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_NEWNET 0x40000000 /* New network namespace */
33 #define SCHED_NORMAL 0
37 /* SCHED_ISO: reserved but not implemented yet */
46 #include <asm/param.h> /* for HZ */
48 #include <linux/capability.h>
49 #include <linux/threads.h>
50 #include <linux/kernel.h>
51 #include <linux/types.h>
52 #include <linux/timex.h>
53 #include <linux/jiffies.h>
54 #include <linux/rbtree.h>
55 #include <linux/thread_info.h>
56 #include <linux/cpumask.h>
57 #include <linux/errno.h>
58 #include <linux/nodemask.h>
59 #include <linux/mm_types.h>
61 #include <asm/system.h>
62 #include <asm/semaphore.h>
64 #include <asm/ptrace.h>
65 #include <asm/cputime.h>
67 #include <linux/smp.h>
68 #include <linux/sem.h>
69 #include <linux/signal.h>
70 #include <linux/securebits.h>
71 #include <linux/fs_struct.h>
72 #include <linux/compiler.h>
73 #include <linux/completion.h>
74 #include <linux/pid.h>
75 #include <linux/percpu.h>
76 #include <linux/topology.h>
77 #include <linux/proportions.h>
78 #include <linux/seccomp.h>
79 #include <linux/rcupdate.h>
80 #include <linux/futex.h>
81 #include <linux/rtmutex.h>
83 #include <linux/time.h>
84 #include <linux/param.h>
85 #include <linux/resource.h>
86 #include <linux/timer.h>
87 #include <linux/hrtimer.h>
88 #include <linux/task_io_accounting.h>
89 #include <linux/kobject.h>
91 #include <asm/processor.h>
94 struct futex_pi_state
;
98 * List of flags we want to share for kernel threads,
99 * if only because they are not used by them anyway.
101 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
104 * These are the constant used to fake the fixed-point load-average
105 * counting. Some notes:
106 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
107 * a load-average precision of 10 bits integer + 11 bits fractional
108 * - if you want to count load-averages more often, you need more
109 * precision, or rounding will get you. With 2-second counting freq,
110 * the EXP_n values would be 1981, 2034 and 2043 if still using only
113 extern unsigned long avenrun
[]; /* Load averages */
115 #define FSHIFT 11 /* nr of bits of precision */
116 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
117 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
118 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
119 #define EXP_5 2014 /* 1/exp(5sec/5min) */
120 #define EXP_15 2037 /* 1/exp(5sec/15min) */
122 #define CALC_LOAD(load,exp,n) \
124 load += n*(FIXED_1-exp); \
127 extern unsigned long total_forks
;
128 extern int nr_threads
;
129 DECLARE_PER_CPU(unsigned long, process_counts
);
130 extern int nr_processes(void);
131 extern unsigned long nr_running(void);
132 extern unsigned long nr_uninterruptible(void);
133 extern unsigned long nr_active(void);
134 extern unsigned long nr_iowait(void);
135 extern unsigned long weighted_cpuload(const int cpu
);
140 #ifdef CONFIG_SCHED_DEBUG
141 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
142 extern void proc_sched_set_task(struct task_struct
*p
);
144 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
);
147 proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
)
150 static inline void proc_sched_set_task(struct task_struct
*p
)
154 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
)
160 * Task state bitmask. NOTE! These bits are also
161 * encoded in fs/proc/array.c: get_task_state().
163 * We have two separate sets of flags: task->state
164 * is about runnability, while task->exit_state are
165 * about the task exiting. Confusing, but this way
166 * modifying one set can't modify the other one by
169 #define TASK_RUNNING 0
170 #define TASK_INTERRUPTIBLE 1
171 #define TASK_UNINTERRUPTIBLE 2
172 #define TASK_STOPPED 4
173 #define TASK_TRACED 8
174 /* in tsk->exit_state */
175 #define EXIT_ZOMBIE 16
177 /* in tsk->state again */
180 #define __set_task_state(tsk, state_value) \
181 do { (tsk)->state = (state_value); } while (0)
182 #define set_task_state(tsk, state_value) \
183 set_mb((tsk)->state, (state_value))
186 * set_current_state() includes a barrier so that the write of current->state
187 * is correctly serialised wrt the caller's subsequent test of whether to
190 * set_current_state(TASK_UNINTERRUPTIBLE);
191 * if (do_i_need_to_sleep())
194 * If the caller does not need such serialisation then use __set_current_state()
196 #define __set_current_state(state_value) \
197 do { current->state = (state_value); } while (0)
198 #define set_current_state(state_value) \
199 set_mb(current->state, (state_value))
201 /* Task command name length */
202 #define TASK_COMM_LEN 16
204 #include <linux/spinlock.h>
207 * This serializes "schedule()" and also protects
208 * the run-queue from deletions/modifications (but
209 * _adding_ to the beginning of the run-queue has
212 extern rwlock_t tasklist_lock
;
213 extern spinlock_t mmlist_lock
;
217 extern void sched_init(void);
218 extern void sched_init_smp(void);
219 extern void init_idle(struct task_struct
*idle
, int cpu
);
220 extern void init_idle_bootup_task(struct task_struct
*idle
);
222 extern cpumask_t nohz_cpu_mask
;
223 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
224 extern int select_nohz_load_balancer(int cpu
);
226 static inline int select_nohz_load_balancer(int cpu
)
233 * Only dump TASK_* tasks. (0 for all tasks)
235 extern void show_state_filter(unsigned long state_filter
);
237 static inline void show_state(void)
239 show_state_filter(0);
242 extern void show_regs(struct pt_regs
*);
245 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
246 * task), SP is the stack pointer of the first frame that should be shown in the back
247 * trace (or NULL if the entire call-chain of the task should be shown).
