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 extern void sched_show_task(struct task_struct
*p
);
263 #ifdef CONFIG_DETECT_SOFTLOCKUP
264 extern void softlockup_tick(void);
265 extern void spawn_softlockup_task(void);
266 extern void touch_softlockup_watchdog(void);
267 extern void touch_all_softlockup_watchdogs(void);
268 extern int softlockup_thresh
;
269 extern unsigned long sysctl_hung_task_check_count
;
270 extern unsigned long sysctl_hung_task_timeout_secs
;
271 extern long sysctl_hung_task_warnings
;
273 static inline void softlockup_tick(void)
276 static inline void spawn_softlockup_task(void)
279 static inline void touch_softlockup_watchdog(void)
282 static inline void touch_all_softlockup_watchdogs(void)
288 /* Attach to any functions which should be ignored in wchan output. */
289 #define __sched __attribute__((__section__(".sched.text")))
291 /* Linker adds these: start and end of __sched functions */
292 extern char __sched_text_start
[], __sched_text_end
[];
294 /* Is this address in the __sched functions? */
295 extern int in_sched_functions(unsigned long addr
);
297 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
298 extern signed long FASTCALL(schedule_timeout(signed long timeout
));
299 extern signed long schedule_timeout_interruptible(signed long timeout
);
300 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
301 asmlinkage
void schedule(void);
304 struct user_namespace
;
306 /* Maximum number of active map areas.. This is a random (large) number */
307 #define DEFAULT_MAX_MAP_COUNT 65536
309 extern int sysctl_max_map_count
;
311 #include <linux/aio.h>
314 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
315 unsigned long, unsigned long);
317 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
318 unsigned long len
, unsigned long pgoff
,
319 unsigned long flags
);
320 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
321 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
323 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
325 * The mm counters are not protected by its page_table_lock,
326 * so must be incremented atomically.
328 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
329 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
330 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
331 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
332 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
334 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
336 * The mm counters are protected by its page_table_lock,
337 * so can be incremented directly.
339 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
340 #define get_mm_counter(mm, member) ((mm)->_##member)
341 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
342 #define inc_mm_counter(mm, member) (mm)->_##member++
343 #define dec_mm_counter(mm, member) (mm)->_##member--
345 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
347 #define get_mm_rss(mm) \
348 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
349 #define update_hiwater_rss(mm) do { \
350 unsigned long _rss = get_mm_rss(mm); \
351 if ((mm)->hiwater_rss < _rss) \
352 (mm)->hiwater_rss = _rss; \
354 #define update_hiwater_vm(mm) do { \
355 if ((mm)->hiwater_vm < (mm)->total_vm) \
356 (mm)->hiwater_vm = (mm)->total_vm; \
359 extern void set_dumpable(struct mm_struct
*mm
, int value
);
360 extern int get_dumpable(struct mm_struct
*mm
);
364 #define MMF_DUMPABLE 0 /* core dump is permitted */
365 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
366 #define MMF_DUMPABLE_BITS 2
368 /* coredump filter bits */
369 #define MMF_DUMP_ANON_PRIVATE 2
370 #define MMF_DUMP_ANON_SHARED 3
371 #define MMF_DUMP_MAPPED_PRIVATE 4
372 #define MMF_DUMP_MAPPED_SHARED 5
373 #define MMF_DUMP_ELF_HEADERS 6
374 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
375 #define MMF_DUMP_FILTER_BITS 5
376 #define MMF_DUMP_FILTER_MASK \
377 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
378 #define MMF_DUMP_FILTER_DEFAULT \
379 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
381 struct sighand_struct
{
383 struct k_sigaction action
[_NSIG
];
385 wait_queue_head_t signalfd_wqh
;
388 struct pacct_struct
{
391 unsigned long ac_mem
;
392 cputime_t ac_utime
, ac_stime
;
393 unsigned long ac_minflt
, ac_majflt
;
397 * NOTE! "signal_struct" does not have it's own
398 * locking, because a shared signal_struct always
399 * implies a shared sighand_struct, so locking
400 * sighand_struct is always a proper superset of
401 * the locking of signal_struct.
403 struct signal_struct
{
407 wait_queue_head_t wait_chldexit
; /* for wait4() */
409 /* current thread group signal load-balancing target: */
410 struct task_struct
*curr_target
;
412 /* shared signal handling: */
413 struct sigpending shared_pending
;
415 /* thread group exit support */
418 * - notify group_exit_task when ->count is equal to notify_count
419 * - everyone except group_exit_task is stopped during signal delivery
420 * of fatal signals, group_exit_task processes the signal.
