7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 #define CLONE_NEWNET 0x40000000 /* New network namespace */
34 #define SCHED_NORMAL 0
38 /* SCHED_ISO: reserved but not implemented yet */
47 #include <asm/param.h> /* for HZ */
49 #include <linux/capability.h>
50 #include <linux/threads.h>
51 #include <linux/kernel.h>
52 #include <linux/types.h>
53 #include <linux/timex.h>
54 #include <linux/jiffies.h>
55 #include <linux/rbtree.h>
56 #include <linux/thread_info.h>
57 #include <linux/cpumask.h>
58 #include <linux/errno.h>
59 #include <linux/nodemask.h>
60 #include <linux/mm_types.h>
62 #include <asm/system.h>
63 #include <asm/semaphore.h>
65 #include <asm/ptrace.h>
66 #include <asm/cputime.h>
68 #include <linux/smp.h>
69 #include <linux/sem.h>
70 #include <linux/signal.h>
71 #include <linux/securebits.h>
72 #include <linux/fs_struct.h>
73 #include <linux/compiler.h>
74 #include <linux/completion.h>
75 #include <linux/pid.h>
76 #include <linux/percpu.h>
77 #include <linux/topology.h>
78 #include <linux/proportions.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>
90 #include <linux/kobject.h>
92 #include <asm/processor.h>
95 struct futex_pi_state
;
99 * List of flags we want to share for kernel threads,
100 * if only because they are not used by them anyway.
102 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
105 * These are the constant used to fake the fixed-point load-average
106 * counting. Some notes:
107 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
108 * a load-average precision of 10 bits integer + 11 bits fractional
109 * - if you want to count load-averages more often, you need more
110 * precision, or rounding will get you. With 2-second counting freq,
111 * the EXP_n values would be 1981, 2034 and 2043 if still using only
114 extern unsigned long avenrun
[]; /* Load averages */
116 #define FSHIFT 11 /* nr of bits of precision */
117 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
118 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
119 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
120 #define EXP_5 2014 /* 1/exp(5sec/5min) */
121 #define EXP_15 2037 /* 1/exp(5sec/15min) */
123 #define CALC_LOAD(load,exp,n) \
125 load += n*(FIXED_1-exp); \
128 extern unsigned long total_forks
;
129 extern int nr_threads
;
130 DECLARE_PER_CPU(unsigned long, process_counts
);
131 extern int nr_processes(void);
132 extern unsigned long nr_running(void);
133 extern unsigned long nr_uninterruptible(void);
134 extern unsigned long nr_active(void);
135 extern unsigned long nr_iowait(void);
136 extern unsigned long weighted_cpuload(const int cpu
);
141 #ifdef CONFIG_SCHED_DEBUG
142 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
143 extern void proc_sched_set_task(struct task_struct
*p
);
145 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
);
148 proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
)
151 static inline void proc_sched_set_task(struct task_struct
*p
)
155 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
)
161 * Task state bitmask. NOTE! These bits are also
162 * encoded in fs/proc/array.c: get_task_state().
164 * We have two separate sets of flags: task->state
165 * is about runnability, while task->exit_state are
166 * about the task exiting. Confusing, but this way
167 * modifying one set can't modify the other one by
170 #define TASK_RUNNING 0
171 #define TASK_INTERRUPTIBLE 1
172 #define TASK_UNINTERRUPTIBLE 2
173 #define TASK_STOPPED 4
174 #define TASK_TRACED 8
175 /* in tsk->exit_state */
176 #define EXIT_ZOMBIE 16
178 /* in tsk->state again */
181 #define __set_task_state(tsk, state_value) \
182 do { (tsk)->state = (state_value); } while (0)
183 #define set_task_state(tsk, state_value) \
184 set_mb((tsk)->state, (state_value))
187 * set_current_state() includes a barrier so that the write of current->state
188 * is correctly serialised wrt the caller's subsequent test of whether to
191 * set_current_state(TASK_UNINTERRUPTIBLE);
192 * if (do_i_need_to_sleep())
195 * If the caller does not need such serialisation then use __set_current_state()
197 #define __set_current_state(state_value) \
198 do { current->state = (state_value); } while (0)
199 #define set_current_state(state_value) \
200 set_mb(current->state, (state_value))
202 /* Task command name length */
203 #define TASK_COMM_LEN 16
205 #include <linux/spinlock.h>
208 * This serializes "schedule()" and also protects
209 * the run-queue from deletions/modifications (but
210 * _adding_ to the beginning of the run-queue has
213 extern rwlock_t tasklist_lock
;
214 extern spinlock_t mmlist_lock
;
218 extern void sched_init(void);
219 extern void sched_init_smp(void);
220 extern void init_idle(struct task_struct
*idle
, int cpu
);
221 extern void init_idle_bootup_task(struct task_struct
*idle
);
223 extern cpumask_t nohz_cpu_mask
;
224 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
225 extern int select_nohz_load_balancer(int cpu
);
227 static inline int select_nohz_load_balancer(int cpu
)
234 * Only dump TASK_* tasks. (0 for all tasks)
236 extern void show_state_filter(unsigned long state_filter
);
238 static inline void show_state(void)
240 show_state_filter(0);
243 extern void show_regs(struct pt_regs
*);
246 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
247 * task), SP is the stack pointer of the first frame that should be shown in the back
248 * trace (or NULL if the entire call-chain of the task should be shown).
