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/fs_struct.h>
72 #include <linux/compiler.h>
73 #include <linux/completion.h>
74 #include <linux/pid.h>
75 #include <linux/percpu.h>
76 #include <linux/topology.h>
77 #include <linux/proportions.h>
78 #include <linux/seccomp.h>
79 #include <linux/rcupdate.h>
80 #include <linux/rtmutex.h>
82 #include <linux/time.h>
83 #include <linux/param.h>
84 #include <linux/resource.h>
85 #include <linux/timer.h>
86 #include <linux/hrtimer.h>
87 #include <linux/task_io_accounting.h>
88 #include <linux/kobject.h>
89 #include <linux/latencytop.h>
91 #include <asm/processor.h>
95 struct futex_pi_state;
96 struct robust_list_head;
100 * List of flags we want to share for kernel threads,
101 * if only because they are not used by them anyway.
103 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
106 * These are the constant used to fake the fixed-point load-average
107 * counting. Some notes:
108 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
109 * a load-average precision of 10 bits integer + 11 bits fractional
110 * - if you want to count load-averages more often, you need more
111 * precision, or rounding will get you. With 2-second counting freq,
112 * the EXP_n values would be 1981, 2034 and 2043 if still using only
115 extern unsigned long avenrun[]; /* Load averages */
117 #define FSHIFT 11 /* nr of bits of precision */
118 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
119 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
120 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
121 #define EXP_5 2014 /* 1/exp(5sec/5min) */
122 #define EXP_15 2037 /* 1/exp(5sec/15min) */
124 #define CALC_LOAD(load,exp,n) \
126 load += n*(FIXED_1-exp); \
129 extern unsigned long total_forks;
130 extern int nr_threads;
131 DECLARE_PER_CPU(unsigned long, process_counts);
132 extern int nr_processes(void);
133 extern unsigned long nr_running(void);
134 extern unsigned long nr_uninterruptible(void);
135 extern unsigned long nr_active(void);
136 extern unsigned long nr_iowait(void);
137 extern unsigned long weighted_cpuload(const int cpu);
142 #ifdef CONFIG_SCHED_DEBUG
143 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
144 extern void proc_sched_set_task(struct task_struct *p);
146 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
149 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
152 static inline void proc_sched_set_task(struct task_struct *p)
156 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
161 extern unsigned long long time_sync_thresh;
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 * Pending notifications to parent.
562 #define SIGNAL_CLD_STOPPED 0x00000010
563 #define SIGNAL_CLD_CONTINUED 0x00000020
564 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
566 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
568 /* If true, all threads except ->group_exit_task have pending SIGKILL */
569 static inline int signal_group_exit(const struct signal_struct *sig)
571 return (sig->flags & SIGNAL_GROUP_EXIT) ||
572 (sig->group_exit_task != NULL);
576 * Some day this will be a full-fledged user tracking system..
579 atomic_t __count; /* reference count */
580 atomic_t processes; /* How many processes does this user have? */
581 atomic_t files; /* How many open files does this user have? */
582 atomic_t sigpending; /* How many pending signals does this user have? */
583 #ifdef CONFIG_INOTIFY_USER
584 atomic_t inotify_watches; /* How many inotify watches does this user have? */
585 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
587 #ifdef CONFIG_POSIX_MQUEUE
588 /* protected by mq_lock */
589 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
591 unsigned long locked_shm; /* How many pages of mlocked shm ? */
594 struct key *uid_keyring; /* UID specific keyring */
595 struct key *session_keyring; /* UID's default session keyring */
598 /* Hash table maintenance information */
599 struct hlist_node uidhash_node;
602 #ifdef CONFIG_USER_SCHED
603 struct task_group *tg;
606 struct work_struct work;
611 extern int uids_sysfs_init(void);
613 extern struct user_struct *find_user(uid_t);
615 extern struct user_struct root_user;
616 #define INIT_USER (&root_user)
618 struct backing_dev_info;
619 struct reclaim_state;
621 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
623 /* cumulative counters */
624 unsigned long pcount; /* # of times run on this cpu */
625 unsigned long long cpu_time, /* time spent on the cpu */
626 run_delay; /* time spent waiting on a runqueue */
629 unsigned long long last_arrival,/* when we last ran on a cpu */
630 last_queued; /* when we were last queued to run */
631 #ifdef CONFIG_SCHEDSTATS
633 unsigned int bkl_count;
636 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
638 #ifdef CONFIG_SCHEDSTATS
639 extern const struct file_operations proc_schedstat_operations;
640 #endif /* CONFIG_SCHEDSTATS */
642 #ifdef CONFIG_TASK_DELAY_ACCT
643 struct task_delay_info {
645 unsigned int flags; /* Private per-task flags */
647 /* For each stat XXX, add following, aligned appropriately
649 * struct timespec XXX_start, XXX_end;
653 * Atomicity of updates to XXX_delay, XXX_count protected by
654 * single lock above (split into XXX_lock if contention is an issue).
