4 #include <linux/auxvec.h> /* For AT_VECTOR_SIZE */
9 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
10 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
11 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
12 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
13 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
14 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
15 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
16 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
17 #define CLONE_THREAD 0x00010000 /* Same thread group? */
18 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
19 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
20 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
21 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
22 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
23 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
24 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
25 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
26 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
27 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
28 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
29 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
30 #define CLONE_NEWNET 0x40000000 /* New network namespace */
35 #define SCHED_NORMAL 0
39 /* SCHED_ISO: reserved but not implemented yet */
48 #include <asm/param.h> /* for HZ */
50 #include <linux/capability.h>
51 #include <linux/threads.h>
52 #include <linux/kernel.h>
53 #include <linux/types.h>
54 #include <linux/timex.h>
55 #include <linux/jiffies.h>
56 #include <linux/rbtree.h>
57 #include <linux/thread_info.h>
58 #include <linux/cpumask.h>
59 #include <linux/errno.h>
60 #include <linux/nodemask.h>
62 #include <asm/system.h>
63 #include <asm/semaphore.h>
65 #include <asm/ptrace.h>
67 #include <asm/cputime.h>
69 #include <linux/smp.h>
70 #include <linux/sem.h>
71 #include <linux/signal.h>
72 #include <linux/securebits.h>
73 #include <linux/fs_struct.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/seccomp.h>
80 #include <linux/rcupdate.h>
81 #include <linux/futex.h>
82 #include <linux/rtmutex.h>
84 #include <linux/time.h>
85 #include <linux/param.h>
86 #include <linux/resource.h>
87 #include <linux/timer.h>
88 #include <linux/hrtimer.h>
89 #include <linux/task_io_accounting.h>
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);
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)
322 typedef atomic_long_t mm_counter_t
;
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--
334 typedef unsigned long mm_counter_t
;
336 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
338 #define get_mm_rss(mm) \
339 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
340 #define update_hiwater_rss(mm) do { \
341 unsigned long _rss = get_mm_rss(mm); \
342 if ((mm)->hiwater_rss < _rss) \
343 (mm)->hiwater_rss = _rss; \
345 #define update_hiwater_vm(mm) do { \
346 if ((mm)->hiwater_vm < (mm)->total_vm) \
347 (mm)->hiwater_vm = (mm)->total_vm; \
350 extern void set_dumpable(struct mm_struct
*mm
, int value
);
351 extern int get_dumpable(struct mm_struct
*mm
);
355 #define MMF_DUMPABLE 0 /* core dump is permitted */
356 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
357 #define MMF_DUMPABLE_BITS 2
359 /* coredump filter bits */
360 #define MMF_DUMP_ANON_PRIVATE 2
361 #define MMF_DUMP_ANON_SHARED 3
362 #define MMF_DUMP_MAPPED_PRIVATE 4
363 #define MMF_DUMP_MAPPED_SHARED 5
364 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
365 #define MMF_DUMP_FILTER_BITS 4
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))
372 struct vm_area_struct
* mmap
; /* list of VMAs */
373 struct rb_root mm_rb
;
374 struct vm_area_struct
* mmap_cache
; /* last find_vma result */
375 unsigned long (*get_unmapped_area
) (struct file
*filp
,
376 unsigned long addr
, unsigned long len
,
377 unsigned long pgoff
, unsigned long flags
);
378 void (*unmap_area
) (struct mm_struct
*mm
, unsigned long addr
);
379 unsigned long mmap_base
; /* base of mmap area */
380 unsigned long task_size
; /* size of task vm space */
381 unsigned long cached_hole_size
; /* if non-zero, the largest hole below free_area_cache */
382 unsigned long free_area_cache
; /* first hole of size cached_hole_size or larger */
384 atomic_t mm_users
; /* How many users with user space? */
385 atomic_t mm_count
; /* How many references to "struct mm_struct" (users count as 1) */
386 int map_count
; /* number of VMAs */
387 struct rw_semaphore mmap_sem
;
388 spinlock_t page_table_lock
; /* Protects page tables and some counters */
390 struct list_head mmlist
; /* List of maybe swapped mm's. These are globally strung
391 * together off init_mm.mmlist, and are protected
395 /* Special counters, in some configurations protected by the
396 * page_table_lock, in other configurations by being atomic.
398 mm_counter_t _file_rss
;
399 mm_counter_t _anon_rss
;
401 unsigned long hiwater_rss
; /* High-watermark of RSS usage */
402 unsigned long hiwater_vm
; /* High-water virtual memory usage */
404 unsigned long total_vm
, locked_vm
, shared_vm
, exec_vm
;
405 unsigned long stack_vm
, reserved_vm
, def_flags
, nr_ptes
;
406 unsigned long start_code
, end_code
, start_data
, end_data
;
407 unsigned long start_brk
, brk
, start_stack
;
408 unsigned long arg_start
, arg_end
, env_start
, env_end
;
410 unsigned long saved_auxv
[AT_VECTOR_SIZE
]; /* for /proc/PID/auxv */
412 cpumask_t cpu_vm_mask
;
414 /* Architecture-specific MM context */
415 mm_context_t context
;
417 /* Swap token stuff */
419 * Last value of global fault stamp as seen by this process.
420 * In other words, this value gives an indication of how long
421 * it has been since this task got the token.