422 struct task_struct
*group_exit_task
;
425 /* thread group stop support, overloads group_exit_code too */
426 int group_stop_count
;
427 unsigned int flags
; /* see SIGNAL_* flags below */
429 /* POSIX.1b Interval Timers */
430 struct list_head posix_timers
;
432 /* ITIMER_REAL timer for the process */
433 struct hrtimer real_timer
;
434 struct task_struct
*tsk
;
435 ktime_t it_real_incr
;
437 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
438 cputime_t it_prof_expires
, it_virt_expires
;
439 cputime_t it_prof_incr
, it_virt_incr
;
441 /* job control IDs */
444 * pgrp and session fields are deprecated.
445 * use the task_session_Xnr and task_pgrp_Xnr routines below
449 pid_t pgrp __deprecated
;
453 struct pid
*tty_old_pgrp
;
456 pid_t session __deprecated
;
460 /* boolean value for session group leader */
463 struct tty_struct
*tty
; /* NULL if no tty */
466 * Cumulative resource counters for dead threads in the group,
467 * and for reaped dead child processes forked by this group.
468 * Live threads maintain their own counters and add to these
469 * in __exit_signal, except for the group leader.
471 cputime_t utime
, stime
, cutime
, cstime
;
474 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
475 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
476 unsigned long inblock
, oublock
, cinblock
, coublock
;
479 * Cumulative ns of scheduled CPU time for dead threads in the
480 * group, not including a zombie group leader. (This only differs
481 * from jiffies_to_ns(utime + stime) if sched_clock uses something
482 * other than jiffies.)
484 unsigned long long sum_sched_runtime
;
487 * We don't bother to synchronize most readers of this at all,
488 * because there is no reader checking a limit that actually needs
489 * to get both rlim_cur and rlim_max atomically, and either one
490 * alone is a single word that can safely be read normally.
491 * getrlimit/setrlimit use task_lock(current->group_leader) to
492 * protect this instead of the siglock, because they really
493 * have no need to disable irqs.
495 struct rlimit rlim
[RLIM_NLIMITS
];
497 struct list_head cpu_timers
[3];
499 /* keep the process-shared keyrings here so that they do the right
500 * thing in threads created with CLONE_THREAD */
502 struct key
*session_keyring
; /* keyring inherited over fork */
503 struct key
*process_keyring
; /* keyring private to this process */
505 #ifdef CONFIG_BSD_PROCESS_ACCT
506 struct pacct_struct pacct
; /* per-process accounting information */
508 #ifdef CONFIG_TASKSTATS
509 struct taskstats
*stats
;
513 struct tty_audit_buf
*tty_audit_buf
;
517 /* Context switch must be unlocked if interrupts are to be enabled */
518 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
519 # define __ARCH_WANT_UNLOCKED_CTXSW
523 * Bits in flags field of signal_struct.
525 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
526 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
527 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
528 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
531 * Some day this will be a full-fledged user tracking system..
534 atomic_t __count
; /* reference count */
535 atomic_t processes
; /* How many processes does this user have? */
536 atomic_t files
; /* How many open files does this user have? */
537 atomic_t sigpending
; /* How many pending signals does this user have? */
538 #ifdef CONFIG_INOTIFY_USER
539 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
540 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
542 #ifdef CONFIG_POSIX_MQUEUE
543 /* protected by mq_lock */
544 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
546 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
549 struct key
*uid_keyring
; /* UID specific keyring */
550 struct key
*session_keyring
; /* UID's default session keyring */
553 /* Hash table maintenance information */
554 struct hlist_node uidhash_node
;
557 #ifdef CONFIG_FAIR_USER_SCHED
558 struct task_group
*tg
;
561 struct work_struct work
;
566 extern int uids_sysfs_init(void);
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
);
830 int (*select_task_rq
)(struct task_struct
*p
, int sync
);
832 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
);
834 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
835 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
838 unsigned long (*load_balance
) (struct rq
*this_rq
, int this_cpu
,
839 struct rq
*busiest
, unsigned long max_load_move
,
840 struct sched_domain
*sd
, enum cpu_idle_type idle
,
841 int *all_pinned
, int *this_best_prio
