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/securebits.h>
72 #include <linux/fs_struct.h>
73 #include <linux/compiler.h>
74 #include <linux/completion.h>
75 #include <linux/pid.h>
76 #include <linux/percpu.h>
77 #include <linux/topology.h>
78 #include <linux/proportions.h>
79 #include <linux/seccomp.h>
80 #include <linux/rcupdate.h>
81 #include <linux/rtmutex.h>
83 #include <linux/time.h>
84 #include <linux/param.h>
85 #include <linux/resource.h>
86 #include <linux/timer.h>
87 #include <linux/hrtimer.h>
88 #include <linux/task_io_accounting.h>
89 #include <linux/kobject.h>
90 #include <linux/latencytop.h>
92 #include <asm/processor.h>
96 struct futex_pi_state;
97 struct robust_list_head;
101 * List of flags we want to share for kernel threads,
102 * if only because they are not used by them anyway.
104 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
107 * These are the constant used to fake the fixed-point load-average
108 * counting. Some notes:
109 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
110 * a load-average precision of 10 bits integer + 11 bits fractional
111 * - if you want to count load-averages more often, you need more
112 * precision, or rounding will get you. With 2-second counting freq,
113 * the EXP_n values would be 1981, 2034 and 2043 if still using only
116 extern unsigned long avenrun[]; /* Load averages */
118 #define FSHIFT 11 /* nr of bits of precision */
119 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
120 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
121 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
122 #define EXP_5 2014 /* 1/exp(5sec/5min) */
123 #define EXP_15 2037 /* 1/exp(5sec/15min) */
125 #define CALC_LOAD(load,exp,n) \
127 load += n*(FIXED_1-exp); \
130 extern unsigned long total_forks;
131 extern int nr_threads;
132 DECLARE_PER_CPU(unsigned long, process_counts);
133 extern int nr_processes(void);
134 extern unsigned long nr_running(void);
135 extern unsigned long nr_uninterruptible(void);
136 extern unsigned long nr_active(void);
137 extern unsigned long nr_iowait(void);
138 extern unsigned long weighted_cpuload(const int cpu);
143 #ifdef CONFIG_SCHED_DEBUG
144 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
145 extern void proc_sched_set_task(struct task_struct *p);
147 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
150 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
153 static inline void proc_sched_set_task(struct task_struct *p)
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
163 * Task state bitmask. NOTE! These bits are also
164 * encoded in fs/proc/array.c: get_task_state().
166 * We have two separate sets of flags: task->state
167 * is about runnability, while task->exit_state are
168 * about the task exiting. Confusing, but this way
169 * modifying one set can't modify the other one by
172 #define TASK_RUNNING 0
173 #define TASK_INTERRUPTIBLE 1
174 #define TASK_UNINTERRUPTIBLE 2
175 #define __TASK_STOPPED 4
176 #define __TASK_TRACED 8
177 /* in tsk->exit_state */
178 #define EXIT_ZOMBIE 16
180 /* in tsk->state again */
182 #define TASK_WAKEKILL 128
184 /* Convenience macros for the sake of set_task_state */
185 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
186 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
187 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
189 /* Convenience macros for the sake of wake_up */
190 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
191 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
193 /* get_task_state() */
194 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
195 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
198 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
199 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
200 #define task_is_stopped_or_traced(task) \
201 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
202 #define task_contributes_to_load(task) \
203 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
205 #define __set_task_state(tsk, state_value) \
206 do { (tsk)->state = (state_value); } while (0)
207 #define set_task_state(tsk, state_value) \
208 set_mb((tsk)->state, (state_value))
211 * set_current_state() includes a barrier so that the write of current->state
212 * is correctly serialised wrt the caller's subsequent test of whether to
215 * set_current_state(TASK_UNINTERRUPTIBLE);
216 * if (do_i_need_to_sleep())
219 * If the caller does not need such serialisation then use __set_current_state()
221 #define __set_current_state(state_value) \
222 do { current->state = (state_value); } while (0)
223 #define set_current_state(state_value) \
224 set_mb(current->state, (state_value))
226 /* Task command name length */
227 #define TASK_COMM_LEN 16
229 #include <linux/spinlock.h>
232 * This serializes "schedule()" and also protects
233 * the run-queue from deletions/modifications (but
234 * _adding_ to the beginning of the run-queue has
237 extern rwlock_t tasklist_lock;
238 extern spinlock_t mmlist_lock;
242 extern void sched_init(void);
243 extern void sched_init_smp(void);
244 extern asmlinkage void schedule_tail(struct task_struct *prev);
245 extern void init_idle(struct task_struct *idle, int cpu);
246 extern void init_idle_bootup_task(struct task_struct *idle);
248 extern cpumask_t nohz_cpu_mask;
249 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
250 extern int select_nohz_load_balancer(int cpu);
252 static inline int select_nohz_load_balancer(int cpu)
258 extern unsigned long rt_needs_cpu(int cpu);
261 * Only dump TASK_* tasks. (0 for all tasks)
263 extern void show_state_filter(unsigned long state_filter);
265 static inline void show_state(void)
267 show_state_filter(0);
270 extern void show_regs(struct pt_regs *);
273 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
274 * task), SP is the stack pointer of the first frame that should be shown in the back
275 * trace (or NULL if the entire call-chain of the task should be shown).
277 extern void show_stack(struct task_struct *task, unsigned long *sp);
279 void io_schedule(void);
280 long io_schedule_timeout(long timeout);
282 extern void cpu_init (void);
283 extern void trap_init(void);
284 extern void account_process_tick(struct task_struct *task, int user);
285 extern void update_process_times(int user);
286 extern void scheduler_tick(void);
287 extern void hrtick_resched(void);
289 extern void sched_show_task(struct task_struct *p);
291 #ifdef CONFIG_DETECT_SOFTLOCKUP
292 extern void softlockup_tick(void);
293 extern void spawn_softlockup_task(void);
294 extern void touch_softlockup_watchdog(void);
295 extern void touch_all_softlockup_watchdogs(void);
296 extern unsigned long softlockup_thresh;
297 extern unsigned long sysctl_hung_task_check_count;
298 extern unsigned long sysctl_hung_task_timeout_secs;
299 extern unsigned long sysctl_hung_task_warnings;
301 static inline void softlockup_tick(void)
304 static inline void spawn_softlockup_task(void)
307 static inline void touch_softlockup_watchdog(void)
310 static inline void touch_all_softlockup_watchdogs(void)
316 /* Attach to any functions which should be ignored in wchan output. */
317 #define __sched __attribute__((__section__(".sched.text")))
319 /* Linker adds these: start and end of __sched functions */
320 extern char __sched_text_start[], __sched_text_end[];
322 /* Is this address in the __sched functions? */
323 extern int in_sched_functions(unsigned long addr);
325 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
326 extern signed long schedule_timeout(signed long timeout);
327 extern signed long schedule_timeout_interruptible(signed long timeout);
328 extern signed long schedule_timeout_killable(signed long timeout);
329 extern signed long schedule_timeout_uninterruptible(signed long timeout);
330 asmlinkage void schedule(void);
333 struct user_namespace;
335 /* Maximum number of active map areas.. This is a random (large) number */
336 #define DEFAULT_MAX_MAP_COUNT 65536
338 extern int sysctl_max_map_count;
340 #include <linux/aio.h>
343 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
344 unsigned long, unsigned long);
346 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
347 unsigned long len, unsigned long pgoff,
348 unsigned long flags);
349 extern void arch_unmap_area(struct mm_struct *, unsigned long);
350 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
352 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
354 * The mm counters are not protected by its page_table_lock,
355 * so must be incremented atomically.