250 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
252 void io_schedule(void);
253 long io_schedule_timeout(long timeout
);
255 extern void cpu_init (void);
256 extern void trap_init(void);
257 extern void account_process_tick(struct task_struct
*task
, int user
);
258 extern void update_process_times(int user
);
259 extern void scheduler_tick(void);
261 #ifdef CONFIG_DETECT_SOFTLOCKUP
262 extern void softlockup_tick(void);
263 extern void spawn_softlockup_task(void);
264 extern void touch_softlockup_watchdog(void);
265 extern void touch_all_softlockup_watchdogs(void);
266 extern int softlockup_thresh
;
268 static inline void softlockup_tick(void)
271 static inline void spawn_softlockup_task(void)
274 static inline void touch_softlockup_watchdog(void)
277 static inline void touch_all_softlockup_watchdogs(void)
283 /* Attach to any functions which should be ignored in wchan output. */
284 #define __sched __attribute__((__section__(".sched.text")))
286 /* Linker adds these: start and end of __sched functions */
287 extern char __sched_text_start
[], __sched_text_end
[];
289 /* Is this address in the __sched functions? */
290 extern int in_sched_functions(unsigned long addr
);
292 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
293 extern signed long FASTCALL(schedule_timeout(signed long timeout
));
294 extern signed long schedule_timeout_interruptible(signed long timeout
);
295 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
296 asmlinkage
void schedule(void);
299 struct user_namespace
;
301 /* Maximum number of active map areas.. This is a random (large) number */
302 #define DEFAULT_MAX_MAP_COUNT 65536
304 extern int sysctl_max_map_count
;
306 #include <linux/aio.h>
309 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
310 unsigned long, unsigned long);
312 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
313 unsigned long len
, unsigned long pgoff
,
314 unsigned long flags
);
315 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
316 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
318 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
320 * The mm counters are not protected by its page_table_lock,
321 * so must be incremented atomically.
323 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
324 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
325 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
326 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
327 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
329 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
331 * The mm counters are protected by its page_table_lock,
332 * so can be incremented directly.
334 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
335 #define get_mm_counter(mm, member) ((mm)->_##member)
336 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
337 #define inc_mm_counter(mm, member) (mm)->_##member++
338 #define dec_mm_counter(mm, member) (mm)->_##member--
340 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
342 #define get_mm_rss(mm) \
343 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
344 #define update_hiwater_rss(mm) do { \
345 unsigned long _rss = get_mm_rss(mm); \
346 if ((mm)->hiwater_rss < _rss) \
347 (mm)->hiwater_rss = _rss; \
349 #define update_hiwater_vm(mm) do { \
350 if ((mm)->hiwater_vm < (mm)->total_vm) \
351 (mm)->hiwater_vm = (mm)->total_vm; \
354 extern void set_dumpable(struct mm_struct
*mm
, int value
);
355 extern int get_dumpable(struct mm_struct
*mm
);
359 #define MMF_DUMPABLE 0 /* core dump is permitted */
360 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
361 #define MMF_DUMPABLE_BITS 2
363 /* coredump filter bits */
364 #define MMF_DUMP_ANON_PRIVATE 2
365 #define MMF_DUMP_ANON_SHARED 3
366 #define MMF_DUMP_MAPPED_PRIVATE 4
367 #define MMF_DUMP_MAPPED_SHARED 5
368 #define MMF_DUMP_ELF_HEADERS 6
369 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
370 #define MMF_DUMP_FILTER_BITS 5
371 #define MMF_DUMP_FILTER_MASK \
372 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
373 #define MMF_DUMP_FILTER_DEFAULT \
374 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
376 struct sighand_struct
{
378 struct k_sigaction action
[_NSIG
];
380 wait_queue_head_t signalfd_wqh
;
383 struct pacct_struct
{
386 unsigned long ac_mem
;
387 cputime_t ac_utime
, ac_stime
;
388 unsigned long ac_minflt
, ac_majflt
;
392 * NOTE! "signal_struct" does not have it's own
393 * locking, because a shared signal_struct always
394 * implies a shared sighand_struct, so locking
395 * sighand_struct is always a proper superset of
396 * the locking of signal_struct.
398 struct signal_struct
{
402 wait_queue_head_t wait_chldexit
; /* for wait4() */
404 /* current thread group signal load-balancing target: */
405 struct task_struct
*curr_target
;
407 /* shared signal handling: */
408 struct sigpending shared_pending
;
410 /* thread group exit support */
413 * - notify group_exit_task when ->count is equal to notify_count
414 * - everyone except group_exit_task is stopped during signal delivery
415 * of fatal signals, group_exit_task processes the signal.
417 struct task_struct
*group_exit_task
;
420 /* thread group stop support, overloads group_exit_code too */
421 int group_stop_count
;
422 unsigned int flags
; /* see SIGNAL_* flags below */
424 /* POSIX.1b Interval Timers */
425 struct list_head posix_timers
;
427 /* ITIMER_REAL timer for the process */
428 struct hrtimer real_timer
;
429 struct task_struct
*tsk
;
430 ktime_t it_real_incr
;
432 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
433 cputime_t it_prof_expires
, it_virt_expires
;
434 cputime_t it_prof_incr
, it_virt_incr
;
436 /* job control IDs */
439 * pgrp and session fields are deprecated.
440 * use the task_session_Xnr and task_pgrp_Xnr routines below
444 pid_t pgrp __deprecated
;
448 struct pid
*tty_old_pgrp
;
451 pid_t session __deprecated
;
455 /* boolean value for session group leader */
458 struct tty_struct
*tty
; /* NULL if no tty */
461 * Cumulative resource counters for dead threads in the group,
462 * and for reaped dead child processes forked by this group.
463 * Live threads maintain their own counters and add to these
464 * in __exit_signal, except for the group leader.
466 cputime_t utime
, stime
, cutime
, cstime
;
469 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
470 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
471 unsigned long inblock
, oublock
, cinblock
, coublock
;
474 * Cumulative ns of scheduled CPU time for dead threads in the
475 * group, not including a zombie group leader. (This only differs
476 * from jiffies_to_ns(utime + stime) if sched_clock uses something
477 * other than jiffies.)
479 unsigned long long sum_sched_runtime
;
482 * We don't bother to synchronize most readers of this at all,
483 * because there is no reader checking a limit that actually needs
484 * to get both rlim_cur and rlim_max atomically, and either one
485 * alone is a single word that can safely be read normally.
486 * getrlimit/setrlimit use task_lock(current->group_leader) to
487 * protect this instead of the siglock, because they really
488 * have no need to disable irqs.
490 struct rlimit rlim
[RLIM_NLIMITS
];
492 struct list_head cpu_timers
[3];
494 /* keep the process-shared keyrings here so that they do the right
495 * thing in threads created with CLONE_THREAD */
497 struct key
*session_keyring
; /* keyring inherited over fork */
498 struct key
*process_keyring
; /* keyring private to this process */
500 #ifdef CONFIG_BSD_PROCESS_ACCT
501 struct pacct_struct pacct
; /* per-process accounting information */
503 #ifdef CONFIG_TASKSTATS
504 struct taskstats
*stats
;
508 struct tty_audit_buf
*tty_audit_buf
;
512 /* Context switch must be unlocked if interrupts are to be enabled */
513 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
514 # define __ARCH_WANT_UNLOCKED_CTXSW
518 * Bits in flags field of signal_struct.
