7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 #define CLONE_NEWNET 0x40000000 /* New network namespace */
30 #define CLONE_IO 0x80000000 /* Clone io context */
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>
61 #include <linux/mm_types.h>
63 #include <asm/system.h>
65 #include <asm/ptrace.h>
66 #include <asm/cputime.h>
68 #include <linux/smp.h>
69 #include <linux/sem.h>
70 #include <linux/signal.h>
71 #include <linux/path.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/rtmutex.h>
82 #include <linux/time.h>
83 #include <linux/param.h>
84 #include <linux/resource.h>
85 #include <linux/timer.h>
86 #include <linux/hrtimer.h>
87 #include <linux/task_io_accounting.h>
88 #include <linux/kobject.h>
89 #include <linux/latencytop.h>
90 #include <linux/cred.h>
92 #include <asm/processor.h>
96 struct futex_pi_state
;
97 struct robust_list_head
;
101 struct perf_counter_context
;
104 * List of flags we want to share for kernel threads,
105 * if only because they are not used by them anyway.
107 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
110 * These are the constant used to fake the fixed-point load-average
111 * counting. Some notes:
112 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
113 * a load-average precision of 10 bits integer + 11 bits fractional
114 * - if you want to count load-averages more often, you need more
115 * precision, or rounding will get you. With 2-second counting freq,
116 * the EXP_n values would be 1981, 2034 and 2043 if still using only
119 extern unsigned long avenrun
[]; /* Load averages */
121 #define FSHIFT 11 /* nr of bits of precision */
122 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
123 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
124 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
125 #define EXP_5 2014 /* 1/exp(5sec/5min) */
126 #define EXP_15 2037 /* 1/exp(5sec/15min) */
128 #define CALC_LOAD(load,exp,n) \
130 load += n*(FIXED_1-exp); \
133 extern unsigned long total_forks
;
134 extern int nr_threads
;
135 DECLARE_PER_CPU(unsigned long, process_counts
);
136 extern int nr_processes(void);
137 extern unsigned long nr_running(void);
138 extern unsigned long nr_uninterruptible(void);
139 extern unsigned long nr_active(void);
140 extern unsigned long nr_iowait(void);
141 extern u64
cpu_nr_migrations(int cpu
);
143 extern unsigned long get_parent_ip(unsigned long addr
);
148 #ifdef CONFIG_SCHED_DEBUG
149 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
150 extern void proc_sched_set_task(struct task_struct
*p
);
152 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
);
155 proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
)
158 static inline void proc_sched_set_task(struct task_struct
*p
)
162 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
)
167 extern unsigned long long time_sync_thresh
;
170 * Task state bitmask. NOTE! These bits are also
171 * encoded in fs/proc/array.c: get_task_state().
173 * We have two separate sets of flags: task->state
174 * is about runnability, while task->exit_state are
175 * about the task exiting. Confusing, but this way
176 * modifying one set can't modify the other one by
179 #define TASK_RUNNING 0
180 #define TASK_INTERRUPTIBLE 1
181 #define TASK_UNINTERRUPTIBLE 2
182 #define __TASK_STOPPED 4
183 #define __TASK_TRACED 8
184 /* in tsk->exit_state */
185 #define EXIT_ZOMBIE 16
187 /* in tsk->state again */
189 #define TASK_WAKEKILL 128
191 /* Convenience macros for the sake of set_task_state */
192 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
193 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
194 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
196 /* Convenience macros for the sake of wake_up */
197 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
198 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
200 /* get_task_state() */
201 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
202 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
205 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
206 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
207 #define task_is_stopped_or_traced(task) \
208 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
209 #define task_contributes_to_load(task) \
210 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
211 (task->flags & PF_FROZEN) == 0)
213 #define __set_task_state(tsk, state_value) \
214 do { (tsk)->state = (state_value); } while (0)
215 #define set_task_state(tsk, state_value) \
216 set_mb((tsk)->state, (state_value))
219 * set_current_state() includes a barrier so that the write of current->state
220 * is correctly serialised wrt the caller's subsequent test of whether to
223 * set_current_state(TASK_UNINTERRUPTIBLE);
224 * if (do_i_need_to_sleep())
227 * If the caller does not need such serialisation then use __set_current_state()
229 #define __set_current_state(state_value) \
230 do { current->state = (state_value); } while (0)
231 #define set_current_state(state_value) \
232 set_mb(current->state, (state_value))
234 /* Task command name length */
235 #define TASK_COMM_LEN 16
237 #include <linux/spinlock.h>
240 * This serializes "schedule()" and also protects
241 * the run-queue from deletions/modifications (but
242 * _adding_ to the beginning of the run-queue has
245 extern rwlock_t tasklist_lock
;
246 extern spinlock_t mmlist_lock
;
250 extern void sched_init(void);
251 extern void sched_init_smp(void);
252 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
253 extern void init_idle(struct task_struct
*idle
, int cpu
);
254 extern void init_idle_bootup_task(struct task_struct
*idle
);
256 extern int runqueue_is_locked(void);
257 extern void task_rq_unlock_wait(struct task_struct
*p
);
259 extern cpumask_var_t nohz_cpu_mask
;
260 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
261 extern int select_nohz_load_balancer(int cpu
);
263 static inline int select_nohz_load_balancer(int cpu
)
270 * Only dump TASK_* tasks. (0 for all tasks)
272 extern void show_state_filter(unsigned long state_filter
);
274 static inline void show_state(void)
276 show_state_filter(0);
279 extern void show_regs(struct pt_regs
*);
282 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
283 * task), SP is the stack pointer of the first frame that should be shown in the back
284 * trace (or NULL if the entire call-chain of the task should be shown).
286 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
288 void io_schedule(void);
289 long io_schedule_timeout(long timeout
);
291 extern void cpu_init (void);
292 extern void trap_init(void);
293 extern void update_process_times(int user
);
294 extern void scheduler_tick(void);
296 extern void sched_show_task(struct task_struct
*p
);
298 #ifdef CONFIG_DETECT_SOFTLOCKUP
299 extern void softlockup_tick(void);
300 extern void touch_softlockup_watchdog(void);
301 extern void touch_all_softlockup_watchdogs(void);
302 extern int proc_dosoftlockup_thresh(struct ctl_table
*table
, int write
,
303 struct file
*filp
, void __user
*buffer
,
304 size_t *lenp
, loff_t
*ppos
);
305 extern unsigned int softlockup_panic
;
306 extern int softlockup_thresh
;
308 static inline void softlockup_tick(void)
311 static inline void touch_softlockup_watchdog(void)
314 static inline void touch_all_softlockup_watchdogs(void)
319 #ifdef CONFIG_DETECT_HUNG_TASK
320 extern unsigned int sysctl_hung_task_panic
;
321 extern unsigned long sysctl_hung_task_check_count
;
322 extern unsigned long sysctl_hung_task_timeout_secs
;
323 extern unsigned long sysctl_hung_task_warnings
;
324 extern int proc_dohung_task_timeout_secs(struct ctl_table
*table
, int write
,
325 struct file
*filp
, void __user
*buffer
,
326 size_t *lenp
, loff_t
*ppos
);
329 /* Attach to any functions which should be ignored in wchan output. */
330 #define __sched __attribute__((__section__(".sched.text")))
332 /* Linker adds these: start and end of __sched functions */
333 extern char __sched_text_start
[], __sched_text_end
[];
335 /* Is this address in the __sched functions? */
336 extern int in_sched_functions(unsigned long addr
);
338 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
339 extern signed long schedule_timeout(signed long timeout
);
340 extern signed long schedule_timeout_interruptible(signed long timeout
);
341 extern signed long schedule_timeout_killable(signed long timeout
);
342 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
343 asmlinkage
void __schedule(void);
344 asmlinkage
void schedule(void);
345 extern int mutex_spin_on_owner(struct mutex
*lock
, struct thread_info
*owner
);
348 struct user_namespace
;
350 /* Maximum number of active map areas.. This is a random (large) number */
351 #define DEFAULT_MAX_MAP_COUNT 65536
353 extern int sysctl_max_map_count
;
355 #include <linux/aio.h>
358 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
359 unsigned long, unsigned long);
361 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
362 unsigned long len
, unsigned long pgoff
,
363 unsigned long flags
);
364 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
365 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
367 #if USE_SPLIT_PTLOCKS
369 * The mm counters are not protected by its page_table_lock,
370 * so must be incremented atomically.