357 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
358 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
359 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
360 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
361 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
363 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
365 * The mm counters are protected by its page_table_lock,
366 * so can be incremented directly.
368 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
369 #define get_mm_counter(mm, member) ((mm)->_##member)
370 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
371 #define inc_mm_counter(mm, member) (mm)->_##member++
372 #define dec_mm_counter(mm, member) (mm)->_##member--
374 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
376 #define get_mm_rss(mm) \
377 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
378 #define update_hiwater_rss(mm) do { \
379 unsigned long _rss = get_mm_rss(mm); \
380 if ((mm)->hiwater_rss < _rss) \
381 (mm)->hiwater_rss = _rss; \
383 #define update_hiwater_vm(mm) do { \
384 if ((mm)->hiwater_vm < (mm)->total_vm) \
385 (mm)->hiwater_vm = (mm)->total_vm; \
388 extern void set_dumpable(struct mm_struct *mm, int value);
389 extern int get_dumpable(struct mm_struct *mm);
393 #define MMF_DUMPABLE 0 /* core dump is permitted */
394 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
395 #define MMF_DUMPABLE_BITS 2
397 /* coredump filter bits */
398 #define MMF_DUMP_ANON_PRIVATE 2
399 #define MMF_DUMP_ANON_SHARED 3
400 #define MMF_DUMP_MAPPED_PRIVATE 4
401 #define MMF_DUMP_MAPPED_SHARED 5
402 #define MMF_DUMP_ELF_HEADERS 6
403 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
404 #define MMF_DUMP_FILTER_BITS 5
405 #define MMF_DUMP_FILTER_MASK \
406 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
407 #define MMF_DUMP_FILTER_DEFAULT \
408 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
410 struct sighand_struct {
412 struct k_sigaction action[_NSIG];
414 wait_queue_head_t signalfd_wqh;
417 struct pacct_struct {
420 unsigned long ac_mem;
421 cputime_t ac_utime, ac_stime;
422 unsigned long ac_minflt, ac_majflt;
426 * NOTE! "signal_struct" does not have it's own
427 * locking, because a shared signal_struct always
428 * implies a shared sighand_struct, so locking
429 * sighand_struct is always a proper superset of
430 * the locking of signal_struct.
432 struct signal_struct {
436 wait_queue_head_t wait_chldexit; /* for wait4() */
438 /* current thread group signal load-balancing target: */
439 struct task_struct *curr_target;
441 /* shared signal handling: */
442 struct sigpending shared_pending;
444 /* thread group exit support */
447 * - notify group_exit_task when ->count is equal to notify_count
448 * - everyone except group_exit_task is stopped during signal delivery
449 * of fatal signals, group_exit_task processes the signal.
451 struct task_struct *group_exit_task;
454 /* thread group stop support, overloads group_exit_code too */
455 int group_stop_count;
456 unsigned int flags; /* see SIGNAL_* flags below */
458 /* POSIX.1b Interval Timers */
459 struct list_head posix_timers;
461 /* ITIMER_REAL timer for the process */
462 struct hrtimer real_timer;
463 struct pid *leader_pid;
464 ktime_t it_real_incr;
466 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
467 cputime_t it_prof_expires, it_virt_expires;
468 cputime_t it_prof_incr, it_virt_incr;
470 /* job control IDs */
473 * pgrp and session fields are deprecated.
474 * use the task_session_Xnr and task_pgrp_Xnr routines below
478 pid_t pgrp __deprecated;
482 struct pid *tty_old_pgrp;
485 pid_t session __deprecated;
489 /* boolean value for session group leader */
492 struct tty_struct *tty; /* NULL if no tty */
495 * Cumulative resource counters for dead threads in the group,
496 * and for reaped dead child processes forked by this group.
497 * Live threads maintain their own counters and add to these
498 * in __exit_signal, except for the group leader.
500 cputime_t utime, stime, cutime, cstime;
503 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
504 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
505 unsigned long inblock, oublock, cinblock, coublock;
508 * Cumulative ns of scheduled CPU time for dead threads in the
509 * group, not including a zombie group leader. (This only differs
510 * from jiffies_to_ns(utime + stime) if sched_clock uses something
511 * other than jiffies.)
513 unsigned long long sum_sched_runtime;
516 * We don't bother to synchronize most readers of this at all,
517 * because there is no reader checking a limit that actually needs
518 * to get both rlim_cur and rlim_max atomically, and either one
519 * alone is a single word that can safely be read normally.
520 * getrlimit/setrlimit use task_lock(current->group_leader) to
521 * protect this instead of the siglock, because they really
522 * have no need to disable irqs.
524 struct rlimit rlim[RLIM_NLIMITS];
526 struct list_head cpu_timers[3];
528 /* keep the process-shared keyrings here so that they do the right
529 * thing in threads created with CLONE_THREAD */
531 struct key *session_keyring; /* keyring inherited over fork */
532 struct key *process_keyring; /* keyring private to this process */
534 #ifdef CONFIG_BSD_PROCESS_ACCT
535 struct pacct_struct pacct; /* per-process accounting information */
537 #ifdef CONFIG_TASKSTATS
538 struct taskstats *stats;
542 struct tty_audit_buf *tty_audit_buf;
546 /* Context switch must be unlocked if interrupts are to be enabled */
547 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
548 # define __ARCH_WANT_UNLOCKED_CTXSW
552 * Bits in flags field of signal_struct.