);
843 int (*move_one_task
) (struct rq
*this_rq
, int this_cpu
,
844 struct rq
*busiest
, struct sched_domain
*sd
,
845 enum cpu_idle_type idle
);
848 void (*set_curr_task
) (struct rq
*rq
);
849 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
);
850 void (*task_new
) (struct rq
*rq
, struct task_struct
*p
);
851 void (*set_cpus_allowed
)(struct task_struct
*p
, cpumask_t
*newmask
);
855 unsigned long weight
, inv_weight
;
859 * CFS stats for a schedulable entity (task, task-group etc)
861 * Current field usage histogram:
868 struct sched_entity
{
869 struct load_weight load
; /* for load-balancing */
870 struct rb_node run_node
;
874 u64 sum_exec_runtime
;
876 u64 prev_sum_exec_runtime
;
878 #ifdef CONFIG_SCHEDSTATS
884 s64 sum_sleep_runtime
;
892 u64 nr_migrations_cold
;
893 u64 nr_failed_migrations_affine
;
894 u64 nr_failed_migrations_running
;
895 u64 nr_failed_migrations_hot
;
896 u64 nr_forced_migrations
;
897 u64 nr_forced2_migrations
;
901 u64 nr_wakeups_migrate
;
902 u64 nr_wakeups_local
;
903 u64 nr_wakeups_remote
;
904 u64 nr_wakeups_affine
;
905 u64 nr_wakeups_affine_attempts
;
906 u64 nr_wakeups_passive
;
910 #ifdef CONFIG_FAIR_GROUP_SCHED
911 struct sched_entity
*parent
;
912 /* rq on which this entity is (to be) queued: */
913 struct cfs_rq
*cfs_rq
;
914 /* rq "owned" by this entity/group: */
920 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
923 unsigned int flags
; /* per process flags, defined below */
926 int lock_depth
; /* BKL lock depth */
929 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
934 int prio
, static_prio
, normal_prio
;
935 struct list_head run_list
;
936 const struct sched_class
*sched_class
;
937 struct sched_entity se
;
939 #ifdef CONFIG_PREEMPT_NOTIFIERS
940 /* list of struct preempt_notifier: */
941 struct hlist_head preempt_notifiers
;
944 unsigned short ioprio
;
946 * fpu_counter contains the number of consecutive context switches
947 * that the FPU is used. If this is over a threshold, the lazy fpu
948 * saving becomes unlazy to save the trap. This is an unsigned char
949 * so that after 256 times the counter wraps and the behavior turns
950 * lazy again; this to deal with bursty apps that only use FPU for
953 unsigned char fpu_counter
;
954 s8 oomkilladj
; /* OOM kill score adjustment (bit shift). */
955 #ifdef CONFIG_BLK_DEV_IO_TRACE
956 unsigned int btrace_seq
;
960 cpumask_t cpus_allowed
;
962 unsigned int time_slice
;
964 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
965 struct sched_info sched_info
;
968 struct list_head tasks
;
970 * ptrace_list/ptrace_children forms the list of my children
971 * that were stolen by a ptracer.
973 struct list_head ptrace_children
;
974 struct list_head ptrace_list
;
976 struct mm_struct
*mm
, *active_mm
;
979 struct linux_binfmt
*binfmt
;
981 int exit_code
, exit_signal
;
982 int pdeath_signal
; /* The signal sent when the parent dies */
984 unsigned int personality
;
989 #ifdef CONFIG_CC_STACKPROTECTOR
990 /* Canary value for the -fstack-protector gcc feature */
991 unsigned long stack_canary
;
994 * pointers to (original) parent process, youngest child, younger sibling,
995 * older sibling, respectively. (p->father can be replaced with
998 struct task_struct
*real_parent
; /* real parent process (when being debugged) */
999 struct task_struct
*parent
; /* parent process */
1001 * children/sibling forms the list of my children plus the
1002 * tasks I'm ptracing.
1004 struct list_head children
; /* list of my children */
1005 struct list_head sibling
; /* linkage in my parent's children list */
1006 struct task_struct
*group_leader
; /* threadgroup leader */
1008 /* PID/PID hash table linkage. */
1009 struct pid_link pids
[PIDTYPE_MAX
];
1010 struct list_head thread_group
;
1012 struct completion
*vfork_done
; /* for vfork() */
1013 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1014 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1016 unsigned int rt_priority
;
1017 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1019 cputime_t prev_utime
, prev_stime
;
1020 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1021 struct timespec start_time
; /* monotonic time */
1022 struct timespec real_start_time
; /* boot based time */
1023 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1024 unsigned long min_flt
, maj_flt
;
1026 cputime_t it_prof_expires
, it_virt_expires
;
1027 unsigned long long it_sched_expires
;
1028 struct list_head cpu_timers
[3];
1030 /* process credentials */
1031 uid_t uid
,euid
,suid
,fsuid
;
1032 gid_t gid
,egid
,sgid
,fsgid
;
