sum_sleep_runtime;
925 #ifdef CONFIG_FAIR_GROUP_SCHED
926 struct sched_entity *parent;
927 /* rq on which this entity is (to be) queued: */
928 struct cfs_rq *cfs_rq;
929 /* rq "owned" by this entity/group: */
935 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
938 unsigned int flags; /* per process flags, defined below */
941 int lock_depth; /* BKL lock depth */
944 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
949 int prio, static_prio, normal_prio;
950 struct list_head run_list;
951 const struct sched_class *sched_class;
952 struct sched_entity se;
954 #ifdef CONFIG_PREEMPT_NOTIFIERS
955 /* list of struct preempt_notifier: */
956 struct hlist_head preempt_notifiers;
959 unsigned short ioprio;
960 #ifdef CONFIG_BLK_DEV_IO_TRACE
961 unsigned int btrace_seq;
965 cpumask_t cpus_allowed;
966 unsigned int time_slice;
968 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
969 struct sched_info sched_info;
972 struct list_head tasks;
974 * ptrace_list/ptrace_children forms the list of my children
975 * that were stolen by a ptracer.
977 struct list_head ptrace_children;
978 struct list_head ptrace_list;
980 struct mm_struct *mm, *active_mm;
983 struct linux_binfmt *binfmt;
985 int exit_code, exit_signal;
986 int pdeath_signal; /* The signal sent when the parent dies */
988 unsigned int personality;
993 #ifdef CONFIG_CC_STACKPROTECTOR
994 /* Canary value for the -fstack-protector gcc feature */
995 unsigned long stack_canary;
998 * pointers to (original) parent process, youngest child, younger sibling,
999 * older sibling, respectively. (p->father can be replaced with
1002 struct task_struct *real_parent; /* real parent process (when being debugged) */
1003 struct task_struct *parent; /* parent process */
1005 * children/sibling forms the list of my children plus the
1006 * tasks I'm ptracing.
1008 struct list_head children; /* list of my children */
1009 struct list_head sibling; /* linkage in my parent's children list */
1010 struct task_struct *group_leader; /* threadgroup leader */
1012 /* PID/PID hash table linkage. */
1013 struct pid_link pids[PIDTYPE_MAX];
1014 struct list_head thread_group;
1016 struct completion *vfork_done; /* for vfork() */
1017 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1018 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1020 unsigned int rt_priority;
1021 cputime_t utime, stime;
1022 unsigned long nvcsw, nivcsw; /* context switch counts */
1023 struct timespec start_time; /* monotonic time */
1024 struct timespec real_start_time; /* boot based time */
1025 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1026 unsigned long min_flt, maj_flt;
1028 cputime_t it_prof_expires, it_virt_expires;
1029 unsigned long long it_sched_expires;
1030 struct list_head cpu_timers[3];
1032 /* process credentials */
1033 uid_t uid,euid,suid,fsuid;
1034 gid_t gid,egid,sgid,fsgid;
1035 struct group_info *group_info;
1036 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
1037 unsigned keep_capabilities:1;
1038 struct user_struct *user;
1040 struct key *request_key_auth; /* assumed request_key authority */
1041 struct key *thread_keyring; /* keyring private to this thread */
1042 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1045 * fpu_counter contains the number of consecutive context switches
1046 * that the FPU is used. If this is over a threshold, the lazy fpu
1047 * saving becomes unlazy to save the trap. This is an unsigned char
1048 * so that after 256 times the counter wraps and the behavior turns
1049 * lazy again; this to deal with bursty apps that only use FPU for
1052 unsigned char fpu_counter;
1053 int oomkilladj; /* OOM kill score adjustment (bit shift). */
1054 char comm[TASK_COMM_LEN]; /* executable name excluding path
1055 - access with [gs]et_task_comm (which lock
1056 it with task_lock())
