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 update_process_times(int user
);
258 extern void scheduler_tick(void);
260 #ifdef CONFIG_DETECT_SOFTLOCKUP
261 extern void softlockup_tick(void);
262 extern void spawn_softlockup_task(void);
263 extern void touch_softlockup_watchdog(void);
264 extern void touch_all_softlockup_watchdogs(void);
265 extern int softlockup_thresh
;
267 static inline void softlockup_tick(void)
270 static inline void spawn_softlockup_task(void)
273 static inline void touch_softlockup_watchdog(void)
276 static inline void touch_all_softlockup_watchdogs(void)
282 /* Attach to any functions which should be ignored in wchan output. */
283 #define __sched __attribute__((__section__(".sched.text")))
284 /* Is this address in the __sched functions? */
285 extern int in_sched_functions(unsigned long addr
);
287 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
288 extern signed long FASTCALL(schedule_timeout(signed long timeout
));
289 extern signed long schedule_timeout_interruptible(signed long timeout
);
290 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
291 asmlinkage
void schedule(void);
294 struct user_namespace
;
296 /* Maximum number of active map areas.. This is a random (large) number */
297 #define DEFAULT_MAX_MAP_COUNT 65536
299 extern int sysctl_max_map_count
;
301 #include <linux/aio.h>
304 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
305 unsigned long, unsigned long);
307 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
308 unsigned long len
, unsigned long pgoff
,
309 unsigned long flags
);
310 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
311 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
313 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
315 * The mm counters are not protected by its page_table_lock,
316 * so must be incremented atomically.
318 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
319 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
320 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
321 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
322 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
324 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
326 * The mm counters are protected by its page_table_lock,
327 * so can be incremented directly.
329 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
330 #define get_mm_counter(mm, member) ((mm)->_##member)
331 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
332 #define inc_mm_counter(mm, member) (mm)->_##member++
333 #define dec_mm_counter(mm, member) (mm)->_##member--
335 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
337 #define get_mm_rss(mm) \
338 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
339 #define update_hiwater_rss(mm) do { \
340 unsigned long _rss = get_mm_rss(mm); \
341 if ((mm)->hiwater_rss < _rss) \
342 (mm)->hiwater_rss = _rss; \
344 #define update_hiwater_vm(mm) do { \
345 if ((mm)->hiwater_vm < (mm)->total_vm) \
346 (mm)->hiwater_vm = (mm)->total_vm; \
349 extern void set_dumpable(struct mm_struct
*mm
, int value
);
350 extern int get_dumpable(struct mm_struct
*mm
);
354 #define MMF_DUMPABLE 0 /* core dump is permitted */
355 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
356 #define MMF_DUMPABLE_BITS 2
358 /* coredump filter bits */
359 #define MMF_DUMP_ANON_PRIVATE 2
360 #define MMF_DUMP_ANON_SHARED 3
361 #define MMF_DUMP_MAPPED_PRIVATE 4
362 #define MMF_DUMP_MAPPED_SHARED 5
363 #define MMF_DUMP_ELF_HEADERS 6
364 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
365 #define MMF_DUMP_FILTER_BITS 5
366 #define MMF_DUMP_FILTER_MASK \
367 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
368 #define MMF_DUMP_FILTER_DEFAULT \
369 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
371 struct sighand_struct
{
373 struct k_sigaction action
[_NSIG
];
375 wait_queue_head_t signalfd_wqh
;
378 struct pacct_struct
{
381 unsigned long ac_mem
;
382 cputime_t ac_utime
, ac_stime
;
383 unsigned long ac_minflt
, ac_majflt
;
387 * NOTE! "signal_struct" does not have it's own
388 * locking, because a shared signal_struct always
389 * implies a shared sighand_struct, so locking
390 * sighand_struct is always a proper superset of
391 * the locking of signal_struct.
393 struct signal_struct
{
397 wait_queue_head_t wait_chldexit
; /* for wait4() */
399 /* current thread group signal load-balancing target: */
400 struct task_struct
*curr_target
;
402 /* shared signal handling: */
403 struct sigpending shared_pending
;
405 /* thread group exit support */
408 * - notify group_exit_task when ->count is equal to notify_count
409 * - everyone except group_exit_task is stopped during signal delivery
410 * of fatal signals, group_exit_task processes the signal.
412 struct task_struct
*group_exit_task
;
415 /* thread group stop support, overloads group_exit_code too */
416 int group_stop_count
;
417 unsigned int flags
; /* see SIGNAL_* flags below */
419 /* POSIX.1b Interval Timers */
420 struct list_head posix_timers
;
422 /* ITIMER_REAL timer for the process */
423 struct hrtimer real_timer
;
424 struct task_struct
*tsk
;
425 ktime_t it_real_incr
;
427 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
428 cputime_t it_prof_expires
, it_virt_expires
;
429 cputime_t it_prof_incr
, it_virt_incr
;
431 /* job control IDs */
434 * pgrp and session fields are deprecated.
435 * use the task_session_Xnr and task_pgrp_Xnr routines below
439 pid_t pgrp __deprecated
;
443 struct pid
*tty_old_pgrp
;
446 pid_t session __deprecated
;
450 /* boolean value for session group leader */
453 struct tty_struct
*tty
; /* NULL if no tty */
456 * Cumulative resource counters for dead threads in the group,
457 * and for reaped dead child processes forked by this group.
458 * Live threads maintain their own counters and add to these
459 * in __exit_signal, except for the group leader.