249 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
251 void io_schedule(void);
252 long io_schedule_timeout(long timeout
);
254 extern void cpu_init (void);
255 extern void trap_init(void);
256 extern void update_process_times(int user
);
257 extern void scheduler_tick(void);
259 #ifdef CONFIG_DETECT_SOFTLOCKUP
260 extern void softlockup_tick(void);
261 extern void spawn_softlockup_task(void);
262 extern void touch_softlockup_watchdog(void);
263 extern void touch_all_softlockup_watchdogs(void);
264 extern int softlockup_thresh
;
266 static inline void softlockup_tick(void)
269 static inline void spawn_softlockup_task(void)
272 static inline void touch_softlockup_watchdog(void)
275 static inline void touch_all_softlockup_watchdogs(void)
281 /* Attach to any functions which should be ignored in wchan output. */
282 #define __sched __attribute__((__section__(".sched.text")))
283 /* Is this address in the __sched functions? */
284 extern int in_sched_functions(unsigned long addr
);
286 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
287 extern signed long FASTCALL(schedule_timeout(signed long timeout
));
288 extern signed long schedule_timeout_interruptible(signed long timeout
);
289 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
290 asmlinkage
void schedule(void);
293 struct user_namespace
;
295 /* Maximum number of active map areas.. This is a random (large) number */
296 #define DEFAULT_MAX_MAP_COUNT 65536
298 extern int sysctl_max_map_count
;
300 #include <linux/aio.h>
303 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
304 unsigned long, unsigned long);
306 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
307 unsigned long len
, unsigned long pgoff
,
308 unsigned long flags
);
309 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
310 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
312 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
314 * The mm counters are not protected by its page_table_lock,
315 * so must be incremented atomically.
317 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
318 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
319 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
320 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
321 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
323 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
325 * The mm counters are protected by its page_table_lock,
326 * so can be incremented directly.
328 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
329 #define get_mm_counter(mm, member) ((mm)->_##member)
330 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
331 #define inc_mm_counter(mm, member) (mm)->_##member++
332 #define dec_mm_counter(mm, member) (mm)->_##member--
334 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
336 #define get_mm_rss(mm) \
337 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
338 #define update_hiwater_rss(mm) do { \
339 unsigned long _rss = get_mm_rss(mm); \
340 if ((mm)->hiwater_rss < _rss) \
341 (mm)->hiwater_rss = _rss; \
343 #define update_hiwater_vm(mm) do { \
344 if ((mm)->hiwater_vm < (mm)->total_vm) \
345 (mm)->hiwater_vm = (mm)->total_vm; \
348 extern void set_dumpable(struct mm_struct
*mm
, int value
);
349 extern int get_dumpable(struct mm_struct
*mm
);
353 #define MMF_DUMPABLE 0 /* core dump is permitted */
354 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
355 #define MMF_DUMPABLE_BITS 2
357 /* coredump filter bits */
358 #define MMF_DUMP_ANON_PRIVATE 2
359 #define MMF_DUMP_ANON_SHARED 3
360 #define MMF_DUMP_MAPPED_PRIVATE 4
361 #define MMF_DUMP_MAPPED_SHARED 5
362 #define MMF_DUMP_ELF_HEADERS 6
363 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
364 #define MMF_DUMP_FILTER_BITS 5
365 #define MMF_DUMP_FILTER_MASK \
366 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
367 #define MMF_DUMP_FILTER_DEFAULT \
368 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
370 struct sighand_struct
{
372 struct k_sigaction action
[_NSIG
];
374 wait_queue_head_t signalfd_wqh
;
377 struct pacct_struct
{
380 unsigned long ac_mem
;
381 cputime_t ac_utime
, ac_stime
;
382 unsigned long ac_minflt
, ac_majflt
;
386 * NOTE! "signal_struct" does not have it's own
387 * locking, because a shared signal_struct always
388 * implies a shared sighand_struct, so locking
389 * sighand_struct is always a proper superset of
390 * the locking of signal_struct.
392 struct signal_struct
{
396 wait_queue_head_t wait_chldexit
; /* for wait4() */
398 /* current thread group signal load-balancing target: */
399 struct task_struct
*curr_target
;
401 /* shared signal handling: */
402 struct sigpending shared_pending
;
404 /* thread group exit support */
407 * - notify group_exit_task when ->count is equal to notify_count
408 * - everyone except group_exit_task is stopped during signal delivery
409 * of fatal signals, group_exit_task processes the signal.
411 struct task_struct
*group_exit_task
;
414 /* thread group stop support, overloads group_exit_code too */
415 int group_stop_count
;
416 unsigned int flags
; /* see SIGNAL_* flags below */
418 /* POSIX.1b Interval Timers */
419 struct list_head posix_timers
;
421 /* ITIMER_REAL timer for the process */
422 struct hrtimer real_timer
;
423 struct task_struct
*tsk
;
424 ktime_t it_real_incr
;
426 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
427 cputime_t it_prof_expires
, it_virt_expires
;
428 cputime_t it_prof_incr
, it_virt_incr
;
430 /* job control IDs */
432 struct pid
*tty_old_pgrp
;
435 pid_t session __deprecated
;
439 /* boolean value for session group leader */
442 struct tty_struct
*tty
; /* NULL if no tty */
445 * Cumulative resource counters for dead threads in the group,
446 * and for reaped dead child processes forked by this group.
447 * Live threads maintain their own counters and add to these
448 * in __exit_signal, except for the group leader.
450 cputime_t utime
, stime
, cutime
, cstime
;
453 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
454 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
455 unsigned long inblock
, oublock
, cinblock
, coublock
;
458 * Cumulative ns of scheduled CPU time for dead threads in the
459 * group, not including a zombie group leader. (This only differs
460 * from jiffies_to_ns(utime + stime) if sched_clock uses something
461 * other than jiffies.)
463 unsigned long long sum_sched_runtime
;
466 * We don't bother to synchronize most readers of this at all,
467 * because there is no reader checking a limit that actually needs
468 * to get both rlim_cur and rlim_max atomically, and either one
469 * alone is a single word that can safely be read normally.
470 * getrlimit/setrlimit use task_lock(current->group_leader) to
471 * protect this instead of the siglock, because they really
472 * have no need to disable irqs.