1033 struct group_info
*group_info
;
1034 kernel_cap_t cap_effective
, cap_inheritable
, cap_permitted
;
1035 unsigned keep_capabilities
:1;
1036 struct user_struct
*user
;
1038 struct key
*request_key_auth
; /* assumed request_key authority */
1039 struct key
*thread_keyring
; /* keyring private to this thread */
1040 unsigned char jit_keyring
; /* default keyring to attach requested keys to */
1042 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1043 - access with [gs]et_task_comm (which lock
1044 it with task_lock())
1045 - initialized normally by flush_old_exec */
1046 /* file system info */
1047 int link_count
, total_link_count
;
1048 #ifdef CONFIG_SYSVIPC
1050 struct sysv_sem sysvsem
;
1052 #ifdef CONFIG_DETECT_SOFTLOCKUP
1053 /* hung task detection */
1054 unsigned long last_switch_timestamp
;
1055 unsigned long last_switch_count
;
1057 /* CPU-specific state of this task */
1058 struct thread_struct thread
;
1059 /* filesystem information */
1060 struct fs_struct
*fs
;
1061 /* open file information */
1062 struct files_struct
*files
;
1064 struct nsproxy
*nsproxy
;
1065 /* signal handlers */
1066 struct signal_struct
*signal
;
1067 struct sighand_struct
*sighand
;
1069 sigset_t blocked
, real_blocked
;
1070 sigset_t saved_sigmask
; /* To be restored with TIF_RESTORE_SIGMASK */
1071 struct sigpending pending
;
1073 unsigned long sas_ss_sp
;
1075 int (*notifier
)(void *priv
);
1076 void *notifier_data
;
1077 sigset_t
*notifier_mask
;
1078 #ifdef CONFIG_SECURITY
1081 struct audit_context
*audit_context
;
1084 /* Thread group tracking */
1087 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1088 spinlock_t alloc_lock
;
1090 /* Protection of the PI data structures: */
1093 #ifdef CONFIG_RT_MUTEXES
1094 /* PI waiters blocked on a rt_mutex held by this task */
1095 struct plist_head pi_waiters
;
1096 /* Deadlock detection and priority inheritance handling */
1097 struct rt_mutex_waiter
*pi_blocked_on
;
1100 #ifdef CONFIG_DEBUG_MUTEXES
1101 /* mutex deadlock detection */
1102 struct mutex_waiter
*blocked_on
;
1104 #ifdef CONFIG_TRACE_IRQFLAGS
1105 unsigned int irq_events
;
1106 int hardirqs_enabled
;
1107 unsigned long hardirq_enable_ip
;
1108 unsigned int hardirq_enable_event
;
1109 unsigned long hardirq_disable_ip
;
1110 unsigned int hardirq_disable_event
;
1111 int softirqs_enabled
;
1112 unsigned long softirq_disable_ip
;
1113 unsigned int softirq_disable_event
;
1114 unsigned long softirq_enable_ip
;
1115 unsigned int softirq_enable_event
;
1116 int hardirq_context
;
1117 int softirq_context
;
1119 #ifdef CONFIG_LOCKDEP
1120 # define MAX_LOCK_DEPTH 30UL
1123 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1124 unsigned int lockdep_recursion
;
1127 /* journalling filesystem info */
1130 /* stacked block device info */
1131 struct bio
*bio_list
, **bio_tail
;
1134 struct reclaim_state
*reclaim_state
;
1136 struct backing_dev_info
*backing_dev_info
;
1138 struct io_context
*io_context
;
1140 unsigned long ptrace_message
;
1141 siginfo_t
*last_siginfo
; /* For ptrace use. */
1142 #ifdef CONFIG_TASK_XACCT
1143 /* i/o counters(bytes read/written, #syscalls */
1144 u64 rchar
, wchar
, syscr
, syscw
;
1146 struct task_io_accounting ioac
;
1147 #if defined(CONFIG_TASK_XACCT)
1148 u64 acct_rss_mem1
; /* accumulated rss usage */
1149 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1150 cputime_t acct_stimexpd
;/* stime since last update */
1153 struct mempolicy
*mempolicy
;
1156 #ifdef CONFIG_CPUSETS
1157 nodemask_t mems_allowed
;
1158 int cpuset_mems_generation
;
1159 int cpuset_mem_spread_rotor
;
1161 #ifdef CONFIG_CGROUPS
1162 /* Control Group info protected by css_set_lock */
1163 struct css_set
*cgroups
;
1164 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1165 struct list_head cg_list
;
1168 struct robust_list_head __user
*robust_list
;
1169 #ifdef CONFIG_COMPAT
1170 struct compat_robust_list_head __user
*compat_robust_list
;
1172 struct list_head pi_state_list
;
1173 struct futex_pi_state
*pi_state_cache
;
1175 atomic_t fs_excl
; /* holding fs exclusive resources */
1176 struct rcu_head rcu
;
1179 * cache last used pipe for splice
1181 struct pipe_inode_info
*splice_pipe
;
1182 #ifdef CONFIG_TASK_DELAY_ACCT
1183 struct task_delay_info
*delays
;
1185 #ifdef CONFIG_FAULT_INJECTION
1188 struct prop_local_single dirties
;
1192 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1193 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1194 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1195 * values are inverted: lower p->prio value means higher priority.