461 cputime_t utime
, stime
, cutime
, cstime
;
464 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
465 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
466 unsigned long inblock
, oublock
, cinblock
, coublock
;
469 * Cumulative ns of scheduled CPU time for dead threads in the
470 * group, not including a zombie group leader. (This only differs
471 * from jiffies_to_ns(utime + stime) if sched_clock uses something
472 * other than jiffies.)
474 unsigned long long sum_sched_runtime
;
477 * We don't bother to synchronize most readers of this at all,
478 * because there is no reader checking a limit that actually needs
479 * to get both rlim_cur and rlim_max atomically, and either one
480 * alone is a single word that can safely be read normally.
481 * getrlimit/setrlimit use task_lock(current->group_leader) to
482 * protect this instead of the siglock, because they really
483 * have no need to disable irqs.
485 struct rlimit rlim
[RLIM_NLIMITS
];
487 struct list_head cpu_timers
[3];
489 /* keep the process-shared keyrings here so that they do the right
490 * thing in threads created with CLONE_THREAD */
492 struct key
*session_keyring
; /* keyring inherited over fork */
493 struct key
*process_keyring
; /* keyring private to this process */
495 #ifdef CONFIG_BSD_PROCESS_ACCT
496 struct pacct_struct pacct
; /* per-process accounting information */
498 #ifdef CONFIG_TASKSTATS
499 struct taskstats
*stats
;
503 struct tty_audit_buf
*tty_audit_buf
;
507 /* Context switch must be unlocked if interrupts are to be enabled */
508 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
509 # define __ARCH_WANT_UNLOCKED_CTXSW
513 * Bits in flags field of signal_struct.
515 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
516 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
517 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
518 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
521 * Some day this will be a full-fledged user tracking system..
524 atomic_t __count
; /* reference count */
525 atomic_t processes
; /* How many processes does this user have? */
526 atomic_t files
; /* How many open files does this user have? */
527 atomic_t sigpending
; /* How many pending signals does this user have? */
528 #ifdef CONFIG_INOTIFY_USER
529 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
530 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
532 #ifdef CONFIG_POSIX_MQUEUE
533 /* protected by mq_lock */
534 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
536 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
539 struct key
*uid_keyring
; /* UID specific keyring */
540 struct key
*session_keyring
; /* UID's default session keyring */
543 /* Hash table maintenance information */
544 struct hlist_node uidhash_node
;
547 #ifdef CONFIG_FAIR_USER_SCHED
548 struct task_group
*tg
;
551 struct subsys_attribute user_attr
;
552 struct work_struct work
;
557 #ifdef CONFIG_FAIR_USER_SCHED
558 extern int uids_kobject_init(void);
560 static inline int uids_kobject_init(void) { return 0; }
563 extern struct user_struct
*find_user(uid_t
);
565 extern struct user_struct root_user
;
566 #define INIT_USER (&root_user)
568 struct backing_dev_info
;
569 struct reclaim_state
;
571 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
573 /* cumulative counters */
574 unsigned long pcount
; /* # of times run on this cpu */
575 unsigned long long cpu_time
, /* time spent on the cpu */
576 run_delay
; /* time spent waiting on a runqueue */
579 unsigned long long last_arrival
,/* when we last ran on a cpu */
580 last_queued
; /* when we were last queued to run */
581 #ifdef CONFIG_SCHEDSTATS
583 unsigned int bkl_count
;
586 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
588 #ifdef CONFIG_SCHEDSTATS
589 extern const struct file_operations proc_schedstat_operations
;
590 #endif /* CONFIG_SCHEDSTATS */
592 #ifdef CONFIG_TASK_DELAY_ACCT
593 struct task_delay_info
{
595 unsigned int flags
; /* Private per-task flags */
597 /* For each stat XXX, add following, aligned appropriately
599 * struct timespec XXX_start, XXX_end;
603 * Atomicity of updates to XXX_delay, XXX_count protected by
604 * single lock above (split into XXX_lock if contention is an issue).
608 * XXX_count is incremented on every XXX operation, the delay
609 * associated with the operation is added to XXX_delay.
610 * XXX_delay contains the accumulated delay time in nanoseconds.
612 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
613 u64 blkio_delay
; /* wait for sync block io completion */
614 u64 swapin_delay
; /* wait for swapin block io completion */
615 u32 blkio_count
; /* total count of the number of sync block */
616 /* io operations performed */
617 u32 swapin_count
; /* total count of the number of swapin block */
618 /* io operations performed */
620 #endif /* CONFIG_TASK_DELAY_ACCT */
622 static inline int sched_info_on(void)
624 #ifdef CONFIG_SCHEDSTATS
626 #elif defined(CONFIG_TASK_DELAY_ACCT)
627 extern int delayacct_on
;
642 * sched-domains (multiprocessor balancing) declarations:
646 * Increase resolution of nice-level calculations:
648 #define SCHED_LOAD_SHIFT 10
649 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
651 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
654 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
655 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
656 #define SD_BALANCE_EXEC 4 /* Balance on exec */
657 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
658 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
659 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
660 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
661 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
662 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
663 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
664 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
666 #define BALANCE_FOR_MC_POWER \
667 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
669 #define BALANCE_FOR_PKG_POWER \
670 ((sched_mc_power_savings || sched_smt_power_savings) ? \
671 SD_POWERSAVINGS_BALANCE : 0)
673 #define test_sd_parent(sd, flag) ((sd->parent && \
674 (sd->parent->flags & flag)) ? 1 : 0)
678 struct sched_group
*next
; /* Must be a circular list */
682 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
683 * single CPU. This is read only (except for setup, hotplug CPU).
