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>
64 #include <asm/semaphore.h>
66 #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/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.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>
91 #include <linux/latencytop.h>
93 #include <asm/processor.h>
97 struct futex_pi_state
;
98 struct robust_list_head
;
102 * List of flags we want to share for kernel threads,
103 * if only because they are not used by them anyway.
105 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
108 * These are the constant used to fake the fixed-point load-average
109 * counting. Some notes:
110 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
111 * a load-average precision of 10 bits integer + 11 bits fractional
112 * - if you want to count load-averages more often, you need more
113 * precision, or rounding will get you. With 2-second counting freq,
114 * the EXP_n values would be 1981, 2034 and 2043 if still using only
117 extern unsigned long avenrun
[]; /* Load averages */
119 #define FSHIFT 11 /* nr of bits of precision */
120 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
121 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
122 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
123 #define EXP_5 2014 /* 1/exp(5sec/5min) */
124 #define EXP_15 2037 /* 1/exp(5sec/15min) */
126 #define CALC_LOAD(load,exp,n) \
128 load += n*(FIXED_1-exp); \
131 extern unsigned long total_forks
;
132 extern int nr_threads
;
133 DECLARE_PER_CPU(unsigned long, process_counts
);
134 extern int nr_processes(void);
135 extern unsigned long nr_running(void);
136 extern unsigned long nr_uninterruptible(void);
137 extern unsigned long nr_active(void);
138 extern unsigned long nr_iowait(void);
139 extern unsigned long weighted_cpuload(const int cpu
);
144 #ifdef CONFIG_SCHED_DEBUG
145 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
146 extern void proc_sched_set_task(struct task_struct
*p
);
148 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
);
151 proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
)
154 static inline void proc_sched_set_task(struct task_struct
*p
)
158 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
)
164 * Task state bitmask. NOTE! These bits are also
165 * encoded in fs/proc/array.c: get_task_state().
167 * We have two separate sets of flags: task->state
168 * is about runnability, while task->exit_state are
169 * about the task exiting. Confusing, but this way
170 * modifying one set can't modify the other one by
173 #define TASK_RUNNING 0
174 #define TASK_INTERRUPTIBLE 1
175 #define TASK_UNINTERRUPTIBLE 2
176 #define __TASK_STOPPED 4
177 #define __TASK_TRACED 8
178 /* in tsk->exit_state */
179 #define EXIT_ZOMBIE 16
181 /* in tsk->state again */
183 #define TASK_WAKEKILL 128
185 /* Convenience macros for the sake of set_task_state */
186 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
187 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
188 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
190 /* Convenience macros for the sake of wake_up */
191 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
192 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
194 /* get_task_state() */
195 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
196 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
199 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
200 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
201 #define task_is_stopped_or_traced(task) \
202 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
203 #define task_contributes_to_load(task) \
204 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
206 #define __set_task_state(tsk, state_value) \
207 do { (tsk)->state = (state_value); } while (0)
208 #define set_task_state(tsk, state_value) \
209 set_mb((tsk)->state, (state_value))
212 * set_current_state() includes a barrier so that the write of current->state
213 * is correctly serialised wrt the caller's subsequent test of whether to
216 * set_current_state(TASK_UNINTERRUPTIBLE);
217 * if (do_i_need_to_sleep())
220 * If the caller does not need such serialisation then use __set_current_state()
222 #define __set_current_state(state_value) \
223 do { current->state = (state_value); } while (0)
224 #define set_current_state(state_value) \
225 set_mb(current->state, (state_value))
227 /* Task command name length */
228 #define TASK_COMM_LEN 16
230 #include <linux/spinlock.h>
233 * This serializes "schedule()" and also protects
234 * the run-queue from deletions/modifications (but
235 * _adding_ to the beginning of the run-queue has
238 extern rwlock_t tasklist_lock
;
239 extern spinlock_t mmlist_lock
;
243 extern void sched_init(void);
244 extern void sched_init_smp(void);
245 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
246 extern void init_idle(struct task_struct
*idle
, int cpu
);
247 extern void init_idle_bootup_task(struct task_struct
*idle
);
249 extern cpumask_t nohz_cpu_mask
;
250 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
251 extern int select_nohz_load_balancer(int cpu
);
253 static inline int select_nohz_load_balancer(int cpu
)
259 extern unsigned long rt_needs_cpu(int cpu
);
262 * Only dump TASK_* tasks. (0 for all tasks)
264 extern void show_state_filter(unsigned long state_filter
);
266 static inline void show_state(void)
268 show_state_filter(0);
271 extern void show_regs(struct pt_regs
*);
274 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
275 * task), SP is the stack pointer of the first frame that should be shown in the back
276 * trace (or NULL if the entire call-chain of the task should be shown).
278 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
280 void io_schedule(void);
281 long io_schedule_timeout(long timeout
);
283 extern void cpu_init (void);
284 extern void trap_init(void);
285 extern void account_process_tick(struct task_struct
*task
, int user
);
286 extern void update_process_times(int user
);
287 extern void scheduler_tick(void);
288 extern void hrtick_resched(void);
290 extern void sched_show_task(struct task_struct
*p
);
292 #ifdef CONFIG_DETECT_SOFTLOCKUP
293 extern void softlockup_tick(void);
294 extern void spawn_softlockup_task(void);
295 extern void touch_softlockup_watchdog(void);
296 extern void touch_all_softlockup_watchdogs(void);
297 extern unsigned long softlockup_thresh
;
298 extern unsigned long sysctl_hung_task_check_count
;
299 extern unsigned long sysctl_hung_task_timeout_secs
;
300 extern unsigned long sysctl_hung_task_warnings
;
302 static inline void softlockup_tick(void)
305 static inline void spawn_softlockup_task(void)
308 static inline void touch_softlockup_watchdog(void)
311 static inline void touch_all_softlockup_watchdogs(void)
317 /* Attach to any functions which should be ignored in wchan output. */
318 #define __sched __attribute__((__section__(".sched.text")))
320 /* Linker adds these: start and end of __sched functions */
321 extern char __sched_text_start
[], __sched_text_end
[];
323 /* Is this address in the __sched functions? */
324 extern int in_sched_functions(unsigned long addr
);
326 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
327 extern signed long schedule_timeout(signed long timeout
);
328 extern signed long schedule_timeout_interruptible(signed long timeout
);
329 extern signed long schedule_timeout_killable(signed long timeout
);
330 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
331 asmlinkage
void schedule(void);
334 struct user_namespace
;
336 /* Maximum number of active map areas.. This is a random (large) number */
337 #define DEFAULT_MAX_MAP_COUNT 65536
339 extern int sysctl_max_map_count
;
341 #include <linux/aio.h>
344 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
345 unsigned long, unsigned long);
347 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
348 unsigned long len
, unsigned long pgoff
,
349 unsigned long flags
);
350 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
351 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
353 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
355 * The mm counters are not protected by its page_table_lock,
356 * so must be incremented atomically.