474 struct rlimit rlim
[RLIM_NLIMITS
];
476 struct list_head cpu_timers
[3];
478 /* keep the process-shared keyrings here so that they do the right
479 * thing in threads created with CLONE_THREAD */
481 struct key
*session_keyring
; /* keyring inherited over fork */
482 struct key
*process_keyring
; /* keyring private to this process */
484 #ifdef CONFIG_BSD_PROCESS_ACCT
485 struct pacct_struct pacct
; /* per-process accounting information */
487 #ifdef CONFIG_TASKSTATS
488 struct taskstats
*stats
;
492 struct tty_audit_buf
*tty_audit_buf
;
496 /* Context switch must be unlocked if interrupts are to be enabled */
497 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
498 # define __ARCH_WANT_UNLOCKED_CTXSW
502 * Bits in flags field of signal_struct.
504 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
505 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
506 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
507 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
510 * Some day this will be a full-fledged user tracking system..
513 atomic_t __count
; /* reference count */
514 atomic_t processes
; /* How many processes does this user have? */
515 atomic_t files
; /* How many open files does this user have? */
516 atomic_t sigpending
; /* How many pending signals does this user have? */
517 #ifdef CONFIG_INOTIFY_USER
518 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
519 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
521 #ifdef CONFIG_POSIX_MQUEUE
522 /* protected by mq_lock */
523 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
525 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
528 struct key
*uid_keyring
; /* UID specific keyring */
529 struct key
*session_keyring
; /* UID's default session keyring */
532 /* Hash table maintenance information */
533 struct hlist_node uidhash_node
;
536 #ifdef CONFIG_FAIR_USER_SCHED
537 struct task_group
*tg
;
540 struct subsys_attribute user_attr
;
541 struct work_struct work
;
546 #ifdef CONFIG_FAIR_USER_SCHED
547 extern int uids_kobject_init(void);
549 static inline int uids_kobject_init(void) { return 0; }
552 extern struct user_struct
*find_user(uid_t
);
554 extern struct user_struct root_user
;
555 #define INIT_USER (&root_user)
557 struct backing_dev_info
;
558 struct reclaim_state
;
560 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
562 /* cumulative counters */
563 unsigned long pcount
; /* # of times run on this cpu */
564 unsigned long long cpu_time
, /* time spent on the cpu */
565 run_delay
; /* time spent waiting on a runqueue */
568 unsigned long long last_arrival
,/* when we last ran on a cpu */
569 last_queued
; /* when we were last queued to run */
570 #ifdef CONFIG_SCHEDSTATS
572 unsigned int bkl_count
;
575 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
577 #ifdef CONFIG_SCHEDSTATS
578 extern const struct file_operations proc_schedstat_operations
;
579 #endif /* CONFIG_SCHEDSTATS */
581 #ifdef CONFIG_TASK_DELAY_ACCT
582 struct task_delay_info
{
584 unsigned int flags
; /* Private per-task flags */
586 /* For each stat XXX, add following, aligned appropriately
588 * struct timespec XXX_start, XXX_end;
592 * Atomicity of updates to XXX_delay, XXX_count protected by
593 * single lock above (split into XXX_lock if contention is an issue).
597 * XXX_count is incremented on every XXX operation, the delay
598 * associated with the operation is added to XXX_delay.
599 * XXX_delay contains the accumulated delay time in nanoseconds.
601 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
602 u64 blkio_delay
; /* wait for sync block io completion */
603 u64 swapin_delay
; /* wait for swapin block io completion */
604 u32 blkio_count
; /* total count of the number of sync block */
605 /* io operations performed */
606 u32 swapin_count
; /* total count of the number of swapin block */
607 /* io operations performed */
609 #endif /* CONFIG_TASK_DELAY_ACCT */
611 static inline int sched_info_on(void)
613 #ifdef CONFIG_SCHEDSTATS
615 #elif defined(CONFIG_TASK_DELAY_ACCT)
616 extern int delayacct_on
;
631 * sched-domains (multiprocessor balancing) declarations:
635 * Increase resolution of nice-level calculations:
637 #define SCHED_LOAD_SHIFT 10
638 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
640 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
643 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
644 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
645 #define SD_BALANCE_EXEC 4 /* Balance on exec */
646 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
647 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
648 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
649 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
650 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
651 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
652 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
653 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
655 #define BALANCE_FOR_MC_POWER \
656 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
658 #define BALANCE_FOR_PKG_POWER \
659 ((sched_mc_power_savings || sched_smt_power_savings) ? \
660 SD_POWERSAVINGS_BALANCE : 0)
662 #define test_sd_parent(sd, flag) ((sd->parent && \
663 (sd->parent->flags & flag)) ? 1 : 0)
667 struct sched_group
*next
; /* Must be a circular list */
671 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
672 * single CPU. This is read only (except for setup, hotplug CPU).