372 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
373 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
374 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
375 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
376 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
378 #else /* !USE_SPLIT_PTLOCKS */
380 * The mm counters are protected by its page_table_lock,
381 * so can be incremented directly.
383 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
384 #define get_mm_counter(mm, member) ((mm)->_##member)
385 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
386 #define inc_mm_counter(mm, member) (mm)->_##member++
387 #define dec_mm_counter(mm, member) (mm)->_##member--
389 #endif /* !USE_SPLIT_PTLOCKS */
391 #define get_mm_rss(mm) \
392 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
393 #define update_hiwater_rss(mm) do { \
394 unsigned long _rss = get_mm_rss(mm); \
395 if ((mm)->hiwater_rss < _rss) \
396 (mm)->hiwater_rss = _rss; \
398 #define update_hiwater_vm(mm) do { \
399 if ((mm)->hiwater_vm < (mm)->total_vm) \
400 (mm)->hiwater_vm = (mm)->total_vm; \
403 static inline unsigned long get_mm_hiwater_rss(struct mm_struct
*mm
)
405 return max(mm
->hiwater_rss
, get_mm_rss(mm
));
408 static inline unsigned long get_mm_hiwater_vm(struct mm_struct
*mm
)
410 return max(mm
->hiwater_vm
, mm
->total_vm
);
413 extern void set_dumpable(struct mm_struct
*mm
, int value
);
414 extern int get_dumpable(struct mm_struct
*mm
);
418 #define MMF_DUMPABLE 0 /* core dump is permitted */
419 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
420 #define MMF_DUMPABLE_BITS 2
422 /* coredump filter bits */
423 #define MMF_DUMP_ANON_PRIVATE 2
424 #define MMF_DUMP_ANON_SHARED 3
425 #define MMF_DUMP_MAPPED_PRIVATE 4
426 #define MMF_DUMP_MAPPED_SHARED 5
427 #define MMF_DUMP_ELF_HEADERS 6
428 #define MMF_DUMP_HUGETLB_PRIVATE 7
429 #define MMF_DUMP_HUGETLB_SHARED 8
430 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
431 #define MMF_DUMP_FILTER_BITS 7
432 #define MMF_DUMP_FILTER_MASK \
433 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
434 #define MMF_DUMP_FILTER_DEFAULT \
435 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
436 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
438 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
439 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
441 # define MMF_DUMP_MASK_DEFAULT_ELF 0
444 struct sighand_struct
{
446 struct k_sigaction action
[_NSIG
];
448 wait_queue_head_t signalfd_wqh
;
451 struct pacct_struct
{
454 unsigned long ac_mem
;
455 cputime_t ac_utime
, ac_stime
;
456 unsigned long ac_minflt
, ac_majflt
;
460 * struct task_cputime - collected CPU time counts
461 * @utime: time spent in user mode, in &cputime_t units
462 * @stime: time spent in kernel mode, in &cputime_t units
463 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
465 * This structure groups together three kinds of CPU time that are
466 * tracked for threads and thread groups. Most things considering
467 * CPU time want to group these counts together and treat all three
468 * of them in parallel.
470 struct task_cputime
{
473 unsigned long long sum_exec_runtime
;
475 /* Alternate field names when used to cache expirations. */
476 #define prof_exp stime
477 #define virt_exp utime
478 #define sched_exp sum_exec_runtime
480 #define INIT_CPUTIME \
481 (struct task_cputime) { \
482 .utime = cputime_zero, \
483 .stime = cputime_zero, \
484 .sum_exec_runtime = 0, \
488 * struct thread_group_cputimer - thread group interval timer counts
489 * @cputime: thread group interval timers.
490 * @running: non-zero when there are timers running and
491 * @cputime receives updates.
492 * @lock: lock for fields in this struct.
494 * This structure contains the version of task_cputime, above, that is
495 * used for thread group CPU timer calculations.
497 struct thread_group_cputimer
{
498 struct task_cputime cputime
;
504 * NOTE! "signal_struct" does not have it's own
505 * locking, because a shared signal_struct always
506 * implies a shared sighand_struct, so locking
507 * sighand_struct is always a proper superset of
508 * the locking of signal_struct.
510 struct signal_struct
{
514 wait_queue_head_t wait_chldexit
; /* for wait4() */
516 /* current thread group signal load-balancing target: */
517 struct task_struct
*curr_target
;
519 /* shared signal handling: */
520 struct sigpending shared_pending
;
522 /* thread group exit support */
525 * - notify group_exit_task when ->count is equal to notify_count
526 * - everyone except group_exit_task is stopped during signal delivery
527 * of fatal signals, group_exit_task processes the signal.
530 struct task_struct
*group_exit_task
;
532 /* thread group stop support, overloads group_exit_code too */
533 int group_stop_count
;
534 unsigned int flags
; /* see SIGNAL_* flags below */
536 /* POSIX.1b Interval Timers */
537 struct list_head posix_timers
;
539 /* ITIMER_REAL timer for the process */
540 struct hrtimer real_timer
;
541 struct pid
*leader_pid
;
542 ktime_t it_real_incr
;
544 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
545 cputime_t it_prof_expires
, it_virt_expires
;
546 cputime_t it_prof_incr
, it_virt_incr
;
549 * Thread group totals for process CPU timers.
550 * See thread_group_cputimer(), et al, for details.
552 struct thread_group_cputimer cputimer
;
554 /* Earliest-expiration cache. */
555 struct task_cputime cputime_expires
;
557 struct list_head cpu_timers
[3];
559 struct pid
*tty_old_pgrp
;
561 /* boolean value for session group leader */
564 struct tty_struct
*tty
; /* NULL if no tty */
567 * Cumulative resource counters for dead threads in the group,
568 * and for reaped dead child processes forked by this group.
569 * Live threads maintain their own counters and add to these
570 * in __exit_signal, except for the group leader.
572 cputime_t utime
, stime
, cutime
, cstime
;
575 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
576 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
577 unsigned long inblock
, oublock
, cinblock
, coublock
;
578 struct task_io_accounting ioac
;
581 * Cumulative ns of schedule CPU time fo dead threads in the
582 * group, not including a zombie group leader, (This only differs
583 * from jiffies_to_ns(utime + stime) if sched_clock uses something
584 * other than jiffies.)
586 unsigned long long sum_sched_runtime
;
589 * We don't bother to synchronize most readers of this at all,
590 * because there is no reader checking a limit that actually needs
591 * to get both rlim_cur and rlim_max atomically, and either one
592 * alone is a single word that can safely be read normally.
593 * getrlimit/setrlimit use task_lock(current->group_leader) to
594 * protect this instead of the siglock, because they really
595 * have no need to disable irqs.