684 * Note : Never change cpu_power without recompute its reciprocal
686 unsigned int __cpu_power
;
688 * reciprocal value of cpu_power to avoid expensive divides
689 * (see include/linux/reciprocal_div.h)
691 u32 reciprocal_cpu_power
;
694 struct sched_domain
{
695 /* These fields must be setup */
696 struct sched_domain
*parent
; /* top domain must be null terminated */
697 struct sched_domain
*child
; /* bottom domain must be null terminated */
698 struct sched_group
*groups
; /* the balancing groups of the domain */
699 cpumask_t span
; /* span of all CPUs in this domain */
700 unsigned long min_interval
; /* Minimum balance interval ms */
701 unsigned long max_interval
; /* Maximum balance interval ms */
702 unsigned int busy_factor
; /* less balancing by factor if busy */
703 unsigned int imbalance_pct
; /* No balance until over watermark */
704 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
705 unsigned int busy_idx
;
706 unsigned int idle_idx
;
707 unsigned int newidle_idx
;
708 unsigned int wake_idx
;
709 unsigned int forkexec_idx
;
710 int flags
; /* See SD_* */
712 /* Runtime fields. */
713 unsigned long last_balance
; /* init to jiffies. units in jiffies */
714 unsigned int balance_interval
; /* initialise to 1. units in ms. */
715 unsigned int nr_balance_failed
; /* initialise to 0 */
717 #ifdef CONFIG_SCHEDSTATS
718 /* load_balance() stats */
719 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
720 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
721 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
722 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
723 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
724 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
725 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
726 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
728 /* Active load balancing */
729 unsigned int alb_count
;
730 unsigned int alb_failed
;
731 unsigned int alb_pushed
;
733 /* SD_BALANCE_EXEC stats */
734 unsigned int sbe_count
;
735 unsigned int sbe_balanced
;
736 unsigned int sbe_pushed
;
738 /* SD_BALANCE_FORK stats */
739 unsigned int sbf_count
;
740 unsigned int sbf_balanced
;
741 unsigned int sbf_pushed
;
743 /* try_to_wake_up() stats */
744 unsigned int ttwu_wake_remote
;
745 unsigned int ttwu_move_affine
;
746 unsigned int ttwu_move_balance
;
750 extern void partition_sched_domains(int ndoms_new
, cpumask_t
*doms_new
);
752 #endif /* CONFIG_SMP */
755 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
756 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
757 * task of nice 0 or enough lower priority tasks to bring up the
760 static inline int above_background_load(void)
764 for_each_online_cpu(cpu
) {
765 if (weighted_cpuload(cpu
) >= SCHED_LOAD_SCALE
)
771 struct io_context
; /* See blkdev.h */
772 #define NGROUPS_SMALL 32
773 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
777 gid_t small_block
[NGROUPS_SMALL
];
783 * get_group_info() must be called with the owning task locked (via task_lock())
784 * when task != current. The reason being that the vast majority of callers are
785 * looking at current->group_info, which can not be changed except by the
786 * current task. Changing current->group_info requires the task lock, too.
788 #define get_group_info(group_info) do { \
789 atomic_inc(&(group_info)->usage); \
792 #define put_group_info(group_info) do { \
793 if (atomic_dec_and_test(&(group_info)->usage)) \
794 groups_free(group_info); \
797 extern struct group_info
*groups_alloc(int gidsetsize
);
798 extern void groups_free(struct group_info
*group_info
);
799 extern int set_current_groups(struct group_info
*group_info
);
800 extern int groups_search(struct group_info
*group_info
, gid_t grp
);
801 /* access the groups "array" with this macro */
802 #define GROUP_AT(gi, i) \
803 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
805 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
806 extern void prefetch_stack(struct task_struct
*t
);
808 static inline void prefetch_stack(struct task_struct
*t
) { }
811 struct audit_context
; /* See audit.c */
813 struct pipe_inode_info
;
814 struct uts_namespace
;
820 const struct sched_class
*next
;
822 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int wakeup
);
823 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int sleep
);
824 void (*yield_task
) (struct rq
*rq
);
826 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
);
828 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
829 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
832 unsigned long (*load_balance
) (struct rq
*this_rq
, int this_cpu
,
833 struct rq
*busiest
, unsigned long max_load_move
,
834 struct sched_domain
*sd
, enum cpu_idle_type idle
,
835 int *all_pinned
, int *this_best_prio
);
837 int (*move_one_task
) (struct rq
*this_rq
, int this_cpu
,
838 struct rq
*busiest
, struct sched_domain
*sd
,
839 enum cpu_idle_type idle
);
842 void (*set_curr_task
) (struct rq
*rq
);
843 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
);
844 void (*task_new
) (struct rq
*rq
, struct task_struct
*p
);
848 unsigned long weight
, inv_weight
;
852 * CFS stats for a schedulable entity (task, task-group etc)
854 * Current field usage histogram:
861 struct sched_entity
{
862 struct load_weight load
; /* for load-balancing */
863 struct rb_node run_node
;
868 u64 sum_exec_runtime
;
870 u64 prev_sum_exec_runtime
;
872 #ifdef CONFIG_SCHEDSTATS
878 s64 sum_sleep_runtime
;
886 u64 nr_migrations_cold
;
887 u64 nr_failed_migrations_affine
;
888 u64 nr_failed_migrations_running
;
889 u64 nr_failed_migrations_hot
;
890 u64 nr_forced_migrations
;
891 u64 nr_forced2_migrations
;
895 u64 nr_wakeups_migrate
;
896 u64 nr_wakeups_local
;
897 u64 nr_wakeups_remote
;
898 u64 nr_wakeups_affine
;
899 u64 nr_wakeups_affine_attempts
;
900 u64 nr_wakeups_passive
;
904 #ifdef CONFIG_FAIR_GROUP_SCHED
905 struct sched_entity
*parent
;
906 /* rq on which this entity is (to be) queued: */
907 struct cfs_rq
*cfs_rq
;
908 /* rq "owned" by this entity/group: */
914 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
917 unsigned int flags
; /* per process flags, defined below */
920 int lock_depth
; /* BKL lock depth */
923 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
928 int prio
, static_prio
, normal_prio
;
929 struct list_head run_list
;
930 const struct sched_class
*sched_class
;
931 struct sched_entity se
;
933 #ifdef CONFIG_PREEMPT_NOTIFIERS
934 /* list of struct preempt_notifier: */
935 struct hlist_head preempt_notifiers
;
938 unsigned short ioprio
;
940 * fpu_counter contains the number of consecutive context switches
941 * that the FPU is used. If this is over a threshold, the lazy fpu
942 * saving becomes unlazy to save the trap. This is an unsigned char
943 * so that after 256 times the counter wraps and the behavior turns
944 * lazy again; this to deal with bursty apps that only use FPU for
947 unsigned char fpu_counter
;
948 s8 oomkilladj
; /* OOM kill score adjustment (bit shift). */
949 #ifdef CONFIG_BLK_DEV_IO_TRACE
950 unsigned int btrace_seq
;
954 cpumask_t cpus_allowed
;
955 unsigned int time_slice
;
957 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
958 struct sched_info sched_info
;
961 struct list_head tasks
;
963 * ptrace_list/ptrace_children forms the list of my children
964 * that were stolen by a ptracer.