358 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
359 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
360 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
361 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
362 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
364 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
366 * The mm counters are protected by its page_table_lock,
367 * so can be incremented directly.
369 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
370 #define get_mm_counter(mm, member) ((mm)->_##member)
371 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
372 #define inc_mm_counter(mm, member) (mm)->_##member++
373 #define dec_mm_counter(mm, member) (mm)->_##member--
375 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
377 #define get_mm_rss(mm) \
378 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
379 #define update_hiwater_rss(mm) do { \
380 unsigned long _rss = get_mm_rss(mm); \
381 if ((mm)->hiwater_rss < _rss) \
382 (mm)->hiwater_rss = _rss; \
384 #define update_hiwater_vm(mm) do { \
385 if ((mm)->hiwater_vm < (mm)->total_vm) \
386 (mm)->hiwater_vm = (mm)->total_vm; \
389 extern void set_dumpable(struct mm_struct
*mm
, int value
);
390 extern int get_dumpable(struct mm_struct
*mm
);
394 #define MMF_DUMPABLE 0 /* core dump is permitted */
395 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
396 #define MMF_DUMPABLE_BITS 2
398 /* coredump filter bits */
399 #define MMF_DUMP_ANON_PRIVATE 2
400 #define MMF_DUMP_ANON_SHARED 3
401 #define MMF_DUMP_MAPPED_PRIVATE 4
402 #define MMF_DUMP_MAPPED_SHARED 5
403 #define MMF_DUMP_ELF_HEADERS 6
404 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
405 #define MMF_DUMP_FILTER_BITS 5
406 #define MMF_DUMP_FILTER_MASK \
407 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
408 #define MMF_DUMP_FILTER_DEFAULT \
409 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
411 struct sighand_struct
{
413 struct k_sigaction action
[_NSIG
];
415 wait_queue_head_t signalfd_wqh
;
418 struct pacct_struct
{
421 unsigned long ac_mem
;
422 cputime_t ac_utime
, ac_stime
;
423 unsigned long ac_minflt
, ac_majflt
;
427 * NOTE! "signal_struct" does not have it's own
428 * locking, because a shared signal_struct always
429 * implies a shared sighand_struct, so locking
430 * sighand_struct is always a proper superset of
431 * the locking of signal_struct.
433 struct signal_struct
{
437 wait_queue_head_t wait_chldexit
; /* for wait4() */
439 /* current thread group signal load-balancing target: */
440 struct task_struct
*curr_target
;
442 /* shared signal handling: */
443 struct sigpending shared_pending
;
445 /* thread group exit support */
448 * - notify group_exit_task when ->count is equal to notify_count
449 * - everyone except group_exit_task is stopped during signal delivery
450 * of fatal signals, group_exit_task processes the signal.
452 struct task_struct
*group_exit_task
;
455 /* thread group stop support, overloads group_exit_code too */
456 int group_stop_count
;
457 unsigned int flags
; /* see SIGNAL_* flags below */
459 /* POSIX.1b Interval Timers */
460 struct list_head posix_timers
;
462 /* ITIMER_REAL timer for the process */
463 struct hrtimer real_timer
;
464 struct pid
*leader_pid
;
465 ktime_t it_real_incr
;
467 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
468 cputime_t it_prof_expires
, it_virt_expires
;
469 cputime_t it_prof_incr
, it_virt_incr
;
471 /* job control IDs */
474 * pgrp and session fields are deprecated.
475 * use the task_session_Xnr and task_pgrp_Xnr routines below
479 pid_t pgrp __deprecated
;
483 struct pid
*tty_old_pgrp
;
486 pid_t session __deprecated
;
490 /* boolean value for session group leader */
493 struct tty_struct
*tty
; /* NULL if no tty */
496 * Cumulative resource counters for dead threads in the group,
497 * and for reaped dead child processes forked by this group.
498 * Live threads maintain their own counters and add to these
499 * in __exit_signal, except for the group leader.
501 cputime_t utime
, stime
, cutime
, cstime
;
504 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
505 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
506 unsigned long inblock
, oublock
, cinblock
, coublock
;
509 * Cumulative ns of scheduled CPU time for dead threads in the
510 * group, not including a zombie group leader. (This only differs
511 * from jiffies_to_ns(utime + stime) if sched_clock uses something
512 * other than jiffies.)
514 unsigned long long sum_sched_runtime
;
517 * We don't bother to synchronize most readers of this at all,
518 * because there is no reader checking a limit that actually needs
519 * to get both rlim_cur and rlim_max atomically, and either one
520 * alone is a single word that can safely be read normally.
521 * getrlimit/setrlimit use task_lock(current->group_leader) to
522 * protect this instead of the siglock, because they really
523 * have no need to disable irqs.
525 struct rlimit rlim
[RLIM_NLIMITS
];
527 struct list_head cpu_timers
[3];
529 /* keep the process-shared keyrings here so that they do the right
530 * thing in threads created with CLONE_THREAD */
532 struct key
*session_keyring
; /* keyring inherited over fork */
533 struct key
*process_keyring
; /* keyring private to this process */
535 #ifdef CONFIG_BSD_PROCESS_ACCT
536 struct pacct_struct pacct
; /* per-process accounting information */
538 #ifdef CONFIG_TASKSTATS
539 struct taskstats
*stats
;
543 struct tty_audit_buf
*tty_audit_buf
;
547 /* Context switch must be unlocked if interrupts are to be enabled */
548 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
549 # define __ARCH_WANT_UNLOCKED_CTXSW
553 * Bits in flags field of signal_struct.
555 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
556 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
557 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
558 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
560 /* If true, all threads except ->group_exit_task have pending SIGKILL */
561 static inline int signal_group_exit(const struct signal_struct
*sig
)
563 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
564 (sig
->group_exit_task
!= NULL
);
568 * Some day this will be a full-fledged user tracking system..