1057 - initialized normally by flush_old_exec */
1058 /* file system info */
1059 int link_count, total_link_count;
1060 #ifdef CONFIG_SYSVIPC
1062 struct sysv_sem sysvsem;
1064 /* CPU-specific state of this task */
1065 struct thread_struct thread;
1066 /* filesystem information */
1067 struct fs_struct *fs;
1068 /* open file information */
1069 struct files_struct *files;
1071 struct nsproxy *nsproxy;
1072 /* signal handlers */
1073 struct signal_struct *signal;
1074 struct sighand_struct *sighand;
1076 sigset_t blocked, real_blocked;
1077 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
1078 struct sigpending pending;
1080 unsigned long sas_ss_sp;
1082 int (*notifier)(void *priv);
1083 void *notifier_data;
1084 sigset_t *notifier_mask;
1087 struct audit_context *audit_context;
1090 /* Thread group tracking */
1093 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1094 spinlock_t alloc_lock;
1096 /* Protection of the PI data structures: */
1099 #ifdef CONFIG_RT_MUTEXES
1100 /* PI waiters blocked on a rt_mutex held by this task */
1101 struct plist_head pi_waiters;
1102 /* Deadlock detection and priority inheritance handling */
1103 struct rt_mutex_waiter *pi_blocked_on;
1106 #ifdef CONFIG_DEBUG_MUTEXES
1107 /* mutex deadlock detection */
1108 struct mutex_waiter *blocked_on;
1110 #ifdef CONFIG_TRACE_IRQFLAGS
1111 unsigned int irq_events;
1112 int hardirqs_enabled;
1113 unsigned long hardirq_enable_ip;
1114 unsigned int hardirq_enable_event;
1115 unsigned long hardirq_disable_ip;
1116 unsigned int hardirq_disable_event;
1117 int softirqs_enabled;
1118 unsigned long softirq_disable_ip;
1119 unsigned int softirq_disable_event;
1120 unsigned long softirq_enable_ip;
1121 unsigned int softirq_enable_event;
1122 int hardirq_context;
1123 int softirq_context;
1125 #ifdef CONFIG_LOCKDEP
1126 # define MAX_LOCK_DEPTH 30UL
1129 struct held_lock held_locks[MAX_LOCK_DEPTH];
1130 unsigned int lockdep_recursion;
1133 /* journalling filesystem info */
1136 /* stacked block device info */
1137 struct bio *bio_list, **bio_tail;
1140 struct reclaim_state *reclaim_state;
1142 struct backing_dev_info *backing_dev_info;
1144 struct io_context *io_context;
1146 unsigned long ptrace_message;
1147 siginfo_t *last_siginfo; /* For ptrace use. */
1149 * current io wait handle: wait queue entry to use for io waits
1150 * If this thread is processing aio, this points at the waitqueue
1151 * inside the currently handled kiocb. It may be NULL (i.e. default
1152 * to a stack based synchronous wait) if its doing sync IO.
1154 wait_queue_t *io_wait;
1155 #ifdef CONFIG_TASK_XACCT
1156 /* i/o counters(bytes read/written, #syscalls */
1157 u64 rchar, wchar, syscr, syscw;
1159 struct task_io_accounting ioac;
1160 #if defined(CONFIG_TASK_XACCT)
1161 u64 acct_rss_mem1; /* accumulated rss usage */
1162 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1163 cputime_t acct_stimexpd;/* stime since last update */
1166 struct mempolicy *mempolicy;
1169 #ifdef CONFIG_CPUSETS
1170 struct cpuset *cpuset;
1171 nodemask_t mems_allowed;
1172 int cpuset_mems_generation;
1173 int cpuset_mem_spread_rotor;
1175 struct robust_list_head __user *robust_list;
1176 #ifdef CONFIG_COMPAT
1177 struct compat_robust_list_head __user *compat_robust_list;
1179 struct list_head pi_state_list;
1180 struct futex_pi_state *pi_state_cache;
1182 atomic_t fs_excl; /* holding fs exclusive resources */
1183 struct rcu_head rcu;
1186 * cache last used pipe for splice
1188 struct pipe_inode_info *splice_pipe;
1189 #ifdef CONFIG_TASK_DELAY_ACCT
1190 struct task_delay_info *delays;
1192 #ifdef CONFIG_FAULT_INJECTION
1198 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1199 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1200 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1201 * values are inverted: lower p->prio value means higher priority.