966 struct list_head ptrace_children
;
967 struct list_head ptrace_list
;
969 struct mm_struct
*mm
, *active_mm
;
972 struct linux_binfmt
*binfmt
;
974 int exit_code
, exit_signal
;
975 int pdeath_signal
; /* The signal sent when the parent dies */
977 unsigned int personality
;
982 #ifdef CONFIG_CC_STACKPROTECTOR
983 /* Canary value for the -fstack-protector gcc feature */
984 unsigned long stack_canary
;
987 * pointers to (original) parent process, youngest child, younger sibling,
988 * older sibling, respectively. (p->father can be replaced with
991 struct task_struct
*real_parent
; /* real parent process (when being debugged) */
992 struct task_struct
*parent
; /* parent process */
994 * children/sibling forms the list of my children plus the
995 * tasks I'm ptracing.
997 struct list_head children
; /* list of my children */
998 struct list_head sibling
; /* linkage in my parent's children list */
999 struct task_struct
*group_leader
; /* threadgroup leader */
1001 /* PID/PID hash table linkage. */
1002 struct pid_link pids
[PIDTYPE_MAX
];
1003 struct list_head thread_group
;
1005 struct completion
*vfork_done
; /* for vfork() */
1006 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1007 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1009 unsigned int rt_priority
;
1010 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1012 cputime_t prev_utime
, prev_stime
;
1013 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1014 struct timespec start_time
; /* monotonic time */
1015 struct timespec real_start_time
; /* boot based time */
1016 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1017 unsigned long min_flt
, maj_flt
;
1019 cputime_t it_prof_expires
, it_virt_expires
;
1020 unsigned long long it_sched_expires
;
1021 struct list_head cpu_timers
[3];
1023 /* process credentials */
1024 uid_t uid
,euid
,suid
,fsuid
;
1025 gid_t gid
,egid
,sgid
,fsgid
;
1026 struct group_info
*group_info
;
1027 kernel_cap_t cap_effective
, cap_inheritable
, cap_permitted
;
1028 unsigned keep_capabilities
:1;
1029 struct user_struct
*user
;
1031 struct key
*request_key_auth
; /* assumed request_key authority */
1032 struct key
*thread_keyring
; /* keyring private to this thread */
1033 unsigned char jit_keyring
; /* default keyring to attach requested keys to */
1035 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1036 - access with [gs]et_task_comm (which lock
1037 it with task_lock())
1038 - initialized normally by flush_old_exec */
1039 /* file system info */
1040 int link_count
, total_link_count
;
1041 #ifdef CONFIG_SYSVIPC
1043 struct sysv_sem sysvsem
;
1045 /* CPU-specific state of this task */
1046 struct thread_struct thread
;
1047 /* filesystem information */
1048 struct fs_struct
*fs
;
1049 /* open file information */
1050 struct files_struct
*files
;
1052 struct nsproxy
*nsproxy
;
1053 /* signal handlers */
1054 struct signal_struct
*signal
;
1055 struct sighand_struct
*sighand
;
1057 sigset_t blocked
, real_blocked
;
1058 sigset_t saved_sigmask
; /* To be restored with TIF_RESTORE_SIGMASK */
1059 struct sigpending pending
;
1061 unsigned long sas_ss_sp
;
1063 int (*notifier
)(void *priv
);
1064 void *notifier_data
;
1065 sigset_t
*notifier_mask
;
1066 #ifdef CONFIG_SECURITY
1069 struct audit_context
*audit_context
;
1072 /* Thread group tracking */
1075 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1076 spinlock_t alloc_lock
;
1078 /* Protection of the PI data structures: */
1081 #ifdef CONFIG_RT_MUTEXES
1082 /* PI waiters blocked on a rt_mutex held by this task */
1083 struct plist_head pi_waiters
;
1084 /* Deadlock detection and priority inheritance handling */
1085 struct rt_mutex_waiter
*pi_blocked_on
;
1088 #ifdef CONFIG_DEBUG_MUTEXES
1089 /* mutex deadlock detection */
1090 struct mutex_waiter
*blocked_on
;
1092 #ifdef CONFIG_TRACE_IRQFLAGS
1093 unsigned int irq_events
;
1094 int hardirqs_enabled
;
1095 unsigned long hardirq_enable_ip
;
1096 unsigned int hardirq_enable_event
;
1097 unsigned long hardirq_disable_ip
;
1098 unsigned int hardirq_disable_event
;
1099 int softirqs_enabled
;
1100 unsigned long softirq_disable_ip
;
1101 unsigned int softirq_disable_event
;
1102 unsigned long softirq_enable_ip
;
1103 unsigned int softirq_enable_event
;
1104 int hardirq_context
;
1105 int softirq_context
;
1107 #ifdef CONFIG_LOCKDEP
1108 # define MAX_LOCK_DEPTH 30UL
1111 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1112 unsigned int lockdep_recursion
;
1115 /* journalling filesystem info */
1118 /* stacked block device info */
1119 struct bio
*bio_list
, **bio_tail
;
1122 struct reclaim_state
*reclaim_state
;
1124 struct backing_dev_info
*backing_dev_info
;
1126 struct io_context
*io_context
;
1128 unsigned long ptrace_message
;
1129 siginfo_t
*last_siginfo
; /* For ptrace use. */
1130 #ifdef CONFIG_TASK_XACCT
1131 /* i/o counters(bytes read/written, #syscalls */
1132 u64 rchar
, wchar
, syscr