571 atomic_t __count
; /* reference count */
572 atomic_t processes
; /* How many processes does this user have? */
573 atomic_t files
; /* How many open files does this user have? */
574 atomic_t sigpending
; /* How many pending signals does this user have? */
575 #ifdef CONFIG_INOTIFY_USER
576 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
577 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
579 #ifdef CONFIG_POSIX_MQUEUE
580 /* protected by mq_lock */
581 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
583 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
586 struct key
*uid_keyring
; /* UID specific keyring */
587 struct key
*session_keyring
; /* UID's default session keyring */
590 /* Hash table maintenance information */
591 struct hlist_node uidhash_node
;
594 #ifdef CONFIG_USER_SCHED
595 struct task_group
*tg
;
598 struct work_struct work
;
603 extern int uids_sysfs_init(void);
605 extern struct user_struct
*find_user(uid_t
);
607 extern struct user_struct root_user
;
608 #define INIT_USER (&root_user)
610 struct backing_dev_info
;
611 struct reclaim_state
;
613 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
615 /* cumulative counters */
616 unsigned long pcount
; /* # of times run on this cpu */
617 unsigned long long cpu_time
, /* time spent on the cpu */
618 run_delay
; /* time spent waiting on a runqueue */
621 unsigned long long last_arrival
,/* when we last ran on a cpu */
622 last_queued
; /* when we were last queued to run */
623 #ifdef CONFIG_SCHEDSTATS
625 unsigned int bkl_count
;
628 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
630 #ifdef CONFIG_SCHEDSTATS
631 extern const struct file_operations proc_schedstat_operations
;
632 #endif /* CONFIG_SCHEDSTATS */
634 #ifdef CONFIG_TASK_DELAY_ACCT
635 struct task_delay_info
{
637 unsigned int flags
; /* Private per-task flags */
639 /* For each stat XXX, add following, aligned appropriately
641 * struct timespec XXX_start, XXX_end;
645 * Atomicity of updates to XXX_delay, XXX_count protected by
646 * single lock above (split into XXX_lock if contention is an issue).
650 * XXX_count is incremented on every XXX operation, the delay
651 * associated with the operation is added to XXX_delay.
652 * XXX_delay contains the accumulated delay time in nanoseconds.
654 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
655 u64 blkio_delay
; /* wait for sync block io completion */
656 u64 swapin_delay
; /* wait for swapin block io completion */
657 u32 blkio_count
; /* total count of the number of sync block */
658 /* io operations performed */
659 u32 swapin_count
; /* total count of the number of swapin block */
660 /* io operations performed */
662 #endif /* CONFIG_TASK_DELAY_ACCT */
664 static inline int sched_info_on(void)
666 #ifdef CONFIG_SCHEDSTATS
668 #elif defined(CONFIG_TASK_DELAY_ACCT)
669 extern int delayacct_on
;
684 * sched-domains (multiprocessor balancing) declarations:
688 * Increase resolution of nice-level calculations:
690 #define SCHED_LOAD_SHIFT 10
691 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
693 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
696 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
697 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
698 #define SD_BALANCE_EXEC 4 /* Balance on exec */
699 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
700 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
701 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
702 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
703 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
704 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
705 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
706 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
708 #define BALANCE_FOR_MC_POWER \
709 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
711 #define BALANCE_FOR_PKG_POWER \
712 ((sched_mc_power_savings || sched_smt_power_savings) ? \
713 SD_POWERSAVINGS_BALANCE : 0)
715 #define test_sd_parent(sd, flag) ((sd->parent && \
716 (sd->parent->flags & flag)) ? 1 : 0)
720 struct sched_group
*next
; /* Must be a circular list */
724 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
725 * single CPU. This is read only (except for setup, hotplug CPU).
726 * Note : Never change cpu_power without recompute its reciprocal
728 unsigned int __cpu_power
;
730 * reciprocal value of cpu_power to avoid expensive divides
731 * (see include/linux/reciprocal_div.h)
733 u32 reciprocal_cpu_power
;
736 struct sched_domain
{
737 /* These fields must be setup */
738 struct sched_domain
*parent
; /* top domain must be null terminated */
739 struct sched_domain
*child
; /* bottom domain must be null terminated */
740 struct sched_group
*groups
; /* the balancing groups of the domain */
741 cpumask_t span
; /* span of all CPUs in this domain */
742 unsigned long min_interval
; /* Minimum balance interval ms */
743 unsigned long max_interval
; /* Maximum balance interval ms */
744 unsigned int busy_factor
; /* less balancing by factor if busy */
745 unsigned int imbalance_pct
; /* No balance until over watermark */
746 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
747 unsigned int busy_idx
;
748 unsigned int idle_idx
;
749 unsigned int newidle_idx
;
750 unsigned int wake_idx
;
751 unsigned int forkexec_idx
;
752 int flags
; /* See SD_* */
754 /* Runtime fields. */
755 unsigned long last_balance
; /* init to jiffies. units in jiffies */
756 unsigned int balance_interval
; /* initialise to 1. units in ms. */
757 unsigned int nr_balance_failed
; /* initialise to 0 */
759 #ifdef CONFIG_SCHEDSTATS
760 /* load_balance() stats */
761 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
762 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
763 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
764 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
765 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
766 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
767 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
768 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
770 /* Active load balancing */
771 unsigned int alb_count
;
772 unsigned int alb_failed
;
773 unsigned int alb_pushed
;
775 /* SD_BALANCE_EXEC stats */
776 unsigned int sbe_count
;
777 unsigned int sbe_balanced
;
778 unsigned int sbe_pushed
;
780 /* SD_BALANCE_FORK stats */
781 unsigned int sbf_count
;
782 unsigned int sbf_balanced
;
783 unsigned int sbf_pushed
;
785 /* try_to_wake_up() stats */
786 unsigned int ttwu_wake_remote
;
787 unsigned int ttwu_move_affine
;
788 unsigned int ttwu_move_balance
;
792 extern void partition_sched_domains(int ndoms_new
, cpumask_t
*doms_new
);
794 #endif /* CONFIG_SMP */
797 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
798 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
799 * task of nice 0 or enough lower priority tasks to bring up the
802 static inline int above_background_load(void)
806 for_each_online_cpu(cpu
) {
807 if (weighted_cpuload(cpu
) >= SCHED_LOAD_SCALE
)
813 struct io_context
; /* See blkdev.h */
814 #define NGROUPS_SMALL 32
815 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
819 gid_t small_block
[NGROUPS_SMALL
];
825 * get_group_info() must be called with the owning task locked (via task_lock())
826 * when task != current. The reason being that the vast majority of callers are
827 * looking at current->group_info, which can not be changed except by the
828 * current task. Changing current->group_info requires the task lock, too.