520 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
521 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
522 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
523 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
526 * Some day this will be a full-fledged user tracking system..
529 atomic_t __count
; /* reference count */
530 atomic_t processes
; /* How many processes does this user have? */
531 atomic_t files
; /* How many open files does this user have? */
532 atomic_t sigpending
; /* How many pending signals does this user have? */
533 #ifdef CONFIG_INOTIFY_USER
534 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
535 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
537 #ifdef CONFIG_POSIX_MQUEUE
538 /* protected by mq_lock */
539 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
541 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
544 struct key
*uid_keyring
; /* UID specific keyring */
545 struct key
*session_keyring
; /* UID's default session keyring */
548 /* Hash table maintenance information */
549 struct hlist_node uidhash_node
;
552 #ifdef CONFIG_FAIR_USER_SCHED
553 struct task_group
*tg
;
556 struct subsys_attribute user_attr
;
557 struct work_struct work
;
562 #ifdef CONFIG_FAIR_USER_SCHED
563 extern int uids_kobject_init(void);
565 static inline int uids_kobject_init(void) { return 0; }
568 extern struct user_struct
*find_user(uid_t
);
570 extern struct user_struct root_user
;
571 #define INIT_USER (&root_user)
573 struct backing_dev_info
;
574 struct reclaim_state
;
576 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
578 /* cumulative counters */
579 unsigned long pcount
; /* # of times run on this cpu */
580 unsigned long long cpu_time
, /* time spent on the cpu */
581 run_delay
; /* time spent waiting on a runqueue */
584 unsigned long long last_arrival
,/* when we last ran on a cpu */
585 last_queued
; /* when we were last queued to run */
586 #ifdef CONFIG_SCHEDSTATS
588 unsigned int bkl_count
;
591 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
593 #ifdef CONFIG_SCHEDSTATS
594 extern const struct file_operations proc_schedstat_operations
;
595 #endif /* CONFIG_SCHEDSTATS */
597 #ifdef CONFIG_TASK_DELAY_ACCT
598 struct task_delay_info
{
600 unsigned int flags
; /* Private per-task flags */
602 /* For each stat XXX, add following, aligned appropriately
604 * struct timespec XXX_start, XXX_end;
608 * Atomicity of updates to XXX_delay, XXX_count protected by
609 * single lock above (split into XXX_lock if contention is an issue).
613 * XXX_count is incremented on every XXX operation, the delay
614 * associated with the operation is added to XXX_delay.
615 * XXX_delay contains the accumulated delay time in nanoseconds.
617 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
618 u64 blkio_delay
; /* wait for sync block io completion */
619 u64 swapin_delay
; /* wait for swapin block io completion */
620 u32 blkio_count
; /* total count of the number of sync block */
621 /* io operations performed */
622 u32 swapin_count
; /* total count of the number of swapin block */
623 /* io operations performed */
625 #endif /* CONFIG_TASK_DELAY_ACCT */
627 static inline int sched_info_on(void)
629 #ifdef CONFIG_SCHEDSTATS
631 #elif defined(CONFIG_TASK_DELAY_ACCT)
632 extern int delayacct_on
;
647 * sched-domains (multiprocessor balancing) declarations:
651 * Increase resolution of nice-level calculations:
653 #define SCHED_LOAD_SHIFT 10
654 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
656 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
659 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
660 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
661 #define SD_BALANCE_EXEC 4 /* Balance on exec */
662 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
663 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
664 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
665 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
666 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
667 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
668 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
669 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
671 #define BALANCE_FOR_MC_POWER \
672 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
674 #define BALANCE_FOR_PKG_POWER \
675 ((sched_mc_power_savings || sched_smt_power_savings) ? \
676 SD_POWERSAVINGS_BALANCE : 0)
678 #define test_sd_parent(sd, flag) ((sd->parent && \
679 (sd->parent->flags & flag)) ? 1 : 0)
683 struct sched_group
*next
; /* Must be a circular list */
687 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
688 * single CPU. This is read only (except for setup, hotplug CPU).
689 * Note : Never change cpu_power without recompute its reciprocal
691 unsigned int __cpu_power
;
693 * reciprocal value of cpu_power to avoid expensive divides
694 * (see include/linux/reciprocal_div.h)
696 u32 reciprocal_cpu_power
;
699 struct sched_domain
{
700 /* These fields must be setup */
701 struct sched_domain
*parent
; /* top domain must be null terminated */
702 struct sched_domain
*child
; /* bottom domain must be null terminated */
703 struct sched_group
*groups
; /* the balancing groups of the domain */
704 cpumask_t span
; /* span of all CPUs in this domain */
705 unsigned long min_interval
; /* Minimum balance interval ms */
706 unsigned long max_interval
; /* Maximum balance interval ms */
707 unsigned int busy_factor
; /* less balancing by factor if busy */
708 unsigned int imbalance_pct
; /* No balance until over watermark */
709 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
710 unsigned int busy_idx
;
711 unsigned int idle_idx
;
712 unsigned int newidle_idx
;
713 unsigned int wake_idx
;
714 unsigned int forkexec_idx
;
715 int flags
; /* See SD_* */
717 /* Runtime fields. */
718 unsigned long last_balance
; /* init to jiffies. units in jiffies */
719 unsigned int balance_interval
; /* initialise to 1. units in ms. */
720 unsigned int nr_balance_failed
; /* initialise to 0 */
722 #ifdef CONFIG_SCHEDSTATS
723 /* load_balance() stats */
724 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
725 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
726 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
727 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
728 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
729 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
730 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
731 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
733 /* Active load balancing */
734 unsigned int alb_count
;
735 unsigned int alb_failed
;
736 unsigned int alb_pushed
;
738 /* SD_BALANCE_EXEC stats */
739 unsigned int sbe_count
;
740 unsigned int sbe_balanced
;
741 unsigned int sbe_pushed
;
743 /* SD_BALANCE_FORK stats */
744 unsigned int sbf_count
;
745 unsigned int sbf_balanced
;
746 unsigned int sbf_pushed
;
748 /* try_to_wake_up() stats */
749 unsigned int ttwu_wake_remote
;
750 unsigned int ttwu_move_affine
;
751 unsigned int ttwu_move_balance
;
755 extern void partition_sched_domains(int ndoms_new
, cpumask_t
*doms_new
);
757 #endif /* CONFIG_SMP */
760 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
761 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
762 * task of nice 0 or enough lower priority tasks to bring up the
765 static inline int above_background_load(void)
769 for_each_online_cpu(cpu
) {
770 if (weighted_cpuload(cpu
) >= SCHED_LOAD_SCALE
)
776 struct io_context
; /* See blkdev.h */
777 #define NGROUPS_SMALL 32
778 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
782 gid_t small_block
[NGROUPS_SMALL
];
788 * get_group_info() must be called with the owning task locked (via task_lock())
789 * when task != current. The reason being that the vast majority of callers are
790 * looking at current->group_info, which can not be changed except by the
791 * current task. Changing current->group_info requires the task lock, too.