1197 * The MAX_USER_RT_PRIO value allows the actual maximum
1198 * RT priority to be separate from the value exported to
1199 * user-space. This allows kernel threads to set their
1200 * priority to a value higher than any user task. Note:
1201 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1204 #define MAX_USER_RT_PRIO 100
1205 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1207 #define MAX_PRIO (MAX_RT_PRIO + 40)
1208 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1210 static inline int rt_prio(int prio
)
1212 if (unlikely(prio
< MAX_RT_PRIO
))
1217 static inline int rt_task(struct task_struct
*p
)
1219 return rt_prio(p
->prio
);
1222 static inline void set_task_session(struct task_struct
*tsk
, pid_t session
)
1224 tsk
->signal
->__session
= session
;
1227 static inline void set_task_pgrp(struct task_struct
*tsk
, pid_t pgrp
)
1229 tsk
->signal
->__pgrp
= pgrp
;
1232 static inline struct pid
*task_pid(struct task_struct
*task
)
1234 return task
->pids
[PIDTYPE_PID
].pid
;
1237 static inline struct pid
*task_tgid(struct task_struct
*task
)
1239 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1242 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1244 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1247 static inline struct pid
*task_session(struct task_struct
*task
)
1249 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1252 struct pid_namespace
;
1255 * the helpers to get the task's different pids as they are seen
1256 * from various namespaces
1258 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1259 * task_xid_vnr() : virtual id, i.e. the id seen from the namespace the task
1260 * belongs to. this only makes sence when called in the
1261 * context of the task that belongs to the same namespace;
1262 * task_xid_nr_ns() : id seen from the ns specified;
1264 * set_task_vxid() : assigns a virtual id to a task;
1266 * see also pid_nr() etc in include/linux/pid.h
1269 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1274 pid_t
task_pid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1276 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1278 return pid_vnr(task_pid(tsk
));
1282 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1287 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1289 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1291 return pid_vnr(task_tgid(tsk
));
1295 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1297 return tsk
->signal
->__pgrp
;
1300 pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1302 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1304 return pid_vnr(task_pgrp(tsk
));
1308 static inline pid_t
task_session_nr(struct task_struct
*tsk
)
1310 return tsk
->signal
->__session
;
1313 pid_t
task_session_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1315 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1317 return pid_vnr(task_session(tsk
));
1322 * pid_alive - check that a task structure is not stale
1323 * @p: Task structure to be checked.
1325 * Test if a process is not yet dead (at most zombie state)
1326 * If pid_alive fails, then pointers within the task structure
1327 * can be stale and must not be dereferenced.
1329 static inline int pid_alive(struct task_struct
*p
)
1331 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1335 * is_global_init - check if a task structure is init
1336 * @tsk: Task structure to be checked.
1338 * Check if a task structure is the first user space task the kernel created.
1340 static inline int is_global_init(struct task_struct
*tsk
)
1342 return tsk
->pid
== 1;
1346 * is_container_init:
1347 * check whether in the task is init in its own pid namespace.
1349 extern int is_container_init(struct task_struct
*tsk
);
1351 extern struct pid
*cad_pid
;
1353 extern void free_task(struct task_struct
*tsk
);
1354 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1356 extern void __put_task_struct(struct task_struct
*t
);
1358 static inline void put_task_struct(struct task_struct
*t
)
1360 if (atomic_dec_and_test(&t
->usage
))
1361 __put_task_struct(t
);
1367 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1368 /* Not implemented yet, only for 486*/
1369 #define PF_STARTING 0x00000002 /* being created */
1370 #define PF_EXITING 0x00000004 /* getting shut down */
1371 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1372 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1373 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1374 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1375 #define PF_DUMPCORE 0x00000200 /* dumped core */
1376 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1377 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1378 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1379 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1380 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1381 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1382 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1383 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1384 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1385 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1386 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1387 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1388 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1389 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1390 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1391 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1392 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1393 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1396 * Only the _current_ task can read/write to tsk->flags, but other
1397 * tasks can access tsk->flags in readonly mode for example
1398 * with tsk_used_math (like during threaded core dumping).
1399 * There is however an exception to this rule during ptrace
1400 * or during fork: the ptracer task is allowed to write to the
1401 * child->flags of its traced child (same goes for fork, the parent
1402 * can write to the child->flags), because we're guaranteed the
1403 * child is not running and in turn not changing child->flags
1404 * at the same time the parent does it.