554 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
555 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
556 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
557 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
559 /* If true, all threads except ->group_exit_task have pending SIGKILL */
560 static inline int signal_group_exit(const struct signal_struct *sig)
562 return (sig->flags & SIGNAL_GROUP_EXIT) ||
563 (sig->group_exit_task != NULL);
567 * Some day this will be a full-fledged user tracking system..
570 atomic_t __count; /* reference count */
571 atomic_t processes; /* How many processes does this user have? */
572 atomic_t files; /* How many open files does this user have? */
573 atomic_t sigpending; /* How many pending signals does this user have? */
574 #ifdef CONFIG_INOTIFY_USER
575 atomic_t inotify_watches; /* How many inotify watches does this user have? */
576 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
578 #ifdef CONFIG_POSIX_MQUEUE
579 /* protected by mq_lock */
580 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
582 unsigned long locked_shm; /* How many pages of mlocked shm ? */
585 struct key *uid_keyring; /* UID specific keyring */
586 struct key *session_keyring; /* UID's default session keyring */
589 /* Hash table maintenance information */
590 struct hlist_node uidhash_node;
593 #ifdef CONFIG_USER_SCHED
594 struct task_group *tg;
597 struct work_struct work;
602 extern int uids_sysfs_init(void);
604 extern struct user_struct *find_user(uid_t);
606 extern struct user_struct root_user;
607 #define INIT_USER (&root_user)
609 struct backing_dev_info;
610 struct reclaim_state;
612 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
614 /* cumulative counters */
615 unsigned long pcount; /* # of times run on this cpu */
616 unsigned long long cpu_time, /* time spent on the cpu */
617 run_delay; /* time spent waiting on a runqueue */
620 unsigned long long last_arrival,/* when we last ran on a cpu */
621 last_queued; /* when we were last queued to run */
622 #ifdef CONFIG_SCHEDSTATS
624 unsigned int bkl_count;
627 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
629 #ifdef CONFIG_SCHEDSTATS
630 extern const struct file_operations proc_schedstat_operations;
631 #endif /* CONFIG_SCHEDSTATS */
633 #ifdef CONFIG_TASK_DELAY_ACCT
634 struct task_delay_info {
636 unsigned int flags; /* Private per-task flags */
638 /* For each stat XXX, add following, aligned appropriately
640 * struct timespec XXX_start, XXX_end;
644 * Atomicity of updates to XXX_delay, XXX_count protected by
645 * single lock above (split into XXX_lock if contention is an issue).
649 * XXX_count is incremented on every XXX operation, the delay
650 * associated with the operation is added to XXX_delay.
651 * XXX_delay contains the accumulated delay time in nanoseconds.
653 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
654 u64 blkio_delay; /* wait for sync block io completion */
655 u64 swapin_delay; /* wait for swapin block io completion */
656 u32 blkio_count; /* total count of the number of sync block */
657 /* io operations performed */
658 u32 swapin_count; /* total count of the number of swapin block */
659 /* io operations performed */
661 #endif /* CONFIG_TASK_DELAY_ACCT */
663 static inline int sched_info_on(void)
665 #ifdef CONFIG_SCHEDSTATS
667 #elif defined(CONFIG_TASK_DELAY_ACCT)
668 extern int delayacct_on;
683 * sched-domains (multiprocessor balancing) declarations:
687 * Increase resolution of nice-level calculations:
689 #define SCHED_LOAD_SHIFT 10
690 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
692 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
695 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
696 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
697 #define SD_BALANCE_EXEC 4 /* Balance on exec */
698 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
699 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
700 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
701 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
702 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
703 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
704 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
705 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
706 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
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 enum sched_domain_level {
746 struct sched_domain_attr {
747 int relax_domain_level;
750 #define SD_ATTR_INIT (struct sched_domain_attr) { \
751 .relax_domain_level = -1, \
754 struct sched_domain {
755 /* These fields must be setup */
756 struct sched_domain *parent; /* top domain must be null terminated */
757 struct sched_domain *child; /* bottom domain must be null terminated */
758 struct sched_group *groups; /* the balancing groups of the domain */
759 cpumask_t span; /* span of all CPUs in this domain */
760 int first_cpu; /* cache of the first cpu in this domain */
761 unsigned long min_interval; /* Minimum balance interval ms */
762 unsigned long max_interval; /* Maximum balance interval ms */
763 unsigned int busy_factor; /* less balancing by factor if busy */
764 unsigned int imbalance_pct; /* No balance until over watermark */
765 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
766 unsigned int busy_idx;
767 unsigned int idle_idx;
768 unsigned int newidle_idx;
769 unsigned int wake_idx;
770 unsigned int forkexec_idx;
771 int flags; /* See SD_* */
772 enum sched_domain_level level;
774 /* Runtime fields. */
775 unsigned long last_balance; /* init to jiffies. units in jiffies */
776 unsigned int balance_interval; /* initialise to 1. units in ms. */
777 unsigned int nr_balance_failed; /* initialise to 0 */
779 #ifdef CONFIG_SCHEDSTATS
780 /* load_balance() stats */
781 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
782 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
783 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
784 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
785 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
786 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
787 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
788 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
790 /* Active load balancing */
791 unsigned int alb_count;
792 unsigned int alb_failed;
793 unsigned int alb_pushed;
795 /* SD_BALANCE_EXEC stats */
796 unsigned int sbe_count;
797 unsigned int sbe_balanced;
798 unsigned int sbe_pushed;
800 /* SD_BALANCE_FORK stats */
801 unsigned int sbf_count;
802 unsigned int sbf_balanced;
803 unsigned int sbf_pushed;
805 /* try_to_wake_up() stats */
806 unsigned int ttwu_wake_remote;
807 unsigned int ttwu_move_affine;
808 unsigned int ttwu_move_balance;
812 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new,
813 struct sched_domain_attr *dattr_new);
814 extern int arch_reinit_sched_domains(void);
816 #endif /* CONFIG_SMP */
819 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
820 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
821 * task of nice 0 or enough lower priority tasks to bring up the
824 static inline int above_background_load(void)
828 for_each_online_cpu(cpu) {
829 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
835 struct io_context; /* See blkdev.h */
836 #define NGROUPS_SMALL 32
837 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
841 gid_t small_block[NGROUPS_SMALL];
847 * get_group_info() must be called with the owning task locked (via task_lock())
848 * when task != current. The reason being that the vast majority of callers are
849 * looking at current->group_info, which can not be changed except by the
850 * current task. Changing current->group_info requires the task lock, too.