422 * Look at mm/thrash.c
424 unsigned int faultstamp
;
425 unsigned int token_priority
;
426 unsigned int last_interval
;
428 unsigned long flags
; /* Must use atomic bitops to access the bits */
430 /* coredumping support */
432 struct completion
*core_startup_done
, core_done
;
435 rwlock_t ioctx_list_lock
;
436 struct kioctx
*ioctx_list
;
439 struct sighand_struct
{
441 struct k_sigaction action
[_NSIG
];
443 wait_queue_head_t signalfd_wqh
;
446 struct pacct_struct
{
449 unsigned long ac_mem
;
450 cputime_t ac_utime
, ac_stime
;
451 unsigned long ac_minflt
, ac_majflt
;
455 * NOTE! "signal_struct" does not have it's own
456 * locking, because a shared signal_struct always
457 * implies a shared sighand_struct, so locking
458 * sighand_struct is always a proper superset of
459 * the locking of signal_struct.
461 struct signal_struct
{
465 wait_queue_head_t wait_chldexit
; /* for wait4() */
467 /* current thread group signal load-balancing target: */
468 struct task_struct
*curr_target
;
470 /* shared signal handling: */
471 struct sigpending shared_pending
;
473 /* thread group exit support */
476 * - notify group_exit_task when ->count is equal to notify_count
477 * - everyone except group_exit_task is stopped during signal delivery
478 * of fatal signals, group_exit_task processes the signal.
480 struct task_struct
*group_exit_task
;
483 /* thread group stop support, overloads group_exit_code too */
484 int group_stop_count
;
485 unsigned int flags
; /* see SIGNAL_* flags below */
487 /* POSIX.1b Interval Timers */
488 struct list_head posix_timers
;
490 /* ITIMER_REAL timer for the process */
491 struct hrtimer real_timer
;
492 struct task_struct
*tsk
;
493 ktime_t it_real_incr
;
495 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
496 cputime_t it_prof_expires
, it_virt_expires
;
497 cputime_t it_prof_incr
, it_virt_incr
;
499 /* job control IDs */
501 struct pid
*tty_old_pgrp
;
504 pid_t session __deprecated
;
508 /* boolean value for session group leader */
511 struct tty_struct
*tty
; /* NULL if no tty */
514 * Cumulative resource counters for dead threads in the group,
515 * and for reaped dead child processes forked by this group.
516 * Live threads maintain their own counters and add to these
517 * in __exit_signal, except for the group leader.
519 cputime_t utime
, stime
, cutime
, cstime
;
520 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
521 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
522 unsigned long inblock
, oublock
, cinblock
, coublock
;
525 * Cumulative ns of scheduled CPU time for dead threads in the
526 * group, not including a zombie group leader. (This only differs
527 * from jiffies_to_ns(utime + stime) if sched_clock uses something
528 * other than jiffies.)
530 unsigned long long sum_sched_runtime
;
533 * We don't bother to synchronize most readers of this at all,
534 * because there is no reader checking a limit that actually needs
535 * to get both rlim_cur and rlim_max atomically, and either one
536 * alone is a single word that can safely be read normally.
537 * getrlimit/setrlimit use task_lock(current->group_leader) to
538 * protect this instead of the siglock, because they really
539 * have no need to disable irqs.
541 struct rlimit rlim
[RLIM_NLIMITS
];
543 struct list_head cpu_timers
[3];
545 /* keep the process-shared keyrings here so that they do the right
546 * thing in threads created with CLONE_THREAD */
548 struct key
*session_keyring
; /* keyring inherited over fork */
549 struct key
*process_keyring
; /* keyring private to this process */
551 #ifdef CONFIG_BSD_PROCESS_ACCT
552 struct pacct_struct pacct
; /* per-process accounting information */
554 #ifdef CONFIG_TASKSTATS
555 struct taskstats
*stats
;
559 struct tty_audit_buf
*tty_audit_buf
;
563 /* Context switch must be unlocked if interrupts are to be enabled */
564 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
565 # define __ARCH_WANT_UNLOCKED_CTXSW
569 * Bits in flags field of signal_struct.
571 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
572 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
573 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
574 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
577 * Some day this will be a full-fledged user tracking system..
580 atomic_t __count
; /* reference count */
581 atomic_t processes
; /* How many processes does this user have? */
582 atomic_t files
; /* How many open files does this user have? */
583 atomic_t sigpending
; /* How many pending signals does this user have? */
584 #ifdef CONFIG_INOTIFY_USER
585 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
586 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
588 /* protected by mq_lock */
589 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
590 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
593 struct key
*uid_keyring
; /* UID specific keyring */
594 struct key
*session_keyring
; /* UID's default session keyring */
597 /* Hash table maintenance information */
598 struct hlist_node uidhash_node
;
601 #ifdef CONFIG_FAIR_USER_SCHED
602 struct task_group
*tg
;
604 struct subsys_attribute user_attr
;
605 struct work_struct work
;
609 #ifdef CONFIG_FAIR_USER_SCHED
610 extern int uids_kobject_init(void);
612 static inline int uids_kobject_init(void) { return 0; }
615 extern struct user_struct
*find_user(uid_t
);
617 extern struct user_struct root_user
;
618 #define INIT_USER (&root_user)
620 struct backing_dev_info
;
621 struct reclaim_state
;
623 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
625 /* cumulative counters */
626 unsigned long pcount
; /* # of times run on this cpu */
627 unsigned long long cpu_time
, /* time spent on the cpu */
628 run_delay
; /* time spent waiting on a runqueue */
631 unsigned long long last_arrival
,/* when we last ran on a cpu */
632 last_queued
; /* when we were last queued to run */
633 #ifdef CONFIG_SCHEDSTATS
635 unsigned long bkl_count
;
638 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
640 #ifdef CONFIG_SCHEDSTATS
641 extern const struct file_operations proc_schedstat_operations
;
642 #endif /* CONFIG_SCHEDSTATS */
644 #ifdef CONFIG_TASK_DELAY_ACCT
645 struct task_delay_info
{
647 unsigned int flags
; /* Private per-task flags */
649 /* For each stat XXX, add following, aligned appropriately
651 * struct timespec XXX_start, XXX_end;
655 * Atomicity of updates to XXX_delay, XXX_count protected by
656 * single lock above (split into XXX_lock if contention is an issue).