597 struct rlimit rlim
[RLIM_NLIMITS
];
599 #ifdef CONFIG_BSD_PROCESS_ACCT
600 struct pacct_struct pacct
; /* per-process accounting information */
602 #ifdef CONFIG_TASKSTATS
603 struct taskstats
*stats
;
607 struct tty_audit_buf
*tty_audit_buf
;
611 /* Context switch must be unlocked if interrupts are to be enabled */
612 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
613 # define __ARCH_WANT_UNLOCKED_CTXSW
617 * Bits in flags field of signal_struct.
619 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
620 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
621 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
622 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
624 * Pending notifications to parent.
626 #define SIGNAL_CLD_STOPPED 0x00000010
627 #define SIGNAL_CLD_CONTINUED 0x00000020
628 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
630 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
632 /* If true, all threads except ->group_exit_task have pending SIGKILL */
633 static inline int signal_group_exit(const struct signal_struct
*sig
)
635 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
636 (sig
->group_exit_task
!= NULL
);
640 * Some day this will be a full-fledged user tracking system..
643 atomic_t __count
; /* reference count */
644 atomic_t processes
; /* How many processes does this user have? */
645 atomic_t files
; /* How many open files does this user have? */
646 atomic_t sigpending
; /* How many pending signals does this user have? */
647 #ifdef CONFIG_INOTIFY_USER
648 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
649 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
652 atomic_t epoll_watches
; /* The number of file descriptors currently watched */
654 #ifdef CONFIG_POSIX_MQUEUE
655 /* protected by mq_lock */
656 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
658 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
661 struct key
*uid_keyring
; /* UID specific keyring */
662 struct key
*session_keyring
; /* UID's default session keyring */
665 /* Hash table maintenance information */
666 struct hlist_node uidhash_node
;
668 struct user_namespace
*user_ns
;
670 #ifdef CONFIG_USER_SCHED
671 struct task_group
*tg
;
674 struct work_struct work
;
678 #ifdef CONFIG_PERF_COUNTERS
679 atomic_long_t locked_vm
;
683 extern int uids_sysfs_init(void);
685 extern struct user_struct
*find_user(uid_t
);
687 extern struct user_struct root_user
;
688 #define INIT_USER (&root_user)
691 struct backing_dev_info
;
692 struct reclaim_state
;
694 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
696 /* cumulative counters */
697 unsigned long pcount
; /* # of times run on this cpu */
698 unsigned long long run_delay
; /* time spent waiting on a runqueue */
701 unsigned long long last_arrival
,/* when we last ran on a cpu */
702 last_queued
; /* when we were last queued to run */
703 #ifdef CONFIG_SCHEDSTATS
705 unsigned int bkl_count
;
708 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
710 #ifdef CONFIG_TASK_DELAY_ACCT
711 struct task_delay_info
{
713 unsigned int flags
; /* Private per-task flags */
715 /* For each stat XXX, add following, aligned appropriately
717 * struct timespec XXX_start, XXX_end;
721 * Atomicity of updates to XXX_delay, XXX_count protected by
722 * single lock above (split into XXX_lock if contention is an issue).
726 * XXX_count is incremented on every XXX operation, the delay
727 * associated with the operation is added to XXX_delay.
728 * XXX_delay contains the accumulated delay time in nanoseconds.
730 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
731 u64 blkio_delay
; /* wait for sync block io completion */
732 u64 swapin_delay
; /* wait for swapin block io completion */
733 u32 blkio_count
; /* total count of the number of sync block */
734 /* io operations performed */
735 u32 swapin_count
; /* total count of the number of swapin block */
736 /* io operations performed */
738 struct timespec freepages_start
, freepages_end
;
739 u64 freepages_delay
; /* wait for memory reclaim */
740 u32 freepages_count
; /* total count of memory reclaim */
742 #endif /* CONFIG_TASK_DELAY_ACCT */
744 static inline int sched_info_on(void)
746 #ifdef CONFIG_SCHEDSTATS
748 #elif defined(CONFIG_TASK_DELAY_ACCT)
749 extern int delayacct_on
;
764 * sched-domains (multiprocessor balancing) declarations:
768 * Increase resolution of nice-level calculations:
770 #define SCHED_LOAD_SHIFT 10
771 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
773 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
776 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
777 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
778 #define SD_BALANCE_EXEC 4 /* Balance on exec */
779 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
780 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
781 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
782 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
783 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
784 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
785 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
786 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
787 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
789 enum powersavings_balance_level
{
790 POWERSAVINGS_BALANCE_NONE
= 0, /* No power saving load balance */
791 POWERSAVINGS_BALANCE_BASIC
, /* Fill one thread/core/package
792 * first for long running threads
794 POWERSAVINGS_BALANCE_WAKEUP
, /* Also bias task wakeups to semi-idle
795 * cpu package for power savings
797 MAX_POWERSAVINGS_BALANCE_LEVELS
800 extern int sched_mc_power_savings
, sched_smt_power_savings
;
802 static inline int sd_balance_for_mc_power(void)
804 if (sched_smt_power_savings
)
805 return SD_POWERSAVINGS_BALANCE
;
810 static inline int sd_balance_for_package_power(void)
812 if (sched_mc_power_savings
| sched_smt_power_savings
)
813 return SD_POWERSAVINGS_BALANCE
;
819 * Optimise SD flags for power savings:
820 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
821 * Keep default SD flags if sched_{smt,mc}_power_saving=0
824 static inline int sd_power_saving_flags(void)
826 if (sched_mc_power_savings
| sched_smt_power_savings
)
827 return SD_BALANCE_NEWIDLE
;
833 struct sched_group
*next
; /* Must be a circular list */
836 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
837 * single CPU. This is read only (except for setup, hotplug CPU).