852 #define get_group_info(group_info) do { \
853 atomic_inc(&(group_info)->usage); \
856 #define put_group_info(group_info) do { \
857 if (atomic_dec_and_test(&(group_info)->usage)) \
858 groups_free(group_info); \
861 extern struct group_info *groups_alloc(int gidsetsize);
862 extern void groups_free(struct group_info *group_info);
863 extern int set_current_groups(struct group_info *group_info);
864 extern int groups_search(struct group_info *group_info, gid_t grp);
865 /* access the groups "array" with this macro */
866 #define GROUP_AT(gi, i) \
867 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
869 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
870 extern void prefetch_stack(struct task_struct *t);
872 static inline void prefetch_stack(struct task_struct *t) { }
875 struct audit_context; /* See audit.c */
877 struct pipe_inode_info;
878 struct uts_namespace;
884 const struct sched_class *next;
886 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
887 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
888 void (*yield_task) (struct rq *rq);
889 int (*select_task_rq)(struct task_struct *p, int sync);
891 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
893 struct task_struct * (*pick_next_task) (struct rq *rq);
894 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
897 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
898 struct rq *busiest, unsigned long max_load_move,
899 struct sched_domain *sd, enum cpu_idle_type idle,
900 int *all_pinned, int *this_best_prio);
902 int (*move_one_task) (struct rq *this_rq, int this_cpu,
903 struct rq *busiest, struct sched_domain *sd,
904 enum cpu_idle_type idle);
905 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
906 void (*post_schedule) (struct rq *this_rq);
907 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
910 void (*set_curr_task) (struct rq *rq);
911 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
912 void (*task_new) (struct rq *rq, struct task_struct *p);
913 void (*set_cpus_allowed)(struct task_struct *p,
914 const cpumask_t *newmask);
916 void (*join_domain)(struct rq *rq);
917 void (*leave_domain)(struct rq *rq);
919 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
921 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
923 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
924 int oldprio, int running);
926 #ifdef CONFIG_FAIR_GROUP_SCHED
927 void (*moved_group) (struct task_struct *p);
932 unsigned long weight, inv_weight;
936 * CFS stats for a schedulable entity (task, task-group etc)
938 * Current field usage histogram:
945 struct sched_entity {
946 struct load_weight load; /* for load-balancing */
947 struct rb_node run_node;
948 struct list_head group_node;
952 u64 sum_exec_runtime;
954 u64 prev_sum_exec_runtime;
959 #ifdef CONFIG_SCHEDSTATS
967 s64 sum_sleep_runtime;
975 u64 nr_migrations_cold;
976 u64 nr_failed_migrations_affine;
977 u64 nr_failed_migrations_running;
978 u64 nr_failed_migrations_hot;
979 u64 nr_forced_migrations;
980 u64 nr_forced2_migrations;
984 u64 nr_wakeups_migrate;
985 u64 nr_wakeups_local;
986 u64 nr_wakeups_remote;
987 u64 nr_wakeups_affine;
988 u64 nr_wakeups_affine_attempts;
989 u64 nr_wakeups_passive;
993 #ifdef CONFIG_FAIR_GROUP_SCHED
994 struct sched_entity *parent;
995 /* rq on which this entity is (to be) queued: */
996 struct cfs_rq *cfs_rq;
997 /* rq "owned" by this entity/group: */
1002 struct sched_rt_entity {
1003 struct list_head run_list;
1004 unsigned int time_slice;
1005 unsigned long timeout;
1006 int nr_cpus_allowed;
1008 struct sched_rt_entity *back;
1009 #ifdef CONFIG_RT_GROUP_SCHED
1010 struct sched_rt_entity *parent;
1011 /* rq on which this entity is (to be) queued: */
1012 struct rt_rq *rt_rq;
1013 /* rq "owned" by this entity/group: */
1018 struct task_struct {
1019 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1022 unsigned int flags; /* per process flags, defined below */
1023 unsigned int ptrace;
1025 int lock_depth; /* BKL lock depth */
1028 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1033 int prio, static_prio, normal_prio;
1034 const struct sched_class *sched_class;
1035 struct sched_entity se;
1036 struct sched_rt_entity rt;
1038 #ifdef CONFIG_PREEMPT_NOTIFIERS
1039 /* list of struct preempt_notifier: */
1040 struct hlist_head preempt_notifiers;
1044 * fpu_counter contains the number of consecutive context switches
1045 * that the FPU is used. If this is over a threshold, the lazy fpu
1046 * saving becomes unlazy to save the trap. This is an unsigned char
1047 * so that after 256 times the counter wraps and the behavior turns
1048 * lazy again; this to deal with bursty apps that only use FPU for
1051 unsigned char fpu_counter;
1052 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1053 #ifdef CONFIG_BLK_DEV_IO_TRACE
1054 unsigned int btrace_seq;
1057 unsigned int policy;
1058 cpumask_t cpus_allowed;
1060 #ifdef CONFIG_PREEMPT_RCU
1061 int rcu_read_lock_nesting;
1062 int rcu_flipctr_idx;
1063 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1065 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1066 struct sched_info sched_info;
1069 struct list_head tasks;
1071 * ptrace_list/ptrace_children forms the list of my children
1072 * that were stolen by a ptracer.
1074 struct list_head ptrace_children;
1075 struct list_head ptrace_list;
1077 struct mm_struct *mm, *active_mm;
1080 struct linux_binfmt *binfmt;
1082 int exit_code, exit_signal;
1083 int pdeath_signal; /* The signal sent when the parent dies */
1085 unsigned int personality;
1086 unsigned did_exec:1;
1090 #ifdef CONFIG_CC_STACKPROTECTOR
1091 /* Canary value for the -fstack-protector gcc feature */
1092 unsigned long stack_canary;
1095 * pointers to (original) parent process, youngest child, younger sibling,
1096 * older sibling, respectively. (p->father can be replaced with
1099 struct task_struct *real_parent; /* real parent process (when being debugged) */
1100 struct task_struct *parent; /* parent process */
1102 * children/sibling forms the list of my children plus the
1103 * tasks I'm ptracing.