1203 * The MAX_USER_RT_PRIO value allows the actual maximum
1204 * RT priority to be separate from the value exported to
1205 * user-space. This allows kernel threads to set their
1206 * priority to a value higher than any user task. Note:
1207 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1210 #define MAX_USER_RT_PRIO 100
1211 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1213 #define MAX_PRIO (MAX_RT_PRIO + 40)
1214 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1216 static inline int rt_prio(int prio)
1218 if (unlikely(prio < MAX_RT_PRIO))
1223 static inline int rt_task(struct task_struct *p)
1225 return rt_prio(p->prio);
1228 static inline pid_t process_group(struct task_struct *tsk)
1230 return tsk->signal->pgrp;
1233 static inline pid_t signal_session(struct signal_struct *sig)
1235 return sig->__session;
1238 static inline pid_t process_session(struct task_struct *tsk)
1240 return signal_session(tsk->signal);
1243 static inline void set_signal_session(struct signal_struct *sig, pid_t session)
1245 sig->__session = session;
1248 static inline struct pid *task_pid(struct task_struct *task)
1250 return task->pids[PIDTYPE_PID].pid;
1253 static inline struct pid *task_tgid(struct task_struct *task)
1255 return task->group_leader->pids[PIDTYPE_PID].pid;
1258 static inline struct pid *task_pgrp(struct task_struct *task)
1260 return task->group_leader->pids[PIDTYPE_PGID].pid;
1263 static inline struct pid *task_session(struct task_struct *task)
1265 return task->group_leader->pids[PIDTYPE_SID].pid;
1269 * pid_alive - check that a task structure is not stale
1270 * @p: Task structure to be checked.
1272 * Test if a process is not yet dead (at most zombie state)
1273 * If pid_alive fails, then pointers within the task structure
1274 * can be stale and must not be dereferenced.
1276 static inline int pid_alive(struct task_struct *p)
1278 return p->pids[PIDTYPE_PID].pid != NULL;
1282 * is_init - check if a task structure is init
1283 * @tsk: Task structure to be checked.
1285 * Check if a task structure is the first user space task the kernel created.
1287 static inline int is_init(struct task_struct *tsk)
1289 return tsk->pid == 1;
1292 extern struct pid *cad_pid;
1294 extern void free_task(struct task_struct *tsk);
1295 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1297 extern void __put_task_struct(struct task_struct *t);
1299 static inline void put_task_struct(struct task_struct *t)
1301 if (atomic_dec_and_test(&t->usage))
1302 __put_task_struct(t);
1308 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1309 /* Not implemented yet, only for 486*/
1310 #define PF_STARTING 0x00000002 /* being created */
1311 #define PF_EXITING 0x00000004 /* getting shut down */
1312 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1313 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1314 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1315 #define PF_DUMPCORE 0x00000200 /* dumped core */
1316 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1317 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1318 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1319 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1320 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1321 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1322 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1323 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1324 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1325 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1326 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1327 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1328 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1329 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1330 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1331 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1332 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1333 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1336 * Only the _current_ task can read/write to tsk->flags, but other
1337 * tasks can access tsk->flags in readonly mode for example
1338 * with tsk_used_math (like during threaded core dumping).
1339 * There is however an exception to this rule during ptrace
1340 * or during fork: the ptracer task is allowed to write to the
1341 * child->flags of its traced child (same goes for fork, the parent
1342 * can write to the child->flags), because we're guaranteed the
1343 * child is not running and in turn not changing child->flags
1344 * at the same time the parent does it.