830 #define get_group_info(group_info) do { \
831 atomic_inc(&(group_info)->usage); \
834 #define put_group_info(group_info) do { \
835 if (atomic_dec_and_test(&(group_info)->usage)) \
836 groups_free(group_info); \
839 extern struct group_info
*groups_alloc(int gidsetsize
);
840 extern void groups_free(struct group_info
*group_info
);
841 extern int set_current_groups(struct group_info
*group_info
);
842 extern int groups_search(struct group_info
*group_info
, gid_t grp
);
843 /* access the groups "array" with this macro */
844 #define GROUP_AT(gi, i) \
845 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
847 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
848 extern void prefetch_stack(struct task_struct
*t
);
850 static inline void prefetch_stack(struct task_struct
*t
) { }
853 struct audit_context
; /* See audit.c */
855 struct pipe_inode_info
;
856 struct uts_namespace
;
862 const struct sched_class
*next
;
864 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int wakeup
);
865 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int sleep
);
866 void (*yield_task
) (struct rq
*rq
);
867 int (*select_task_rq
)(struct task_struct
*p
, int sync
);
869 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
);
871 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
872 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
875 unsigned long (*load_balance
) (struct rq
*this_rq
, int this_cpu
,
876 struct rq
*busiest
, unsigned long max_load_move
,
877 struct sched_domain
*sd
, enum cpu_idle_type idle
,
878 int *all_pinned
, int *this_best_prio
);
880 int (*move_one_task
) (struct rq
*this_rq
, int this_cpu
,
881 struct rq
*busiest
, struct sched_domain
*sd
,
882 enum cpu_idle_type idle
);
883 void (*pre_schedule
) (struct rq
*this_rq
, struct task_struct
*task
);
884 void (*post_schedule
) (struct rq
*this_rq
);
885 void (*task_wake_up
) (struct rq
*this_rq
, struct task_struct
*task
);
888 void (*set_curr_task
) (struct rq
*rq
);
889 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
, int queued
);
890 void (*task_new
) (struct rq
*rq
, struct task_struct
*p
);
891 void (*set_cpus_allowed
)(struct task_struct
*p
, cpumask_t
*newmask
);
893 void (*join_domain
)(struct rq
*rq
);
894 void (*leave_domain
)(struct rq
*rq
);
896 void (*switched_from
) (struct rq
*this_rq
, struct task_struct
*task
,
898 void (*switched_to
) (struct rq
*this_rq
, struct task_struct
*task
,
900 void (*prio_changed
) (struct rq
*this_rq
, struct task_struct
*task
,
901 int oldprio
, int running
);
903 #ifdef CONFIG_FAIR_GROUP_SCHED
904 void (*moved_group
) (struct task_struct
*p
);
909 unsigned long weight
, inv_weight
;
913 * CFS stats for a schedulable entity (task, task-group etc)
915 * Current field usage histogram:
922 struct sched_entity
{
923 struct load_weight load
; /* for load-balancing */
924 struct rb_node run_node
;
928 u64 sum_exec_runtime
;
930 u64 prev_sum_exec_runtime
;
932 #ifdef CONFIG_SCHEDSTATS
940 s64 sum_sleep_runtime
;
948 u64 nr_migrations_cold
;
949 u64 nr_failed_migrations_affine
;
950 u64 nr_failed_migrations_running
;
951 u64 nr_failed_migrations_hot
;
952 u64 nr_forced_migrations
;
953 u64 nr_forced2_migrations
;
957 u64 nr_wakeups_migrate
;
958 u64 nr_wakeups_local
;
959 u64 nr_wakeups_remote
;
960 u64 nr_wakeups_affine
;
961 u64 nr_wakeups_affine_attempts
;
962 u64 nr_wakeups_passive
;
966 #ifdef CONFIG_FAIR_GROUP_SCHED
967 struct sched_entity
*parent
;
968 /* rq on which this entity is (to be) queued: */
969 struct cfs_rq
*cfs_rq
;
970 /* rq "owned" by this entity/group: */
975 struct sched_rt_entity
{
976 struct list_head run_list
;
977 unsigned int time_slice
;
978 unsigned long timeout
;
981 #ifdef CONFIG_RT_GROUP_SCHED
982 struct sched_rt_entity
*parent
;
983 /* rq on which this entity is (to be) queued: */
985 /* rq "owned" by this entity/group: */
991 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
994 unsigned int flags
; /* per process flags, defined below */
997 int lock_depth
; /* BKL lock depth */
1000 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1005 int prio
, static_prio
, normal_prio
;
1006 const struct sched_class
*sched_class
;
1007 struct sched_entity se
;
1008 struct sched_rt_entity rt
;
1010 #ifdef CONFIG_PREEMPT_NOTIFIERS
1011 /* list of struct preempt_notifier: */
1012 struct hlist_head preempt_notifiers
;
1016 * fpu_counter contains the number of consecutive context switches
1017 * that the FPU is used. If this is over a threshold, the lazy fpu
1018 * saving becomes unlazy to save the trap. This is an unsigned char
1019 * so that after 256 times the counter wraps and the behavior turns
1020 * lazy again; this to deal with bursty apps that only use FPU for
1023 unsigned char fpu_counter
;
1024 s8 oomkilladj
; /* OOM kill score adjustment (bit shift). */
1025 #ifdef CONFIG_BLK_DEV_IO_TRACE
1026 unsigned int btrace_seq
;
1029 unsigned int policy
;
1030 cpumask_t cpus_allowed
;
1032 #ifdef CONFIG_PREEMPT_RCU
1033 int rcu_read_lock_nesting
;
1034 int rcu_flipctr_idx
;
1035 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1037 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1038 struct sched_info sched_info
;
1041 struct list_head tasks
;
1043 * ptrace_list/ptrace_children forms the list of my children
1044 * that were stolen by a ptracer.
1046 struct list_head ptrace_children
;
1047 struct list_head ptrace_list
;
1049 struct mm_struct
*mm
, *active_mm
;
1052 struct linux_binfmt
*binfmt
;
1054 int exit_code
, exit_signal
;
1055 int pdeath_signal
; /* The signal sent when the parent dies */
1057 unsigned int personality
;
1058 unsigned did_exec
:1;
1062 #ifdef CONFIG_CC_STACKPROTECTOR
1063 /* Canary value for the -fstack-protector gcc feature */
1064 unsigned long stack_canary
;
1067 * pointers to (original) parent process, youngest child, younger sibling,
1068 * older sibling, respectively. (p->father can be replaced with
1071 struct task_struct
*real_parent
; /* real parent process (when being debugged) */
1072 struct task_struct
*parent
; /* parent process */
1074 * children/sibling forms the list of my children plus the
1075 * tasks I'm ptracing.