658 * XXX_count is incremented on every XXX operation, the delay
659 * associated with the operation is added to XXX_delay.
660 * XXX_delay contains the accumulated delay time in nanoseconds.
662 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
663 u64 blkio_delay; /* wait for sync block io completion */
664 u64 swapin_delay; /* wait for swapin block io completion */
665 u32 blkio_count; /* total count of the number of sync block */
666 /* io operations performed */
667 u32 swapin_count; /* total count of the number of swapin block */
668 /* io operations performed */
670 #endif /* CONFIG_TASK_DELAY_ACCT */
672 static inline int sched_info_on(void)
674 #ifdef CONFIG_SCHEDSTATS
676 #elif defined(CONFIG_TASK_DELAY_ACCT)
677 extern int delayacct_on;
692 * sched-domains (multiprocessor balancing) declarations:
696 * Increase resolution of nice-level calculations:
698 #define SCHED_LOAD_SHIFT 10
699 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
701 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
704 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
705 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
706 #define SD_BALANCE_EXEC 4 /* Balance on exec */
707 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
708 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
709 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
710 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
711 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
712 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
713 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
714 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
715 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
717 #define BALANCE_FOR_MC_POWER \
718 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
720 #define BALANCE_FOR_PKG_POWER \
721 ((sched_mc_power_savings || sched_smt_power_savings) ? \
722 SD_POWERSAVINGS_BALANCE : 0)
724 #define test_sd_parent(sd, flag) ((sd->parent && \
725 (sd->parent->flags & flag)) ? 1 : 0)
729 struct sched_group *next; /* Must be a circular list */
733 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
734 * single CPU. This is read only (except for setup, hotplug CPU).
735 * Note : Never change cpu_power without recompute its reciprocal
737 unsigned int __cpu_power;
739 * reciprocal value of cpu_power to avoid expensive divides
740 * (see include/linux/reciprocal_div.h)
742 u32 reciprocal_cpu_power;
745 enum sched_domain_level {
755 struct sched_domain_attr {
756 int relax_domain_level;
759 #define SD_ATTR_INIT (struct sched_domain_attr) { \
760 .relax_domain_level = -1, \
763 struct sched_domain {
764 /* These fields must be setup */
765 struct sched_domain *parent; /* top domain must be null terminated */
766 struct sched_domain *child; /* bottom domain must be null terminated */
767 struct sched_group *groups; /* the balancing groups of the domain */
768 cpumask_t span; /* span of all CPUs in this domain */
769 int first_cpu; /* cache of the first cpu in this domain */
770 unsigned long min_interval; /* Minimum balance interval ms */
771 unsigned long max_interval; /* Maximum balance interval ms */
772 unsigned int busy_factor; /* less balancing by factor if busy */
773 unsigned int imbalance_pct; /* No balance until over watermark */
774 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
775 unsigned int busy_idx;
776 unsigned int idle_idx;
777 unsigned int newidle_idx;
778 unsigned int wake_idx;
779 unsigned int forkexec_idx;
780 int flags; /* See SD_* */
781 enum sched_domain_level level;
783 /* Runtime fields. */
784 unsigned long last_balance; /* init to jiffies. units in jiffies */
785 unsigned int balance_interval; /* initialise to 1. units in ms. */
786 unsigned int nr_balance_failed; /* initialise to 0 */
788 #ifdef CONFIG_SCHEDSTATS
789 /* load_balance() stats */
790 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
791 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