1346 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1347 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1348 #define clear_used_math() clear_stopped_child_used_math(current)
1349 #define set_used_math() set_stopped_child_used_math(current)
1350 #define conditional_stopped_child_used_math(condition, child) \
1351 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1352 #define conditional_used_math(condition) \
1353 conditional_stopped_child_used_math(condition, current)
1354 #define copy_to_stopped_child_used_math(child) \
1355 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1356 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1357 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1358 #define used_math() tsk_used_math(current)
1361 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1363 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1365 if (!cpu_isset(0, new_mask))
1371 extern unsigned long long sched_clock(void);
1374 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1375 * clock constructed from sched_clock():
1377 extern unsigned long long cpu_clock(int cpu);
1379 extern unsigned long long
1380 task_sched_runtime(struct task_struct *task);
1382 /* sched_exec is called by processes performing an exec */
1384 extern void sched_exec(void);
1386 #define sched_exec() {}
1389 extern void sched_clock_idle_sleep_event(void);
1390 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1392 #ifdef CONFIG_HOTPLUG_CPU
1393 extern void idle_task_exit(void);
1395 static inline void idle_task_exit(void) {}
1398 extern void sched_idle_next(void);
1400 #ifdef CONFIG_SCHED_DEBUG
1401 extern unsigned int sysctl_sched_latency;
1402 extern unsigned int sysctl_sched_nr_latency;
1403 extern unsigned int sysctl_sched_wakeup_granularity;
1404 extern unsigned int sysctl_sched_batch_wakeup_granularity;
1405 extern unsigned int sysctl_sched_child_runs_first;
1406 extern unsigned int sysctl_sched_features;
1409 extern unsigned int sysctl_sched_compat_yield;
1411 #ifdef CONFIG_RT_MUTEXES
1412 extern int rt_mutex_getprio(struct task_struct *p);
1413 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1414 extern void rt_mutex_adjust_pi(struct task_struct *p);
1416 static inline int rt_mutex_getprio(struct task_struct *p)
1418 return p->normal_prio;
1420 # define rt_mutex_adjust_pi(p) do { } while (0)
1423 extern void set_user_nice(struct task_struct *p, long nice);
1424 extern int task_prio(const struct task_struct *p);
1425 extern int task_nice(const struct task_struct *p);
1426 extern int can_nice(const struct task_struct *p, const int nice);
1427 extern int task_curr(const struct task_struct *p);
1428 extern int idle_cpu(int cpu);
1429 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1430 extern struct task_struct *idle_task(int cpu);
1431 extern struct task_struct *curr_task(int cpu);
1432 extern void set_curr_task(int cpu, struct task_struct *p);
1437 * The default (Linux) execution domain.
1439 extern struct exec_domain default_exec_domain;
1441 union thread_union {
1442 struct thread_info thread_info;
1443 unsigned long stack[THREAD_SIZE/sizeof(long)];
1446 #ifndef __HAVE_ARCH_KSTACK_END
1447 static inline int kstack_end(void *addr)
1449 /* Reliable end of stack detection:
1450 * Some APM bios versions misalign the stack
1452 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1456 extern union thread_union init_thread_union;
1457 extern struct task_struct init_task;
1459 extern struct mm_struct init_mm;
1461 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1462 extern struct task_struct *find_task_by_pid_type(int type, int pid);
1463 extern void __set_special_pids(pid_t session, pid_t pgrp);
1465 /* per-UID process charging. */
1466 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1467 static inline struct user_struct *get_uid(struct user_struct *u)
1469 atomic_inc(&u->__count);
1472 extern void free_uid(struct user_struct *);
1473 extern void switch_uid(struct user_struct *);
1474 extern void release_uids(struct user_namespace *ns);
1476 #include <asm/current.h>
1478 extern void do_timer(unsigned long ticks);
1480 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1481 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1482 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1483 unsigned long clone_flags));
1485 extern void kick_process(struct task_struct *tsk);
1487 static inline void kick_process(struct task_struct *tsk) { }
1489 extern void sched_fork(struct task_struct *p, int clone_flags);
1490 extern void sched_dead(struct task_struct *p);
1492 extern int in_group_p(gid_t);
1493 extern int in_egroup_p(gid_t);
1495 extern void proc_caches_init(void);
1496 extern void flush_signals(struct task_struct *);
1497 extern void ignore_signals(struct task_struct *);
1498 extern void flush_signal_handlers(struct task_struct *, int force_default);
1499 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1501 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1503 unsigned long flags;
1506 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1507 ret = dequeue_signal(tsk, mask, info);
1508 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1513 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1515 extern void unblock_all_signals(void);
1516 extern void release_task(struct task_struct * p);