673 * Note : Never change cpu_power without recompute its reciprocal
675 unsigned int __cpu_power
;
677 * reciprocal value of cpu_power to avoid expensive divides
678 * (see include/linux/reciprocal_div.h)
680 u32 reciprocal_cpu_power
;
683 struct sched_domain
{
684 /* These fields must be setup */
685 struct sched_domain
*parent
; /* top domain must be null terminated */
686 struct sched_domain
*child
; /* bottom domain must be null terminated */
687 struct sched_group
*groups
; /* the balancing groups of the domain */
688 cpumask_t span
; /* span of all CPUs in this domain */
689 unsigned long min_interval
; /* Minimum balance interval ms */
690 unsigned long max_interval
; /* Maximum balance interval ms */
691 unsigned int busy_factor
; /* less balancing by factor if busy */
692 unsigned int imbalance_pct
; /* No balance until over watermark */
693 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
694 unsigned int busy_idx
;
695 unsigned int idle_idx
;
696 unsigned int newidle_idx
;
697 unsigned int wake_idx
;
698 unsigned int forkexec_idx
;
699 int flags
; /* See SD_* */
701 /* Runtime fields. */
702 unsigned long last_balance
; /* init to jiffies. units in jiffies */
703 unsigned int balance_interval
; /* initialise to 1. units in ms. */
704 unsigned int nr_balance_failed
; /* initialise to 0 */
706 #ifdef CONFIG_SCHEDSTATS
707 /* load_balance() stats */
708 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
709 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
710 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
711 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
712 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
713 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
714 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
715 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
717 /* Active load balancing */
718 unsigned int alb_count
;
719 unsigned int alb_failed
;
720 unsigned int alb_pushed
;
722 /* SD_BALANCE_EXEC stats */
723 unsigned int sbe_count
;
724 unsigned int sbe_balanced
;
725 unsigned int sbe_pushed
;
727 /* SD_BALANCE_FORK stats */
728 unsigned int sbf_count
;
729 unsigned int sbf_balanced
;
730 unsigned int sbf_pushed
;
732 /* try_to_wake_up() stats */
733 unsigned int ttwu_wake_remote
;
734 unsigned int ttwu_move_affine
;
735 unsigned int ttwu_move_balance
;
739 #endif /* CONFIG_SMP */
742 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
743 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
744 * task of nice 0 or enough lower priority tasks to bring up the
747 static inline int above_background_load(void)
751 for_each_online_cpu(cpu
) {
752 if (weighted_cpuload(cpu
) >= SCHED_LOAD_SCALE
)
758 struct io_context
; /* See blkdev.h */
761 #define NGROUPS_SMALL 32
762 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
766 gid_t small_block
[NGROUPS_SMALL
];
772 * get_group_info() must be called with the owning task locked (via task_lock())
773 * when task != current. The reason being that the vast majority of callers are
774 * looking at current->group_info, which can not be changed except by the
775 * current task. Changing current->group_info requires the task lock, too.
777 #define get_group_info(group_info) do { \
778 atomic_inc(&(group_info)->usage); \
781 #define put_group_info(group_info) do { \
782 if (atomic_dec_and_test(&(group_info)->usage)) \
783 groups_free(group_info); \
786 extern struct group_info
*groups_alloc(int gidsetsize
);
787 extern void groups_free(struct group_info
*group_info
);
788 extern int set_current_groups(struct group_info
*group_info
);
789 extern int groups_search(struct group_info
*group_info
, gid_t grp
);
790 /* access the groups "array" with this macro */
791 #define GROUP_AT(gi, i) \
792 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
794 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
795 extern void prefetch_stack(struct task_struct
*t
);
797 static inline void prefetch_stack(struct task_struct
*t
) { }
800 struct audit_context
; /* See audit.c */
802 struct pipe_inode_info
;
803 struct uts_namespace
;
809 const struct sched_class
*next
;
811 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int wakeup
);
812 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int sleep
);
813 void (*yield_task
) (struct rq
*rq
);
815 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
);
817 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
818 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
820 unsigned long (*load_balance
) (struct rq
*this_rq
, int this_cpu
,
822 unsigned long max_nr_move
, unsigned long max_load_move
,
823 struct sched_domain
*sd
, enum cpu_idle_type idle
,
824 int *all_pinned
, int *this_best_prio
);
826 void (*set_curr_task
) (struct rq
*rq
);
827 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
);
828 void (*task_new
) (struct rq
*rq
, struct task_struct
*p
);
832 unsigned long weight
, inv_weight
;
836 * CFS stats for a schedulable entity (task, task-group etc)
838 * Current field usage histogram:
845 struct sched_entity
{
846 struct load_weight load
; /* for load-balancing */
847 struct rb_node run_node
;
852 u64 sum_exec_runtime
;
854 u64 prev_sum_exec_runtime
;
856 #ifdef CONFIG_SCHEDSTATS
862 s64 sum_sleep_runtime
;
870 u64 nr_migrations_cold
;
871 u64 nr_failed_migrations_affine
;
872 u64 nr_failed_migrations_running
;
873 u64 nr_failed_migrations_hot
;
874 u64 nr_forced_migrations
;
875 u64 nr_forced2_migrations
;
879 u64 nr_wakeups_migrate
;
880 u64 nr_wakeups_local
;
881 u64 nr_wakeups_remote
;
882 u64 nr_wakeups_affine
;
883 u64 nr_wakeups_affine_attempts
;
884 u64 nr_wakeups_passive
;
888 #ifdef CONFIG_FAIR_GROUP_SCHED
889 struct sched_entity
*parent
;
890 /* rq on which this entity is (to be) queued: */
891 struct cfs_rq
*cfs_rq
;
892 /* rq "owned" by this entity/group: */
897 #ifdef CONFIG_CGROUPS
899 #define SUBSYS(_x) _x ## _subsys_id,
900 enum cgroup_subsys_id
{
901 #include <linux/cgroup_subsys.h>
906 /* A css_set is a structure holding pointers to a set of
907 * cgroup_subsys_state objects.
912 /* Set of subsystem states, one for each subsystem. NULL for
913 * subsystems that aren't part of this hierarchy. These
914 * pointers reduce the number of dereferences required to get
915 * from a task to its state for a given cgroup, but result
916 * in increased space usage if tasks are in wildly different
917 * groupings across different hierarchies. This array is
918 * immutable after creation */
919 struct cgroup_subsys_state
*subsys
[CGROUP_SUBSYS_COUNT
];
923 #endif /* CONFIG_CGROUPS */
926 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
929 unsigned int flags
; /* per process flags, defined below */
932 int lock_depth
; /* BKL lock depth */
935 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
940 int prio
, static_prio
, normal_prio
;
941 struct list_head run_list
;
942 const struct sched_class
*sched_class
;
943 struct sched_entity se
;
945 #ifdef CONFIG_PREEMPT_NOTIFIERS
946 /* list of struct preempt_notifier: */
947 struct hlist_head preempt_notifiers
;
950 unsigned short ioprio
;
952 * fpu_counter contains the number of consecutive context switches
953 * that the FPU is used. If this is over a threshold, the lazy fpu
954 * saving becomes unlazy to save the trap. This is an unsigned char
955 * so that after 256 times the counter wraps and the behavior turns
956 * lazy again; this to deal with bursty apps that only use FPU for
959 unsigned char fpu_counter
;
960 s8 oomkilladj
; /* OOM kill score adjustment (bit shift). */
961 #ifdef CONFIG_BLK_DEV_IO_TRACE
962 unsigned int btrace_seq
;
966 cpumask_t cpus_allowed
;
967 unsigned int time_slice
;
969 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
970 struct sched_info sched_info
;
973 struct list_head tasks
;
975 * ptrace_list/ptrace_children forms the list of my children
976 * that were stolen by a ptracer.