660 * XXX_count is incremented on every XXX operation, the delay
661 * associated with the operation is added to XXX_delay.
662 * XXX_delay contains the accumulated delay time in nanoseconds.
664 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
665 u64 blkio_delay
; /* wait for sync block io completion */
666 u64 swapin_delay
; /* wait for swapin block io completion */
667 u32 blkio_count
; /* total count of the number of sync block */
668 /* io operations performed */
669 u32 swapin_count
; /* total count of the number of swapin block */
670 /* io operations performed */
672 #endif /* CONFIG_TASK_DELAY_ACCT */
674 static inline int sched_info_on(void)
676 #ifdef CONFIG_SCHEDSTATS
678 #elif defined(CONFIG_TASK_DELAY_ACCT)
679 extern int delayacct_on
;
694 * sched-domains (multiprocessor balancing) declarations:
698 * Increase resolution of nice-level calculations:
700 #define SCHED_LOAD_SHIFT 10
701 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
703 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
706 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
707 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
708 #define SD_BALANCE_EXEC 4 /* Balance on exec */
709 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
710 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
711 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
712 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
713 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
714 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
715 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
716 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
718 #define BALANCE_FOR_MC_POWER \
719 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
721 #define BALANCE_FOR_PKG_POWER \
722 ((sched_mc_power_savings || sched_smt_power_savings) ? \
723 SD_POWERSAVINGS_BALANCE : 0)
725 #define test_sd_parent(sd, flag) ((sd->parent && \
726 (sd->parent->flags & flag)) ? 1 : 0)
730 struct sched_group
*next
; /* Must be a circular list */
734 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
735 * single CPU. This is read only (except for setup, hotplug CPU).
736 * Note : Never change cpu_power without recompute its reciprocal
738 unsigned int __cpu_power
;
740 * reciprocal value of cpu_power to avoid expensive divides
741 * (see include/linux/reciprocal_div.h)
743 u32 reciprocal_cpu_power
;
746 struct sched_domain
{
747 /* These fields must be setup */
748 struct sched_domain
*parent
; /* top domain must be null terminated */
749 struct sched_domain
*child
; /* bottom domain must be null terminated */
750 struct sched_group
*groups
; /* the balancing groups of the domain */
751 cpumask_t span
; /* span of all CPUs in this domain */
752 unsigned long min_interval
; /* Minimum balance interval ms */
753 unsigned long max_interval
; /* Maximum balance interval ms */
754 unsigned int busy_factor
; /* less balancing by factor if busy */
755 unsigned int imbalance_pct
; /* No balance until over watermark */
756 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
757 unsigned int busy_idx
;
758 unsigned int idle_idx
;
759 unsigned int newidle_idx
;
760 unsigned int wake_idx
;
761 unsigned int forkexec_idx
;
762 int flags
; /* See SD_* */
764 /* Runtime fields. */
765 unsigned long last_balance
; /* init to jiffies. units in jiffies */
766 unsigned int balance_interval
; /* initialise to 1. units in ms. */
767 unsigned int nr_balance_failed
; /* initialise to 0 */
769 #ifdef CONFIG_SCHEDSTATS
770 /* load_balance() stats */
771 unsigned long lb_count
[CPU_MAX_IDLE_TYPES
];
772 unsigned long lb_failed
[CPU_MAX_IDLE_TYPES
];
773 unsigned long lb_balanced
[CPU_MAX_IDLE_TYPES
];
774 unsigned long lb_imbalance
[CPU_MAX_IDLE_TYPES
];
775 unsigned long lb_gained
[CPU_MAX_IDLE_TYPES
];
776 unsigned long lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
777 unsigned long lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
778 unsigned long lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
780 /* Active load balancing */
781 unsigned long alb_count
;
782 unsigned long alb_failed
;
783 unsigned long alb_pushed
;
785 /* SD_BALANCE_EXEC stats */
786 unsigned long sbe_count
;
787 unsigned long sbe_balanced
;
788 unsigned long sbe_pushed
;
790 /* SD_BALANCE_FORK stats */
791 unsigned long sbf_count
;
792 unsigned long sbf_balanced
;
793 unsigned long sbf_pushed
;
795 /* try_to_wake_up() stats */
796 unsigned long ttwu_wake_remote
;
797 unsigned long ttwu_move_affine
;
798 unsigned long ttwu_move_balance
;
802 extern int partition_sched_domains(cpumask_t
*partition1
,
803 cpumask_t
*partition2
);
805 #endif /* CONFIG_SMP */
808 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
809 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
810 * task of nice 0 or enough lower priority tasks to bring up the
813 static inline int above_background_load(void)
817 for_each_online_cpu(cpu
) {
818 if (weighted_cpuload(cpu
) >= SCHED_LOAD_SCALE
)
824 struct io_context
; /* See blkdev.h */
827 #define NGROUPS_SMALL 32
828 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
832 gid_t small_block
[NGROUPS_SMALL
];
838 * get_group_info() must be called with the owning task locked (via task_lock())
839 * when task != current. The reason being that the vast majority of callers are
840 * looking at current->group_info, which can not be changed except by the
841 * current task. Changing current->group_info requires the task lock, too.