1105 struct list_head children; /* list of my children */
1106 struct list_head sibling; /* linkage in my parent's children list */
1107 struct task_struct *group_leader; /* threadgroup leader */
1109 /* PID/PID hash table linkage. */
1110 struct pid_link pids[PIDTYPE_MAX];
1111 struct list_head thread_group;
1113 struct completion *vfork_done; /* for vfork() */
1114 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1115 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1117 unsigned int rt_priority;
1118 cputime_t utime, stime, utimescaled, stimescaled;
1120 cputime_t prev_utime, prev_stime;
1121 unsigned long nvcsw, nivcsw; /* context switch counts */
1122 struct timespec start_time; /* monotonic time */
1123 struct timespec real_start_time; /* boot based time */
1124 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1125 unsigned long min_flt, maj_flt;
1127 cputime_t it_prof_expires, it_virt_expires;
1128 unsigned long long it_sched_expires;
1129 struct list_head cpu_timers[3];
1131 /* process credentials */
1132 uid_t uid,euid,suid,fsuid;
1133 gid_t gid,egid,sgid,fsgid;
1134 struct group_info *group_info;
1135 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset;
1136 unsigned keep_capabilities:1;
1137 struct user_struct *user;
1139 struct key *request_key_auth; /* assumed request_key authority */
1140 struct key *thread_keyring; /* keyring private to this thread */
1141 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1143 char comm[TASK_COMM_LEN]; /* executable name excluding path
1144 - access with [gs]et_task_comm (which lock
1145 it with task_lock())
1146 - initialized normally by flush_old_exec */
1147 /* file system info */
1148 int link_count, total_link_count;
1149 #ifdef CONFIG_SYSVIPC
1151 struct sysv_sem sysvsem;
1153 #ifdef CONFIG_DETECT_SOFTLOCKUP
1154 /* hung task detection */
1155 unsigned long last_switch_timestamp;
1156 unsigned long last_switch_count;
1158 /* CPU-specific state of this task */
1159 struct thread_struct thread;
1160 /* filesystem information */
1161 struct fs_struct *fs;
1162 /* open file information */
1163 struct files_struct *files;
1165 struct nsproxy *nsproxy;
1166 /* signal handlers */
1167 struct signal_struct *signal;
1168 struct sighand_struct *sighand;
1170 sigset_t blocked, real_blocked;
1171 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
1172 struct sigpending pending;
1174 unsigned long sas_ss_sp;
1176 int (*notifier)(void *priv);
1177 void *notifier_data;
1178 sigset_t *notifier_mask;
1179 #ifdef CONFIG_SECURITY
1182 struct audit_context *audit_context;
1183 #ifdef CONFIG_AUDITSYSCALL
1185 unsigned int sessionid;
1189 /* Thread group tracking */
1192 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1193 spinlock_t alloc_lock;
1195 /* Protection of the PI data structures: */
1198 #ifdef CONFIG_RT_MUTEXES
1199 /* PI waiters blocked on a rt_mutex held by this task */
1200 struct plist_head pi_waiters;
1201 /* Deadlock detection and priority inheritance handling */
1202 struct rt_mutex_waiter *pi_blocked_on;
1205 #ifdef CONFIG_DEBUG_MUTEXES
1206 /* mutex deadlock detection */
1207 struct mutex_waiter *blocked_on;
1209 #ifdef CONFIG_TRACE_IRQFLAGS
1210 unsigned int irq_events;
1211 int hardirqs_enabled;
1212 unsigned long hardirq_enable_ip;
1213 unsigned int hardirq_enable_event;
1214 unsigned long hardirq_disable_ip;
1215 unsigned int hardirq_disable_event;
1216 int softirqs_enabled;
1217 unsigned long softirq_disable_ip;
1218 unsigned int softirq_disable_event;
1219 unsigned long softirq_enable_ip;
1220 unsigned int softirq_enable_event;
1221 int hardirq_context;
1222 int softirq_context;
1224 #ifdef CONFIG_LOCKDEP
1225 # define MAX_LOCK_DEPTH 48UL
1228 struct held_lock held_locks[MAX_LOCK_DEPTH];
1229 unsigned int lockdep_recursion;
1232 /* journalling filesystem info */
1235 /* stacked block device info */
1236 struct bio *bio_list, **bio_tail;
1239 struct reclaim_state *reclaim_state;
1241 struct backing_dev_info *backing_dev_info;
1243 struct io_context *io_context;
1245 unsigned long ptrace_message;
1246 siginfo_t *last_siginfo; /* For ptrace use. */
1247 #ifdef CONFIG_TASK_XACCT
1248 /* i/o counters(bytes read/written, #syscalls */
1249 u64 rchar, wchar, syscr, syscw;
1251 struct task_io_accounting ioac;
1252 #if defined(CONFIG_TASK_XACCT)
1253 u64 acct_rss_mem1; /* accumulated rss usage */
1254 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1255 cputime_t acct_stimexpd;/* stime since last update */
1258 struct mempolicy *mempolicy;
1261 #ifdef CONFIG_CPUSETS
1262 nodemask_t mems_allowed;
1263 int cpuset_mems_generation;
1264 int cpuset_mem_spread_rotor;
1266 #ifdef CONFIG_CGROUPS
1267 /* Control Group info protected by css_set_lock */
1268 struct css_set *cgroups;
1269 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1270 struct list_head cg_list;
1273 struct robust_list_head __user *robust_list;
1274 #ifdef CONFIG_COMPAT
1275 struct compat_robust_list_head __user *compat_robust_list;
1277 struct list_head pi_state_list;
1278 struct futex_pi_state *pi_state_cache;
1280 atomic_t fs_excl; /* holding fs exclusive resources */
1281 struct rcu_head rcu;
1284 * cache last used pipe for splice
1286 struct pipe_inode_info *splice_pipe;
1287 #ifdef CONFIG_TASK_DELAY_ACCT
1288 struct task_delay_info *delays;
1290 #ifdef CONFIG_FAULT_INJECTION
1293 struct prop_local_single dirties;
1294 #ifdef CONFIG_LATENCYTOP
1295 int latency_record_count;
1296 struct latency_record latency_record[LT_SAVECOUNT];
1301 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1302 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1303 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1304 * values are inverted: lower p->prio value means higher priority.