1077 struct list_head children
; /* list of my children */
1078 struct list_head sibling
; /* linkage in my parent's children list */
1079 struct task_struct
*group_leader
; /* threadgroup leader */
1081 /* PID/PID hash table linkage. */
1082 struct pid_link pids
[PIDTYPE_MAX
];
1083 struct list_head thread_group
;
1085 struct completion
*vfork_done
; /* for vfork() */
1086 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1087 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1089 unsigned int rt_priority
;
1090 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1092 cputime_t prev_utime
, prev_stime
;
1093 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1094 struct timespec start_time
; /* monotonic time */
1095 struct timespec real_start_time
; /* boot based time */
1096 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1097 unsigned long min_flt
, maj_flt
;
1099 cputime_t it_prof_expires
, it_virt_expires
;
1100 unsigned long long it_sched_expires
;
1101 struct list_head cpu_timers
[3];
1103 /* process credentials */
1104 uid_t uid
,euid
,suid
,fsuid
;
1105 gid_t gid
,egid
,sgid
,fsgid
;
1106 struct group_info
*group_info
;
1107 kernel_cap_t cap_effective
, cap_inheritable
, cap_permitted
, cap_bset
;
1108 unsigned keep_capabilities
:1;
1109 struct user_struct
*user
;
1111 struct key
*request_key_auth
; /* assumed request_key authority */
1112 struct key
*thread_keyring
; /* keyring private to this thread */
1113 unsigned char jit_keyring
; /* default keyring to attach requested keys to */
1115 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1116 - access with [gs]et_task_comm (which lock
1117 it with task_lock())
1118 - initialized normally by flush_old_exec */
1119 /* file system info */
1120 int link_count
, total_link_count
;
1121 #ifdef CONFIG_SYSVIPC
1123 struct sysv_sem sysvsem
;
1125 #ifdef CONFIG_DETECT_SOFTLOCKUP
1126 /* hung task detection */
1127 unsigned long last_switch_timestamp
;
1128 unsigned long last_switch_count
;
1130 /* CPU-specific state of this task */
1131 struct thread_struct thread
;
1132 /* filesystem information */
1133 struct fs_struct
*fs
;
1134 /* open file information */
1135 struct files_struct
*files
;
1137 struct nsproxy
*nsproxy
;
1138 /* signal handlers */
1139 struct signal_struct
*signal
;
1140 struct sighand_struct
*sighand
;
1142 sigset_t blocked
, real_blocked
;
1143 sigset_t saved_sigmask
; /* To be restored with TIF_RESTORE_SIGMASK */
1144 struct sigpending pending
;
1146 unsigned long sas_ss_sp
;
1148 int (*notifier
)(void *priv
);
1149 void *notifier_data
;
1150 sigset_t
*notifier_mask
;
1151 #ifdef CONFIG_SECURITY
1154 struct audit_context
*audit_context
;
1155 #ifdef CONFIG_AUDITSYSCALL
1157 unsigned int sessionid
;
1161 /* Thread group tracking */
1164 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1165 spinlock_t alloc_lock
;
1167 /* Protection of the PI data structures: */
1170 #ifdef CONFIG_RT_MUTEXES
1171 /* PI waiters blocked on a rt_mutex held by this task */
1172 struct plist_head pi_waiters
;
1173 /* Deadlock detection and priority inheritance handling */
1174 struct rt_mutex_waiter
*pi_blocked_on
;
1177 #ifdef CONFIG_DEBUG_MUTEXES
1178 /* mutex deadlock detection */
1179 struct mutex_waiter
*blocked_on
;
1181 #ifdef CONFIG_TRACE_IRQFLAGS
1182 unsigned int irq_events
;
1183 int hardirqs_enabled
;
1184 unsigned long hardirq_enable_ip
;
1185 unsigned int hardirq_enable_event
;
1186 unsigned long hardirq_disable_ip
;
1187 unsigned int hardirq_disable_event
;
1188 int softirqs_enabled
;
1189 unsigned long softirq_disable_ip
;
1190 unsigned int softirq_disable_event
;
1191 unsigned long softirq_enable_ip
;
1192 unsigned int softirq_enable_event
;
1193 int hardirq_context
;
1194 int softirq_context
;
1196 #ifdef CONFIG_LOCKDEP
1197 # define MAX_LOCK_DEPTH 48UL
1200 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1201 unsigned int lockdep_recursion
;
1204 /* journalling filesystem info */
1207 /* stacked block device info */
1208 struct bio
*bio_list
, **bio_tail
;
1211 struct reclaim_state
*reclaim_state
;
1213 struct backing_dev_info
*backing_dev_info
;
1215 struct io_context
*io_context
;
1217 unsigned long ptrace_message
;
1218 siginfo_t
*last_siginfo
; /* For ptrace use. */
1219 #ifdef CONFIG_TASK_XACCT
1220 /* i/o counters(bytes read/written, #syscalls */
1221 u64 rchar
, wchar
, syscr
, syscw
;
1223 struct task_io_accounting ioac
;
1224 #if defined(CONFIG_TASK_XACCT)
1225 u64 acct_rss_mem1
; /* accumulated rss usage */
1226 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1227 cputime_t acct_stimexpd
;/* stime since last update */
1230 struct mempolicy
*mempolicy
;
1233 #ifdef CONFIG_CPUSETS
1234 nodemask_t mems_allowed
;
1235 int cpuset_mems_generation
;
1236 int cpuset_mem_spread_rotor
;
1238 #ifdef CONFIG_CGROUPS
1239 /* Control Group info protected by css_set_lock */
1240 struct css_set
*cgroups
;
1241 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1242 struct list_head cg_list
;
1245 struct robust_list_head __user
*robust_list
;
1246 #ifdef CONFIG_COMPAT
1247 struct compat_robust_list_head __user
*compat_robust_list
;
1249 struct list_head pi_state_list
;
1250 struct futex_pi_state
*pi_state_cache
;
1252 atomic_t fs_excl
; /* holding fs exclusive resources */
1253 struct rcu_head rcu
;
1256 * cache last used pipe for splice
1258 struct pipe_inode_info
*splice_pipe
;
1259 #ifdef CONFIG_TASK_DELAY_ACCT
1260 struct task_delay_info
*delays
;
1262 #ifdef CONFIG_FAULT_INJECTION
1265 struct prop_local_single dirties
;
1266 #ifdef CONFIG_LATENCYTOP
1267 int latency_record_count
;
1268 struct latency_record latency_record
[LT_SAVECOUNT
];
1273 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1274 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1275 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1276 * values are inverted: lower p->prio value means higher priority.