1517 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1518 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1519 extern int force_sigsegv(int, struct task_struct *);
1520 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1521 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1522 extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1523 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1524 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1525 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1526 extern int kill_pid(struct pid *pid, int sig, int priv);
1527 extern int kill_proc_info(int, struct siginfo *, pid_t);
1528 extern void do_notify_parent(struct task_struct *, int);
1529 extern void force_sig(int, struct task_struct *);
1530 extern void force_sig_specific(int, struct task_struct *);
1531 extern int send_sig(int, struct task_struct *, int);
1532 extern void zap_other_threads(struct task_struct *p);
1533 extern int kill_proc(pid_t, int, int);
1534 extern struct sigqueue *sigqueue_alloc(void);
1535 extern void sigqueue_free(struct sigqueue *);
1536 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1537 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
1538 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1539 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1541 static inline int kill_cad_pid(int sig, int priv)
1543 return kill_pid(cad_pid, sig, priv);
1546 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1547 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1548 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1549 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1551 static inline int is_si_special(const struct siginfo *info)
1553 return info <= SEND_SIG_FORCED;
1556 /* True if we are on the alternate signal stack. */
1558 static inline int on_sig_stack(unsigned long sp)
1560 return (sp - current->sas_ss_sp < current->sas_ss_size);
1563 static inline int sas_ss_flags(unsigned long sp)
1565 return (current->sas_ss_size == 0 ? SS_DISABLE
1566 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1570 * Routines for handling mm_structs
1572 extern struct mm_struct * mm_alloc(void);
1574 /* mmdrop drops the mm and the page tables */
1575 extern void FASTCALL(__mmdrop(struct mm_struct *));
1576 static inline void mmdrop(struct mm_struct * mm)
1578 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1582 /* mmput gets rid of the mappings and all user-space */
1583 extern void mmput(struct mm_struct *);
1584 /* Grab a reference to a task's mm, if it is not already going away */
1585 extern struct mm_struct *get_task_mm(struct task_struct *task);
1586 /* Remove the current tasks stale references to the old mm_struct */
1587 extern void mm_release(struct task_struct *, struct mm_struct *);
1589 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1590 extern void flush_thread(void);
1591 extern void exit_thread(void);
1593 extern void exit_files(struct task_struct *);
1594 extern void __cleanup_signal(struct signal_struct *);
1595 extern void __cleanup_sighand(struct sighand_struct *);
1596 extern void exit_itimers(struct signal_struct *);
1598 extern NORET_TYPE void do_group_exit(int);
1600 extern void daemonize(const char *, ...);
1601 extern int allow_signal(int);
1602 extern int disallow_signal(int);
1604 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1605 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1606 struct task_struct *fork_idle(int);
1608 extern void set_task_comm(struct task_struct *tsk, char *from);
1609 extern void get_task_comm(char *to, struct task_struct *tsk);
1612 extern void wait_task_inactive(struct task_struct * p);
1614 #define wait_task_inactive(p) do { } while (0)
1617 #define remove_parent(p) list_del_init(&(p)->sibling)
1618 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1620 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1622 #define for_each_process(p) \
1623 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1626 * Careful: do_each_thread/while_each_thread is a double loop so
1627 * 'break' will not work as expected - use goto instead.
1629 #define do_each_thread(g, t) \
1630 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1632 #define while_each_thread(g, t) \
1633 while ((t = next_thread(t)) != g)
1635 /* de_thread depends on thread_group_leader not being a pid based check */
1636 #define thread_group_leader(p) (p == p->group_leader)
1638 /* Do to the insanities of de_thread it is possible for a process
1639 * to have the pid of the thread group leader without actually being
1640 * the thread group leader. For iteration through the pids in proc
1641 * all we care about is that we have a task with the appropriate
1642 * pid, we don't actually care if we have the right task.
1644 static inline int has_group_leader_pid(struct task_struct *p)
1646 return p->pid == p->tgid;
1649 static inline struct task_struct *next_thread(const struct task_struct *p)
1651 return list_entry(rcu_dereference(p->thread_group.next),
1652 struct task_struct, thread_group);
1655 static inline int thread_group_empty(struct task_struct *p)
1657 return list_empty(&p->thread_group);
1660 #define delay_group_leader(p) \
1661 (thread_group_leader(p) && !thread_group_empty(p))
1664 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1665 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1666 * pins the final release of task.io_context. Also protects ->cpuset.