1306 * The MAX_USER_RT_PRIO value allows the actual maximum
1307 * RT priority to be separate from the value exported to
1308 * user-space. This allows kernel threads to set their
1309 * priority to a value higher than any user task. Note:
1310 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1313 #define MAX_USER_RT_PRIO 100
1314 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1316 #define MAX_PRIO (MAX_RT_PRIO + 40)
1317 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1319 static inline int rt_prio(int prio)
1321 if (unlikely(prio < MAX_RT_PRIO))
1326 static inline int rt_task(struct task_struct *p)
1328 return rt_prio(p->prio);
1331 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1333 tsk->signal->__session = session;
1336 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1338 tsk->signal->__pgrp = pgrp;
1341 static inline struct pid *task_pid(struct task_struct *task)
1343 return task->pids[PIDTYPE_PID].pid;
1346 static inline struct pid *task_tgid(struct task_struct *task)
1348 return task->group_leader->pids[PIDTYPE_PID].pid;
1351 static inline struct pid *task_pgrp(struct task_struct *task)
1353 return task->group_leader->pids[PIDTYPE_PGID].pid;
1356 static inline struct pid *task_session(struct task_struct *task)
1358 return task->group_leader->pids[PIDTYPE_SID].pid;
1361 struct pid_namespace;
1364 * the helpers to get the task's different pids as they are seen
1365 * from various namespaces
1367 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1368 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1370 * task_xid_nr_ns() : id seen from the ns specified;
1372 * set_task_vxid() : assigns a virtual id to a task;
1374 * see also pid_nr() etc in include/linux/pid.h
1377 static inline pid_t task_pid_nr(struct task_struct *tsk)
1382 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1384 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1386 return pid_vnr(task_pid(tsk));
1390 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1395 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1397 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1399 return pid_vnr(task_tgid(tsk));
1403 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1405 return tsk->signal->__pgrp;
1408 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1410 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1412 return pid_vnr(task_pgrp(tsk));
1416 static inline pid_t task_session_nr(struct task_struct *tsk)
1418 return tsk->signal->__session;
1421 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1423 static inline pid_t task_session_vnr(struct task_struct *tsk)
1425 return pid_vnr(task_session(tsk));
1430 * pid_alive - check that a task structure is not stale
1431 * @p: Task structure to be checked.
1433 * Test if a process is not yet dead (at most zombie state)
1434 * If pid_alive fails, then pointers within the task structure
1435 * can be stale and must not be dereferenced.
1437 static inline int pid_alive(struct task_struct *p)
1439 return p->pids[PIDTYPE_PID].pid != NULL;
1443 * is_global_init - check if a task structure is init
1444 * @tsk: Task structure to be checked.
1446 * Check if a task structure is the first user space task the kernel created.
1448 static inline int is_global_init(struct task_struct *tsk)
1450 return tsk->pid == 1;
1454 * is_container_init:
1455 * check whether in the task is init in its own pid namespace.
1457 extern int is_container_init(struct task_struct *tsk);
1459 extern struct pid *cad_pid;
1461 extern void free_task(struct task_struct *tsk);
1462 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1464 extern void __put_task_struct(struct task_struct *t);
1466 static inline void put_task_struct(struct task_struct *t)
1468 if (atomic_dec_and_test(&t->usage))
1469 __put_task_struct(t);
1475 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1476 /* Not implemented yet, only for 486*/
1477 #define PF_STARTING 0x00000002 /* being created */
1478 #define PF_EXITING 0x00000004 /* getting shut down */
1479 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1480 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1481 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1482 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1483 #define PF_DUMPCORE 0x00000200 /* dumped core */
1484 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1485 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1486 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1487 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1488 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1489 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1490 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1491 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1492 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1493 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1494 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1495 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1496 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1497 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1498 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1499 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1500 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1501 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1504 * Only the _current_ task can read/write to tsk->flags, but other
1505 * tasks can access tsk->flags in readonly mode for example
1506 * with tsk_used_math (like during threaded core dumping).
1507 * There is however an exception to this rule during ptrace
1508 * or during fork: the ptracer task is allowed to write to the
1509 * child->flags of its traced child (same goes for fork, the parent
1510 * can write to the child->flags), because we're guaranteed the
1511 * child is not running and in turn not changing child->flags
1512 * at the same time the parent does it.
1514 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1515 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1516 #define clear_used_math() clear_stopped_child_used_math(current)
1517 #define set_used_math() set_stopped_child_used_math(current)
1518 #define conditional_stopped_child_used_math(condition, child) \
1519 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1520 #define conditional_used_math(condition) \
1521 conditional_stopped_child_used_math(condition, current)
1522 #define copy_to_stopped_child_used_math(child) \
1523 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1524 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1525 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1526 #define used_math() tsk_used_math(current)
1529 extern int set_cpus_allowed_ptr(struct task_struct *p,
1530 const cpumask_t *new_mask);
1532 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1533 const cpumask_t *new_mask)
1535 if (!cpu_isset(0, *new_mask))
1540 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1542 return set_cpus_allowed_ptr(p, &new_mask);
1545 extern unsigned long long sched_clock(void);
1548 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1549 * clock constructed from sched_clock():
1551 extern unsigned long long cpu_clock(int cpu);
1553 extern unsigned long long
1554 task_sched_runtime(struct task_struct *task);
1556 /* sched_exec is called by processes performing an exec */
1558 extern void sched_exec(void);
1560 #define sched_exec() {}
1563 extern void sched_clock_idle_sleep_event(void);
1564 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1566 #ifdef CONFIG_HOTPLUG_CPU
1567 extern void idle_task_exit(void);
1569 static inline void idle_task_exit(void) {}
1572 extern void sched_idle_next(void);
1574 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1575 extern void wake_up_idle_cpu(int cpu);
1577 static inline void wake_up_idle_cpu(int cpu) { }
1580 #ifdef CONFIG_SCHED_DEBUG
1581 extern unsigned int sysctl_sched_latency;
1582 extern unsigned int sysctl_sched_min_granularity;
1583 extern unsigned int sysctl_sched_wakeup_granularity;
1584 extern unsigned int sysctl_sched_child_runs_first;
1585 extern unsigned int sysctl_sched_features;
1586 extern unsigned int sysctl_sched_migration_cost;
1587 extern unsigned int sysctl_sched_nr_migrate;
1589 int sched_nr_latency_handler(struct ctl_table *table, int write,
1590 struct file *file, void __user *buffer, size_t *length,
1593 extern unsigned int sysctl_sched_rt_period;
1594 extern int sysctl_sched_rt_runtime;
1596 int sched_rt_handler(struct ctl_table *table, int write,
1597 struct file *filp, void __user *buffer, size_t *lenp,
1600 extern unsigned int sysctl_sched_compat_yield;
1602 #ifdef CONFIG_RT_MUTEXES
1603 extern int rt_mutex_getprio(struct task_struct *p);
1604 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1605 extern void rt_mutex_adjust_pi(struct task_struct *p);
1607 static inline int rt_mutex_getprio(struct task_struct *p)
1609 return p->normal_prio;
1611 # define rt_mutex_adjust_pi(p) do { } while (0)
1614 extern void set_user_nice(struct task_struct *p, long nice);
1615 extern int task_prio(const struct task_struct *p);
1616 extern int task_nice(const struct task_struct *p);
1617 extern int can_nice(const struct task_struct *p, const int nice);
1618 extern int task_curr(const struct task_struct *p);
1619 extern int idle_cpu(int cpu);
1620 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1621 extern struct task_struct *idle_task(int cpu);
1622 extern struct task_struct *curr_task(int cpu);
1623 extern void set_curr_task(int cpu, struct task_struct *p);
1628 * The default (Linux) execution domain.