1406 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1407 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1408 #define clear_used_math() clear_stopped_child_used_math(current)
1409 #define set_used_math() set_stopped_child_used_math(current)
1410 #define conditional_stopped_child_used_math(condition, child) \
1411 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1412 #define conditional_used_math(condition) \
1413 conditional_stopped_child_used_math(condition, current)
1414 #define copy_to_stopped_child_used_math(child) \
1415 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1416 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1417 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1418 #define used_math() tsk_used_math(current)
1421 extern int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
);
1423 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1425 if (!cpu_isset(0, new_mask
))
1431 extern unsigned long long sched_clock(void);
1434 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1435 * clock constructed from sched_clock():
1437 extern unsigned long long cpu_clock(int cpu
);
1439 extern unsigned long long
1440 task_sched_runtime(struct task_struct
*task
);
1442 /* sched_exec is called by processes performing an exec */
1444 extern void sched_exec(void);
1446 #define sched_exec() {}
1449 extern void sched_clock_idle_sleep_event(void);
1450 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1452 #ifdef CONFIG_HOTPLUG_CPU
1453 extern void idle_task_exit(void);
1455 static inline void idle_task_exit(void) {}
1458 extern void sched_idle_next(void);
1460 #ifdef CONFIG_SCHED_DEBUG
1461 extern unsigned int sysctl_sched_latency
;
1462 extern unsigned int sysctl_sched_min_granularity
;
1463 extern unsigned int sysctl_sched_wakeup_granularity
;
1464 extern unsigned int sysctl_sched_batch_wakeup_granularity
;
1465 extern unsigned int sysctl_sched_child_runs_first
;
1466 extern unsigned int sysctl_sched_features
;
1467 extern unsigned int sysctl_sched_migration_cost
;
1468 extern unsigned int sysctl_sched_nr_migrate
;
1469 #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP)
1470 extern unsigned int sysctl_sched_min_bal_int_shares
;
1471 extern unsigned int sysctl_sched_max_bal_int_shares
;
1474 int sched_nr_latency_handler(struct ctl_table
*table
, int write
,
1475 struct file
*file
, void __user
*buffer
, size_t *length
,
1479 extern unsigned int sysctl_sched_compat_yield
;
1481 #ifdef CONFIG_RT_MUTEXES
1482 extern int rt_mutex_getprio(struct task_struct
*p
);
1483 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
1484 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
1486 static inline int rt_mutex_getprio(struct task_struct
*p
)
1488 return p
->normal_prio
;
1490 # define rt_mutex_adjust_pi(p) do { } while (0)
1493 extern void set_user_nice(struct task_struct
*p
, long nice
);
1494 extern int task_prio(const struct task_struct
*p
);
1495 extern int task_nice(const struct task_struct
*p
);
1496 extern int can_nice(const struct task_struct
*p
, const int nice
);
1497 extern int task_curr(const struct task_struct
*p
);
1498 extern int idle_cpu(int cpu
);
1499 extern int sched_setscheduler(struct task_struct
*, int, struct sched_param
*);
1500 extern struct task_struct
*idle_task(int cpu
);
1501 extern struct task_struct
*curr_task(int cpu
);
1502 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1507 * The default (Linux) execution domain.
1509 extern struct exec_domain default_exec_domain
;
1511 union thread_union
{
1512 struct thread_info thread_info
;
1513 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1516 #ifndef __HAVE_ARCH_KSTACK_END
1517 static inline int kstack_end(void *addr
)
1519 /* Reliable end of stack detection:
1520 * Some APM bios versions misalign the stack
1522 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1526 extern union thread_union init_thread_union
;
1527 extern struct task_struct init_task
;
1529 extern struct mm_struct init_mm
;
1531 extern struct pid_namespace init_pid_ns
;
1534 * find a task by one of its numerical ids
1536 * find_task_by_pid_type_ns():
1537 * it is the most generic call - it finds a task by all id,
1538 * type and namespace specified
1539 * find_task_by_pid_ns():
1540 * finds a task by its pid in the specified namespace
1541 * find_task_by_vpid():
1542 * finds a task by its virtual pid
1543 * find_task_by_pid():
1544 * finds a task by its global pid
1546 * see also find_pid() etc in include/linux/pid.h
1549 extern struct task_struct
*find_task_by_pid_type_ns(int type
, int pid
,
1550 struct pid_namespace
*ns
);
1552 extern struct task_struct
*find_task_by_pid(pid_t nr
);
1553 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1554 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
1555 struct pid_namespace
*ns
);
1557 extern void __set_special_pids(pid_t session
, pid_t pgrp
);
1559 /* per-UID process charging. */
1560 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
1561 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1563 atomic_inc(&u
->__count
);
1566 extern void free_uid(struct user_struct
*);
1567 extern void switch_uid(struct user_struct
*);
1568 extern void release_uids(struct user_namespace
*ns
);
1570 #include <asm/current.h>
1572 extern void do_timer(unsigned long ticks
);
1574 extern int FASTCALL(wake_up_state(struct task_struct
* tsk
, unsigned int state
));
1575 extern int FASTCALL(wake_up_process(struct task_struct
* tsk
));
1576 extern void FASTCALL(wake_up_new_task(struct task_struct
* tsk
,
1577 unsigned long clone_flags
));
1579 extern void kick_process(struct task_struct
*tsk
);
1581 static inline void kick_process(struct task_struct
*tsk
) { }
1583 extern void sched_fork(struct task_struct
*p