793 #define get_group_info(group_info) do { \
794 atomic_inc(&(group_info)->usage); \
797 #define put_group_info(group_info) do { \
798 if (atomic_dec_and_test(&(group_info)->usage)) \
799 groups_free(group_info); \
802 extern struct group_info
*groups_alloc(int gidsetsize
);
803 extern void groups_free(struct group_info
*group_info
);
804 extern int set_current_groups(struct group_info
*group_info
);
805 extern int groups_search(struct group_info
*group_info
, gid_t grp
);
806 /* access the groups "array" with this macro */
807 #define GROUP_AT(gi, i) \
808 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
810 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
811 extern void prefetch_stack(struct task_struct
*t
);
813 static inline void prefetch_stack(struct task_struct
*t
) { }
816 struct audit_context
; /* See audit.c */
818 struct pipe_inode_info
;
819 struct uts_namespace
;
825 const struct sched_class
*next
;
827 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int wakeup
);
828 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int sleep
);
829 void (*yield_task
) (struct rq
*rq
);
831 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
);
833 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
834 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
837 unsigned long (*load_balance
) (struct rq
*this_rq
, int this_cpu
,
838 struct rq
*busiest
, unsigned long max_load_move
,
839 struct sched_domain
*sd
, enum cpu_idle_type idle
,
840 int *all_pinned
, int *this_best_prio
);
842 int (*move_one_task
) (struct rq
*this_rq
, int this_cpu
,
843 struct rq
*busiest
, struct sched_domain
*sd
,
844 enum cpu_idle_type idle
);
847 void (*set_curr_task
) (struct rq
*rq
);
848 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
);
849 void (*task_new
) (struct rq
*rq
, struct task_struct
*p
);
853 unsigned long weight
, inv_weight
;
857 * CFS stats for a schedulable entity (task, task-group etc)
859 * Current field usage histogram:
866 struct sched_entity
{
867 struct load_weight load
; /* for load-balancing */
868 struct rb_node run_node
;
872 u64 sum_exec_runtime
;
874 u64 prev_sum_exec_runtime
;
876 #ifdef CONFIG_SCHEDSTATS
882 s64 sum_sleep_runtime
;
890 u64 nr_migrations_cold
;
891 u64 nr_failed_migrations_affine
;
892 u64 nr_failed_migrations_running
;
893 u64 nr_failed_migrations_hot
;
894 u64 nr_forced_migrations
;
895 u64 nr_forced2_migrations
;
899 u64 nr_wakeups_migrate
;
900 u64 nr_wakeups_local
;
901 u64 nr_wakeups_remote
;
902 u64 nr_wakeups_affine
;
903 u64 nr_wakeups_affine_attempts
;
904 u64 nr_wakeups_passive
;
908 #ifdef CONFIG_FAIR_GROUP_SCHED
909 struct sched_entity
*parent
;
910 /* rq on which this entity is (to be) queued: */
911 struct cfs_rq
*cfs_rq
;
912 /* rq "owned" by this entity/group: */
918 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
921 unsigned int flags
; /* per process flags, defined below */
924 int lock_depth
; /* BKL lock depth */
927 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
932 int prio
, static_prio
, normal_prio
;
933 struct list_head run_list
;
934 const struct sched_class
*sched_class
;
935 struct sched_entity se
;
937 #ifdef CONFIG_PREEMPT_NOTIFIERS
938 /* list of struct preempt_notifier: */
939 struct hlist_head preempt_notifiers
;
942 unsigned short ioprio
;
944 * fpu_counter contains the number of consecutive context switches
945 * that the FPU is used. If this is over a threshold, the lazy fpu
946 * saving becomes unlazy to save the trap. This is an unsigned char
947 * so that after 256 times the counter wraps and the behavior turns
948 * lazy again; this to deal with bursty apps that only use FPU for
951 unsigned char fpu_counter
;
952 s8 oomkilladj
; /* OOM kill score adjustment (bit shift). */
953 #ifdef CONFIG_BLK_DEV_IO_TRACE
954 unsigned int btrace_seq
;
958 cpumask_t cpus_allowed
;
959 unsigned int time_slice
;
961 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
962 struct sched_info sched_info
;
965 struct list_head tasks
;
967 * ptrace_list/ptrace_children forms the list of my children
968 * that were stolen by a ptracer.
970 struct list_head ptrace_children
;
971 struct list_head ptrace_list
;
973 struct mm_struct
*mm
, *active_mm
;
976 struct linux_binfmt
*binfmt
;
978 int exit_code
, exit_signal
;
979 int pdeath_signal
; /* The signal sent when the parent dies */
981 unsigned int personality
;
986 #ifdef CONFIG_CC_STACKPROTECTOR
987 /* Canary value for the -fstack-protector gcc feature */
988 unsigned long stack_canary
;
991 * pointers to (original) parent process, youngest child, younger sibling,
992 * older sibling, respectively. (p->father can be replaced with
995 struct task_struct
*real_parent
; /* real parent process (when being debugged) */
996 struct task_struct
*parent
; /* parent process */
998 * children/sibling forms the list of my children plus the
999 * tasks I'm ptracing.