792 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
793 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
794 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
795 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
796 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
797 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
799 /* Active load balancing */
800 unsigned int alb_count;
801 unsigned int alb_failed;
802 unsigned int alb_pushed;
804 /* SD_BALANCE_EXEC stats */
805 unsigned int sbe_count;
806 unsigned int sbe_balanced;
807 unsigned int sbe_pushed;
809 /* SD_BALANCE_FORK stats */
810 unsigned int sbf_count;
811 unsigned int sbf_balanced;
812 unsigned int sbf_pushed;
814 /* try_to_wake_up() stats */
815 unsigned int ttwu_wake_remote;
816 unsigned int ttwu_move_affine;
817 unsigned int ttwu_move_balance;
821 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new,
822 struct sched_domain_attr *dattr_new);
823 extern int arch_reinit_sched_domains(void);
825 #endif /* CONFIG_SMP */
828 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
829 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
830 * task of nice 0 or enough lower priority tasks to bring up the
833 static inline int above_background_load(void)
837 for_each_online_cpu(cpu) {
838 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
844 struct io_context; /* See blkdev.h */
845 #define NGROUPS_SMALL 32
846 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
850 gid_t small_block[NGROUPS_SMALL];
856 * get_group_info() must be called with the owning task locked (via task_lock())
857 * when task != current. The reason being that the vast majority of callers are
858 * looking at current->group_info, which can not be changed except by the
859 * current task. Changing current->group_info requires the task lock, too.
861 #define get_group_info(group_info) do { \
862 atomic_inc(&(group_info)->usage); \
865 #define put_group_info(group_info) do { \
866 if (atomic_dec_and_test(&(group_info)->usage)) \
867 groups_free(group_info); \
870 extern struct group_info *groups_alloc(int gidsetsize);
871 extern void groups_free(struct group_info *group_info);
872 extern int set_current_groups(struct group_info *group_info);
873 extern int groups_search(struct group_info *group_info, gid_t grp);
874 /* access the groups "array" with this macro */
875 #define GROUP_AT(gi, i) \
876 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
878 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
879 extern void prefetch_stack(struct task_struct *t);
881 static inline void prefetch_stack(struct task_struct *t) { }
884 struct audit_context; /* See audit.c */
886 struct pipe_inode_info;
887 struct uts_namespace;
893 const struct sched_class *next;
895 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
896 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
897 void (*yield_task) (struct rq *rq);
898 int (*select_task_rq)(struct task_struct *p, int sync);
900 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
902 struct task_struct * (*pick_next_task) (struct rq *rq);
903 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
906 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
907 struct rq *busiest, unsigned long max_load_move,
908 struct sched_domain *sd, enum cpu_idle_type idle,
909 int *all_pinned, int *this_best_prio);
911 int (*move_one_task) (struct rq *this_rq, int this_cpu,
912 struct rq *busiest, struct sched_domain *sd,
913 enum cpu_idle_type idle);
914 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
915 void (*post_schedule) (struct rq *this_rq);
916 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
919 void (*set_curr_task) (struct rq *rq);
920 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
921 void (*task_new) (struct rq *rq, struct task_struct *p);
922 void (*set_cpus_allowed)(struct task_struct *p,
923 const cpumask_t *newmask);
925 void (*join_domain)(struct rq *rq);