838 * Note : Never change cpu_power without recompute its reciprocal
840 unsigned int __cpu_power
;
842 * reciprocal value of cpu_power to avoid expensive divides
843 * (see include/linux/reciprocal_div.h)
845 u32 reciprocal_cpu_power
;
847 unsigned long cpumask
[];
850 static inline struct cpumask
*sched_group_cpus(struct sched_group
*sg
)
852 return to_cpumask(sg
->cpumask
);
855 enum sched_domain_level
{
865 struct sched_domain_attr
{
866 int relax_domain_level
;
869 #define SD_ATTR_INIT (struct sched_domain_attr) { \
870 .relax_domain_level = -1, \
873 struct sched_domain
{
874 /* These fields must be setup */
875 struct sched_domain
*parent
; /* top domain must be null terminated */
876 struct sched_domain
*child
; /* bottom domain must be null terminated */
877 struct sched_group
*groups
; /* the balancing groups of the domain */
878 unsigned long min_interval
; /* Minimum balance interval ms */
879 unsigned long max_interval
; /* Maximum balance interval ms */
880 unsigned int busy_factor
; /* less balancing by factor if busy */
881 unsigned int imbalance_pct
; /* No balance until over watermark */
882 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
883 unsigned int busy_idx
;
884 unsigned int idle_idx
;
885 unsigned int newidle_idx
;
886 unsigned int wake_idx
;
887 unsigned int forkexec_idx
;
888 int flags
; /* See SD_* */
889 enum sched_domain_level level
;
891 /* Runtime fields. */
892 unsigned long last_balance
; /* init to jiffies. units in jiffies */
893 unsigned int balance_interval
; /* initialise to 1. units in ms. */
894 unsigned int nr_balance_failed
; /* initialise to 0 */
898 #ifdef CONFIG_SCHEDSTATS
899 /* load_balance() stats */
900 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
901 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
902 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
903 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
904 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
905 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
906 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
907 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
909 /* Active load balancing */
910 unsigned int alb_count
;
911 unsigned int alb_failed
;
912 unsigned int alb_pushed
;
914 /* SD_BALANCE_EXEC stats */
915 unsigned int sbe_count
;
916 unsigned int sbe_balanced
;
917 unsigned int sbe_pushed
;
919 /* SD_BALANCE_FORK stats */
920 unsigned int sbf_count
;
921 unsigned int sbf_balanced
;
922 unsigned int sbf_pushed
;
924 /* try_to_wake_up() stats */
925 unsigned int ttwu_wake_remote
;
926 unsigned int ttwu_move_affine
;
927 unsigned int ttwu_move_balance
;
929 #ifdef CONFIG_SCHED_DEBUG
933 /* span of all CPUs in this domain */
934 unsigned long span
[];
937 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
939 return to_cpumask(sd
->span
);
942 extern void partition_sched_domains(int ndoms_new
, struct cpumask
*doms_new
,
943 struct sched_domain_attr
*dattr_new
);
945 /* Test a flag in parent sched domain */
946 static inline int test_sd_parent(struct sched_domain
*sd
, int flag
)
948 if (sd
->parent
&& (sd
->parent
->flags
& flag
))
954 #else /* CONFIG_SMP */
956 struct sched_domain_attr
;
959 partition_sched_domains(int ndoms_new
, struct cpumask
*doms_new
,
960 struct sched_domain_attr
*dattr_new
)
963 #endif /* !CONFIG_SMP */
965 struct io_context
; /* See blkdev.h */
968 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
969 extern void prefetch_stack(struct task_struct
*t
);
971 static inline void prefetch_stack(struct task_struct
*t
) { }
974 struct audit_context
; /* See audit.c */
976 struct pipe_inode_info
;
977 struct uts_namespace
;
983 const struct sched_class
*next
;
985 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int wakeup
);
986 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int sleep
);
987 void (*yield_task
) (struct rq
*rq
);
989 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
, int sync
);
991 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
992 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
995 int (*select_task_rq
)(struct task_struct
*p
, int sync
);
997 unsigned long (*load_balance
) (struct rq
*this_rq
, int this_cpu
,
998 struct rq
*busiest
, unsigned long max_load_move
,
999 struct sched_domain
*sd
, enum cpu_idle_type idle
,
1000 int *all_pinned
, int *this_best_prio
);
1002 int (*move_one_task
) (struct rq
*this_rq
, int this_cpu
,
1003 struct rq
*busiest
, struct sched_domain
*sd
,
1004 enum cpu_idle_type idle
);
1005 void (*pre_schedule
) (struct rq
*this_rq
, struct task_struct
*task
);
1006 int (*needs_post_schedule
) (struct rq
*this_rq
);
1007 void (*post_schedule
) (struct rq
*this_rq
);
1008 void (*task_wake_up
) (struct rq
*this_rq
, struct task_struct
*task
);
1010 void (*set_cpus_allowed
)(struct task_struct
*p
,
1011 const struct cpumask
*newmask
);
1013 void (*rq_online
)(struct rq
*rq
);
1014 void (*rq_offline
)(struct rq
*rq
);
1017 void (*set_curr_task
) (struct rq
*rq
);
1018 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
, int queued
);
1019 void (*task_new
) (struct rq
*rq
, struct task_struct
*p
);
1021 void (*switched_from
) (struct rq
*this_rq
, struct task_struct
*task
,
1023 void (*switched_to
) (struct rq
*this_rq
, struct task_struct
*task
,
1025 void (*prio_changed
) (struct rq
*this_rq
, struct task_struct
*task
,
1026 int oldprio
, int running
);
1028 #ifdef CONFIG_FAIR_GROUP_SCHED
1029 void (*moved_group
) (struct task_struct
*p
);
1033 struct load_weight
{
1034 unsigned long weight
, inv_weight
;
1038 * CFS stats for a schedulable entity (task, task-group etc)
1040 * Current field usage histogram:
1047 struct sched_entity
{
1048 struct load_weight load
; /* for load-balancing */
1049 struct rb_node run_node
;
1050 struct list_head group_node
;
1054 u64 sum_exec_runtime
;
1056 u64 prev_sum_exec_runtime
;
1066 #ifdef CONFIG_SCHEDSTATS
1074 s64 sum_sleep_runtime
;
1081 u64 nr_migrations_cold
;
1082 u64 nr_failed_migrations_affine
;
1083 u64 nr_failed_migrations_running
;
1084 u64 nr_failed_migrations_hot
;
1085 u64 nr_forced_migrations
;
1086 u64 nr_forced2_migrations
;
1089 u64 nr_wakeups_sync
;
1090 u64 nr_wakeups_migrate
;
1091 u64 nr_wakeups_local
;
1092 u64 nr_wakeups_remote
;
1093 u64 nr_wakeups_affine
;
1094 u64 nr_wakeups_affine_attempts
;
1095 u64 nr_wakeups_passive
;
1096 u64 nr_wakeups_idle
;
1099 #ifdef CONFIG_FAIR_GROUP_SCHED
1100 struct sched_entity
*parent
;
1101 /* rq on which this entity is (to be) queued: */
1102 struct cfs_rq
*cfs_rq
;
1103 /* rq "owned" by this entity/group: */
1104 struct cfs_rq
*my_q
;
1108 struct sched_rt_entity
{
1109 struct list_head run_list
;
1110 unsigned long timeout
;
1111 unsigned int time_slice
;
1112 int nr_cpus_allowed
;
1114 struct sched_rt_entity
*back
;
1115 #ifdef CONFIG_RT_GROUP_SCHED
1116 struct sched_rt_entity
*parent
;
1117 /* rq on which this entity is (to be) queued: */
1118 struct rt_rq
*rt_rq
;
1119 /* rq "owned" by this entity/group: */
1124 struct task_struct
{
1125 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1128 unsigned int flags
; /* per process flags, defined below */
1129 unsigned int ptrace
;
1131 int lock_depth
; /* BKL lock depth */
1134 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1139 int prio
, static_prio
, normal_prio
;
1140 unsigned int rt_priority
;
1141 const struct sched_class
*sched_class
;
1142 struct sched_entity se
;
1143 struct sched_rt_entity rt
;
1145 #ifdef CONFIG_PREEMPT_NOTIFIERS
1146 /* list of struct preempt_notifier: */
1147 struct hlist_head preempt_notifiers
;
1151 * fpu_counter contains the number of consecutive context switches
1152 * that the FPU is used. If this is over a threshold, the lazy fpu
1153 * saving becomes unlazy to save the trap. This is an unsigned char
1154 * so that after 256 times the counter wraps and the behavior turns
1155 * lazy again; this to deal with bursty apps that only use FPU for
1158 unsigned char fpu_counter
;
1159 s8 oomkilladj
; /* OOM kill score adjustment (bit shift). */
1160 #ifdef CONFIG_BLK_DEV_IO_TRACE
1161 unsigned int btrace_seq
;
1164 unsigned int policy
;
1165 cpumask_t cpus_allowed
;
1167 #ifdef CONFIG_PREEMPT_RCU
1168 int rcu_read_lock_nesting
;
1169 int rcu_flipctr_idx
;
1170 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1172 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1173 struct sched_info sched_info
;
1176 struct list_head tasks
;
1177 struct plist_node pushable_tasks
;
1179 struct mm_struct
*mm
, *active_mm
;
1182 struct linux_binfmt
*binfmt
;
1184 int exit_code
, exit_signal
;
1185 int pdeath_signal
; /* The signal sent when the parent dies */
1187 unsigned int personality
;
1188 unsigned did_exec
:1;
1189 unsigned in_execve
:1; /* Tell the LSMs that the process is doing an
1194 /* Canary value for the -fstack-protector gcc feature */
1195 unsigned long stack_canary
;
1198 * pointers to (original) parent process, youngest child, younger sibling,
1199 * older sibling, respectively. (p->father can be replaced with
1200 * p->real_parent->pid)
1202 struct task_struct
*real_parent
; /* real parent process */
1203 struct task_struct
*parent
; /* recipient of SIGCHLD, wait4() reports */
1205 * children/sibling forms the list of my natural children
1207 struct list_head children
; /* list of my children */
1208 struct list_head sibling
; /* linkage in my parent's children list */
1209 struct task_struct
*group_leader
; /* threadgroup leader */
1212 * ptraced is the list of tasks this task is using ptrace on.