843 #define get_group_info(group_info) do { \
844 atomic_inc(&(group_info)->usage); \
847 #define put_group_info(group_info) do { \
848 if (atomic_dec_and_test(&(group_info)->usage)) \
849 groups_free(group_info); \
852 extern struct group_info
*groups_alloc(int gidsetsize
);
853 extern void groups_free(struct group_info
*group_info
);
854 extern int set_current_groups(struct group_info
*group_info
);
855 extern int groups_search(struct group_info
*group_info
, gid_t grp
);
856 /* access the groups "array" with this macro */
857 #define GROUP_AT(gi, i) \
858 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
860 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
861 extern void prefetch_stack(struct task_struct
*t
);
863 static inline void prefetch_stack(struct task_struct
*t
) { }
866 struct audit_context
; /* See audit.c */
868 struct pipe_inode_info
;
869 struct uts_namespace
;
875 const struct sched_class
*next
;
877 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int wakeup
);
878 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int sleep
);
879 void (*yield_task
) (struct rq
*rq
);
881 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
);
883 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
884 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
886 unsigned long (*load_balance
) (struct rq
*this_rq
, int this_cpu
,
888 unsigned long max_nr_move
, unsigned long max_load_move
,
889 struct sched_domain
*sd
, enum cpu_idle_type idle
,
890 int *all_pinned
, int *this_best_prio
);
892 void (*set_curr_task
) (struct rq
*rq
);
893 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
);
894 void (*task_new
) (struct rq
*rq
, struct task_struct
*p
);
898 unsigned long weight
, inv_weight
;
902 * CFS stats for a schedulable entity (task, task-group etc)
904 * Current field usage histogram:
911 struct sched_entity
{
912 struct load_weight load
; /* for load-balancing */
913 struct rb_node run_node
;
918 u64 sum_exec_runtime
;
920 u64 prev_sum_exec_runtime
;
922 #ifdef CONFIG_SCHEDSTATS
928 s64 sum_sleep_runtime
;
936 u64 nr_migrations_cold
;
937 u64 nr_failed_migrations_affine
;
938 u64 nr_failed_migrations_running
;
939 u64 nr_failed_migrations_hot
;
940 u64 nr_forced_migrations
;
941 u64 nr_forced2_migrations
;
945 u64 nr_wakeups_migrate
;
946 u64 nr_wakeups_local
;
947 u64 nr_wakeups_remote
;
948 u64 nr_wakeups_affine
;
949 u64 nr_wakeups_affine_attempts
;
950 u64 nr_wakeups_passive
;
954 #ifdef CONFIG_FAIR_GROUP_SCHED
955 struct sched_entity
*parent
;
956 /* rq on which this entity is (to be) queued: */
957 struct cfs_rq
*cfs_rq
;
958 /* rq "owned" by this entity/group: */
964 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
967 unsigned int flags
; /* per process flags, defined below */
970 int lock_depth
; /* BKL lock depth */
973 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
978 int prio
, static_prio
, normal_prio
;
979 struct list_head run_list
;
980 const struct sched_class
*sched_class
;
981 struct sched_entity se
;
983 #ifdef CONFIG_PREEMPT_NOTIFIERS
984 /* list of struct preempt_notifier: */
985 struct hlist_head preempt_notifiers
;
988 unsigned short ioprio
;
989 #ifdef CONFIG_BLK_DEV_IO_TRACE
990 unsigned int btrace_seq
;
994 cpumask_t cpus_allowed
;
995 unsigned int time_slice
;
997 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
998 struct sched_info sched_info
;
1001 struct list_head tasks
;
1003 * ptrace_list/ptrace_children forms the list of my children
1004 * that were stolen by a ptracer.
1006 struct list_head ptrace_children
;
1007 struct list_head ptrace_list
;
1009 struct mm_struct
*mm
, *active_mm
;
1012 struct linux_binfmt
*binfmt
;
1014 int exit_code
, exit_signal
;
1015 int pdeath_signal
; /* The signal sent when the parent dies */
1017 unsigned int personality
;
1018 unsigned did_exec
:1;
1022 #ifdef CONFIG_CC_STACKPROTECTOR
1023 /* Canary value for the -fstack-protector gcc feature */
1024 unsigned long stack_canary
;
1027 * pointers to (original) parent process, youngest child, younger sibling,
1028 * older sibling, respectively. (p->father can be replaced with
1031 struct task_struct
*real_parent
; /* real parent process (when being debugged) */
1032 struct task_struct
*parent
; /* parent process */
1034 * children/sibling forms the list of my children plus the
1035 * tasks I'm ptracing.