926 void (*leave_domain)(struct rq *rq);
928 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
930 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
932 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
933 int oldprio, int running);
935 #ifdef CONFIG_FAIR_GROUP_SCHED
936 void (*moved_group) (struct task_struct *p);
941 unsigned long weight, inv_weight;
945 * CFS stats for a schedulable entity (task, task-group etc)
947 * Current field usage histogram:
954 struct sched_entity {
955 struct load_weight load; /* for load-balancing */
956 struct rb_node run_node;
957 struct list_head group_node;
961 u64 sum_exec_runtime;
963 u64 prev_sum_exec_runtime;
968 #ifdef CONFIG_SCHEDSTATS
976 s64 sum_sleep_runtime;
984 u64 nr_migrations_cold;
985 u64 nr_failed_migrations_affine;
986 u64 nr_failed_migrations_running;
987 u64 nr_failed_migrations_hot;
988 u64 nr_forced_migrations;
989 u64 nr_forced2_migrations;
993 u64 nr_wakeups_migrate;
994 u64 nr_wakeups_local;
995 u64 nr_wakeups_remote;
996 u64 nr_wakeups_affine;
997 u64 nr_wakeups_affine_attempts;
998 u64 nr_wakeups_passive;
1002 #ifdef CONFIG_FAIR_GROUP_SCHED
1003 struct sched_entity *parent;
1004 /* rq on which this entity is (to be) queued: */
1005 struct cfs_rq *cfs_rq;
1006 /* rq "owned" by this entity/group: */
1007 struct cfs_rq *my_q;
1011 struct sched_rt_entity {
1012 struct list_head run_list;
1013 unsigned int time_slice;
1014 unsigned long timeout;
1015 int nr_cpus_allowed;
1017 struct sched_rt_entity *back;
1018 #ifdef CONFIG_RT_GROUP_SCHED
1019 struct sched_rt_entity *parent;
1020 /* rq on which this entity is (to be) queued: */
1021 struct rt_rq *rt_rq;
1022 /* rq "owned" by this entity/group: */
1027 struct task_struct {
1028 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1031 unsigned int flags; /* per process flags, defined below */
1032 unsigned int ptrace;
1034 int lock_depth; /* BKL lock depth */
1037 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1042 int prio, static_prio, normal_prio;
1043 const struct sched_class *sched_class;
1044 struct sched_entity se;
1045 struct sched_rt_entity rt;
1047 #ifdef CONFIG_PREEMPT_NOTIFIERS
1048 /* list of struct preempt_notifier: */
1049 struct hlist_head preempt_notifiers;
1053 * fpu_counter contains the number of consecutive context switches
1054 * that the FPU is used. If this is over a threshold, the lazy fpu
1055 * saving becomes unlazy to save the trap. This is an unsigned char
1056 * so that after 256 times the counter wraps and the behavior turns
1057 * lazy again; this to deal with bursty apps that only use FPU for
1060 unsigned char fpu_counter;
1061 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1062 #ifdef CONFIG_BLK_DEV_IO_TRACE
1063 unsigned int btrace_seq;
1066 unsigned int policy;
1067 cpumask_t cpus_allowed;
1069 #ifdef CONFIG_PREEMPT_RCU
1070 int rcu_read_lock_nesting;
1071 int rcu_flipctr_idx;
1072 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1074 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1075 struct sched_info sched_info;
1078 struct list_head tasks;
1080 * ptrace_list/ptrace_children forms the list of my children
1081 * that were stolen by a ptracer.
1083 struct list_head ptrace_children;
1084 struct list_head ptrace_list;
1086 struct mm_struct *mm, *active_mm;
1089 struct linux_binfmt *binfmt;
1091 int exit_code, exit_signal;
1092 int pdeath_signal; /* The signal sent when the parent dies */
1094 unsigned int personality;
1095 unsigned did_exec:1;
1099 #ifdef CONFIG_CC_STACKPROTECTOR
1100 /* Canary value for the -fstack-protector gcc feature */
1101 unsigned long stack_canary;
1104 * pointers to (original) parent process, youngest child, younger sibling,
1105 * older sibling, respectively. (p->father can be replaced with
1108 struct task_struct *real_parent; /* real parent process (when being debugged) */
1109 struct task_struct *parent; /* parent process */
1111 * children/sibling forms the list of my children plus the
1112 * tasks I'm ptracing.