1213 * This includes both natural children and PTRACE_ATTACH targets.
1214 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1216 struct list_head ptraced
;
1217 struct list_head ptrace_entry
;
1219 #ifdef CONFIG_X86_PTRACE_BTS
1221 * This is the tracer handle for the ptrace BTS extension.
1222 * This field actually belongs to the ptracer task.
1224 struct bts_tracer
*bts
;
1226 * The buffer to hold the BTS data.
1230 #endif /* CONFIG_X86_PTRACE_BTS */
1232 /* PID/PID hash table linkage. */
1233 struct pid_link pids
[PIDTYPE_MAX
];
1234 struct list_head thread_group
;
1236 struct completion
*vfork_done
; /* for vfork() */
1237 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1238 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1240 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1242 cputime_t prev_utime
, prev_stime
;
1243 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1244 struct timespec start_time
; /* monotonic time */
1245 struct timespec real_start_time
; /* boot based time */
1246 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1247 unsigned long min_flt
, maj_flt
;
1249 struct task_cputime cputime_expires
;
1250 struct list_head cpu_timers
[3];
1252 /* process credentials */
1253 const struct cred
*real_cred
; /* objective and real subjective task
1254 * credentials (COW) */
1255 const struct cred
*cred
; /* effective (overridable) subjective task
1256 * credentials (COW) */
1257 struct mutex cred_exec_mutex
; /* execve vs ptrace cred calculation mutex */
1259 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1260 - access with [gs]et_task_comm (which lock
1261 it with task_lock())
1262 - initialized normally by flush_old_exec */
1263 /* file system info */
1264 int link_count
, total_link_count
;
1265 #ifdef CONFIG_SYSVIPC
1267 struct sysv_sem sysvsem
;
1269 #ifdef CONFIG_DETECT_HUNG_TASK
1270 /* hung task detection */
1271 unsigned long last_switch_count
;
1273 /* CPU-specific state of this task */
1274 struct thread_struct thread
;
1275 /* filesystem information */
1276 struct fs_struct
*fs
;
1277 /* open file information */
1278 struct files_struct
*files
;
1280 struct nsproxy
*nsproxy
;
1281 /* signal handlers */
1282 struct signal_struct
*signal
;
1283 struct sighand_struct
*sighand
;
1285 sigset_t blocked
, real_blocked
;
1286 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1287 struct sigpending pending
;
1289 unsigned long sas_ss_sp
;
1291 int (*notifier
)(void *priv
);
1292 void *notifier_data
;
1293 sigset_t
*notifier_mask
;
1294 struct audit_context
*audit_context
;
1295 #ifdef CONFIG_AUDITSYSCALL
1297 unsigned int sessionid
;
1301 /* Thread group tracking */
1304 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1305 spinlock_t alloc_lock
;
1307 #ifdef CONFIG_GENERIC_HARDIRQS
1308 /* IRQ handler threads */
1309 struct irqaction
*irqaction
;
1312 /* Protection of the PI data structures: */
1315 #ifdef CONFIG_RT_MUTEXES
1316 /* PI waiters blocked on a rt_mutex held by this task */
1317 struct plist_head pi_waiters
;
1318 /* Deadlock detection and priority inheritance handling */
1319 struct rt_mutex_waiter
*pi_blocked_on
;
1322 #ifdef CONFIG_DEBUG_MUTEXES
1323 /* mutex deadlock detection */
1324 struct mutex_waiter
*blocked_on
;
1326 #ifdef CONFIG_TRACE_IRQFLAGS
1327 unsigned int irq_events
;
1328 int hardirqs_enabled
;
1329 unsigned long hardirq_enable_ip
;
1330 unsigned int hardirq_enable_event
;
1331 unsigned long hardirq_disable_ip
;
1332 unsigned int hardirq_disable_event
;
1333 int softirqs_enabled
;
1334 unsigned long softirq_disable_ip
;
1335 unsigned int softirq_disable_event
;
1336 unsigned long softirq_enable_ip
;
1337 unsigned int softirq_enable_event
;
1338 int hardirq_context
;
1339 int softirq_context
;
1341 #ifdef CONFIG_LOCKDEP
1342 # define MAX_LOCK_DEPTH 48UL
1345 unsigned int lockdep_recursion
;
1346 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1347 gfp_t lockdep_reclaim_gfp
;
1350 /* journalling filesystem info */
1353 /* stacked block device info */
1354 struct bio
*bio_list
, **bio_tail
;
1357 struct reclaim_state
*reclaim_state
;
1359 struct backing_dev_info
*backing_dev_info
;
1361 struct io_context
*io_context
;
1363 unsigned long ptrace_message
;
1364 siginfo_t
*last_siginfo
; /* For ptrace use. */
1365 struct task_io_accounting ioac
;
1366 #if defined(CONFIG_TASK_XACCT)
1367 u64 acct_rss_mem1
; /* accumulated rss usage */
1368 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1369 cputime_t acct_timexpd
; /* stime + utime since last update */
1371 #ifdef CONFIG_CPUSETS
1372 nodemask_t mems_allowed
;
1373 int cpuset_mems_generation
;
1374 int cpuset_mem_spread_rotor
;
1376 #ifdef CONFIG_CGROUPS
1377 /* Control Group info protected by css_set_lock */
1378 struct css_set
*cgroups
;
1379 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1380 struct list_head cg_list
;
1383 struct robust_list_head __user
*robust_list
;
1384 #ifdef CONFIG_COMPAT
1385 struct compat_robust_list_head __user
*compat_robust_list
;
1387 struct list_head pi_state_list
;
1388 struct futex_pi_state
*pi_state_cache
;
1390 #ifdef CONFIG_PERF_COUNTERS
1391 struct perf_counter_context
*perf_counter_ctxp
;
1394 struct mempolicy
*mempolicy
;
1397 atomic_t fs_excl
; /* holding fs exclusive resources */
1398 struct rcu_head rcu
;
1401 * cache last used pipe for splice
1403 struct pipe_inode_info
*splice_pipe
;
1404 #ifdef CONFIG_TASK_DELAY_ACCT
1405 struct task_delay_info
*delays
;
1407 #ifdef CONFIG_FAULT_INJECTION
1410 struct prop_local_single dirties
;
1411 #ifdef CONFIG_LATENCYTOP
1412 int latency_record_count
;
1413 struct latency_record latency_record
[LT_SAVECOUNT
];
1416 * time slack values; these are used to round up poll() and
1417 * select() etc timeout values. These are in nanoseconds.
1419 unsigned long timer_slack_ns
;
1420 unsigned long default_timer_slack_ns
;
1422 struct list_head
*scm_work_list
;
1423 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1424 /* Index of current stored adress in ret_stack */
1426 /* Stack of return addresses for return function tracing */
1427 struct ftrace_ret_stack
*ret_stack
;
1428 /* time stamp for last schedule */
1429 unsigned long long ftrace_timestamp
;
1431 * Number of functions that haven't been traced
1432 * because of depth overrun.