, syscw
;
1134 struct task_io_accounting ioac
;
1135 #if defined(CONFIG_TASK_XACCT)
1136 u64 acct_rss_mem1
; /* accumulated rss usage */
1137 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1138 cputime_t acct_stimexpd
;/* stime since last update */
1141 struct mempolicy
*mempolicy
;
1144 #ifdef CONFIG_CPUSETS
1145 nodemask_t mems_allowed
;
1146 int cpuset_mems_generation
;
1147 int cpuset_mem_spread_rotor
;
1149 #ifdef CONFIG_CGROUPS
1150 /* Control Group info protected by css_set_lock */
1151 struct css_set
*cgroups
;
1152 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1153 struct list_head cg_list
;
1156 struct robust_list_head __user
*robust_list
;
1157 #ifdef CONFIG_COMPAT
1158 struct compat_robust_list_head __user
*compat_robust_list
;
1160 struct list_head pi_state_list
;
1161 struct futex_pi_state
*pi_state_cache
;
1163 atomic_t fs_excl
; /* holding fs exclusive resources */
1164 struct rcu_head rcu
;
1167 * cache last used pipe for splice
1169 struct pipe_inode_info
*splice_pipe
;
1170 #ifdef CONFIG_TASK_DELAY_ACCT
1171 struct task_delay_info
*delays
;
1173 #ifdef CONFIG_FAULT_INJECTION
1176 struct prop_local_single dirties
;
1180 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1181 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1182 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1183 * values are inverted: lower p->prio value means higher priority.
1185 * The MAX_USER_RT_PRIO value allows the actual maximum
1186 * RT priority to be separate from the value exported to
1187 * user-space. This allows kernel threads to set their
1188 * priority to a value higher than any user task. Note:
1189 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1192 #define MAX_USER_RT_PRIO 100
1193 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1195 #define MAX_PRIO (MAX_RT_PRIO + 40)
1196 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1198 static inline int rt_prio(int prio
)
1200 if (unlikely(prio
< MAX_RT_PRIO
))
1205 static inline int rt_task(struct task_struct
*p
)
1207 return rt_prio(p
->prio
);
1210 static inline void set_task_session(struct task_struct
*tsk
, pid_t session
)
1212 tsk
->signal
->__session
= session
;
1215 static inline void set_task_pgrp(struct task_struct
*tsk
, pid_t pgrp
)
1217 tsk
->signal
->__pgrp
= pgrp
;
1220 static inline struct pid
*task_pid(struct task_struct
*task
)
1222 return task
->pids
[PIDTYPE_PID
].pid
;
1225 static inline struct pid
*task_tgid(struct task_struct
*task
)
1227 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1230 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1232 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1235 static inline struct pid
*task_session(struct task_struct
*task
)
1237 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1240 struct pid_namespace
;
1243 * the helpers to get the task's different pids as they are seen
1244 * from various namespaces
1246 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1247 * task_xid_vnr() : virtual id, i.e. the id seen from the namespace the task
1248 * belongs to. this only makes sence when called in the
1249 * context of the task that belongs to the same namespace;
1250 * task_xid_nr_ns() : id seen from the ns specified;
1252 * set_task_vxid() : assigns a virtual id to a task;
1254 * task_ppid_nr_ns() : the parent's id as seen from the namespace specified.
1255 * the result depends on the namespace and whether the
1256 * task in question is the namespace's init. e.g. for the
1257 * namespace's init this will return 0 when called from
1258 * the namespace of this init, or appropriate id otherwise.
1261 * see also pid_nr() etc in include/linux/pid.h
1264 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1269 pid_t
task_pid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1271 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1273 return pid_vnr(task_pid(tsk
));
1277 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1282 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1284 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1286 return pid_vnr(task_tgid(tsk
));
1290 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1292 return tsk
->signal
->__pgrp
;
1295 pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1297 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1299 return pid_vnr(task_pgrp(tsk
));
1303 static inline pid_t
task_session_nr(struct task_struct
*tsk
)
1305 return tsk
->signal
->__session
;
1308 pid_t
task_session_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1310 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1312 return pid_vnr(task_session(tsk
));
1316 static inline pid_t
task_ppid_nr_ns(struct task_struct
*tsk
,
1317 struct pid_namespace
*ns
)
1319 return pid_nr_ns(task_pid(rcu_dereference(tsk
->real_parent
)), ns
);
1323 * pid_alive - check that a task structure is not stale
1324 * @p: Task structure to be checked.