1114 struct list_head children; /* list of my children */
1115 struct list_head sibling; /* linkage in my parent's children list */
1116 struct task_struct *group_leader; /* threadgroup leader */
1118 /* PID/PID hash table linkage. */
1119 struct pid_link pids[PIDTYPE_MAX];
1120 struct list_head thread_group;
1122 struct completion *vfork_done; /* for vfork() */
1123 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1124 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1126 unsigned int rt_priority;
1127 cputime_t utime, stime, utimescaled, stimescaled;
1129 cputime_t prev_utime, prev_stime;
1130 unsigned long nvcsw, nivcsw; /* context switch counts */
1131 struct timespec start_time; /* monotonic time */
1132 struct timespec real_start_time; /* boot based time */
1133 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1134 unsigned long min_flt, maj_flt;
1136 cputime_t it_prof_expires, it_virt_expires;
1137 unsigned long long it_sched_expires;
1138 struct list_head cpu_timers[3];
1140 /* process credentials */
1141 uid_t uid,euid,suid,fsuid;
1142 gid_t gid,egid,sgid,fsgid;
1143 struct group_info *group_info;
1144 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset;
1145 unsigned securebits;
1146 struct user_struct *user;
1148 struct key *request_key_auth; /* assumed request_key authority */
1149 struct key *thread_keyring; /* keyring private to this thread */
1150 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1152 char comm[TASK_COMM_LEN]; /* executable name excluding path
1153 - access with [gs]et_task_comm (which lock
1154 it with task_lock())
1155 - initialized normally by flush_old_exec */
1156 /* file system info */
1157 int link_count, total_link_count;
1158 #ifdef CONFIG_SYSVIPC
1160 struct sysv_sem sysvsem;
1162 #ifdef CONFIG_DETECT_SOFTLOCKUP
1163 /* hung task detection */
1164 unsigned long last_switch_timestamp;
1165 unsigned long last_switch_count;
1167 /* CPU-specific state of this task */
1168 struct thread_struct thread;
1169 /* filesystem information */
1170 struct fs_struct *fs;
1171 /* open file information */
1172 struct files_struct *files;
1174 struct nsproxy *nsproxy;
1175 /* signal handlers */
1176 struct signal_struct *signal;
1177 struct sighand_struct *sighand;
1179 sigset_t blocked, real_blocked;
1180 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1181 struct sigpending pending;
1183 unsigned long sas_ss_sp;
1185 int (*notifier)(void *priv);
1186 void *notifier_data;
1187 sigset_t *notifier_mask;
1188 #ifdef CONFIG_SECURITY
1191 struct audit_context *audit_context;
1192 #ifdef CONFIG_AUDITSYSCALL
1194 unsigned int sessionid;
1198 /* Thread group tracking */
1201 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1202 spinlock_t alloc_lock;
1204 /* Protection of the PI data structures: */
1207 #ifdef CONFIG_RT_MUTEXES
1208 /* PI waiters blocked on a rt_mutex held by this task */
1209 struct plist_head pi_waiters;
1210 /* Deadlock detection and priority inheritance handling */
1211 struct rt_mutex_waiter *pi_blocked_on;
1214 #ifdef CONFIG_DEBUG_MUTEXES
1215 /* mutex deadlock detection */
1216 struct mutex_waiter *blocked_on;
1218 #ifdef CONFIG_TRACE_IRQFLAGS
1219 unsigned int irq_events;
1220 int hardirqs_enabled;
1221 unsigned long hardirq_enable_ip;
1222 unsigned int hardirq_enable_event;
1223 unsigned long hardirq_disable_ip;
1224 unsigned int hardirq_disable_event;
1225 int softirqs_enabled;
1226 unsigned long softirq_disable_ip;
1227 unsigned int softirq_disable_event;
1228 unsigned long softirq_enable_ip;
1229 unsigned int softirq_enable_event;
1230 int hardirq_context;
1231 int softirq_context;
1233 #ifdef CONFIG_LOCKDEP
1234 # define MAX_LOCK_DEPTH 48UL
1237 struct held_lock held_locks[MAX_LOCK_DEPTH];
1238 unsigned int lockdep_recursion;
1241 /* journalling filesystem info */
1244 /* stacked block device info */
1245 struct bio *bio_list, **bio_tail;
1248 struct reclaim_state *reclaim_state;
1250 struct backing_dev_info *backing_dev_info;
1252 struct io_context *io_context;
1254 unsigned long ptrace_message;
1255 siginfo_t *last_siginfo; /* For ptrace use. */
1256 #ifdef CONFIG_TASK_XACCT
1257 /* i/o counters(bytes read/written, #syscalls */
1258 u64 rchar, wchar, syscr, syscw;
1260 struct task_io_accounting ioac;
1261 #if defined(CONFIG_TASK_XACCT)
1262 u64 acct_rss_mem1; /* accumulated rss usage */
1263 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1264 cputime_t acct_stimexpd;/* stime since last update */
1267 struct mempolicy *mempolicy;
1270 #ifdef CONFIG_CPUSETS
1271 nodemask_t mems_allowed;