1434 atomic_t trace_overrun
;
1435 /* Pause for the tracing */
1436 atomic_t tracing_graph_pause
;
1438 #ifdef CONFIG_TRACING
1439 /* state flags for use by tracers */
1440 unsigned long trace
;
1444 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1445 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1448 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1449 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1450 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1451 * values are inverted: lower p->prio value means higher priority.
1453 * The MAX_USER_RT_PRIO value allows the actual maximum
1454 * RT priority to be separate from the value exported to
1455 * user-space. This allows kernel threads to set their
1456 * priority to a value higher than any user task. Note:
1457 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1460 #define MAX_USER_RT_PRIO 100
1461 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1463 #define MAX_PRIO (MAX_RT_PRIO + 40)
1464 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1466 static inline int rt_prio(int prio
)
1468 if (unlikely(prio
< MAX_RT_PRIO
))
1473 static inline int rt_task(struct task_struct
*p
)
1475 return rt_prio(p
->prio
);
1478 static inline struct pid
*task_pid(struct task_struct
*task
)
1480 return task
->pids
[PIDTYPE_PID
].pid
;
1483 static inline struct pid
*task_tgid(struct task_struct
*task
)
1485 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1489 * Without tasklist or rcu lock it is not safe to dereference
1490 * the result of task_pgrp/task_session even if task == current,
1491 * we can race with another thread doing sys_setsid/sys_setpgid.
1493 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1495 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1498 static inline struct pid
*task_session(struct task_struct
*task
)
1500 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1503 struct pid_namespace
;
1506 * the helpers to get the task's different pids as they are seen
1507 * from various namespaces
1509 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1510 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1512 * task_xid_nr_ns() : id seen from the ns specified;
1514 * set_task_vxid() : assigns a virtual id to a task;
1516 * see also pid_nr() etc in include/linux/pid.h
1518 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
1519 struct pid_namespace
*ns
);
1521 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1526 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
1527 struct pid_namespace
*ns
)
1529 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1532 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1534 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1538 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1543 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1545 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1547 return pid_vnr(task_tgid(tsk
));
1551 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
1552 struct pid_namespace
*ns
)
1554 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1557 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1559 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1563 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
1564 struct pid_namespace
*ns
)
1566 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1569 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1571 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1574 /* obsolete, do not use */
1575 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1577 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1581 * pid_alive - check that a task structure is not stale
1582 * @p: Task structure to be checked.
1584 * Test if a process is not yet dead (at most zombie state)
1585 * If pid_alive fails, then pointers within the task structure
1586 * can be stale and must not be dereferenced.
1588 static inline int pid_alive(struct task_struct
*p
)
1590 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1594 * is_global_init - check if a task structure is init
1595 * @tsk: Task structure to be checked.
1597 * Check if a task structure is the first user space task the kernel created.
1599 static inline int is_global_init(struct task_struct
*tsk
)
1601 return tsk
->pid
== 1;
1605 * is_container_init:
1606 * check whether in the task is init in its own pid namespace.
1608 extern int is_container_init(struct task_struct
*tsk
);
1610 extern struct pid
*cad_pid
;
1612 extern void free_task(struct task_struct
*tsk
);
1613 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1615 extern void __put_task_struct(struct task_struct
*t
);
1617 static inline void put_task_struct(struct task_struct
*t
)
1619 if (atomic_dec_and_test(&t
->usage
))
1620 __put_task_struct(t
);
1623 extern cputime_t
task_utime(struct task_struct
*p
);
1624 extern cputime_t
task_stime(struct task_struct
*p
);
1625 extern cputime_t
task_gtime(struct task_struct
*p
);
1630 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1631 /* Not implemented yet, only for 486*/
1632 #define PF_STARTING 0x00000002 /* being created */
1633 #define PF_EXITING 0x00000004 /* getting shut down */
1634 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1635 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1636 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1637 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1638 #define PF_DUMPCORE 0x00000200 /* dumped core */
1639 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1640 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1641 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1642 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1643 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1644 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1645 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1646 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1647 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1648 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1649 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1650 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1651 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1652 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1653 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1654 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1655 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1656 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1657 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1658 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1661 * Only the _current_ task can read/write to tsk->flags, but other
1662 * tasks can access tsk->flags in readonly mode for example
1663 * with tsk_used_math (like during threaded core dumping).
1664 * There is however an exception to this rule during ptrace
1665 * or during fork: the ptracer task is allowed to write to the
1666 * child->flags of its traced child (same goes for fork, the parent
1667 * can write to the child->flags), because we're guaranteed the
1668 * child is not running and in turn not changing child->flags
1669 * at the same time the parent does it.
1671 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1672 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1673 #define clear_used_math() clear_stopped_child_used_math(current)
1674 #define set_used_math() set_stopped_child_used_math(current)
1675 #define conditional_stopped_child_used_math(condition, child) \
1676 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1677 #define conditional_used_math(condition) \
1678 conditional_stopped_child_used_math(condition, current)
1679 #define copy_to_stopped_child_used_math(child) \
1680 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1681 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1682 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1683 #define used_math() tsk_used_math(current)
1686 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
1687 const struct cpumask
*new_mask
);
1689 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
1690 const struct cpumask
*new_mask
)
1692 if (!cpumask_test_cpu(0, new_mask
))
1697 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1699 return set_cpus_allowed_ptr(p
, &new_mask
);
1703 * Architectures can set this to 1 if they have specified
1704 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1705 * but then during bootup it turns out that sched_clock()
1706 * is reliable after all:
1708 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1709 extern int sched_clock_stable
;
1712 extern unsigned long long sched_clock(void);
1714 extern void sched_clock_init(void);
1715 extern u64
sched_clock_cpu(int cpu
);
1717 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1718 static inline void sched_clock_tick(void)
1722 static inline void sched_clock_idle_sleep_event(void)
1726 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
1730 extern void sched_clock_tick(void);
1731 extern void sched_clock_idle_sleep_event(void);
1732 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1736 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1737 * clock constructed from sched_clock():
1739 extern unsigned long long cpu_clock(int cpu
);
1741 extern unsigned long long
1742 task_sched_runtime(struct task_struct
*task
);
1743 extern unsigned long long thread_group_sched_runtime(struct task_struct
*task
);
1745 /* sched_exec is called by processes performing an exec */
1747 extern void sched_exec(void);
1749 #define sched_exec() {}
1752 extern void sched_clock_idle_sleep_event(void);
1753 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1755 #ifdef CONFIG_HOTPLUG_CPU
1756 extern void idle_task_exit(void);
1758 static inline void idle_task_exit(void) {}
1761 extern void sched_idle_next(void);
1763 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1764 extern void wake_up_idle_cpu(int cpu
);
1766 static inline void wake_up_idle_cpu(int cpu
) { }
1769 extern unsigned int sysctl_sched_latency
;
1770 extern unsigned int sysctl_sched_min_granularity
;
1771 extern unsigned int sysctl_sched_wakeup_granularity
;
1772 extern unsigned int sysctl_sched_shares_ratelimit
;
1773 extern unsigned int sysctl_sched_shares_thresh
;
1774 #ifdef CONFIG_SCHED_DEBUG
1775 extern unsigned int sysctl_sched_child_runs_first
;
1776 extern unsigned int sysctl_sched_features
;
1777 extern unsigned int sysctl_sched_migration_cost
;
1778 extern unsigned int sysctl_sched_nr_migrate
;
1780 int sched_nr_latency_handler(struct ctl_table
*table
, int write
,
1781 struct file
*file
, void __user
*buffer
, size_t *length
,
1784 extern unsigned int sysctl_sched_rt_period
;
1785 extern int sysctl_sched_rt_runtime
;
1787 int sched_rt_handler(struct ctl_table
*table
, int write
,
1788 struct file
*filp
, void __user
*buffer
, size_t *lenp
,
1791 extern unsigned int sysctl_sched_compat_yield
;
1793 #ifdef CONFIG_RT_MUTEXES
1794 extern int rt_mutex_getprio(struct task_struct
*p
);
1795 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
1796 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
1798 static inline int rt_mutex_getprio(struct task_struct
*p
)
1800 return p
->normal_prio
;
1802 # define rt_mutex_adjust_pi(p) do { } while (0)
1805 extern void set_user_nice(struct task_struct
*p
, long nice
);
1806 extern int task_prio(const struct task_struct
*p
);
1807 extern int task_nice(const struct task_struct
*p
);
1808 extern int can_nice(const struct task_struct
*p
, const int nice
);
1809 extern int task_curr(const struct task_struct
*p
);
1810 extern int idle_cpu(int cpu
);
1811 extern int sched_setscheduler(struct task_struct
*, int, struct sched_param
*);
1812 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
1813 struct sched_param
*);
1814 extern struct task_struct
*idle_task(int cpu
);
1815 extern struct task_struct
*curr_task(int cpu
);
1816 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1821 * The default (Linux) execution domain.