1001 struct list_head children
; /* list of my children */
1002 struct list_head sibling
; /* linkage in my parent's children list */
1003 struct task_struct
*group_leader
; /* threadgroup leader */
1005 /* PID/PID hash table linkage. */
1006 struct pid_link pids
[PIDTYPE_MAX
];
1007 struct list_head thread_group
;
1009 struct completion
*vfork_done
; /* for vfork() */
1010 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1011 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1013 unsigned int rt_priority
;
1014 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1016 cputime_t prev_utime
, prev_stime
;
1017 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1018 struct timespec start_time
; /* monotonic time */
1019 struct timespec real_start_time
; /* boot based time */
1020 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1021 unsigned long min_flt
, maj_flt
;
1023 cputime_t it_prof_expires
, it_virt_expires
;
1024 unsigned long long it_sched_expires
;
1025 struct list_head cpu_timers
[3];
1027 /* process credentials */
1028 uid_t uid
,euid
,suid
,fsuid
;
1029 gid_t gid
,egid
,sgid
,fsgid
;
1030 struct group_info
*group_info
;
1031 kernel_cap_t cap_effective
, cap_inheritable
, cap_permitted
;
1032 unsigned keep_capabilities
:1;
1033 struct user_struct
*user
;
1035 struct key
*request_key_auth
; /* assumed request_key authority */
1036 struct key
*thread_keyring
; /* keyring private to this thread */
1037 unsigned char jit_keyring
; /* default keyring to attach requested keys to */
1039 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1040 - access with [gs]et_task_comm (which lock
1041 it with task_lock())
1042 - initialized normally by flush_old_exec */
1043 /* file system info */
1044 int link_count
, total_link_count
;
1045 #ifdef CONFIG_SYSVIPC
1047 struct sysv_sem sysvsem
;
1049 /* CPU-specific state of this task */
1050 struct thread_struct thread
;
1051 /* filesystem information */
1052 struct fs_struct
*fs
;
1053 /* open file information */
1054 struct files_struct
*files
;
1056 struct nsproxy
*nsproxy
;
1057 /* signal handlers */
1058 struct signal_struct
*signal
;
1059 struct sighand_struct
*sighand
;
1061 sigset_t blocked
, real_blocked
;
1062 sigset_t saved_sigmask
; /* To be restored with TIF_RESTORE_SIGMASK */
1063 struct sigpending pending
;
1065 unsigned long sas_ss_sp
;
1067 int (*notifier
)(void *priv
);
1068 void *notifier_data
;
1069 sigset_t
*notifier_mask
;
1070 #ifdef CONFIG_SECURITY
1073 struct audit_context
*audit_context
;
1076 /* Thread group tracking */
1079 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1080 spinlock_t alloc_lock
;
1082 /* Protection of the PI data structures: */
1085 #ifdef CONFIG_RT_MUTEXES
1086 /* PI waiters blocked on a rt_mutex held by this task */
1087 struct plist_head pi_waiters
;
1088 /* Deadlock detection and priority inheritance handling */
1089 struct rt_mutex_waiter
*pi_blocked_on
;
1092 #ifdef CONFIG_DEBUG_MUTEXES
1093 /* mutex deadlock detection */
1094 struct mutex_waiter
*blocked_on
;
1096 #ifdef CONFIG_TRACE_IRQFLAGS
1097 unsigned int irq_events
;
1098 int hardirqs_enabled
;
1099 unsigned long hardirq_enable_ip
;
1100 unsigned int hardirq_enable_event
;
1101 unsigned long hardirq_disable_ip
;
1102 unsigned int hardirq_disable_event
;
1103 int softirqs_enabled
;
1104 unsigned long softirq_disable_ip
;
1105 unsigned int softirq_disable_event
;
1106 unsigned long softirq_enable_ip
;
1107 unsigned int softirq_enable_event
;
1108 int hardirq_context
;
1109 int softirq_context
;
1111 #ifdef CONFIG_LOCKDEP
1112 # define MAX_LOCK_DEPTH 30UL
1115 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1116 unsigned int lockdep_recursion
;
1119 /* journalling filesystem info */
1122 /* stacked block device info */
1123 struct bio
*bio_list
, **bio_tail
;
1126 struct reclaim_state
*reclaim_state
;
1128 struct backing_dev_info
*backing_dev_info
;
1130 struct io_context
*io_context
;
1132 unsigned long ptrace_message
;
1133 siginfo_t
*last_siginfo
; /* For ptrace use. */
1134 #ifdef CONFIG_TASK_XACCT
1135 /* i/o counters(bytes read/written, #syscalls */
1136 u64 rchar
, wchar
, syscr
, syscw
;
1138 struct task_io_accounting ioac
;
1139 #if defined(CONFIG_TASK_XACCT)
1140 u64 acct_rss_mem1
; /* accumulated rss usage */
1141 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1142 cputime_t acct_stimexpd
;/* stime since last update */
1145 struct mempolicy
*mempolicy
;
1148 #ifdef CONFIG_CPUSETS
1149 nodemask_t mems_allowed
;
1150 int cpuset_mems_generation
;
1151 int cpuset_mem_spread_rotor
;
1153 #ifdef CONFIG_CGROUPS
1154 /* Control Group info protected by css_set_lock */
1155 struct css_set
*cgroups
;
1156 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1157 struct list_head cg_list
;
1160 struct robust_list_head __user
*robust_list
;
1161 #ifdef CONFIG_COMPAT
1162 struct compat_robust_list_head __user
*compat_robust_list
;
1164 struct list_head pi_state_list
;
1165 struct futex_pi_state
*pi_state_cache
;
1167 atomic_t fs_excl
; /* holding fs exclusive resources */
1168 struct rcu_head rcu
;
1171 * cache last used pipe for splice
1173 struct pipe_inode_info
*splice_pipe
;
1174 #ifdef CONFIG_TASK_DELAY_ACCT
1175 struct task_delay_info
*delays
;
1177 #ifdef CONFIG_FAULT_INJECTION
1180 struct prop_local_single dirties
;
1184 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1185 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1186 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1187 * values are inverted: lower p->prio value means higher priority.