1037 struct list_head children
; /* list of my children */
1038 struct list_head sibling
; /* linkage in my parent's children list */
1039 struct task_struct
*group_leader
; /* threadgroup leader */
1041 /* PID/PID hash table linkage. */
1042 struct pid_link pids
[PIDTYPE_MAX
];
1043 struct list_head thread_group
;
1045 struct completion
*vfork_done
; /* for vfork() */
1046 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1047 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1049 unsigned int rt_priority
;
1050 cputime_t utime
, stime
;
1051 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1052 struct timespec start_time
; /* monotonic time */
1053 struct timespec real_start_time
; /* boot based time */
1054 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1055 unsigned long min_flt
, maj_flt
;
1057 cputime_t it_prof_expires
, it_virt_expires
;
1058 unsigned long long it_sched_expires
;
1059 struct list_head cpu_timers
[3];
1061 /* process credentials */
1062 uid_t uid
,euid
,suid
,fsuid
;
1063 gid_t gid
,egid
,sgid
,fsgid
;
1064 struct group_info
*group_info
;
1065 kernel_cap_t cap_effective
, cap_inheritable
, cap_permitted
;
1066 unsigned keep_capabilities
:1;
1067 struct user_struct
*user
;
1069 struct key
*request_key_auth
; /* assumed request_key authority */
1070 struct key
*thread_keyring
; /* keyring private to this thread */
1071 unsigned char jit_keyring
; /* default keyring to attach requested keys to */
1074 * fpu_counter contains the number of consecutive context switches
1075 * that the FPU is used. If this is over a threshold, the lazy fpu
1076 * saving becomes unlazy to save the trap. This is an unsigned char
1077 * so that after 256 times the counter wraps and the behavior turns
1078 * lazy again; this to deal with bursty apps that only use FPU for
1081 unsigned char fpu_counter
;
1082 int oomkilladj
; /* OOM kill score adjustment (bit shift). */
1083 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1084 - access with [gs]et_task_comm (which lock
1085 it with task_lock())
1086 - initialized normally by flush_old_exec */
1087 /* file system info */
1088 int link_count
, total_link_count
;
1089 #ifdef CONFIG_SYSVIPC
1091 struct sysv_sem sysvsem
;
1093 /* CPU-specific state of this task */
1094 struct thread_struct thread
;
1095 /* filesystem information */
1096 struct fs_struct
*fs
;
1097 /* open file information */
1098 struct files_struct
*files
;
1100 struct nsproxy
*nsproxy
;
1101 /* signal handlers */
1102 struct signal_struct
*signal
;
1103 struct sighand_struct
*sighand
;
1105 sigset_t blocked
, real_blocked
;
1106 sigset_t saved_sigmask
; /* To be restored with TIF_RESTORE_SIGMASK */
1107 struct sigpending pending
;
1109 unsigned long sas_ss_sp
;
1111 int (*notifier
)(void *priv
);
1112 void *notifier_data
;
1113 sigset_t
*notifier_mask
;
1116 struct audit_context
*audit_context
;
1119 /* Thread group tracking */
1122 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1123 spinlock_t alloc_lock
;
1125 /* Protection of the PI data structures: */
1128 #ifdef CONFIG_RT_MUTEXES
1129 /* PI waiters blocked on a rt_mutex held by this task */
1130 struct plist_head pi_waiters
;
1131 /* Deadlock detection and priority inheritance handling */
1132 struct rt_mutex_waiter
*pi_blocked_on
;
1135 #ifdef CONFIG_DEBUG_MUTEXES
1136 /* mutex deadlock detection */
1137 struct mutex_waiter
*blocked_on
;
1139 #ifdef CONFIG_TRACE_IRQFLAGS
1140 unsigned int irq_events
;
1141 int hardirqs_enabled
;
1142 unsigned long hardirq_enable_ip
;
1143 unsigned int hardirq_enable_event
;
1144 unsigned long hardirq_disable_ip
;
1145 unsigned int hardirq_disable_event
;
1146 int softirqs_enabled
;
1147 unsigned long softirq_disable_ip
;
1148 unsigned int softirq_disable_event
;
1149 unsigned long softirq_enable_ip
;
1150 unsigned int softirq_enable_event
;
1151 int hardirq_context
;
1152 int softirq_context
;
1154 #ifdef CONFIG_LOCKDEP
1155 # define MAX_LOCK_DEPTH 30UL
1158 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1159 unsigned int lockdep_recursion
;
1162 /* journalling filesystem info */
1165 /* stacked block device info */
1166 struct bio
*bio_list
, **bio_tail
;
1169 struct reclaim_state
*reclaim_state
;
1171 struct backing_dev_info
*backing_dev_info
;
1173 struct io_context
*io_context
;
1175 unsigned long ptrace_message
;
1176 siginfo_t
*last_siginfo
; /* For ptrace use. */
1178 * current io wait handle: wait queue entry to use for io waits
1179 * If this thread is processing aio, this points at the waitqueue
1180 * inside the currently handled kiocb. It may be NULL (i.e. default
1181 * to a stack based synchronous wait) if its doing sync IO.
1183 wait_queue_t
*io_wait
;
1184 #ifdef CONFIG_TASK_XACCT
1185 /* i/o counters(bytes read/written, #syscalls */
1186 u64 rchar
, wchar
, syscr
, syscw
;
1188 struct task_io_accounting ioac
;
1189 #if defined(CONFIG_TASK_XACCT)
1190 u64 acct_rss_mem1
; /* accumulated rss usage */
1191 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1192 cputime_t acct_stimexpd
;/* stime since last update */
1195 struct mempolicy
*mempolicy
;
1198 #ifdef CONFIG_CPUSETS
1199 struct cpuset
*cpuset
;
1200 nodemask_t mems_allowed
;
1201 int cpuset_mems_generation
;
1202 int cpuset_mem_spread_rotor
;
1204 struct robust_list_head __user
*robust_list
;
1205 #ifdef CONFIG_COMPAT
1206 struct compat_robust_list_head __user
*compat_robust_list
;
1208 struct list_head pi_state_list
;
1209 struct futex_pi_state
*pi_state_cache
;
1211 atomic_t fs_excl
; /* holding fs exclusive resources */
1212 struct rcu_head rcu
;
1215 * cache last used pipe for splice
1217 struct pipe_inode_info
*splice_pipe
;
1218 #ifdef CONFIG_TASK_DELAY_ACCT
1219 struct task_delay_info
*delays
;
1221 #ifdef CONFIG_FAULT_INJECTION
1227 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1228 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1229 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1230 * values are inverted: lower p->prio value means higher priority.