, int clone_flags
);
1584 extern void sched_dead(struct task_struct
*p
);
1586 extern int in_group_p(gid_t
);
1587 extern int in_egroup_p(gid_t
);
1589 extern void proc_caches_init(void);
1590 extern void flush_signals(struct task_struct
*);
1591 extern void ignore_signals(struct task_struct
*);
1592 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
1593 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
1595 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
1597 unsigned long flags
;
1600 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
1601 ret
= dequeue_signal(tsk
, mask
, info
);
1602 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
1607 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
1609 extern void unblock_all_signals(void);
1610 extern void release_task(struct task_struct
* p
);
1611 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
1612 extern int send_group_sig_info(int, struct siginfo
*, struct task_struct
*);
1613 extern int force_sigsegv(int, struct task_struct
*);
1614 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
1615 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1616 extern int kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1617 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
1618 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
1619 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
1620 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
1621 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
1622 extern void do_notify_parent(struct task_struct
*, int);
1623 extern void force_sig(int, struct task_struct
*);
1624 extern void force_sig_specific(int, struct task_struct
*);
1625 extern int send_sig(int, struct task_struct
*, int);
1626 extern void zap_other_threads(struct task_struct
*p
);
1627 extern int kill_proc(pid_t
, int, int);
1628 extern struct sigqueue
*sigqueue_alloc(void);
1629 extern void sigqueue_free(struct sigqueue
*);
1630 extern int send_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1631 extern int send_group_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1632 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
1633 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
1635 static inline int kill_cad_pid(int sig
, int priv
)
1637 return kill_pid(cad_pid
, sig
, priv
);
1640 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1641 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1642 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1643 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1645 static inline int is_si_special(const struct siginfo
*info
)
1647 return info
<= SEND_SIG_FORCED
;
1650 /* True if we are on the alternate signal stack. */
1652 static inline int on_sig_stack(unsigned long sp
)
1654 return (sp
- current
->sas_ss_sp
< current
->sas_ss_size
);
1657 static inline int sas_ss_flags(unsigned long sp
)
1659 return (current
->sas_ss_size
== 0 ? SS_DISABLE
1660 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
1664 * Routines for handling mm_structs
1666 extern struct mm_struct
* mm_alloc(void);
1668 /* mmdrop drops the mm and the page tables */
1669 extern void FASTCALL(__mmdrop(struct mm_struct
*));
1670 static inline void mmdrop(struct mm_struct
* mm
)
1672 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
1676 /* mmput gets rid of the mappings and all user-space */
1677 extern void mmput(struct mm_struct
*);
1678 /* Grab a reference to a task's mm, if it is not already going away */
1679 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
1680 /* Remove the current tasks stale references to the old mm_struct */
1681 extern void mm_release(struct task_struct
*, struct mm_struct
*);
1683 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct
*, struct pt_regs
*);
1684 extern void flush_thread(void);
1685 extern void exit_thread(void);
1687 extern void exit_files(struct task_struct
*);
1688 extern void __cleanup_signal(struct signal_struct
*);
1689 extern void __cleanup_sighand(struct sighand_struct
*);
1690 extern void exit_itimers(struct signal_struct
*);
1692 extern NORET_TYPE
void do_group_exit(int);
1694 extern void daemonize(const char *, ...);
1695 extern int allow_signal(int);
1696 extern int disallow_signal(int);
1698 extern int do_execve(char *, char __user
* __user
*, char __user
* __user
*, struct pt_regs
*);
1699 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
1700 struct task_struct
*fork_idle(int);
1702 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
1703 extern void get_task_comm(char *to
, struct task_struct
*tsk
);
1706 extern void wait_task_inactive(struct task_struct
* p
);
1708 #define wait_task_inactive(p) do { } while (0)
1711 #define remove_parent(p) list_del_init(&(p)->sibling)
1712 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1714 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1716 #define for_each_process(p) \
1717 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1720 * Careful: do_each_thread/while_each_thread is a double loop so
1721 * 'break' will not work as expected - use goto instead.
1723 #define do_each_thread(g, t) \
1724 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1726 #define while_each_thread(g, t) \
1727 while ((t = next_thread(t)) != g)
1729 /* de_thread depends on thread_group_leader not being a pid based check */
1730 #define thread_group_leader(p) (p == p->group_leader)
1732 /* Do to the insanities of de_thread it is possible for a process
1733 * to have the pid of the thread group leader without actually being
1734 * the thread group leader. For iteration through the pids in proc
1735 * all we care about is that we have a task with the appropriate
1736 * pid, we don't actually care if we have the right task.