1823 extern struct exec_domain default_exec_domain
;
1825 union thread_union
{
1826 struct thread_info thread_info
;
1827 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1830 #ifndef __HAVE_ARCH_KSTACK_END
1831 static inline int kstack_end(void *addr
)
1833 /* Reliable end of stack detection:
1834 * Some APM bios versions misalign the stack
1836 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1840 extern union thread_union init_thread_union
;
1841 extern struct task_struct init_task
;
1843 extern struct mm_struct init_mm
;
1845 extern struct pid_namespace init_pid_ns
;
1848 * find a task by one of its numerical ids
1850 * find_task_by_pid_type_ns():
1851 * it is the most generic call - it finds a task by all id,
1852 * type and namespace specified
1853 * find_task_by_pid_ns():
1854 * finds a task by its pid in the specified namespace
1855 * find_task_by_vpid():
1856 * finds a task by its virtual pid
1858 * see also find_vpid() etc in include/linux/pid.h
1861 extern struct task_struct
*find_task_by_pid_type_ns(int type
, int pid
,
1862 struct pid_namespace
*ns
);
1864 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1865 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
1866 struct pid_namespace
*ns
);
1868 extern void __set_special_pids(struct pid
*pid
);
1870 /* per-UID process charging. */
1871 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
1872 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1874 atomic_inc(&u
->__count
);
1877 extern void free_uid(struct user_struct
*);
1878 extern void release_uids(struct user_namespace
*ns
);
1880 #include <asm/current.h>
1882 extern void do_timer(unsigned long ticks
);
1884 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
1885 extern int wake_up_process(struct task_struct
*tsk
);
1886 extern void wake_up_new_task(struct task_struct
*tsk
,
1887 unsigned long clone_flags
);
1889 extern void kick_process(struct task_struct
*tsk
);
1891 static inline void kick_process(struct task_struct
*tsk
) { }
1893 extern void sched_fork(struct task_struct
*p
, int clone_flags
);
1894 extern void sched_dead(struct task_struct
*p
);
1896 extern void proc_caches_init(void);
1897 extern void flush_signals(struct task_struct
*);
1898 extern void ignore_signals(struct task_struct
*);
1899 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
1900 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
1902 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
1904 unsigned long flags
;
1907 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
1908 ret
= dequeue_signal(tsk
, mask
, info
);
1909 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
1914 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
1916 extern void unblock_all_signals(void);
1917 extern void release_task(struct task_struct
* p
);
1918 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
1919 extern int force_sigsegv(int, struct task_struct
*);
1920 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
1921 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1922 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
1923 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
1924 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
1925 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
1926 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
1927 extern int do_notify_parent(struct task_struct
*, int);
1928 extern void force_sig(int, struct task_struct
*);
1929 extern void force_sig_specific(int, struct task_struct
*);
1930 extern int send_sig(int, struct task_struct
*, int);
1931 extern void zap_other_threads(struct task_struct
*p
);
1932 extern struct sigqueue
*sigqueue_alloc(void);
1933 extern void sigqueue_free(struct sigqueue
*);
1934 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
1935 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
1936 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
1938 static inline int kill_cad_pid(int sig
, int priv
)
1940 return kill_pid(cad_pid
, sig
, priv
);
1943 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1944 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1945 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1946 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1948 static inline int is_si_special(const struct siginfo
*info
)
1950 return info
<= SEND_SIG_FORCED
;
1953 /* True if we are on the alternate signal stack. */
1955 static inline int on_sig_stack(unsigned long sp
)
1957 return (sp
- current
->sas_ss_sp
< current
->sas_ss_size
);
1960 static inline int sas_ss_flags(unsigned long sp
)
1962 return (current
->sas_ss_size
== 0 ? SS_DISABLE
1963 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
1967 * Routines for handling mm_structs
1969 extern struct mm_struct
* mm_alloc(void);
1971 /* mmdrop drops the mm and the page tables */
1972 extern void __mmdrop(struct mm_struct
*);
1973 static inline void mmdrop(struct mm_struct
* mm
)
1975 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
1979 /* mmput gets rid of the mappings and all user-space */
1980 extern void mmput(struct mm_struct
*);
1981 /* Grab a reference to a task's mm, if it is not already going away */
1982 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
1983 /* Remove the current tasks stale references to the old mm_struct */
1984 extern void mm_release(struct task_struct
*, struct mm_struct
*);
1985 /* Allocate a new mm structure and copy contents from tsk->mm */
1986 extern struct mm_struct
*dup_mm(struct task_struct
*tsk
);
1988 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1989 struct task_struct
*, struct pt_regs
*);
1990 extern void flush_thread(void);
1991 extern void exit_thread(void);
1993 extern void exit_files(struct task_struct
*);
1994 extern void __cleanup_signal(struct signal_struct
*);
1995 extern void __cleanup_sighand(struct sighand_struct
*);
1997 extern void exit_itimers(struct signal_struct
*);
1998 extern void flush_itimer_signals(void);
2000 extern NORET_TYPE
void do_group_exit(int);
2002 extern void daemonize(const char *, ...);
2003 extern int allow_signal(int);
2004 extern int disallow_signal(int);
2006 extern int do_execve(char *, char __user
* __user
*, char __user
* __user
*, struct pt_regs
*);
2007 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
2008 struct task_struct
*fork_idle(int);
2010 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
2011 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2014 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2016 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2023 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2025 #define for_each_process(p) \
2026 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2028 extern bool is_single_threaded(struct task_struct
*);
2031 * Careful: do_each_thread/while_each_thread is a double loop so
2032 * 'break' will not work as expected - use goto instead.
2034 #define do_each_thread(g, t) \
2035 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2037 #define while_each_thread(g, t) \
2038 while ((t = next_thread(t)) != g)
2040 /* de_thread depends on thread_group_leader not being a pid based check */
2041 #define thread_group_leader(p) (p == p->group_leader)
2043 /* Do to the insanities of de_thread it is possible for a process
2044 * to have the pid of the thread group leader without actually being
2045 * the thread group leader. For iteration through the pids in proc
2046 * all we care about is that we have a task with the appropriate
2047 * pid, we don't actually care if we have the right task.