1272 int cpuset_mems_generation;
1273 int cpuset_mem_spread_rotor;
1275 #ifdef CONFIG_CGROUPS
1276 /* Control Group info protected by css_set_lock */
1277 struct css_set *cgroups;
1278 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1279 struct list_head cg_list;
1282 struct robust_list_head __user *robust_list;
1283 #ifdef CONFIG_COMPAT
1284 struct compat_robust_list_head __user *compat_robust_list;
1286 struct list_head pi_state_list;
1287 struct futex_pi_state *pi_state_cache;
1289 atomic_t fs_excl; /* holding fs exclusive resources */
1290 struct rcu_head rcu;
1293 * cache last used pipe for splice
1295 struct pipe_inode_info *splice_pipe;
1296 #ifdef CONFIG_TASK_DELAY_ACCT
1297 struct task_delay_info *delays;
1299 #ifdef CONFIG_FAULT_INJECTION
1302 struct prop_local_single dirties;
1303 #ifdef CONFIG_LATENCYTOP
1304 int latency_record_count;
1305 struct latency_record latency_record[LT_SAVECOUNT];
1310 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1311 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1312 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1313 * values are inverted: lower p->prio value means higher priority.
1315 * The MAX_USER_RT_PRIO value allows the actual maximum
1316 * RT priority to be separate from the value exported to
1317 * user-space. This allows kernel threads to set their
1318 * priority to a value higher than any user task. Note:
1319 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1322 #define MAX_USER_RT_PRIO 100
1323 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1325 #define MAX_PRIO (MAX_RT_PRIO + 40)
1326 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1328 static inline int rt_prio(int prio)
1330 if (unlikely(prio < MAX_RT_PRIO))
1335 static inline int rt_task(struct task_struct *p)
1337 return rt_prio(p->prio);
1340 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1342 tsk->signal->__session = session;
1345 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1347 tsk->signal->__pgrp = pgrp;
1350 static inline struct pid *task_pid(struct task_struct *task)
1352 return task->pids[PIDTYPE_PID].pid;
1355 static inline struct pid *task_tgid(struct task_struct *task)
1357 return task->group_leader->pids[PIDTYPE_PID].pid;
1360 static inline struct pid *task_pgrp(struct task_struct *task)
1362 return task->group_leader->pids[PIDTYPE_PGID].pid;
1365 static inline struct pid *task_session(struct task_struct *task)
1367 return task->group_leader->pids[PIDTYPE_SID].pid;
1370 struct pid_namespace;
1373 * the helpers to get the task's different pids as they are seen
1374 * from various namespaces
1376 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1377 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1379 * task_xid_nr_ns() : id seen from the ns specified;
1381 * set_task_vxid() : assigns a virtual id to a task;
1383 * see also pid_nr() etc in include/linux/pid.h
1386 static inline pid_t task_pid_nr(struct task_struct *tsk)
1391 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1393 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1395 return pid_vnr(task_pid(tsk));
1399 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1404 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1406 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1408 return pid_vnr(task_tgid(tsk));
1412 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1414 return tsk->signal->__pgrp;
1417 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1419 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1421 return pid_vnr(task_pgrp(tsk));
1425 static inline pid_t task_session_nr(struct task_struct *tsk)
1427 return tsk->signal->__session;
1430 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1432 static inline pid_t task_session_vnr(struct task_struct *tsk)
1434 return pid_vnr(task_session(tsk));
1439 * pid_alive - check that a task structure is not stale
1440 * @p: Task structure to be checked.
1442 * Test if a process is not yet dead (at most zombie state)
1443 * If pid_alive fails, then pointers within the task structure
1444 * can be stale and must not be dereferenced.
1446 static inline int pid_alive(struct task_struct *p)
1448 return p->pids[PIDTYPE_PID].pid != NULL;
1452 * is_global_init - check if a task structure is init
1453 * @tsk: Task structure to be checked.
1455 * Check if a task structure is the first user space task the kernel created.
1457 static inline int is_global_init(struct task_struct *tsk)
1459 return tsk->pid == 1;
1463 * is_container_init:
1464 * check whether in the task is init in its own pid namespace.