1278 * The MAX_USER_RT_PRIO value allows the actual maximum
1279 * RT priority to be separate from the value exported to
1280 * user-space. This allows kernel threads to set their
1281 * priority to a value higher than any user task. Note:
1282 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1285 #define MAX_USER_RT_PRIO 100
1286 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1288 #define MAX_PRIO (MAX_RT_PRIO + 40)
1289 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1291 static inline int rt_prio(int prio
)
1293 if (unlikely(prio
< MAX_RT_PRIO
))
1298 static inline int rt_task(struct task_struct
*p
)
1300 return rt_prio(p
->prio
);
1303 static inline void set_task_session(struct task_struct
*tsk
, pid_t session
)
1305 tsk
->signal
->__session
= session
;
1308 static inline void set_task_pgrp(struct task_struct
*tsk
, pid_t pgrp
)
1310 tsk
->signal
->__pgrp
= pgrp
;
1313 static inline struct pid
*task_pid(struct task_struct
*task
)
1315 return task
->pids
[PIDTYPE_PID
].pid
;
1318 static inline struct pid
*task_tgid(struct task_struct
*task
)
1320 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1323 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1325 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1328 static inline struct pid
*task_session(struct task_struct
*task
)
1330 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1333 struct pid_namespace
;
1336 * the helpers to get the task's different pids as they are seen
1337 * from various namespaces
1339 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1340 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1342 * task_xid_nr_ns() : id seen from the ns specified;
1344 * set_task_vxid() : assigns a virtual id to a task;
1346 * see also pid_nr() etc in include/linux/pid.h
1349 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1354 pid_t
task_pid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1356 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1358 return pid_vnr(task_pid(tsk
));
1362 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1367 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1369 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1371 return pid_vnr(task_tgid(tsk
));
1375 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1377 return tsk
->signal
->__pgrp
;
1380 pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1382 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1384 return pid_vnr(task_pgrp(tsk
));
1388 static inline pid_t
task_session_nr(struct task_struct
*tsk
)
1390 return tsk
->signal
->__session
;
1393 pid_t
task_session_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1395 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1397 return pid_vnr(task_session(tsk
));
1402 * pid_alive - check that a task structure is not stale
1403 * @p: Task structure to be checked.
1405 * Test if a process is not yet dead (at most zombie state)
1406 * If pid_alive fails, then pointers within the task structure
1407 * can be stale and must not be dereferenced.
1409 static inline int pid_alive(struct task_struct
*p
)
1411 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1415 * is_global_init - check if a task structure is init
1416 * @tsk: Task structure to be checked.
1418 * Check if a task structure is the first user space task the kernel created.
1420 static inline int is_global_init(struct task_struct
*tsk
)
1422 return tsk
->pid
== 1;
1426 * is_container_init:
1427 * check whether in the task is init in its own pid namespace.
1429 extern int is_container_init(struct task_struct
*tsk
);
1431 extern struct pid
*cad_pid
;
1433 extern void free_task(struct task_struct
*tsk
);
1434 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1436 extern void __put_task_struct(struct task_struct
*t
);
1438 static inline void put_task_struct(struct task_struct
*t
)
1440 if (atomic_dec_and_test(&t
->usage
))
1441 __put_task_struct(t
);
1447 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1448 /* Not implemented yet, only for 486*/
1449 #define PF_STARTING 0x00000002 /* being created */
1450 #define PF_EXITING 0x00000004 /* getting shut down */
1451 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1452 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1453 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1454 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1455 #define PF_DUMPCORE 0x00000200 /* dumped core */
1456 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1457 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1458 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1459 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1460 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1461 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1462 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1463 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1464 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1465 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1466 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1467 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1468 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1469 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1470 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1471 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1472 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1473 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1476 * Only the _current_ task can read/write to tsk->flags, but other
1477 * tasks can access tsk->flags in readonly mode for example
1478 * with tsk_used_math (like during threaded core dumping).
1479 * There is however an exception to this rule during ptrace
1480 * or during fork: the ptracer task is allowed to write to the
1481 * child->flags of its traced child (same goes for fork, the parent
1482 * can write to the child->flags), because we're guaranteed the
1483 * child is not running and in turn not changing child->flags
1484 * at the same time the parent does it.
1486 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1487 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1488 #define clear_used_math() clear_stopped_child_used_math(current)
1489 #define set_used_math() set_stopped_child_used_math(current)
1490 #define conditional_stopped_child_used_math(condition, child) \
1491 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1492 #define conditional_used_math(condition) \
1493 conditional_stopped_child_used_math(condition, current)
1494 #define copy_to_stopped_child_used_math(child) \
1495 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1496 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1497 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1498 #define used_math() tsk_used_math(current)
1501 extern int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
);
1503 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1505 if (!cpu_isset(0, new_mask
))
1511 extern unsigned long long sched_clock(void);
1514 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1515 * clock constructed from sched_clock():
1517 extern unsigned long long cpu_clock(int cpu
);
1519 extern unsigned long long
1520 task_sched_runtime(struct task_struct
*task
);
1522 /* sched_exec is called by processes performing an exec */
1524 extern void sched_exec(void);
1526 #define sched_exec() {}
1529 extern void sched_clock_idle_sleep_event(void);
1530 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1532 #ifdef CONFIG_HOTPLUG_CPU
1533 extern void idle_task_exit(void);
1535 static inline void idle_task_exit(void) {}
1538 extern void sched_idle_next(void);
1540 #ifdef CONFIG_SCHED_DEBUG
1541 extern unsigned int sysctl_sched_latency
;
1542 extern unsigned int sysctl_sched_min_granularity
;
1543 extern unsigned int sysctl_sched_wakeup_granularity
;
1544 extern unsigned int sysctl_sched_batch_wakeup_granularity
;
1545 extern unsigned int sysctl_sched_child_runs_first
;
1546 extern unsigned int sysctl_sched_features
;
1547 extern unsigned int sysctl_sched_migration_cost
;
1548 extern unsigned int sysctl_sched_nr_migrate
;
1550 int sched_nr_latency_handler(struct ctl_table
*table
, int write
,
1551 struct file
*file
, void __user
*buffer
, size_t *length
,
1554 extern unsigned int sysctl_sched_rt_period
;
1555 extern int sysctl_sched_rt_runtime
;
1557 extern unsigned int sysctl_sched_compat_yield
;
1559 #ifdef CONFIG_RT_MUTEXES
1560 extern int rt_mutex_getprio(struct task_struct
*p
);
1561 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
1562 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
1564 static inline int rt_mutex_getprio(struct task_struct
*p
)
1566 return p
->normal_prio
;
1568 # define rt_mutex_adjust_pi(p) do { } while (0)
1571 extern void set_user_nice(struct task_struct
*p
, long nice
);
1572 extern int task_prio(const struct task_struct
*p
);
1573 extern int task_nice(const struct task_struct
*p
);
1574 extern int can_nice(const struct task_struct
*p
, const int nice
);
1575 extern int task_curr(const struct task_struct
*p
);
1576 extern int idle_cpu(int cpu
);
1577 extern int sched_setscheduler(struct task_struct
*, int, struct sched_param
*);
1578 extern struct task_struct
*idle_task(int cpu
);
1579 extern struct task_struct
*curr_task(int cpu
);
1580 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1585 * The default (Linux) execution domain.