1630 extern struct exec_domain default_exec_domain;
1632 union thread_union {
1633 struct thread_info thread_info;
1634 unsigned long stack[THREAD_SIZE/sizeof(long)];
1637 #ifndef __HAVE_ARCH_KSTACK_END
1638 static inline int kstack_end(void *addr)
1640 /* Reliable end of stack detection:
1641 * Some APM bios versions misalign the stack
1643 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1647 extern union thread_union init_thread_union;
1648 extern struct task_struct init_task;
1650 extern struct mm_struct init_mm;
1652 extern struct pid_namespace init_pid_ns;
1655 * find a task by one of its numerical ids
1657 * find_task_by_pid_type_ns():
1658 * it is the most generic call - it finds a task by all id,
1659 * type and namespace specified
1660 * find_task_by_pid_ns():
1661 * finds a task by its pid in the specified namespace
1662 * find_task_by_vpid():
1663 * finds a task by its virtual pid
1664 * find_task_by_pid():
1665 * finds a task by its global pid
1667 * see also find_pid() etc in include/linux/pid.h
1670 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1671 struct pid_namespace *ns);
1673 extern struct task_struct *find_task_by_pid(pid_t nr);
1674 extern struct task_struct *find_task_by_vpid(pid_t nr);
1675 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1676 struct pid_namespace *ns);
1678 extern void __set_special_pids(struct pid *pid);
1680 /* per-UID process charging. */
1681 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1682 static inline struct user_struct *get_uid(struct user_struct *u)
1684 atomic_inc(&u->__count);
1687 extern void free_uid(struct user_struct *);
1688 extern void switch_uid(struct user_struct *);
1689 extern void release_uids(struct user_namespace *ns);
1691 #include <asm/current.h>
1693 extern void do_timer(unsigned long ticks);
1695 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1696 extern int wake_up_process(struct task_struct *tsk);
1697 extern void wake_up_new_task(struct task_struct *tsk,
1698 unsigned long clone_flags);
1700 extern void kick_process(struct task_struct *tsk);
1702 static inline void kick_process(struct task_struct *tsk) { }
1704 extern void sched_fork(struct task_struct *p, int clone_flags);
1705 extern void sched_dead(struct task_struct *p);
1707 extern int in_group_p(gid_t);
1708 extern int in_egroup_p(gid_t);
1710 extern void proc_caches_init(void);
1711 extern void flush_signals(struct task_struct *);
1712 extern void ignore_signals(struct task_struct *);
1713 extern void flush_signal_handlers(struct task_struct *, int force_default);
1714 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1716 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1718 unsigned long flags;
1721 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1722 ret = dequeue_signal(tsk, mask, info);
1723 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1728 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1730 extern void unblock_all_signals(void);
1731 extern void release_task(struct task_struct * p);
1732 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1733 extern int force_sigsegv(int, struct task_struct *);
1734 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1735 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1736 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1737 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1738 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1739 extern int kill_pid(struct pid *pid, int sig, int priv);
1740 extern int kill_proc_info(int, struct siginfo *, pid_t);
1741 extern void do_notify_parent(struct task_struct *, int);
1742 extern void force_sig(int, struct task_struct *);
1743 extern void force_sig_specific(int, struct task_struct *);
1744 extern int send_sig(int, struct task_struct *, int);
1745 extern void zap_other_threads(struct task_struct *p);
1746 extern int kill_proc(pid_t, int, int);
1747 extern struct sigqueue *sigqueue_alloc(void);
1748 extern void sigqueue_free(struct sigqueue *);
1749 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1750 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
1751 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1752 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1754 static inline int kill_cad_pid(int sig, int priv)
1756 return kill_pid(cad_pid, sig, priv);
1759 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1760 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1761 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1762 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1764 static inline int is_si_special(const struct siginfo *info)
1766 return info <= SEND_SIG_FORCED;
1769 /* True if we are on the alternate signal stack. */
1771 static inline int on_sig_stack(unsigned long sp)
1773 return (sp - current->sas_ss_sp < current->sas_ss_size);
1776 static inline int sas_ss_flags(unsigned long sp)
1778 return (current->sas_ss_size == 0 ? SS_DISABLE
1779 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1783 * Routines for handling mm_structs
1785 extern struct mm_struct * mm_alloc(void);
1787 /* mmdrop drops the mm and the page tables */
1788 extern void __mmdrop(struct mm_struct *);
1789 static inline void mmdrop(struct mm_struct * mm)
1791 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1795 /* mmput gets rid of the mappings and all user-space */
1796 extern void mmput(struct mm_struct *);
1797 /* Grab a reference to a task's mm, if it is not already going away */
1798 extern struct mm_struct *get_task_mm(struct task_struct *task);
1799 /* Remove the current tasks stale references to the old mm_struct */
1800 extern void mm_release(struct task_struct *, struct mm_struct *);
1801 /* Allocate a new mm structure and copy contents from tsk->mm */
1802 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1804 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1805 extern void flush_thread(void);
1806 extern void exit_thread(void);
1808 extern void exit_files(struct task_struct *);
1809 extern void __cleanup_signal(struct signal_struct *);
1810 extern void __cleanup_sighand(struct sighand_struct *);
1811 extern void exit_itimers(struct signal_struct *);
1813 extern NORET_TYPE void do_group_exit(int);
1815 extern void daemonize(const char *, ...);
1816 extern int allow_signal(int);
1817 extern int disallow_signal(int);
1819 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1820 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1821 struct task_struct *fork_idle(int);
1823 extern void set_task_comm(struct task_struct *tsk, char *from);
1824 extern char *get_task_comm(char *to, struct task_struct *tsk);
1827 extern void wait_task_inactive(struct task_struct * p);
1829 #define wait_task_inactive(p) do { } while (0)
1832 #define remove_parent(p) list_del_init(&(p)->sibling)
1833 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1835 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1837 #define for_each_process(p) \
1838 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1841 * Careful: do_each_thread/while_each_thread is a double loop so
1842 * 'break' will not work as expected - use goto instead.