1189 * The MAX_USER_RT_PRIO value allows the actual maximum
1190 * RT priority to be separate from the value exported to
1191 * user-space. This allows kernel threads to set their
1192 * priority to a value higher than any user task. Note:
1193 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1196 #define MAX_USER_RT_PRIO 100
1197 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1199 #define MAX_PRIO (MAX_RT_PRIO + 40)
1200 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1202 static inline int rt_prio(int prio
)
1204 if (unlikely(prio
< MAX_RT_PRIO
))
1209 static inline int rt_task(struct task_struct
*p
)
1211 return rt_prio(p
->prio
);
1214 static inline void set_task_session(struct task_struct
*tsk
, pid_t session
)
1216 tsk
->signal
->__session
= session
;
1219 static inline void set_task_pgrp(struct task_struct
*tsk
, pid_t pgrp
)
1221 tsk
->signal
->__pgrp
= pgrp
;
1224 static inline struct pid
*task_pid(struct task_struct
*task
)
1226 return task
->pids
[PIDTYPE_PID
].pid
;
1229 static inline struct pid
*task_tgid(struct task_struct
*task
)
1231 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1234 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1236 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1239 static inline struct pid
*task_session(struct task_struct
*task
)
1241 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1244 struct pid_namespace
;
1247 * the helpers to get the task's different pids as they are seen
1248 * from various namespaces
1250 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1251 * task_xid_vnr() : virtual id, i.e. the id seen from the namespace the task
1252 * belongs to. this only makes sence when called in the
1253 * context of the task that belongs to the same namespace;
1254 * task_xid_nr_ns() : id seen from the ns specified;
1256 * set_task_vxid() : assigns a virtual id to a task;
1258 * see also pid_nr() etc in include/linux/pid.h
1261 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1266 pid_t
task_pid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1268 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1270 return pid_vnr(task_pid(tsk
));
1274 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1279 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1281 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1283 return pid_vnr(task_tgid(tsk
));
1287 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1289 return tsk
->signal
->__pgrp
;
1292 pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1294 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1296 return pid_vnr(task_pgrp(tsk
));
1300 static inline pid_t
task_session_nr(struct task_struct
*tsk
)
1302 return tsk
->signal
->__session
;
1305 pid_t
task_session_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1307 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1309 return pid_vnr(task_session(tsk
));
1314 * pid_alive - check that a task structure is not stale
1315 * @p: Task structure to be checked.
1317 * Test if a process is not yet dead (at most zombie state)
1318 * If pid_alive fails, then pointers within the task structure
1319 * can be stale and must not be dereferenced.
1321 static inline int pid_alive(struct task_struct
*p
)
1323 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1327 * is_global_init - check if a task structure is init
1328 * @tsk: Task structure to be checked.
1330 * Check if a task structure is the first user space task the kernel created.
1332 static inline int is_global_init(struct task_struct
*tsk
)
1334 return tsk
->pid
== 1;
1338 * is_container_init:
1339 * check whether in the task is init in its own pid namespace.
1341 extern int is_container_init(struct task_struct
*tsk
);
1343 extern struct pid
*cad_pid
;
1345 extern void free_task(struct task_struct
*tsk
);
1346 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1348 extern void __put_task_struct(struct task_struct
*t
);
1350 static inline void put_task_struct(struct task_struct
*t
)
1352 if (atomic_dec_and_test(&t
->usage
))
1353 __put_task_struct(t
);
1359 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1360 /* Not implemented yet, only for 486*/
1361 #define PF_STARTING 0x00000002 /* being created */
1362 #define PF_EXITING 0x00000004 /* getting shut down */
1363 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1364 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1365 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1366 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1367 #define PF_DUMPCORE 0x00000200 /* dumped core */
1368 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1369 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1370 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1371 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1372 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1373 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1374 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1375 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1376 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1377 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1378 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1379 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1380 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1381 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1382 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1383 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1384 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1385 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1388 * Only the _current_ task can read/write to tsk->flags, but other
1389 * tasks can access tsk->flags in readonly mode for example
1390 * with tsk_used_math (like during threaded core dumping).
1391 * There is however an exception to this rule during ptrace
1392 * or during fork: the ptracer task is allowed to write to the
1393 * child->flags of its traced child (same goes for fork, the parent
1394 * can write to the child->flags), because we're guaranteed the
1395 * child is not running and in turn not changing child->flags
1396 * at the same time the parent does it.
1398 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1399 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1400 #define clear_used_math() clear_stopped_child_used_math(current)
1401 #define set_used_math() set_stopped_child_used_math(current)
1402 #define conditional_stopped_child_used_math(condition, child) \
1403 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1404 #define conditional_used_math(condition) \
1405 conditional_stopped_child_used_math(condition, current)
1406 #define copy_to_stopped_child_used_math(child) \
1407 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1408 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1409 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1410 #define used_math() tsk_used_math(current)
1413 extern int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
);
1415 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1417 if (!cpu_isset(0, new_mask
))
1423 extern unsigned long long sched_clock(void);
1426 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1427 * clock constructed from sched_clock():
1429 extern unsigned long long cpu_clock(int cpu
);
1431 extern unsigned long long
1432 task_sched_runtime(struct task_struct
*task
);
1434 /* sched_exec is called by processes performing an exec */
1436 extern void sched_exec(void);
1438 #define sched_exec() {}
1441 extern void sched_clock_idle_sleep_event(void);
1442 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1444 #ifdef CONFIG_HOTPLUG_CPU
1445 extern void idle_task_exit(void);
1447 static inline void idle_task_exit(void) {}
1450 extern void sched_idle_next(void);
1452 #ifdef CONFIG_SCHED_DEBUG
1453 extern unsigned int sysctl_sched_latency
;
1454 extern unsigned int sysctl_sched_min_granularity
;
1455 extern unsigned int sysctl_sched_wakeup_granularity
;
1456 extern unsigned int sysctl_sched_batch_wakeup_granularity
;
1457 extern unsigned int sysctl_sched_child_runs_first
;
1458 extern unsigned int sysctl_sched_features
;
1459 extern unsigned int sysctl_sched_migration_cost
;
1460 extern unsigned int sysctl_sched_nr_migrate
;
1462 int sched_nr_latency_handler(struct ctl_table
*table
, int write
,
1463 struct file
*file
, void __user
*buffer
, size_t *length
,
1467 extern unsigned int sysctl_sched_compat_yield
;
1469 #ifdef CONFIG_RT_MUTEXES
1470 extern int rt_mutex_getprio(struct task_struct
*p
);
1471 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
1472 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
1474 static inline int rt_mutex_getprio(struct task_struct
*p
)
1476 return p
->normal_prio
;
1478 # define rt_mutex_adjust_pi(p) do { } while (0)
1481 extern void set_user_nice(struct task_struct
*p
, long nice
);
1482 extern int task_prio(const struct task_struct
*p
);
1483 extern int task_nice(const struct task_struct
*p
);
1484 extern int can_nice(const struct task_struct
*p
, const int nice
);
1485 extern int task_curr(const struct task_struct
*p
);
1486 extern int idle_cpu(int cpu
);
1487 extern int sched_setscheduler(struct task_struct
*, int, struct sched_param
*);
1488 extern struct task_struct
*idle_task(int cpu
);
1489 extern struct task_struct
*curr_task(int cpu
);
1490 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1495 * The default (Linux) execution domain.