1326 * Test if a process is not yet dead (at most zombie state)
1327 * If pid_alive fails, then pointers within the task structure
1328 * can be stale and must not be dereferenced.
1330 static inline int pid_alive(struct task_struct
*p
)
1332 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1336 * is_global_init - check if a task structure is init
1337 * @tsk: Task structure to be checked.
1339 * Check if a task structure is the first user space task the kernel created.
1341 static inline int is_global_init(struct task_struct
*tsk
)
1343 return tsk
->pid
== 1;
1347 * is_container_init:
1348 * check whether in the task is init in its own pid namespace.
1350 extern int is_container_init(struct task_struct
*tsk
);
1352 extern struct pid
*cad_pid
;
1354 extern void free_task(struct task_struct
*tsk
);
1355 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1357 extern void __put_task_struct(struct task_struct
*t
);
1359 static inline void put_task_struct(struct task_struct
*t
)
1361 if (atomic_dec_and_test(&t
->usage
))
1362 __put_task_struct(t
);
1368 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1369 /* Not implemented yet, only for 486*/
1370 #define PF_STARTING 0x00000002 /* being created */
1371 #define PF_EXITING 0x00000004 /* getting shut down */
1372 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1373 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1374 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1375 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1376 #define PF_DUMPCORE 0x00000200 /* dumped core */
1377 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1378 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1379 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1380 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1381 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1382 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1383 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1384 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1385 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1386 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1387 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1388 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1389 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1390 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1391 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1392 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1393 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1394 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1397 * Only the _current_ task can read/write to tsk->flags, but other
1398 * tasks can access tsk->flags in readonly mode for example
1399 * with tsk_used_math (like during threaded core dumping).
1400 * There is however an exception to this rule during ptrace
1401 * or during fork: the ptracer task is allowed to write to the
1402 * child->flags of its traced child (same goes for fork, the parent
1403 * can write to the child->flags), because we're guaranteed the
1404 * child is not running and in turn not changing child->flags
1405 * at the same time the parent does it.
1407 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1408 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1409 #define clear_used_math() clear_stopped_child_used_math(current)
1410 #define set_used_math() set_stopped_child_used_math(current)
1411 #define conditional_stopped_child_used_math(condition, child) \
1412 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1413 #define conditional_used_math(condition) \
1414 conditional_stopped_child_used_math(condition, current)
1415 #define copy_to_stopped_child_used_math(child) \
1416 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1417 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1418 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1419 #define used_math() tsk_used_math(current)
1422 extern int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
);
1424 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1426 if (!cpu_isset(0, new_mask
))
1432 extern unsigned long long sched_clock(void);
1435 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1436 * clock constructed from sched_clock():
1438 extern unsigned long long cpu_clock(int cpu
);
1440 extern unsigned long long
1441 task_sched_runtime(struct task_struct
*task
);
1443 /* sched_exec is called by processes performing an exec */
1445 extern void sched_exec(void);
1447 #define sched_exec() {}
1450 extern void sched_clock_idle_sleep_event(void);
1451 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1453 #ifdef CONFIG_HOTPLUG_CPU
1454 extern void idle_task_exit(void);
1456 static inline void idle_task_exit(void) {}
1459 extern void sched_idle_next(void);
1461 #ifdef CONFIG_SCHED_DEBUG
1462 extern unsigned int sysctl_sched_latency
;
1463 extern unsigned int sysctl_sched_nr_latency
;
1464 extern unsigned int sysctl_sched_wakeup_granularity
;
1465 extern unsigned int sysctl_sched_batch_wakeup_granularity
;
1466 extern unsigned int sysctl_sched_child_runs_first
;
1467 extern unsigned int sysctl_sched_features
;
1468 extern unsigned int sysctl_sched_migration_cost
;
1471 extern unsigned int sysctl_sched_compat_yield
;
1473 #ifdef CONFIG_RT_MUTEXES
1474 extern int rt_mutex_getprio(struct task_struct
*p
);
1475 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
1476 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
1478 static inline int rt_mutex_getprio(struct task_struct
*p
)
1480 return p
->normal_prio
;
1482 # define rt_mutex_adjust_pi(p) do { } while (0)
1485 extern void set_user_nice(struct task_struct
*p
, long nice
);
1486 extern int task_prio(const struct task_struct
*p
);
1487 extern int task_nice(const struct task_struct
*p
);
1488 extern int can_nice(const struct task_struct
*p
, const int nice
);
1489 extern int task_curr(const struct task_struct
*p
);
1490 extern int idle_cpu(int cpu
);
1491 extern int sched_setscheduler(struct task_struct
*, int, struct sched_param
*);
1492 extern struct task_struct
*idle_task(int cpu
);
1493 extern struct task_struct
*curr_task(int cpu
);
1494 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1499 * The default (Linux) execution domain.