978 struct list_head ptrace_children
;
979 struct list_head ptrace_list
;
981 struct mm_struct
*mm
, *active_mm
;
984 struct linux_binfmt
*binfmt
;
986 int exit_code
, exit_signal
;
987 int pdeath_signal
; /* The signal sent when the parent dies */
989 unsigned int personality
;
994 #ifdef CONFIG_CC_STACKPROTECTOR
995 /* Canary value for the -fstack-protector gcc feature */
996 unsigned long stack_canary
;
999 * pointers to (original) parent process, youngest child, younger sibling,
1000 * older sibling, respectively. (p->father can be replaced with
1003 struct task_struct
*real_parent
; /* real parent process (when being debugged) */
1004 struct task_struct
*parent
; /* parent process */
1006 * children/sibling forms the list of my children plus the
1007 * tasks I'm ptracing.
1009 struct list_head children
; /* list of my children */
1010 struct list_head sibling
; /* linkage in my parent's children list */
1011 struct task_struct
*group_leader
; /* threadgroup leader */
1013 /* PID/PID hash table linkage. */
1014 struct pid_link pids
[PIDTYPE_MAX
];
1015 struct list_head thread_group
;
1017 struct completion
*vfork_done
; /* for vfork() */
1018 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1019 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1021 unsigned int rt_priority
;
1022 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1024 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1025 struct timespec start_time
; /* monotonic time */
1026 struct timespec real_start_time
; /* boot based time */
1027 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1028 unsigned long min_flt
, maj_flt
;
1030 cputime_t it_prof_expires
, it_virt_expires
;
1031 unsigned long long it_sched_expires
;
1032 struct list_head cpu_timers
[3];
1034 /* process credentials */
1035 uid_t uid
,euid
,suid
,fsuid
;
1036 gid_t gid
,egid
,sgid
,fsgid
;
1037 struct group_info
*group_info
;
1038 kernel_cap_t cap_effective
, cap_inheritable
, cap_permitted
;
1039 unsigned keep_capabilities
:1;
1040 struct user_struct
*user
;
1042 struct key
*request_key_auth
; /* assumed request_key authority */
1043 struct key
*thread_keyring
; /* keyring private to this thread */
1044 unsigned char jit_keyring
; /* default keyring to attach requested keys to */
1046 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1047 - access with [gs]et_task_comm (which lock
1048 it with task_lock())
1049 - initialized normally by flush_old_exec */
1050 /* file system info */
1051 int link_count
, total_link_count
;
1052 #ifdef CONFIG_SYSVIPC
1054 struct sysv_sem sysvsem
;
1056 /* CPU-specific state of this task */
1057 struct thread_struct thread
;
1058 /* filesystem information */
1059 struct fs_struct
*fs
;
1060 /* open file information */
1061 struct files_struct
*files
;
1063 struct nsproxy
*nsproxy
;
1064 /* signal handlers */
1065 struct signal_struct
*signal
;
1066 struct sighand_struct
*sighand
;
1068 sigset_t blocked
, real_blocked
;
1069 sigset_t saved_sigmask
; /* To be restored with TIF_RESTORE_SIGMASK */
1070 struct sigpending pending
;
1072 unsigned long sas_ss_sp
;
1074 int (*notifier
)(void *priv
);
1075 void *notifier_data
;
1076 sigset_t
*notifier_mask
;
1077 #ifdef CONFIG_SECURITY
1080 struct audit_context
*audit_context
;
1083 /* Thread group tracking */
1086 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1087 spinlock_t alloc_lock
;
1089 /* Protection of the PI data structures: */
1092 #ifdef CONFIG_RT_MUTEXES
1093 /* PI waiters blocked on a rt_mutex held by this task */
1094 struct plist_head pi_waiters
;
1095 /* Deadlock detection and priority inheritance handling */
1096 struct rt_mutex_waiter
*pi_blocked_on
;
1099 #ifdef CONFIG_DEBUG_MUTEXES
1100 /* mutex deadlock detection */
1101 struct mutex_waiter
*blocked_on
;
1103 #ifdef CONFIG_TRACE_IRQFLAGS
1104 unsigned int irq_events
;
1105 int hardirqs_enabled
;
1106 unsigned long hardirq_enable_ip
;
1107 unsigned int hardirq_enable_event
;
1108 unsigned long hardirq_disable_ip
;
1109 unsigned int hardirq_disable_event
;
1110 int softirqs_enabled
;
1111 unsigned long softirq_disable_ip
;
1112 unsigned int softirq_disable_event
;
1113 unsigned long softirq_enable_ip
;
1114 unsigned int softirq_enable_event
;
1115 int hardirq_context
;
1116 int softirq_context
;
1118 #ifdef CONFIG_LOCKDEP
1119 # define MAX_LOCK_DEPTH 30UL
1122 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1123 unsigned int lockdep_recursion
;
1126 /* journalling filesystem info */
1129 /* stacked block device info */
1130 struct bio
*bio_list
, **bio_tail
;
1133 struct reclaim_state
*reclaim_state
;
1135 struct backing_dev_info
*backing_dev_info
;
1137 struct io_context
*io_context
;
1139 unsigned long ptrace_message
;
1140 siginfo_t
*last_siginfo
; /* For ptrace use. */
1141 #ifdef CONFIG_TASK_XACCT
1142 /* i/o counters(bytes read/written, #syscalls */
1143 u64 rchar
, wchar
, syscr
, syscw
;
1145 struct task_io_accounting ioac
;
1146 #if defined(CONFIG_TASK_XACCT)
1147 u64 acct_rss_mem1
; /* accumulated rss usage */
1148 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1149 cputime_t acct_stimexpd
;/* stime since last update */
1152 struct mempolicy
*mempolicy
;
1155 #ifdef CONFIG_CPUSETS
1156 struct cpuset
*cpuset
;
1157 nodemask_t mems_allowed
;
1158 int cpuset_mems_generation
;
1159 int cpuset_mem_spread_rotor
;
1161 #ifdef CONFIG_CGROUPS
1162 struct css_set cgroups
;
1165 struct robust_list_head __user
*robust_list
;
1166 #ifdef CONFIG_COMPAT
1167 struct compat_robust_list_head __user
*compat_robust_list
;
1169 struct list_head pi_state_list
;
1170 struct futex_pi_state
*pi_state_cache
;
1172 atomic_t fs_excl
; /* holding fs exclusive resources */
1173 struct rcu_head rcu
;
1176 * cache last used pipe for splice
1178 struct pipe_inode_info
*splice_pipe
;
1179 #ifdef CONFIG_TASK_DELAY_ACCT
1180 struct task_delay_info
*delays
;
1182 #ifdef CONFIG_FAULT_INJECTION
1185 struct prop_local_single dirties
;
1189 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1190 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1191 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1192 * values are inverted: lower p->prio value means higher priority.