1668 * Nests both inside and outside of read_lock(&tasklist_lock).
1669 * It must not be nested with write_lock_irq(&tasklist_lock),
1670 * neither inside nor outside.
1672 static inline void task_lock(struct task_struct *p)
1674 spin_lock(&p->alloc_lock);
1677 static inline void task_unlock(struct task_struct *p)
1679 spin_unlock(&p->alloc_lock);
1682 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1683 unsigned long *flags);
1685 static inline void unlock_task_sighand(struct task_struct *tsk,
1686 unsigned long *flags)
1688 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1691 #ifndef __HAVE_THREAD_FUNCTIONS
1693 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1694 #define task_stack_page(task) ((task)->stack)
1696 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1698 *task_thread_info(p) = *task_thread_info(org);
1699 task_thread_info(p)->task = p;
1702 static inline unsigned long *end_of_stack(struct task_struct *p)
1704 return (unsigned long *)(task_thread_info(p) + 1);
1709 /* set thread flags in other task's structures
1710 * - see asm/thread_info.h for TIF_xxxx flags available
1712 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1714 set_ti_thread_flag(task_thread_info(tsk), flag);
1717 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1719 clear_ti_thread_flag(task_thread_info(tsk), flag);
1722 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1724 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1727 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1729 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1732 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1734 return test_ti_thread_flag(task_thread_info(tsk), flag);
1737 static inline void set_tsk_need_resched(struct task_struct *tsk)
1739 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1742 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1744 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1747 static inline int signal_pending(struct task_struct *p)
1749 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1752 static inline int need_resched(void)
1754 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1758 * cond_resched() and cond_resched_lock(): latency reduction via
1759 * explicit rescheduling in places that are safe. The return
1760 * value indicates whether a reschedule was done in fact.
1761 * cond_resched_lock() will drop the spinlock before scheduling,
1762 * cond_resched_softirq() will enable bhs before scheduling.
1764 extern int cond_resched(void);
1765 extern int cond_resched_lock(spinlock_t * lock);
1766 extern int cond_resched_softirq(void);
1769 * Does a critical section need to be broken due to another
1772 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1773 # define need_lockbreak(lock) ((lock)->break_lock)
1775 # define need_lockbreak(lock) 0
1779 * Does a critical section need to be broken due to another
1780 * task waiting or preemption being signalled:
1782 static inline int lock_need_resched(spinlock_t *lock)
1784 if (need_lockbreak(lock) || need_resched())
1790 * Reevaluate whether the task has signals pending delivery.
1791 * Wake the task if so.
1792 * This is required every time the blocked sigset_t changes.
1793 * callers must hold sighand->siglock.
1795 extern void recalc_sigpending_and_wake(struct task_struct *t);
1796 extern void recalc_sigpending(void);
1798 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1801 * Wrappers for p->thread_info->cpu access. No-op on UP.
1805 static inline unsigned int task_cpu(const struct task_struct *p)
1807 return task_thread_info(p)->cpu;
1810 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1814 static inline unsigned int task_cpu(const struct task_struct *p)
1819 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1823 #endif /* CONFIG_SMP */
1825 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1826 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1828 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1830 mm->mmap_base = TASK_UNMAPPED_BASE;
1831 mm->get_unmapped_area = arch_get_unmapped_area;
1832 mm->unmap_area = arch_unmap_area;
1836 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1837 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1839 extern int sched_mc_power_savings, sched_smt_power_savings;
1841 extern void normalize_rt_tasks(void);
1843 #ifdef CONFIG_FAIR_GROUP_SCHED
1845 extern struct task_group init_task_group;
1847 extern struct task_group *sched_create_group(void);
1848 extern void sched_destroy_group(struct task_group *tg);
1849 extern void sched_move_task(struct task_struct *tsk);
1850 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
1854 #ifdef CONFIG_TASK_XACCT
1855 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1860 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1865 static inline void inc_syscr(struct task_struct *tsk)
1870 static inline void inc_syscw(struct task_struct *tsk)
1875 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1879 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1883 static inline void inc_syscr(struct task_struct *tsk)
1887 static inline void inc_syscw(struct task_struct *tsk)
1892 #endif /* __KERNEL__ */