1232 * The MAX_USER_RT_PRIO value allows the actual maximum
1233 * RT priority to be separate from the value exported to
1234 * user-space. This allows kernel threads to set their
1235 * priority to a value higher than any user task. Note:
1236 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1239 #define MAX_USER_RT_PRIO 100
1240 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1242 #define MAX_PRIO (MAX_RT_PRIO + 40)
1243 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1245 static inline int rt_prio(int prio
)
1247 if (unlikely(prio
< MAX_RT_PRIO
))
1252 static inline int rt_task(struct task_struct
*p
)
1254 return rt_prio(p
->prio
);
1257 static inline pid_t
process_group(struct task_struct
*tsk
)
1259 return tsk
->signal
->pgrp
;
1262 static inline pid_t
signal_session(struct signal_struct
*sig
)
1264 return sig
->__session
;
1267 static inline pid_t
process_session(struct task_struct
*tsk
)
1269 return signal_session(tsk
->signal
);
1272 static inline void set_signal_session(struct signal_struct
*sig
, pid_t session
)
1274 sig
->__session
= session
;
1277 static inline struct pid
*task_pid(struct task_struct
*task
)
1279 return task
->pids
[PIDTYPE_PID
].pid
;
1282 static inline struct pid
*task_tgid(struct task_struct
*task
)
1284 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1287 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1289 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1292 static inline struct pid
*task_session(struct task_struct
*task
)
1294 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1298 * pid_alive - check that a task structure is not stale
1299 * @p: Task structure to be checked.
1301 * Test if a process is not yet dead (at most zombie state)
1302 * If pid_alive fails, then pointers within the task structure
1303 * can be stale and must not be dereferenced.
1305 static inline int pid_alive(struct task_struct
*p
)
1307 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1311 * is_init - check if a task structure is init
1312 * @tsk: Task structure to be checked.
1314 * Check if a task structure is the first user space task the kernel created.
1316 static inline int is_init(struct task_struct
*tsk
)
1318 return tsk
->pid
== 1;
1321 extern struct pid
*cad_pid
;
1323 extern void free_task(struct task_struct
*tsk
);
1324 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1326 extern void __put_task_struct(struct task_struct
*t
);
1328 static inline void put_task_struct(struct task_struct
*t
)
1330 if (atomic_dec_and_test(&t
->usage
))
1331 __put_task_struct(t
);
1337 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1338 /* Not implemented yet, only for 486*/
1339 #define PF_STARTING 0x00000002 /* being created */
1340 #define PF_EXITING 0x00000004 /* getting shut down */
1341 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1342 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1343 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1344 #define PF_DUMPCORE 0x00000200 /* dumped core */
1345 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1346 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1347 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1348 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1349 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1350 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1351 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1352 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1353 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1354 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1355 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1356 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1357 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1358 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1359 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1360 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1361 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1362 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1365 * Only the _current_ task can read/write to tsk->flags, but other
1366 * tasks can access tsk->flags in readonly mode for example
1367 * with tsk_used_math (like during threaded core dumping).
1368 * There is however an exception to this rule during ptrace
1369 * or during fork: the ptracer task is allowed to write to the
1370 * child->flags of its traced child (same goes for fork, the parent
1371 * can write to the child->flags), because we're guaranteed the
1372 * child is not running and in turn not changing child->flags
1373 * at the same time the parent does it.
1375 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1376 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1377 #define clear_used_math() clear_stopped_child_used_math(current)
1378 #define set_used_math() set_stopped_child_used_math(current)
1379 #define conditional_stopped_child_used_math(condition, child) \
1380 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1381 #define conditional_used_math(condition) \
1382 conditional_stopped_child_used_math(condition, current)
1383 #define copy_to_stopped_child_used_math(child) \
1384 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1385 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1386 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1387 #define used_math() tsk_used_math(current)
1390 extern int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
);
1392 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1394 if (!cpu_isset(0, new_mask
))
1400 extern unsigned long long sched_clock(void);
1403 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1404 * clock constructed from sched_clock():
1406 extern unsigned long long cpu_clock(int cpu
);
1408 extern unsigned long long
1409 task_sched_runtime(struct task_struct
*task
);
1411 /* sched_exec is called by processes performing an exec */
1413 extern void sched_exec(void);
1415 #define sched_exec() {}
1418 extern void sched_clock_idle_sleep_event(void);
1419 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1421 #ifdef CONFIG_HOTPLUG_CPU
1422 extern void idle_task_exit(void);
1424 static inline void idle_task_exit(void) {}
1427 extern void sched_idle_next(void);
1429 #ifdef CONFIG_SCHED_DEBUG
1430 extern unsigned int sysctl_sched_latency
;
1431 extern unsigned int sysctl_sched_nr_latency
;
1432 extern unsigned int sysctl_sched_wakeup_granularity
;
1433 extern unsigned int sysctl_sched_batch_wakeup_granularity
;
1434 extern unsigned int sysctl_sched_child_runs_first
;
1435 extern unsigned int sysctl_sched_features
;
1436 extern unsigned int sysctl_sched_migration_cost
;
1439 extern unsigned int sysctl_sched_compat_yield
;
1441 #ifdef CONFIG_RT_MUTEXES
1442 extern int rt_mutex_getprio(struct task_struct
*p
);
1443 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
1444 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
1446 static inline int rt_mutex_getprio(struct task_struct
*p
)
1448 return p
->normal_prio
;
1450 # define rt_mutex_adjust_pi(p) do { } while (0)
1453 extern void set_user_nice(struct task_struct
*p
, long nice
);
1454 extern int task_prio(const struct task_struct
*p
);
1455 extern int task_nice(const struct task_struct
*p
);
1456 extern int can_nice(const struct task_struct
*p
, const int nice
);
1457 extern int task_curr(const struct task_struct
*p
);
1458 extern int idle_cpu(int cpu
);
1459 extern int sched_setscheduler(struct task_struct
*, int, struct sched_param
*);
1460 extern struct task_struct
*idle_task(int cpu
);
1461 extern struct task_struct
*curr_task(int cpu
);
1462 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1467 * The default (Linux) execution domain.