1194 * The MAX_USER_RT_PRIO value allows the actual maximum
1195 * RT priority to be separate from the value exported to
1196 * user-space. This allows kernel threads to set their
1197 * priority to a value higher than any user task. Note:
1198 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1201 #define MAX_USER_RT_PRIO 100
1202 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1204 #define MAX_PRIO (MAX_RT_PRIO + 40)
1205 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1207 static inline int rt_prio(int prio
)
1209 if (unlikely(prio
< MAX_RT_PRIO
))
1214 static inline int rt_task(struct task_struct
*p
)
1216 return rt_prio(p
->prio
);
1219 static inline pid_t
process_group(struct task_struct
*tsk
)
1221 return tsk
->signal
->pgrp
;
1224 static inline pid_t
signal_session(struct signal_struct
*sig
)
1226 return sig
->__session
;
1229 static inline pid_t
process_session(struct task_struct
*tsk
)
1231 return signal_session(tsk
->signal
);
1234 static inline void set_signal_session(struct signal_struct
*sig
, pid_t session
)
1236 sig
->__session
= session
;
1239 static inline struct pid
*task_pid(struct task_struct
*task
)
1241 return task
->pids
[PIDTYPE_PID
].pid
;
1244 static inline struct pid
*task_tgid(struct task_struct
*task
)
1246 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1249 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1251 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1254 static inline struct pid
*task_session(struct task_struct
*task
)
1256 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1260 * pid_alive - check that a task structure is not stale
1261 * @p: Task structure to be checked.
1263 * Test if a process is not yet dead (at most zombie state)
1264 * If pid_alive fails, then pointers within the task structure
1265 * can be stale and must not be dereferenced.
1267 static inline int pid_alive(struct task_struct
*p
)
1269 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1273 * is_init - check if a task structure is init
1274 * @tsk: Task structure to be checked.
1276 * Check if a task structure is the first user space task the kernel created.
1278 static inline int is_init(struct task_struct
*tsk
)
1280 return tsk
->pid
== 1;
1283 extern struct pid
*cad_pid
;
1285 extern void free_task(struct task_struct
*tsk
);
1286 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1288 extern void __put_task_struct(struct task_struct
*t
);
1290 static inline void put_task_struct(struct task_struct
*t
)
1292 if (atomic_dec_and_test(&t
->usage
))
1293 __put_task_struct(t
);
1299 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1300 /* Not implemented yet, only for 486*/
1301 #define PF_STARTING 0x00000002 /* being created */
1302 #define PF_EXITING 0x00000004 /* getting shut down */
1303 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1304 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1305 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1306 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1307 #define PF_DUMPCORE 0x00000200 /* dumped core */
1308 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1309 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1310 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1311 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1312 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1313 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1314 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1315 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1316 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1317 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1318 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1319 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1320 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1321 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1322 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1323 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1324 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1325 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1328 * Only the _current_ task can read/write to tsk->flags, but other
1329 * tasks can access tsk->flags in readonly mode for example
1330 * with tsk_used_math (like during threaded core dumping).
1331 * There is however an exception to this rule during ptrace
1332 * or during fork: the ptracer task is allowed to write to the
1333 * child->flags of its traced child (same goes for fork, the parent
1334 * can write to the child->flags), because we're guaranteed the
1335 * child is not running and in turn not changing child->flags
1336 * at the same time the parent does it.
1338 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1339 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1340 #define clear_used_math() clear_stopped_child_used_math(current)
1341 #define set_used_math() set_stopped_child_used_math(current)
1342 #define conditional_stopped_child_used_math(condition, child) \
1343 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1344 #define conditional_used_math(condition) \
1345 conditional_stopped_child_used_math(condition, current)
1346 #define copy_to_stopped_child_used_math(child) \
1347 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1348 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1349 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1350 #define used_math() tsk_used_math(current)
1353 extern int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
);
1355 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1357 if (!cpu_isset(0, new_mask
))
1363 extern unsigned long long sched_clock(void);
1366 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1367 * clock constructed from sched_clock():
1369 extern unsigned long long cpu_clock(int cpu
);
1371 extern unsigned long long
1372 task_sched_runtime(struct task_struct
*task
);
1374 /* sched_exec is called by processes performing an exec */
1376 extern void sched_exec(void);
1378 #define sched_exec() {}
1381 extern void sched_clock_idle_sleep_event(void);
1382 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1384 #ifdef CONFIG_HOTPLUG_CPU
1385 extern void idle_task_exit(void);
1387 static inline void idle_task_exit(void) {}
1390 extern void sched_idle_next(void);
1392 #ifdef CONFIG_SCHED_DEBUG
1393 extern unsigned int sysctl_sched_latency
;
1394 extern unsigned int sysctl_sched_nr_latency
;
1395 extern unsigned int sysctl_sched_wakeup_granularity
;
1396 extern unsigned int sysctl_sched_batch_wakeup_granularity
;
1397 extern unsigned int sysctl_sched_child_runs_first
;
1398 extern unsigned int sysctl_sched_features
;
1399 extern unsigned int sysctl_sched_migration_cost
;
1402 extern unsigned int sysctl_sched_compat_yield
;
1404 #ifdef CONFIG_RT_MUTEXES
1405 extern int rt_mutex_getprio(struct task_struct
*p
);
1406 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
1407 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
1409 static inline int rt_mutex_getprio(struct task_struct
*p
)
1411 return p
->normal_prio
;
1413 # define rt_mutex_adjust_pi(p) do { } while (0)
1416 extern void set_user_nice(struct task_struct
*p
, long nice
);
1417 extern int task_prio(const struct task_struct
*p
);
1418 extern int task_nice(const struct task_struct
*p
);
1419 extern int can_nice(const struct task_struct
*p
, const int nice
);
1420 extern int task_curr(const struct task_struct
*p
);
1421 extern int idle_cpu(int cpu
);
1422 extern int sched_setscheduler(struct task_struct
*, int, struct sched_param
*);
1423 extern struct task_struct
*idle_task(int cpu
);
1424 extern struct task_struct
*curr_task(int cpu
);
1425 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1430 * The default (Linux) execution domain.