1587 extern struct exec_domain default_exec_domain
;
1589 union thread_union
{
1590 struct thread_info thread_info
;
1591 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1594 #ifndef __HAVE_ARCH_KSTACK_END
1595 static inline int kstack_end(void *addr
)
1597 /* Reliable end of stack detection:
1598 * Some APM bios versions misalign the stack
1600 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1604 extern union thread_union init_thread_union
;
1605 extern struct task_struct init_task
;
1607 extern struct mm_struct init_mm
;
1609 extern struct pid_namespace init_pid_ns
;
1612 * find a task by one of its numerical ids
1614 * find_task_by_pid_type_ns():
1615 * it is the most generic call - it finds a task by all id,
1616 * type and namespace specified
1617 * find_task_by_pid_ns():
1618 * finds a task by its pid in the specified namespace
1619 * find_task_by_vpid():
1620 * finds a task by its virtual pid
1621 * find_task_by_pid():
1622 * finds a task by its global pid
1624 * see also find_pid() etc in include/linux/pid.h
1627 extern struct task_struct
*find_task_by_pid_type_ns(int type
, int pid
,
1628 struct pid_namespace
*ns
);
1630 extern struct task_struct
*find_task_by_pid(pid_t nr
);
1631 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1632 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
1633 struct pid_namespace
*ns
);
1635 extern void __set_special_pids(struct pid
*pid
);
1637 /* per-UID process charging. */
1638 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
1639 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1641 atomic_inc(&u
->__count
);
1644 extern void free_uid(struct user_struct
*);
1645 extern void switch_uid(struct user_struct
*);
1646 extern void release_uids(struct user_namespace
*ns
);
1648 #include <asm/current.h>
1650 extern void do_timer(unsigned long ticks
);
1652 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
1653 extern int wake_up_process(struct task_struct
*tsk
);
1654 extern void wake_up_new_task(struct task_struct
*tsk
,
1655 unsigned long clone_flags
);
1657 extern void kick_process(struct task_struct
*tsk
);
1659 static inline void kick_process(struct task_struct
*tsk
) { }
1661 extern void sched_fork(struct task_struct
*p
, int clone_flags
);
1662 extern void sched_dead(struct task_struct
*p
);
1664 extern int in_group_p(gid_t
);
1665 extern int in_egroup_p(gid_t
);
1667 extern void proc_caches_init(void);
1668 extern void flush_signals(struct task_struct
*);
1669 extern void ignore_signals(struct task_struct
*);
1670 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
1671 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
1673 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
1675 unsigned long flags
;
1678 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
1679 ret
= dequeue_signal(tsk
, mask
, info
);
1680 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
1685 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
1687 extern void unblock_all_signals(void);
1688 extern void release_task(struct task_struct
* p
);
1689 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
1690 extern int force_sigsegv(int, struct task_struct
*);
1691 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
1692 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
1693 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
1694 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
1695 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
1696 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
1697 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
1698 extern void do_notify_parent(struct task_struct
*, int);
1699 extern void force_sig(int, struct task_struct
*);
1700 extern void force_sig_specific(int, struct task_struct
*);
1701 extern int send_sig(int, struct task_struct
*, int);
1702 extern void zap_other_threads(struct task_struct
*p
);
1703 extern int kill_proc(pid_t
, int, int);
1704 extern struct sigqueue
*sigqueue_alloc(void);
1705 extern void sigqueue_free(struct sigqueue
*);
1706 extern int send_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1707 extern int send_group_sigqueue(int, struct sigqueue
*, struct task_struct
*);
1708 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
1709 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
1711 static inline int kill_cad_pid(int sig
, int priv
)
1713 return kill_pid(cad_pid
, sig
, priv
);
1716 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1717 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1718 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1719 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1721 static inline int is_si_special(const struct siginfo
*info
)
1723 return info
<= SEND_SIG_FORCED
;
1726 /* True if we are on the alternate signal stack. */
1728 static inline int on_sig_stack(unsigned long sp
)
1730 return (sp
- current
->sas_ss_sp
< current
->sas_ss_size
);
1733 static inline int sas_ss_flags(unsigned long sp
)
1735 return (current
->sas_ss_size
== 0 ? SS_DISABLE
1736 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
1740 * Routines for handling mm_structs
1742 extern struct mm_struct
* mm_alloc(void);
1744 /* mmdrop drops the mm and the page tables */
1745 extern void __mmdrop(struct mm_struct
*);
1746 static inline void mmdrop(struct mm_struct
* mm
)
1748 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
1752 /* mmput gets rid of the mappings and all user-space */
1753 extern void mmput(struct mm_struct
*);
1754 /* Grab a reference to a task's mm, if it is not already going away */
1755 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
1756 /* Remove the current tasks stale references to the old mm_struct */
1757 extern void mm_release(struct task_struct
*, struct mm_struct
*);
1759 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct
*, struct pt_regs
*);
1760 extern void flush_thread(void);
1761 extern void exit_thread(void);
1763 extern void exit_files(struct task_struct
*);
1764 extern void __cleanup_signal(struct signal_struct
*);
1765 extern void __cleanup_sighand(struct sighand_struct
*);
1766 extern void exit_itimers(struct signal_struct
*);
1768 extern NORET_TYPE
void do_group_exit(int);
1770 extern void daemonize(const char *, ...);
1771 extern int allow_signal(int);
1772 extern int disallow_signal(int);
1774 extern int do_execve(char *, char __user
* __user
*, char __user
* __user
*, struct pt_regs
*);
1775 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
1776 struct task_struct
*fork_idle(int);
1778 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
1779 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
1782 extern void wait_task_inactive(struct task_struct
* p
);
1784 #define wait_task_inactive(p) do { } while (0)
1787 #define remove_parent(p) list_del_init(&(p)->sibling)
1788 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1790 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1792 #define for_each_process(p) \
1793 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1796 * Careful: do_each_thread/while_each_thread is a double loop so
1797 * 'break' will not work as expected - use goto instead.