1501 extern struct exec_domain default_exec_domain
;
1503 union thread_union
{
1504 struct thread_info thread_info
;
1505 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1508 #ifndef __HAVE_ARCH_KSTACK_END
1509 static inline int kstack_end(void *addr
)
1511 /* Reliable end of stack detection:
1512 * Some APM bios versions misalign the stack
1514 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1518 extern union thread_union init_thread_union
;
1519 extern struct task_struct init_task
;
1521 extern struct mm_struct init_mm
;
1523 extern struct pid_namespace init_pid_ns
;
1526 * find a task by one of its numerical ids
1528 * find_task_by_pid_type_ns():
1529 * it is the most generic call - it finds a task by all id,
1530 * type and namespace specified
1531 * find_task_by_pid_ns():
1532 * finds a task by its pid in the specified namespace
1533 * find_task_by_vpid():
1534 * finds a task by its virtual pid
1535 * find_task_by_pid():
1536 * finds a task by its global pid
1538 * see also find_pid() etc in include/linux/pid.h
1541 extern struct task_struct
*find_task_by_pid_type_ns(int type
, int pid
,
1542 struct pid_namespace
*ns
);
1544 extern struct task_struct
*find_task_by_pid(pid_t nr
);
1545 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1546 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
1547 struct pid_namespace
*ns
);
1549 extern void __set_special_pids(pid_t session
, pid_t pgrp
);
1551 /* per-UID process charging. */
1552 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
1553 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1555 atomic_inc(&u
->__count
);
1558 extern void free_uid(struct user_struct
*);
1559 extern void switch_uid(struct user_struct
*);
1560 extern void release_uids(struct user_namespace
*ns
);
1562 #include <asm/current.h>
1564 extern void do_timer(unsigned long ticks
);
1566 extern int FASTCALL(wake_up_state(struct task_struct
* tsk
, unsigned int state
));
1567 extern int FASTCALL(wake_up_process(struct task_struct
* tsk
));
1568 extern void FASTCALL(wake_up_new_task(struct task_struct
* tsk
,
1569 unsigned long clone_flags
));
1571 extern void kick_process(struct task_struct
*tsk
);
1573 static inline void kick_process(struct task_struct
*tsk
) { }
1575 extern void sched_fork(struct task_struct
*p
, int clone_flags
);
1576 extern void sched_dead(struct task_struct
*p
);
1578 extern int in_group_p(gid_t
);
1579 extern int in_egroup_p(gid_t
);
1581 extern void proc_caches_init(void);
1582 extern void flush_signals(struct task_struct
*);
1583 extern void ignore_signals(struct task_struct
*);
1584 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
1585 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
1587 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
1589 unsigned long flags
;
1592 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
1593 ret
= dequeue_signal(tsk
, mask
, info
);
1594 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
1599 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
1601 extern void unblock_all_signals(void);
1602 extern void release_task(struct task_struct
* p
);
1603 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
1604 extern int send_group_sig_info(int, struct siginfo
*, struct task_struct
*);
1605 extern int force_sigsegv(int, struct task_struct
*);
1606 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
1607 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1608 extern int kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1609 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
1610 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
1611 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
1612 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
1613 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
1614 extern void do_notify_parent(struct task_struct
*, int);
1615 extern void force_sig(int, struct task_struct
*);
1616 extern void force_sig_specific(int, struct task_struct
*);
1617 extern int send_sig(int, struct task_struct
*, int);
1618 extern void zap_other_threads(struct task_struct
*p
);
1619 extern int kill_proc(pid_t
, int, int);
1620 extern struct sigqueue
*sigqueue_alloc(void);
1621 extern void sigqueue_free(struct sigqueue
*);
1622 extern int send_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1623 extern int send_group_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1624 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
1625 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
1627 static inline int kill_cad_pid(int sig
, int priv
)
1629 return kill_pid(cad_pid
, sig
, priv
);
1632 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1633 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1634 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1635 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1637 static inline int is_si_special(const struct siginfo
*info
)
1639 return info
<= SEND_SIG_FORCED
;
1642 /* True if we are on the alternate signal stack. */
1644 static inline int on_sig_stack(unsigned long sp
)
1646 return (sp
- current
->sas_ss_sp
< current
->sas_ss_size
);
1649 static inline int sas_ss_flags(unsigned long sp
)
1651 return (current
->sas_ss_size
== 0 ? SS_DISABLE
1652 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
1656 * Routines for handling mm_structs
1658 extern struct mm_struct
* mm_alloc(void);
1660 /* mmdrop drops the mm and the page tables */
1661 extern void FASTCALL(__mmdrop(struct mm_struct
*));
1662 static inline void mmdrop(struct mm_struct
* mm
)
1664 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
1668 /* mmput gets rid of the mappings and all user-space */
1669 extern void mmput(struct mm_struct
*);
1670 /* Grab a reference to a task's mm, if it is not already going away */
1671 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
1672 /* Remove the current tasks stale references to the old mm_struct */
1673 extern void mm_release(struct task_struct
*, struct mm_struct
*);
1675 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct
*, struct pt_regs
*);
1676 extern void flush_thread(void);
1677 extern void exit_thread(void);
1679 extern void exit_files(struct task_struct
*);
1680 extern void __cleanup_signal(struct signal_struct
*);
1681 extern void __cleanup_sighand(struct sighand_struct
*);
1682 extern void exit_itimers(struct signal_struct
*);
1684 extern NORET_TYPE
void do_group_exit(int);
1686 extern void daemonize(const char *, ...);
1687 extern int allow_signal(int);
1688 extern int disallow_signal(int);
1690 extern int do_execve(char *, char __user
* __user
*, char __user
* __user
*, struct pt_regs
*);
1691 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
1692 struct task_struct
*fork_idle(int);
1694 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
1695 extern void get_task_comm(char *to
, struct task_struct
*tsk
);
1698 extern void wait_task_inactive(struct task_struct
* p
);
1700 #define wait_task_inactive(p) do { } while (0)
1703 #define remove_parent(p) list_del_init(&(p)->sibling)
1704 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1706 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1708 #define for_each_process(p) \
1709 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1712 * Careful: do_each_thread/while_each_thread is a double loop so
1713 * 'break' will not work as expected - use goto instead.