1497 extern struct exec_domain default_exec_domain
;
1499 union thread_union
{
1500 struct thread_info thread_info
;
1501 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1504 #ifndef __HAVE_ARCH_KSTACK_END
1505 static inline int kstack_end(void *addr
)
1507 /* Reliable end of stack detection:
1508 * Some APM bios versions misalign the stack
1510 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1514 extern union thread_union init_thread_union
;
1515 extern struct task_struct init_task
;
1517 extern struct mm_struct init_mm
;
1519 extern struct pid_namespace init_pid_ns
;
1522 * find a task by one of its numerical ids
1524 * find_task_by_pid_type_ns():
1525 * it is the most generic call - it finds a task by all id,
1526 * type and namespace specified
1527 * find_task_by_pid_ns():
1528 * finds a task by its pid in the specified namespace
1529 * find_task_by_vpid():
1530 * finds a task by its virtual pid
1531 * find_task_by_pid():
1532 * finds a task by its global pid
1534 * see also find_pid() etc in include/linux/pid.h
1537 extern struct task_struct
*find_task_by_pid_type_ns(int type
, int pid
,
1538 struct pid_namespace
*ns
);
1540 extern struct task_struct
*find_task_by_pid(pid_t nr
);
1541 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1542 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
1543 struct pid_namespace
*ns
);
1545 extern void __set_special_pids(pid_t session
, pid_t pgrp
);
1547 /* per-UID process charging. */
1548 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
1549 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1551 atomic_inc(&u
->__count
);
1554 extern void free_uid(struct user_struct
*);
1555 extern void switch_uid(struct user_struct
*);
1556 extern void release_uids(struct user_namespace
*ns
);
1558 #include <asm/current.h>
1560 extern void do_timer(unsigned long ticks
);
1562 extern int FASTCALL(wake_up_state(struct task_struct
* tsk
, unsigned int state
));
1563 extern int FASTCALL(wake_up_process(struct task_struct
* tsk
));
1564 extern void FASTCALL(wake_up_new_task(struct task_struct
* tsk
,
1565 unsigned long clone_flags
));
1567 extern void kick_process(struct task_struct
*tsk
);
1569 static inline void kick_process(struct task_struct
*tsk
) { }
1571 extern void sched_fork(struct task_struct
*p
, int clone_flags
);
1572 extern void sched_dead(struct task_struct
*p
);
1574 extern int in_group_p(gid_t
);
1575 extern int in_egroup_p(gid_t
);
1577 extern void proc_caches_init(void);
1578 extern void flush_signals(struct task_struct
*);
1579 extern void ignore_signals(struct task_struct
*);
1580 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
1581 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
1583 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
1585 unsigned long flags
;
1588 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
1589 ret
= dequeue_signal(tsk
, mask
, info
);
1590 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
1595 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
1597 extern void unblock_all_signals(void);
1598 extern void release_task(struct task_struct
* p
);
1599 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
1600 extern int send_group_sig_info(int, struct siginfo
*, struct task_struct
*);
1601 extern int force_sigsegv(int, struct task_struct
*);
1602 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
1603 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1604 extern int kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1605 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
1606 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
1607 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
1608 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
1609 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
1610 extern void do_notify_parent(struct task_struct
*, int);
1611 extern void force_sig(int, struct task_struct
*);
1612 extern void force_sig_specific(int, struct task_struct
*);
1613 extern int send_sig(int, struct task_struct
*, int);
1614 extern void zap_other_threads(struct task_struct
*p
);
1615 extern int kill_proc(pid_t
, int, int);
1616 extern struct sigqueue
*sigqueue_alloc(void);
1617 extern void sigqueue_free(struct sigqueue
*);
1618 extern int send_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1619 extern int send_group_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1620 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
1621 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
1623 static inline int kill_cad_pid(int sig
, int priv
)
1625 return kill_pid(cad_pid
, sig
, priv
);
1628 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1629 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1630 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1631 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1633 static inline int is_si_special(const struct siginfo
*info
)
1635 return info
<= SEND_SIG_FORCED
;
1638 /* True if we are on the alternate signal stack. */
1640 static inline int on_sig_stack(unsigned long sp
)
1642 return (sp
- current
->sas_ss_sp
< current
->sas_ss_size
);
1645 static inline int sas_ss_flags(unsigned long sp
)
1647 return (current
->sas_ss_size
== 0 ? SS_DISABLE
1648 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
1652 * Routines for handling mm_structs
1654 extern struct mm_struct
* mm_alloc(void);
1656 /* mmdrop drops the mm and the page tables */
1657 extern void FASTCALL(__mmdrop(struct mm_struct
*));
1658 static inline void mmdrop(struct mm_struct
* mm
)
1660 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
1664 /* mmput gets rid of the mappings and all user-space */
1665 extern void mmput(struct mm_struct
*);
1666 /* Grab a reference to a task's mm, if it is not already going away */
1667 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
1668 /* Remove the current tasks stale references to the old mm_struct */
1669 extern void mm_release(struct task_struct
*, struct mm_struct
*);
1671 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct
*, struct pt_regs
*);
1672 extern void flush_thread(void);
1673 extern void exit_thread(void);
1675 extern void exit_files(struct task_struct
*);
1676 extern void __cleanup_signal(struct signal_struct
*);
1677 extern void __cleanup_sighand(struct sighand_struct
*);
1678 extern void exit_itimers(struct signal_struct
*);
1680 extern NORET_TYPE
void do_group_exit(int);
1682 extern void daemonize(const char *, ...);
1683 extern int allow_signal(int);
1684 extern int disallow_signal(int);
1686 extern int do_execve(char *, char __user
* __user
*, char __user
* __user
*, struct pt_regs
*);
1687 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
1688 struct task_struct
*fork_idle(int);
1690 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
1691 extern void get_task_comm(char *to
, struct task_struct
*tsk
);
1694 extern void wait_task_inactive(struct task_struct
* p
);
1696 #define wait_task_inactive(p) do { } while (0)
1699 #define remove_parent(p) list_del_init(&(p)->sibling)
1700 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1702 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1704 #define for_each_process(p) \
1705 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1708 * Careful: do_each_thread/while_each_thread is a double loop so
1709 * 'break' will not work as expected - use goto instead.