1469 extern struct exec_domain default_exec_domain
;
1471 union thread_union
{
1472 struct thread_info thread_info
;
1473 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1476 #ifndef __HAVE_ARCH_KSTACK_END
1477 static inline int kstack_end(void *addr
)
1479 /* Reliable end of stack detection:
1480 * Some APM bios versions misalign the stack
1482 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1486 extern union thread_union init_thread_union
;
1487 extern struct task_struct init_task
;
1489 extern struct mm_struct init_mm
;
1491 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1492 extern struct task_struct
*find_task_by_pid_type(int type
, int pid
);
1493 extern void __set_special_pids(pid_t session
, pid_t pgrp
);
1495 /* per-UID process charging. */
1496 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
1497 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1499 atomic_inc(&u
->__count
);
1502 extern void free_uid(struct user_struct
*);
1503 extern void switch_uid(struct user_struct
*);
1504 extern void release_uids(struct user_namespace
*ns
);
1506 #include <asm/current.h>
1508 extern void do_timer(unsigned long ticks
);
1510 extern int FASTCALL(wake_up_state(struct task_struct
* tsk
, unsigned int state
));
1511 extern int FASTCALL(wake_up_process(struct task_struct
* tsk
));
1512 extern void FASTCALL(wake_up_new_task(struct task_struct
* tsk
,
1513 unsigned long clone_flags
));
1515 extern void kick_process(struct task_struct
*tsk
);
1517 static inline void kick_process(struct task_struct
*tsk
) { }
1519 extern void sched_fork(struct task_struct
*p
, int clone_flags
);
1520 extern void sched_dead(struct task_struct
*p
);
1522 extern int in_group_p(gid_t
);
1523 extern int in_egroup_p(gid_t
);
1525 extern void proc_caches_init(void);
1526 extern void flush_signals(struct task_struct
*);
1527 extern void ignore_signals(struct task_struct
*);
1528 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
1529 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
1531 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
1533 unsigned long flags
;
1536 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
1537 ret
= dequeue_signal(tsk
, mask
, info
);
1538 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
1543 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
1545 extern void unblock_all_signals(void);
1546 extern void release_task(struct task_struct
* p
);
1547 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
1548 extern int send_group_sig_info(int, struct siginfo
*, struct task_struct
*);
1549 extern int force_sigsegv(int, struct task_struct
*);
1550 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
1551 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1552 extern int kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1553 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
1554 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
1555 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
1556 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
1557 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
1558 extern void do_notify_parent(struct task_struct
*, int);
1559 extern void force_sig(int, struct task_struct
*);
1560 extern void force_sig_specific(int, struct task_struct
*);
1561 extern int send_sig(int, struct task_struct
*, int);
1562 extern void zap_other_threads(struct task_struct
*p
);
1563 extern int kill_proc(pid_t
, int, int);
1564 extern struct sigqueue
*sigqueue_alloc(void);
1565 extern void sigqueue_free(struct sigqueue
*);
1566 extern int send_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1567 extern int send_group_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1568 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
1569 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
1571 static inline int kill_cad_pid(int sig
, int priv
)
1573 return kill_pid(cad_pid
, sig
, priv
);
1576 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1577 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1578 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1579 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1581 static inline int is_si_special(const struct siginfo
*info
)
1583 return info
<= SEND_SIG_FORCED
;
1586 /* True if we are on the alternate signal stack. */
1588 static inline int on_sig_stack(unsigned long sp
)
1590 return (sp
- current
->sas_ss_sp
< current
->sas_ss_size
);
1593 static inline int sas_ss_flags(unsigned long sp
)
1595 return (current
->sas_ss_size
== 0 ? SS_DISABLE
1596 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
1600 * Routines for handling mm_structs
1602 extern struct mm_struct
* mm_alloc(void);
1604 /* mmdrop drops the mm and the page tables */
1605 extern void FASTCALL(__mmdrop(struct mm_struct
*));
1606 static inline void mmdrop(struct mm_struct
* mm
)
1608 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
1612 /* mmput gets rid of the mappings and all user-space */
1613 extern void mmput(struct mm_struct
*);
1614 /* Grab a reference to a task's mm, if it is not already going away */
1615 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
1616 /* Remove the current tasks stale references to the old mm_struct */
1617 extern void mm_release(struct task_struct
*, struct mm_struct
*);
1619 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct
*, struct pt_regs
*);
1620 extern void flush_thread(void);
1621 extern void exit_thread(void);
1623 extern void exit_files(struct task_struct
*);
1624 extern void __cleanup_signal(struct signal_struct
*);
1625 extern void __cleanup_sighand(struct sighand_struct
*);
1626 extern void exit_itimers(struct signal_struct
*);
1628 extern NORET_TYPE
void do_group_exit(int);
1630 extern void daemonize(const char *, ...);
1631 extern int allow_signal(int);
1632 extern int disallow_signal(int);
1634 extern int do_execve(char *, char __user
* __user
*, char __user
* __user
*, struct pt_regs
*);
1635 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
1636 struct task_struct
*fork_idle(int);
1638 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
1639 extern void get_task_comm(char *to
, struct task_struct
*tsk
);
1642 extern void wait_task_inactive(struct task_struct
* p
);
1644 #define wait_task_inactive(p) do { } while (0)
1647 #define remove_parent(p) list_del_init(&(p)->sibling)
1648 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1650 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1652 #define for_each_process(p) \
1653 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1656 * Careful: do_each_thread/while_each_thread is a double loop so
1657 * 'break' will not work as expected - use goto instead.