1715 #define do_each_thread(g, t) \
1716 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1718 #define while_each_thread(g, t) \
1719 while ((t = next_thread(t)) != g)
1721 /* de_thread depends on thread_group_leader not being a pid based check */
1722 #define thread_group_leader(p) (p == p->group_leader)
1724 /* Do to the insanities of de_thread it is possible for a process
1725 * to have the pid of the thread group leader without actually being
1726 * the thread group leader. For iteration through the pids in proc
1727 * all we care about is that we have a task with the appropriate
1728 * pid, we don't actually care if we have the right task.
1730 static inline int has_group_leader_pid(struct task_struct
*p
)
1732 return p
->pid
== p
->tgid
;
1736 int same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
1738 return p1
->tgid
== p2
->tgid
;
1741 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
1743 return list_entry(rcu_dereference(p
->thread_group
.next
),
1744 struct task_struct
, thread_group
);
1747 static inline int thread_group_empty(struct task_struct
*p
)
1749 return list_empty(&p
->thread_group
);
1752 #define delay_group_leader(p) \
1753 (thread_group_leader(p) && !thread_group_empty(p))
1756 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1757 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1758 * pins the final release of task.io_context. Also protects ->cpuset and
1759 * ->cgroup.subsys[].
1761 * Nests both inside and outside of read_lock(&tasklist_lock).
1762 * It must not be nested with write_lock_irq(&tasklist_lock),
1763 * neither inside nor outside.
1765 static inline void task_lock(struct task_struct
*p
)
1767 spin_lock(&p
->alloc_lock
);
1770 static inline void task_unlock(struct task_struct
*p
)
1772 spin_unlock(&p
->alloc_lock
);
1775 extern struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
1776 unsigned long *flags
);
1778 static inline void unlock_task_sighand(struct task_struct
*tsk
,
1779 unsigned long *flags
)
1781 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
1784 #ifndef __HAVE_THREAD_FUNCTIONS
1786 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1787 #define task_stack_page(task) ((task)->stack)
1789 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
1791 *task_thread_info(p
) = *task_thread_info(org
);
1792 task_thread_info(p
)->task
= p
;
1795 static inline unsigned long *end_of_stack(struct task_struct
*p
)
1797 return (unsigned long *)(task_thread_info(p
) + 1);
1802 /* set thread flags in other task's structures
1803 * - see asm/thread_info.h for TIF_xxxx flags available
1805 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1807 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1810 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1812 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1815 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1817 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1820 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1822 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1825 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1827 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1830 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1832 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1835 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1837 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1840 static inline int signal_pending(struct task_struct
*p
)
1842 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
1845 static inline int need_resched(void)
1847 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
1851 * cond_resched() and cond_resched_lock(): latency reduction via
1852 * explicit rescheduling in places that are safe. The return
1853 * value indicates whether a reschedule was done in fact.
1854 * cond_resched_lock() will drop the spinlock before scheduling,
1855 * cond_resched_softirq() will enable bhs before scheduling.
1857 extern int cond_resched(void);
1858 extern int cond_resched_lock(spinlock_t
* lock
);
1859 extern int cond_resched_softirq(void);
1862 * Does a critical section need to be broken due to another
1865 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1866 # define need_lockbreak(lock) ((lock)->break_lock)
1868 # define need_lockbreak(lock) 0
1872 * Does a critical section need to be broken due to another
1873 * task waiting or preemption being signalled:
1875 static inline int lock_need_resched(spinlock_t
*lock
)
1877 if (need_lockbreak(lock
) || need_resched())
1883 * Reevaluate whether the task has signals pending delivery.
1884 * Wake the task if so.
1885 * This is required every time the blocked sigset_t changes.
1886 * callers must hold sighand->siglock.
1888 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
1889 extern void recalc_sigpending(void);
1891 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
1894 * Wrappers for p->thread_info->cpu access. No-op on UP.
1898 static inline unsigned int task_cpu(const struct task_struct
*p
)
1900 return task_thread_info(p
)->cpu
;
1903 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
1907 static inline unsigned int task_cpu(const struct task_struct
*p
)
1912 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
1916 #endif /* CONFIG_SMP */
1918 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1919 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
1921 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
)
1923 mm
->mmap_base
= TASK_UNMAPPED_BASE
;
1924 mm
->get_unmapped_area
= arch_get_unmapped_area
;
1925 mm
->unmap_area
= arch_unmap_area
;
1929 extern long sched_setaffinity(pid_t pid
, cpumask_t new_mask
);
1930 extern long sched_getaffinity(pid_t pid
, cpumask_t
*mask
);
1932 extern int sched_mc_power_savings
, sched_smt_power_savings
;
1934 extern void normalize_rt_tasks(void);
1936 #ifdef CONFIG_FAIR_GROUP_SCHED
1938 extern struct task_group init_task_group
;
1940 extern struct task_group
*sched_create_group(void);
1941 extern void sched_destroy_group(struct task_group
*tg
);
1942 extern void sched_move_task(struct task_struct
*tsk
);
1943 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
1944 extern unsigned long sched_group_shares(struct task_group
*tg
);
1948 #ifdef CONFIG_TASK_XACCT
1949 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1954 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1959 static inline void inc_syscr(struct task_struct
*tsk
)
1964 static inline void inc_syscw(struct task_struct
*tsk
)
1969 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1973 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1977 static inline void inc_syscr(struct task_struct
*tsk
)
1981 static inline void inc_syscw(struct task_struct
*tsk
)
1986 #endif /* __KERNEL__ */