1711 #define do_each_thread(g, t) \
1712 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1714 #define while_each_thread(g, t) \
1715 while ((t = next_thread(t)) != g)
1717 /* de_thread depends on thread_group_leader not being a pid based check */
1718 #define thread_group_leader(p) (p == p->group_leader)
1720 /* Do to the insanities of de_thread it is possible for a process
1721 * to have the pid of the thread group leader without actually being
1722 * the thread group leader. For iteration through the pids in proc
1723 * all we care about is that we have a task with the appropriate
1724 * pid, we don't actually care if we have the right task.
1726 static inline int has_group_leader_pid(struct task_struct
*p
)
1728 return p
->pid
== p
->tgid
;
1732 int same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
1734 return p1
->tgid
== p2
->tgid
;
1737 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
1739 return list_entry(rcu_dereference(p
->thread_group
.next
),
1740 struct task_struct
, thread_group
);
1743 static inline int thread_group_empty(struct task_struct
*p
)
1745 return list_empty(&p
->thread_group
);
1748 #define delay_group_leader(p) \
1749 (thread_group_leader(p) && !thread_group_empty(p))
1752 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1753 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1754 * pins the final release of task.io_context. Also protects ->cpuset and
1755 * ->cgroup.subsys[].
1757 * Nests both inside and outside of read_lock(&tasklist_lock).
1758 * It must not be nested with write_lock_irq(&tasklist_lock),
1759 * neither inside nor outside.
1761 static inline void task_lock(struct task_struct
*p
)
1763 spin_lock(&p
->alloc_lock
);
1766 static inline void task_unlock(struct task_struct
*p
)
1768 spin_unlock(&p
->alloc_lock
);
1771 extern struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
1772 unsigned long *flags
);
1774 static inline void unlock_task_sighand(struct task_struct
*tsk
,
1775 unsigned long *flags
)
1777 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
1780 #ifndef __HAVE_THREAD_FUNCTIONS
1782 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1783 #define task_stack_page(task) ((task)->stack)
1785 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
1787 *task_thread_info(p
) = *task_thread_info(org
);
1788 task_thread_info(p
)->task
= p
;
1791 static inline unsigned long *end_of_stack(struct task_struct
*p
)
1793 return (unsigned long *)(task_thread_info(p
) + 1);
1798 /* set thread flags in other task's structures
1799 * - see asm/thread_info.h for TIF_xxxx flags available
1801 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1803 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1806 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1808 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1811 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1813 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1816 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1818 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1821 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1823 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1826 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1828 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1831 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1833 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1836 static inline int signal_pending(struct task_struct
*p
)
1838 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
1841 static inline int need_resched(void)
1843 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
1847 * cond_resched() and cond_resched_lock(): latency reduction via
1848 * explicit rescheduling in places that are safe. The return
1849 * value indicates whether a reschedule was done in fact.
1850 * cond_resched_lock() will drop the spinlock before scheduling,
1851 * cond_resched_softirq() will enable bhs before scheduling.
1853 extern int cond_resched(void);
1854 extern int cond_resched_lock(spinlock_t
* lock
);
1855 extern int cond_resched_softirq(void);
1858 * Does a critical section need to be broken due to another
1861 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1862 # define need_lockbreak(lock) ((lock)->break_lock)
1864 # define need_lockbreak(lock) 0
1868 * Does a critical section need to be broken due to another
1869 * task waiting or preemption being signalled:
1871 static inline int lock_need_resched(spinlock_t
*lock
)
1873 if (need_lockbreak(lock
) || need_resched())
1879 * Reevaluate whether the task has signals pending delivery.
1880 * Wake the task if so.
1881 * This is required every time the blocked sigset_t changes.
1882 * callers must hold sighand->siglock.
1884 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
1885 extern void recalc_sigpending(void);
1887 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
1890 * Wrappers for p->thread_info->cpu access. No-op on UP.
1894 static inline unsigned int task_cpu(const struct task_struct
*p
)
1896 return task_thread_info(p
)->cpu
;
1899 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
1903 static inline unsigned int task_cpu(const struct task_struct
*p
)
1908 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
1912 #endif /* CONFIG_SMP */
1914 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1915 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
1917 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
)
1919 mm
->mmap_base
= TASK_UNMAPPED_BASE
;
1920 mm
->get_unmapped_area
= arch_get_unmapped_area
;
1921 mm
->unmap_area
= arch_unmap_area
;
1925 extern long sched_setaffinity(pid_t pid
, cpumask_t new_mask
);
1926 extern long sched_getaffinity(pid_t pid
, cpumask_t
*mask
);
1928 extern int sched_mc_power_savings
, sched_smt_power_savings
;
1930 extern void normalize_rt_tasks(void);
1932 #ifdef CONFIG_FAIR_GROUP_SCHED
1934 extern struct task_group init_task_group
;
1936 extern struct task_group
*sched_create_group(void);
1937 extern void sched_destroy_group(struct task_group
*tg
);
1938 extern void sched_move_task(struct task_struct
*tsk
);
1939 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
1940 extern unsigned long sched_group_shares(struct task_group
*tg
);
1944 #ifdef CONFIG_TASK_XACCT
1945 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1950 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1955 static inline void inc_syscr(struct task_struct
*tsk
)
1960 static inline void inc_syscw(struct task_struct
*tsk
)
1965 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1969 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1973 static inline void inc_syscr(struct task_struct
*tsk
)
1977 static inline void inc_syscw(struct task_struct
*tsk
)
1983 void migration_init(void);
1985 static inline void migration_init(void)
1990 #endif /* __KERNEL__ */