1432 extern struct exec_domain default_exec_domain
;
1434 union thread_union
{
1435 struct thread_info thread_info
;
1436 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1439 #ifndef __HAVE_ARCH_KSTACK_END
1440 static inline int kstack_end(void *addr
)
1442 /* Reliable end of stack detection:
1443 * Some APM bios versions misalign the stack
1445 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1449 extern union thread_union init_thread_union
;
1450 extern struct task_struct init_task
;
1452 extern struct mm_struct init_mm
;
1454 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1455 extern struct task_struct
*find_task_by_pid_type(int type
, int pid
);
1456 extern void __set_special_pids(pid_t session
, pid_t pgrp
);
1458 /* per-UID process charging. */
1459 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
1460 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1462 atomic_inc(&u
->__count
);
1465 extern void free_uid(struct user_struct
*);
1466 extern void switch_uid(struct user_struct
*);
1467 extern void release_uids(struct user_namespace
*ns
);
1469 #include <asm/current.h>
1471 extern void do_timer(unsigned long ticks
);
1473 extern int FASTCALL(wake_up_state(struct task_struct
* tsk
, unsigned int state
));
1474 extern int FASTCALL(wake_up_process(struct task_struct
* tsk
));
1475 extern void FASTCALL(wake_up_new_task(struct task_struct
* tsk
,
1476 unsigned long clone_flags
));
1478 extern void kick_process(struct task_struct
*tsk
);
1480 static inline void kick_process(struct task_struct
*tsk
) { }
1482 extern void sched_fork(struct task_struct
*p
, int clone_flags
);
1483 extern void sched_dead(struct task_struct
*p
);
1485 extern int in_group_p(gid_t
);
1486 extern int in_egroup_p(gid_t
);
1488 extern void proc_caches_init(void);
1489 extern void flush_signals(struct task_struct
*);
1490 extern void ignore_signals(struct task_struct
*);
1491 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
1492 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
1494 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
1496 unsigned long flags
;
1499 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
1500 ret
= dequeue_signal(tsk
, mask
, info
);
1501 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
1506 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
1508 extern void unblock_all_signals(void);
1509 extern void release_task(struct task_struct
* p
);
1510 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
1511 extern int send_group_sig_info(int, struct siginfo
*, struct task_struct
*);
1512 extern int force_sigsegv(int, struct task_struct
*);
1513 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
1514 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1515 extern int kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1516 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
1517 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
1518 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
1519 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
1520 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
1521 extern void do_notify_parent(struct task_struct
*, int);
1522 extern void force_sig(int, struct task_struct
*);
1523 extern void force_sig_specific(int, struct task_struct
*);
1524 extern int send_sig(int, struct task_struct
*, int);
1525 extern void zap_other_threads(struct task_struct
*p
);
1526 extern int kill_proc(pid_t
, int, int);
1527 extern struct sigqueue
*sigqueue_alloc(void);
1528 extern void sigqueue_free(struct sigqueue
*);
1529 extern int send_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1530 extern int send_group_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1531 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
1532 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
1534 static inline int kill_cad_pid(int sig
, int priv
)
1536 return kill_pid(cad_pid
, sig
, priv
);
1539 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1540 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1541 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1542 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1544 static inline int is_si_special(const struct siginfo
*info
)
1546 return info
<= SEND_SIG_FORCED
;
1549 /* True if we are on the alternate signal stack. */
1551 static inline int on_sig_stack(unsigned long sp
)
1553 return (sp
- current
->sas_ss_sp
< current
->sas_ss_size
);
1556 static inline int sas_ss_flags(unsigned long sp
)
1558 return (current
->sas_ss_size
== 0 ? SS_DISABLE
1559 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
1563 * Routines for handling mm_structs
1565 extern struct mm_struct
* mm_alloc(void);
1567 /* mmdrop drops the mm and the page tables */
1568 extern void FASTCALL(__mmdrop(struct mm_struct
*));
1569 static inline void mmdrop(struct mm_struct
* mm
)
1571 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
1575 /* mmput gets rid of the mappings and all user-space */
1576 extern void mmput(struct mm_struct
*);
1577 /* Grab a reference to a task's mm, if it is not already going away */
1578 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
1579 /* Remove the current tasks stale references to the old mm_struct */
1580 extern void mm_release(struct task_struct
*, struct mm_struct
*);
1582 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct
*, struct pt_regs
*);
1583 extern void flush_thread(void);
1584 extern void exit_thread(void);
1586 extern void exit_files(struct task_struct
*);
1587 extern void __cleanup_signal(struct signal_struct
*);
1588 extern void __cleanup_sighand(struct sighand_struct
*);
1589 extern void exit_itimers(struct signal_struct
*);
1591 extern NORET_TYPE
void do_group_exit(int);
1593 extern void daemonize(const char *, ...);
1594 extern int allow_signal(int);
1595 extern int disallow_signal(int);
1597 extern int do_execve(char *, char __user
* __user
*, char __user
* __user
*, struct pt_regs
*);
1598 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
1599 struct task_struct
*fork_idle(int);
1601 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
1602 extern void get_task_comm(char *to
, struct task_struct
*tsk
);
1605 extern void wait_task_inactive(struct task_struct
* p
);
1607 #define wait_task_inactive(p) do { } while (0)
1610 #define remove_parent(p) list_del_init(&(p)->sibling)
1611 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1613 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1615 #define for_each_process(p) \
1616 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1619 * Careful: do_each_thread/while_each_thread is a double loop so
1620 * 'break' will not work as expected - use goto instead.