1659 #define do_each_thread(g, t) \
1660 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1662 #define while_each_thread(g, t) \
1663 while ((t = next_thread(t)) != g)
1665 /* de_thread depends on thread_group_leader not being a pid based check */
1666 #define thread_group_leader(p) (p == p->group_leader)
1668 /* Do to the insanities of de_thread it is possible for a process
1669 * to have the pid of the thread group leader without actually being
1670 * the thread group leader. For iteration through the pids in proc
1671 * all we care about is that we have a task with the appropriate
1672 * pid, we don't actually care if we have the right task.
1674 static inline int has_group_leader_pid(struct task_struct
*p
)
1676 return p
->pid
== p
->tgid
;
1679 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
1681 return list_entry(rcu_dereference(p
->thread_group
.next
),
1682 struct task_struct
, thread_group
);
1685 static inline int thread_group_empty(struct task_struct
*p
)
1687 return list_empty(&p
->thread_group
);
1690 #define delay_group_leader(p) \
1691 (thread_group_leader(p) && !thread_group_empty(p))
1694 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1695 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1696 * pins the final release of task.io_context. Also protects ->cpuset.
1698 * Nests both inside and outside of read_lock(&tasklist_lock).
1699 * It must not be nested with write_lock_irq(&tasklist_lock),
1700 * neither inside nor outside.
1702 static inline void task_lock(struct task_struct
*p
)
1704 spin_lock(&p
->alloc_lock
);
1707 static inline void task_unlock(struct task_struct
*p
)
1709 spin_unlock(&p
->alloc_lock
);
1712 extern struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
1713 unsigned long *flags
);
1715 static inline void unlock_task_sighand(struct task_struct
*tsk
,
1716 unsigned long *flags
)
1718 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
1721 #ifndef __HAVE_THREAD_FUNCTIONS
1723 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1724 #define task_stack_page(task) ((task)->stack)
1726 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
1728 *task_thread_info(p
) = *task_thread_info(org
);
1729 task_thread_info(p
)->task
= p
;
1732 static inline unsigned long *end_of_stack(struct task_struct
*p
)
1734 return (unsigned long *)(task_thread_info(p
) + 1);
1739 /* set thread flags in other task's structures
1740 * - see asm/thread_info.h for TIF_xxxx flags available
1742 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1744 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1747 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1749 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1752 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1754 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1757 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1759 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1762 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1764 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1767 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1769 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1772 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1774 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1777 static inline int signal_pending(struct task_struct
*p
)
1779 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
1782 static inline int need_resched(void)
1784 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
1788 * cond_resched() and cond_resched_lock(): latency reduction via
1789 * explicit rescheduling in places that are safe. The return
1790 * value indicates whether a reschedule was done in fact.
1791 * cond_resched_lock() will drop the spinlock before scheduling,
1792 * cond_resched_softirq() will enable bhs before scheduling.
1794 extern int cond_resched(void);
1795 extern int cond_resched_lock(spinlock_t
* lock
);
1796 extern int cond_resched_softirq(void);
1799 * Does a critical section need to be broken due to another
1802 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1803 # define need_lockbreak(lock) ((lock)->break_lock)
1805 # define need_lockbreak(lock) 0
1809 * Does a critical section need to be broken due to another
1810 * task waiting or preemption being signalled:
1812 static inline int lock_need_resched(spinlock_t
*lock
)
1814 if (need_lockbreak(lock
) || need_resched())
1820 * Reevaluate whether the task has signals pending delivery.
1821 * Wake the task if so.
1822 * This is required every time the blocked sigset_t changes.
1823 * callers must hold sighand->siglock.
1825 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
1826 extern void recalc_sigpending(void);
1828 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
1831 * Wrappers for p->thread_info->cpu access. No-op on UP.
1835 static inline unsigned int task_cpu(const struct task_struct
*p
)
1837 return task_thread_info(p
)->cpu
;
1840 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
1844 static inline unsigned int task_cpu(const struct task_struct
*p
)
1849 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
1853 #endif /* CONFIG_SMP */
1855 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1856 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
1858 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
)
1860 mm
->mmap_base
= TASK_UNMAPPED_BASE
;
1861 mm
->get_unmapped_area
= arch_get_unmapped_area
;
1862 mm
->unmap_area
= arch_unmap_area
;
1866 extern long sched_setaffinity(pid_t pid
, cpumask_t new_mask
);
1867 extern long sched_getaffinity(pid_t pid
, cpumask_t
*mask
);
1869 extern int sched_mc_power_savings
, sched_smt_power_savings
;
1871 extern void normalize_rt_tasks(void);
1873 #ifdef CONFIG_FAIR_GROUP_SCHED
1875 extern struct task_group init_task_group
;
1877 extern struct task_group
*sched_create_group(void);
1878 extern void sched_destroy_group(struct task_group
*tg
);
1879 extern void sched_move_task(struct task_struct
*tsk
);
1880 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
1881 extern unsigned long sched_group_shares(struct task_group
*tg
);
1885 #ifdef CONFIG_TASK_XACCT
1886 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1891 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1896 static inline void inc_syscr(struct task_struct
*tsk
)
1901 static inline void inc_syscw(struct task_struct
*tsk
)
1906 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
1910 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
1914 static inline void inc_syscr(struct task_struct
*tsk
)
1918 static inline void inc_syscw(struct task_struct
*tsk
)
1923 #endif /* __KERNEL__ */