1738 static inline int has_group_leader_pid(struct task_struct
*p
)
1740 return p
->pid
== p
->tgid
;
1744 int same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
1746 return p1
->tgid
== p2
->tgid
;
1749 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
1751 return list_entry(rcu_dereference(p
->thread_group
.next
),
1752 struct task_struct
, thread_group
);
1755 static inline int thread_group_empty(struct task_struct
*p
)
1757 return list_empty(&p
->thread_group
);
1760 #define delay_group_leader(p) \
1761 (thread_group_leader(p) && !thread_group_empty(p))
1764 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1765 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1766 * pins the final release of task.io_context. Also protects ->cpuset and
1767 * ->cgroup.subsys[].
1769 * Nests both inside and outside of read_lock(&tasklist_lock).
1770 * It must not be nested with write_lock_irq(&tasklist_lock),
1771 * neither inside nor outside.
1773 static inline void task_lock(struct task_struct
*p
)
1775 spin_lock(&p
->alloc_lock
);
1778 static inline void task_unlock(struct task_struct
*p
)
1780 spin_unlock(&p
->alloc_lock
);
1783 extern struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
1784 unsigned long *flags
);
1786 static inline void unlock_task_sighand(struct task_struct
*tsk
,
1787 unsigned long *flags
)
1789 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
1792 #ifndef __HAVE_THREAD_FUNCTIONS
1794 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1795 #define task_stack_page(task) ((task)->stack)
1797 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
1799 *task_thread_info(p
) = *task_thread_info(org
);
1800 task_thread_info(p
)->task
= p
;
1803 static inline unsigned long *end_of_stack(struct task_struct
*p
)
1805 return (unsigned long *)(task_thread_info(p
) + 1);
1810 /* set thread flags in other task's structures
1811 * - see asm/thread_info.h for TIF_xxxx flags available
1813 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1815 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1818 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1820 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1823 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1825 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1828 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1830 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1833 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1835 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1838 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1840 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1843 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1845 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1848 static inline int signal_pending(struct task_struct
*p
)
1850 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
1853 static inline int need_resched(void)
1855 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
1859 * cond_resched() and cond_resched_lock(): latency reduction via
1860 * explicit rescheduling in places that are safe. The return
1861 * value indicates whether a reschedule was done in fact.
1862 * cond_resched_lock() will drop the spinlock before scheduling,
1863 * cond_resched_softirq() will enable bhs before scheduling.
1865 extern int cond_resched(void);
1866 extern int cond_resched_lock(spinlock_t
* lock
);
1867 extern int cond_resched_softirq(void);
1870 * Does a critical section need to be broken due to another
1873 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1874 # define need_lockbreak(lock) ((lock)->break_lock)
1876 # define need_lockbreak(lock) 0
1880 * Does a critical section need to be broken due to another
1881 * task waiting or preemption being signalled:
1883 static inline int lock_need_resched(spinlock_t
*lock
)
1885 if (need_lockbreak(lock
) || need_resched())
1891 * Reevaluate whether the task has signals pending delivery.
1892 * Wake the task if so.
1893 * This is required every time the blocked sigset_t changes.
1894 * callers must hold sighand->siglock.
1896 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
1897 extern void recalc_sigpending(void);
1899 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
1902 * Wrappers for p->thread_info->cpu access. No-op on UP.
1906 static inline unsigned int task_cpu(const struct task_struct
*p
)
1908 return task_thread_info(p
)->cpu
;
1911 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
1915 static inline unsigned int task_cpu(const struct task_struct
*p
)
1920 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
1924 #endif /* CONFIG_SMP */
1926 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1927 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
1929 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
)
1931 mm
->mmap_base
= TASK_UNMAPPED_BASE
;
1932 mm
->get_unmapped_area
= arch_get_unmapped_area
;
1933 mm
->unmap_area
= arch_unmap_area
;
1937 extern long sched_setaffinity(pid_t pid
, cpumask_t new_mask
);
1938 extern long sched_getaffinity(pid_t pid
, cpumask_t
*mask
);
1940 extern int sched_mc_power_savings
, sched_smt_power_savings
;
1942 extern void normalize_rt_tasks(void);
1944 #ifdef CONFIG_FAIR_GROUP_SCHED
1946 extern struct task_group init_task_group
;
1948 extern struct task_group
*sched_create_group(void);
1949 extern void sched_destroy_group(struct task_group
*tg
);
1950 extern void sched_move_task(struct task_struct
*tsk
);
1951 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
1952 extern unsigned long sched_group_shares(struct task_group
*tg
);
1956 #ifdef CONFIG_TASK_XACCT
1957 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1962 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1967 static inline void inc_syscr(struct task_struct
*tsk
)
1972 static inline void inc_syscw(struct task_struct
*tsk
)
1977 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1981 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1985 static inline void inc_syscr(struct task_struct
*tsk
)
1989 static inline void inc_syscw(struct task_struct
*tsk
)
1995 void migration_init(void);
1997 static inline void migration_init(void)
2002 #endif /* __KERNEL__ */