1844 #define do_each_thread(g, t) \
1845 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1847 #define while_each_thread(g, t) \
1848 while ((t = next_thread(t)) != g)
1850 /* de_thread depends on thread_group_leader not being a pid based check */
1851 #define thread_group_leader(p) (p == p->group_leader)
1853 /* Do to the insanities of de_thread it is possible for a process
1854 * to have the pid of the thread group leader without actually being
1855 * the thread group leader. For iteration through the pids in proc
1856 * all we care about is that we have a task with the appropriate
1857 * pid, we don't actually care if we have the right task.
1859 static inline int has_group_leader_pid(struct task_struct *p)
1861 return p->pid == p->tgid;
1865 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1867 return p1->tgid == p2->tgid;
1870 static inline struct task_struct *next_thread(const struct task_struct *p)
1872 return list_entry(rcu_dereference(p->thread_group.next),
1873 struct task_struct, thread_group);
1876 static inline int thread_group_empty(struct task_struct *p)
1878 return list_empty(&p->thread_group);
1881 #define delay_group_leader(p) \
1882 (thread_group_leader(p) && !thread_group_empty(p))
1885 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1886 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1887 * pins the final release of task.io_context. Also protects ->cpuset and
1888 * ->cgroup.subsys[].
1890 * Nests both inside and outside of read_lock(&tasklist_lock).
1891 * It must not be nested with write_lock_irq(&tasklist_lock),
1892 * neither inside nor outside.
1894 static inline void task_lock(struct task_struct *p)
1896 spin_lock(&p->alloc_lock);
1899 static inline void task_unlock(struct task_struct *p)
1901 spin_unlock(&p->alloc_lock);
1904 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1905 unsigned long *flags);
1907 static inline void unlock_task_sighand(struct task_struct *tsk,
1908 unsigned long *flags)
1910 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1913 #ifndef __HAVE_THREAD_FUNCTIONS
1915 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1916 #define task_stack_page(task) ((task)->stack)
1918 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1920 *task_thread_info(p) = *task_thread_info(org);
1921 task_thread_info(p)->task = p;
1924 static inline unsigned long *end_of_stack(struct task_struct *p)
1926 return (unsigned long *)(task_thread_info(p) + 1);
1931 extern void thread_info_cache_init(void);
1933 /* set thread flags in other task's structures
1934 * - see asm/thread_info.h for TIF_xxxx flags available
1936 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1938 set_ti_thread_flag(task_thread_info(tsk), flag);
1941 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1943 clear_ti_thread_flag(task_thread_info(tsk), flag);
1946 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1948 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1951 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1953 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1956 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1958 return test_ti_thread_flag(task_thread_info(tsk), flag);
1961 static inline void set_tsk_need_resched(struct task_struct *tsk)
1963 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1966 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1968 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1971 static inline int signal_pending(struct task_struct *p)
1973 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1976 extern int __fatal_signal_pending(struct task_struct *p);
1978 static inline int fatal_signal_pending(struct task_struct *p)
1980 return signal_pending(p) && __fatal_signal_pending(p);
1983 static inline int need_resched(void)
1985 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1989 * cond_resched() and cond_resched_lock(): latency reduction via
1990 * explicit rescheduling in places that are safe. The return
1991 * value indicates whether a reschedule was done in fact.
1992 * cond_resched_lock() will drop the spinlock before scheduling,
1993 * cond_resched_softirq() will enable bhs before scheduling.
1995 #ifdef CONFIG_PREEMPT
1996 static inline int cond_resched(void)
2001 extern int _cond_resched(void);
2002 static inline int cond_resched(void)
2004 return _cond_resched();
2007 extern int cond_resched_lock(spinlock_t * lock);
2008 extern int cond_resched_softirq(void);
2011 * Does a critical section need to be broken due to another
2012 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2013 * but a general need for low latency)
2015 static inline int spin_needbreak(spinlock_t *lock)
2017 #ifdef CONFIG_PREEMPT
2018 return spin_is_contended(lock);
2025 * Reevaluate whether the task has signals pending delivery.
2026 * Wake the task if so.
2027 * This is required every time the blocked sigset_t changes.
2028 * callers must hold sighand->siglock.
2030 extern void recalc_sigpending_and_wake(struct task_struct *t);
2031 extern void recalc_sigpending(void);
2033 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2036 * Wrappers for p->thread_info->cpu access. No-op on UP.
2040 static inline unsigned int task_cpu(const struct task_struct *p)
2042 return task_thread_info(p)->cpu;
2045 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2049 static inline unsigned int task_cpu(const struct task_struct *p)
2054 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2058 #endif /* CONFIG_SMP */
2060 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2061 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2063 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
2065 mm->mmap_base = TASK_UNMAPPED_BASE;
2066 mm->get_unmapped_area = arch_get_unmapped_area;
2067 mm->unmap_area = arch_unmap_area;
2071 extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask);
2072 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2074 extern int sched_mc_power_savings, sched_smt_power_savings;
2076 extern void normalize_rt_tasks(void);
2078 #ifdef CONFIG_GROUP_SCHED
2080 extern struct task_group init_task_group;
2081 #ifdef CONFIG_USER_SCHED
2082 extern struct task_group root_task_group;
2085 extern struct task_group *sched_create_group(struct task_group *parent);
2086 extern void sched_destroy_group(struct task_group *tg);
2087 extern void sched_move_task(struct task_struct *tsk);
2088 #ifdef CONFIG_FAIR_GROUP_SCHED
2089 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2090 extern unsigned long sched_group_shares(struct task_group *tg);
2092 #ifdef CONFIG_RT_GROUP_SCHED
2093 extern int sched_group_set_rt_runtime(struct task_group *tg,
2094 long rt_runtime_us);
2095 extern long sched_group_rt_runtime(struct task_group *tg);
2096 extern int sched_group_set_rt_period(struct task_group *tg,
2098 extern long sched_group_rt_period(struct task_group *tg);
2102 #ifdef CONFIG_TASK_XACCT
2103 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2108 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2113 static inline void inc_syscr(struct task_struct *tsk)
2118 static inline void inc_syscw(struct task_struct *tsk)
2123 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2127 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2131 static inline void inc_syscr(struct task_struct *tsk)
2135 static inline void inc_syscw(struct task_struct *tsk)
2141 void migration_init(void);
2143 static inline void migration_init(void)
2148 #ifndef TASK_SIZE_OF
2149 #define TASK_SIZE_OF(tsk) TASK_SIZE
2152 #endif /* __KERNEL__ */