2049 static inline int has_group_leader_pid(struct task_struct
*p
)
2051 return p
->pid
== p
->tgid
;
2055 int same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2057 return p1
->tgid
== p2
->tgid
;
2060 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2062 return list_entry(rcu_dereference(p
->thread_group
.next
),
2063 struct task_struct
, thread_group
);
2066 static inline int thread_group_empty(struct task_struct
*p
)
2068 return list_empty(&p
->thread_group
);
2071 #define delay_group_leader(p) \
2072 (thread_group_leader(p) && !thread_group_empty(p))
2074 static inline int task_detached(struct task_struct
*p
)
2076 return p
->exit_signal
== -1;
2080 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2081 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2082 * pins the final release of task.io_context. Also protects ->cpuset and
2083 * ->cgroup.subsys[].
2085 * Nests both inside and outside of read_lock(&tasklist_lock).
2086 * It must not be nested with write_lock_irq(&tasklist_lock),
2087 * neither inside nor outside.
2089 static inline void task_lock(struct task_struct
*p
)
2091 spin_lock(&p
->alloc_lock
);
2094 static inline void task_unlock(struct task_struct
*p
)
2096 spin_unlock(&p
->alloc_lock
);
2099 extern struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
2100 unsigned long *flags
);
2102 static inline void unlock_task_sighand(struct task_struct
*tsk
,
2103 unsigned long *flags
)
2105 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2108 #ifndef __HAVE_THREAD_FUNCTIONS
2110 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2111 #define task_stack_page(task) ((task)->stack)
2113 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2115 *task_thread_info(p
) = *task_thread_info(org
);
2116 task_thread_info(p
)->task
= p
;
2119 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2121 return (unsigned long *)(task_thread_info(p
) + 1);
2126 static inline int object_is_on_stack(void *obj
)
2128 void *stack
= task_stack_page(current
);
2130 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2133 extern void thread_info_cache_init(void);
2135 #ifdef CONFIG_DEBUG_STACK_USAGE
2136 static inline unsigned long stack_not_used(struct task_struct
*p
)
2138 unsigned long *n
= end_of_stack(p
);
2140 do { /* Skip over canary */
2144 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
2148 /* set thread flags in other task's structures
2149 * - see asm/thread_info.h for TIF_xxxx flags available
2151 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2153 set_ti_thread_flag(task_thread_info(tsk
), flag
);
2156 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2158 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2161 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2163 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
2166 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2168 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2171 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2173 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
2176 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
2178 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2181 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
2183 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2186 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
2188 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
2191 static inline int signal_pending(struct task_struct
*p
)
2193 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
2196 extern int __fatal_signal_pending(struct task_struct
*p
);
2198 static inline int fatal_signal_pending(struct task_struct
*p
)
2200 return signal_pending(p
) && __fatal_signal_pending(p
);
2203 static inline int signal_pending_state(long state
, struct task_struct
*p
)
2205 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
2207 if (!signal_pending(p
))
2210 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
2213 static inline int need_resched(void)
2215 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
2219 * cond_resched() and cond_resched_lock(): latency reduction via
2220 * explicit rescheduling in places that are safe. The return
2221 * value indicates whether a reschedule was done in fact.
2222 * cond_resched_lock() will drop the spinlock before scheduling,
2223 * cond_resched_softirq() will enable bhs before scheduling.
2225 extern int _cond_resched(void);
2226 #ifdef CONFIG_PREEMPT_BKL
2227 static inline int cond_resched(void)
2232 static inline int cond_resched(void)
2234 return _cond_resched();
2237 extern int cond_resched_lock(spinlock_t
* lock
);
2238 extern int cond_resched_softirq(void);
2239 static inline int cond_resched_bkl(void)
2241 return _cond_resched();
2245 * Does a critical section need to be broken due to another
2246 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2247 * but a general need for low latency)
2249 static inline int spin_needbreak(spinlock_t
*lock
)
2251 #ifdef CONFIG_PREEMPT
2252 return spin_is_contended(lock
);
2259 * Thread group CPU time accounting.
2261 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
2262 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
2264 static inline void thread_group_cputime_init(struct signal_struct
*sig
)
2266 sig
->cputimer
.cputime
= INIT_CPUTIME
;
2267 spin_lock_init(&sig
->cputimer
.lock
);
2268 sig
->cputimer
.running
= 0;
2271 static inline void thread_group_cputime_free(struct signal_struct
*sig
)
2276 * Reevaluate whether the task has signals pending delivery.
2277 * Wake the task if so.
2278 * This is required every time the blocked sigset_t changes.
2279 * callers must hold sighand->siglock.
2281 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
2282 extern void recalc_sigpending(void);
2284 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
2287 * Wrappers for p->thread_info->cpu access. No-op on UP.
2291 static inline unsigned int task_cpu(const struct task_struct
*p
)
2293 return task_thread_info(p
)->cpu
;
2296 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
2300 static inline unsigned int task_cpu(const struct task_struct
*p
)
2305 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
2309 #endif /* CONFIG_SMP */
2311 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
2313 #ifdef CONFIG_TRACING
2315 __trace_special(void *__tr
, void *__data
,
2316 unsigned long arg1
, unsigned long arg2
, unsigned long arg3
);
2319 __trace_special(void *__tr
, void *__data
,
2320 unsigned long arg1
, unsigned long arg2
, unsigned long arg3
)
2325 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
2326 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
2328 extern void normalize_rt_tasks(void);
2330 #ifdef CONFIG_GROUP_SCHED
2332 extern struct task_group init_task_group
;
2333 #ifdef CONFIG_USER_SCHED
2334 extern struct task_group root_task_group
;
2335 extern void set_tg_uid(struct user_struct
*user
);
2338 extern struct task_group
*sched_create_group(struct task_group
*parent
);
2339 extern void sched_destroy_group(struct task_group
*tg
);
2340 extern void sched_move_task(struct task_struct
*tsk
);
2341 #ifdef CONFIG_FAIR_GROUP_SCHED
2342 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
2343 extern unsigned long sched_group_shares(struct task_group
*tg
);
2345 #ifdef CONFIG_RT_GROUP_SCHED
2346 extern int sched_group_set_rt_runtime(struct task_group
*tg
,
2347 long rt_runtime_us
);
2348 extern long sched_group_rt_runtime(struct task_group
*tg
);
2349 extern int sched_group_set_rt_period(struct task_group
*tg
,
2351 extern long sched_group_rt_period(struct task_group
*tg
);
2352 extern int sched_rt_can_attach(struct task_group
*tg
, struct task_struct
*tsk
);
2356 extern int task_can_switch_user(struct user_struct
*up
,
2357 struct task_struct
*tsk
);
2359 #ifdef CONFIG_TASK_XACCT
2360 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2362 tsk
->ioac
.rchar
+= amt
;
2365 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2367 tsk
->ioac
.wchar
+= amt
;
2370 static inline void inc_syscr(struct task_struct
*tsk
)
2375 static inline void inc_syscw(struct task_struct
*tsk
)
2380 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2384 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2388 static inline void inc_syscr(struct task_struct
*tsk
)
2392 static inline void inc_syscw(struct task_struct
*tsk
)
2397 #ifndef TASK_SIZE_OF
2398 #define TASK_SIZE_OF(tsk) TASK_SIZE
2402 * Call the function if the target task is executing on a CPU right now:
2404 extern void task_oncpu_function_call(struct task_struct
*p
,
2405 void (*func
) (void *info
), void *info
);
2408 #ifdef CONFIG_MM_OWNER
2409 extern void mm_update_next_owner(struct mm_struct
*mm
);
2410 extern void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
);
2412 static inline void mm_update_next_owner(struct mm_struct
*mm
)
2416 static inline void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
)
2419 #endif /* CONFIG_MM_OWNER */
2421 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2423 #endif /* __KERNEL__ */