1799 #define do_each_thread(g, t) \
1800 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1802 #define while_each_thread(g, t) \
1803 while ((t = next_thread(t)) != g)
1805 /* de_thread depends on thread_group_leader not being a pid based check */
1806 #define thread_group_leader(p) (p == p->group_leader)
1808 /* Do to the insanities of de_thread it is possible for a process
1809 * to have the pid of the thread group leader without actually being
1810 * the thread group leader. For iteration through the pids in proc
1811 * all we care about is that we have a task with the appropriate
1812 * pid, we don't actually care if we have the right task.
1814 static inline int has_group_leader_pid(struct task_struct
*p
)
1816 return p
->pid
== p
->tgid
;
1820 int same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
1822 return p1
->tgid
== p2
->tgid
;
1825 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
1827 return list_entry(rcu_dereference(p
->thread_group
.next
),
1828 struct task_struct
, thread_group
);
1831 static inline int thread_group_empty(struct task_struct
*p
)
1833 return list_empty(&p
->thread_group
);
1836 #define delay_group_leader(p) \
1837 (thread_group_leader(p) && !thread_group_empty(p))
1840 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1841 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1842 * pins the final release of task.io_context. Also protects ->cpuset and
1843 * ->cgroup.subsys[].
1845 * Nests both inside and outside of read_lock(&tasklist_lock).
1846 * It must not be nested with write_lock_irq(&tasklist_lock),
1847 * neither inside nor outside.
1849 static inline void task_lock(struct task_struct
*p
)
1851 spin_lock(&p
->alloc_lock
);
1854 static inline void task_unlock(struct task_struct
*p
)
1856 spin_unlock(&p
->alloc_lock
);
1859 extern struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
1860 unsigned long *flags
);
1862 static inline void unlock_task_sighand(struct task_struct
*tsk
,
1863 unsigned long *flags
)
1865 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
1868 #ifndef __HAVE_THREAD_FUNCTIONS
1870 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1871 #define task_stack_page(task) ((task)->stack)
1873 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
1875 *task_thread_info(p
) = *task_thread_info(org
);
1876 task_thread_info(p
)->task
= p
;
1879 static inline unsigned long *end_of_stack(struct task_struct
*p
)
1881 return (unsigned long *)(task_thread_info(p
) + 1);
1886 /* set thread flags in other task's structures
1887 * - see asm/thread_info.h for TIF_xxxx flags available
1889 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1891 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1894 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1896 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1899 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1901 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1904 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1906 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1909 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1911 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1914 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1916 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1919 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1921 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1924 static inline int signal_pending(struct task_struct
*p
)
1926 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
1929 extern int __fatal_signal_pending(struct task_struct
*p
);
1931 static inline int fatal_signal_pending(struct task_struct
*p
)
1933 return signal_pending(p
) && __fatal_signal_pending(p
);
1936 static inline int need_resched(void)
1938 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
1942 * cond_resched() and cond_resched_lock(): latency reduction via
1943 * explicit rescheduling in places that are safe. The return
1944 * value indicates whether a reschedule was done in fact.
1945 * cond_resched_lock() will drop the spinlock before scheduling,
1946 * cond_resched_softirq() will enable bhs before scheduling.
1948 #ifdef CONFIG_PREEMPT
1949 static inline int cond_resched(void)
1954 extern int _cond_resched(void);
1955 static inline int cond_resched(void)
1957 return _cond_resched();
1960 extern int cond_resched_lock(spinlock_t
* lock
);
1961 extern int cond_resched_softirq(void);
1964 * Does a critical section need to be broken due to another
1965 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1966 * but a general need for low latency)
1968 static inline int spin_needbreak(spinlock_t
*lock
)
1970 #ifdef CONFIG_PREEMPT
1971 return spin_is_contended(lock
);
1978 * Reevaluate whether the task has signals pending delivery.
1979 * Wake the task if so.
1980 * This is required every time the blocked sigset_t changes.
1981 * callers must hold sighand->siglock.
1983 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
1984 extern void recalc_sigpending(void);
1986 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
1989 * Wrappers for p->thread_info->cpu access. No-op on UP.
1993 static inline unsigned int task_cpu(const struct task_struct
*p
)
1995 return task_thread_info(p
)->cpu
;
1998 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
2002 static inline unsigned int task_cpu(const struct task_struct
*p
)
2007 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
2011 #endif /* CONFIG_SMP */
2013 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2014 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
2016 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
)
2018 mm
->mmap_base
= TASK_UNMAPPED_BASE
;
2019 mm
->get_unmapped_area
= arch_get_unmapped_area
;
2020 mm
->unmap_area
= arch_unmap_area
;
2024 extern long sched_setaffinity(pid_t pid
, cpumask_t new_mask
);
2025 extern long sched_getaffinity(pid_t pid
, cpumask_t
*mask
);
2027 extern int sched_mc_power_savings
, sched_smt_power_savings
;
2029 extern void normalize_rt_tasks(void);
2031 #ifdef CONFIG_GROUP_SCHED
2033 extern struct task_group init_task_group
;
2035 extern struct task_group
*sched_create_group(void);
2036 extern void sched_destroy_group(struct task_group
*tg
);
2037 extern void sched_move_task(struct task_struct
*tsk
);
2038 #ifdef CONFIG_FAIR_GROUP_SCHED
2039 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
2040 extern unsigned long sched_group_shares(struct task_group
*tg
);
2042 #ifdef CONFIG_RT_GROUP_SCHED
2043 extern int sched_group_set_rt_runtime(struct task_group
*tg
,
2044 long rt_runtime_us
);
2045 extern long sched_group_rt_runtime(struct task_group
*tg
);
2049 #ifdef CONFIG_TASK_XACCT
2050 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2055 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2060 static inline void inc_syscr(struct task_struct
*tsk
)
2065 static inline void inc_syscw(struct task_struct
*tsk
)
2070 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2074 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2078 static inline void inc_syscr(struct task_struct
*tsk
)
2082 static inline void inc_syscw(struct task_struct
*tsk
)
2088 void migration_init(void);
2090 static inline void migration_init(void)
2095 #ifndef TASK_SIZE_OF
2096 #define TASK_SIZE_OF(tsk) TASK_SIZE
2099 #endif /* __KERNEL__ */