1622 #define do_each_thread(g, t) \
1623 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1625 #define while_each_thread(g, t) \
1626 while ((t = next_thread(t)) != g)
1628 /* de_thread depends on thread_group_leader not being a pid based check */
1629 #define thread_group_leader(p) (p == p->group_leader)
1631 /* Do to the insanities of de_thread it is possible for a process
1632 * to have the pid of the thread group leader without actually being
1633 * the thread group leader. For iteration through the pids in proc
1634 * all we care about is that we have a task with the appropriate
1635 * pid, we don't actually care if we have the right task.
1637 static inline int has_group_leader_pid(struct task_struct
*p
)
1639 return p
->pid
== p
->tgid
;
1642 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
1644 return list_entry(rcu_dereference(p
->thread_group
.next
),
1645 struct task_struct
, thread_group
);
1648 static inline int thread_group_empty(struct task_struct
*p
)
1650 return list_empty(&p
->thread_group
);
1653 #define delay_group_leader(p) \
1654 (thread_group_leader(p) && !thread_group_empty(p))
1657 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1658 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1659 * pins the final release of task.io_context. Also protects ->cpuset and
1660 * ->cgroup.subsys[].
1662 * Nests both inside and outside of read_lock(&tasklist_lock).
1663 * It must not be nested with write_lock_irq(&tasklist_lock),
1664 * neither inside nor outside.
1666 static inline void task_lock(struct task_struct
*p
)
1668 spin_lock(&p
->alloc_lock
);
1671 static inline void task_unlock(struct task_struct
*p
)
1673 spin_unlock(&p
->alloc_lock
);
1676 extern struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
1677 unsigned long *flags
);
1679 static inline void unlock_task_sighand(struct task_struct
*tsk
,
1680 unsigned long *flags
)
1682 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
1685 #ifndef __HAVE_THREAD_FUNCTIONS
1687 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1688 #define task_stack_page(task) ((task)->stack)
1690 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
1692 *task_thread_info(p
) = *task_thread_info(org
);
1693 task_thread_info(p
)->task
= p
;
1696 static inline unsigned long *end_of_stack(struct task_struct
*p
)
1698 return (unsigned long *)(task_thread_info(p
) + 1);
1703 /* set thread flags in other task's structures
1704 * - see asm/thread_info.h for TIF_xxxx flags available
1706 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1708 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1711 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1713 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1716 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1718 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1721 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1723 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1726 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1728 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1731 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1733 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1736 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1738 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1741 static inline int signal_pending(struct task_struct
*p
)
1743 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
1746 static inline int need_resched(void)
1748 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
1752 * cond_resched() and cond_resched_lock(): latency reduction via
1753 * explicit rescheduling in places that are safe. The return
1754 * value indicates whether a reschedule was done in fact.
1755 * cond_resched_lock() will drop the spinlock before scheduling,
1756 * cond_resched_softirq() will enable bhs before scheduling.
1758 extern int cond_resched(void);
1759 extern int cond_resched_lock(spinlock_t
* lock
);
1760 extern int cond_resched_softirq(void);
1763 * Does a critical section need to be broken due to another
1766 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1767 # define need_lockbreak(lock) ((lock)->break_lock)
1769 # define need_lockbreak(lock) 0
1773 * Does a critical section need to be broken due to another
1774 * task waiting or preemption being signalled:
1776 static inline int lock_need_resched(spinlock_t
*lock
)
1778 if (need_lockbreak(lock
) || need_resched())
1784 * Reevaluate whether the task has signals pending delivery.
1785 * Wake the task if so.
1786 * This is required every time the blocked sigset_t changes.
1787 * callers must hold sighand->siglock.
1789 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
1790 extern void recalc_sigpending(void);
1792 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
1795 * Wrappers for p->thread_info->cpu access. No-op on UP.
1799 static inline unsigned int task_cpu(const struct task_struct
*p
)
1801 return task_thread_info(p
)->cpu
;
1804 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
1808 static inline unsigned int task_cpu(const struct task_struct
*p
)
1813 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
1817 #endif /* CONFIG_SMP */
1819 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1820 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
1822 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
)
1824 mm
->mmap_base
= TASK_UNMAPPED_BASE
;
1825 mm
->get_unmapped_area
= arch_get_unmapped_area
;
1826 mm
->unmap_area
= arch_unmap_area
;
1830 extern long sched_setaffinity(pid_t pid
, cpumask_t new_mask
);
1831 extern long sched_getaffinity(pid_t pid
, cpumask_t
*mask
);
1833 extern int sched_mc_power_savings
, sched_smt_power_savings
;
1835 extern void normalize_rt_tasks(void);
1837 #ifdef CONFIG_FAIR_GROUP_SCHED
1839 extern struct task_group init_task_group
;
1841 extern struct task_group
*sched_create_group(void);
1842 extern void sched_destroy_group(struct task_group
*tg
);
1843 extern void sched_move_task(struct task_struct
*tsk
);
1844 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
1845 extern unsigned long sched_group_shares(struct task_group
*tg
);
1849 #ifdef CONFIG_TASK_XACCT
1850 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1855 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1860 static inline void inc_syscr(struct task_struct
*tsk
)
1865 static inline void inc_syscw(struct task_struct
*tsk
)
1870 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1874 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1878 static inline void inc_syscr(struct task_struct
*tsk
)
1882 static inline void inc_syscw(struct task_struct
*tsk
)
1887 #endif /* __KERNEL__ */