1466 extern int is_container_init(struct task_struct *tsk);
1468 extern struct pid *cad_pid;
1470 extern void free_task(struct task_struct *tsk);
1471 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1473 extern void __put_task_struct(struct task_struct *t);
1475 static inline void put_task_struct(struct task_struct *t)
1477 if (atomic_dec_and_test(&t->usage))
1478 __put_task_struct(t);
1484 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1485 /* Not implemented yet, only for 486*/
1486 #define PF_STARTING 0x00000002 /* being created */
1487 #define PF_EXITING 0x00000004 /* getting shut down */
1488 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1489 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1490 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1491 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1492 #define PF_DUMPCORE 0x00000200 /* dumped core */
1493 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1494 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1495 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1496 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1497 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1498 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1499 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1500 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1501 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1502 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1503 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1504 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1505 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1506 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1507 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1508 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1509 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1510 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1513 * Only the _current_ task can read/write to tsk->flags, but other
1514 * tasks can access tsk->flags in readonly mode for example
1515 * with tsk_used_math (like during threaded core dumping).
1516 * There is however an exception to this rule during ptrace
1517 * or during fork: the ptracer task is allowed to write to the
1518 * child->flags of its traced child (same goes for fork, the parent
1519 * can write to the child->flags), because we're guaranteed the
1520 * child is not running and in turn not changing child->flags
1521 * at the same time the parent does it.
1523 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1524 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1525 #define clear_used_math() clear_stopped_child_used_math(current)
1526 #define set_used_math() set_stopped_child_used_math(current)
1527 #define conditional_stopped_child_used_math(condition, child) \
1528 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1529 #define conditional_used_math(condition) \
1530 conditional_stopped_child_used_math(condition, current)
1531 #define copy_to_stopped_child_used_math(child) \
1532 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1533 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1534 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1535 #define used_math() tsk_used_math(current)
1538 extern int set_cpus_allowed_ptr(struct task_struct *p,
1539 const cpumask_t *new_mask);
1541 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1542 const cpumask_t *new_mask)
1544 if (!cpu_isset(0, *new_mask))
1549 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1551 return set_cpus_allowed_ptr(p, &new_mask);
1554 extern unsigned long long sched_clock(void);
1556 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1557 static inline void sched_clock_init(void)
1561 static inline u64 sched_clock_cpu(int cpu)
1563 return sched_clock();
1566 static inline void sched_clock_tick(void)
1570 static inline void sched_clock_idle_sleep_event(void)
1574 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1578 extern void sched_clock_init(void);
1579 extern u64 sched_clock_cpu(int cpu);
1580 extern void sched_clock_tick(void);
1581 extern void sched_clock_idle_sleep_event(void);
1582 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1586 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1587 * clock constructed from sched_clock():
1589 extern unsigned long long cpu_clock(int cpu);
1591 extern unsigned long long
1592 task_sched_runtime(struct task_struct *task);
1594 /* sched_exec is called by processes performing an exec */
1596 extern void sched_exec(void);
1598 #define sched_exec() {}
1601 extern void sched_clock_idle_sleep_event(void);
1602 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1604 #ifdef CONFIG_HOTPLUG_CPU
1605 extern void idle_task_exit(void);
1607 static inline void idle_task_exit(void) {}
1610 extern void sched_idle_next(void);
1612 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1613 extern void wake_up_idle_cpu(int cpu);
1615 static inline void wake_up_idle_cpu(int cpu) { }
1618 #ifdef CONFIG_SCHED_DEBUG
1619 extern unsigned int sysctl_sched_latency;
1620 extern unsigned int sysctl_sched_min_granularity;
1621 extern unsigned int sysctl_sched_wakeup_granularity;
1622 extern unsigned int sysctl_sched_child_runs_first;
1623 extern unsigned int sysctl_sched_features;
1624 extern unsigned int sysctl_sched_migration_cost;
1625 extern unsigned int sysctl_sched_nr_migrate;
1627 int sched_nr_latency_handler(struct ctl_table *table, int write,
1628 struct file *file, void __user *buffer, size_t *length,
1631 extern unsigned int sysctl_sched_rt_period;
1632 extern int sysctl_sched_rt_runtime;
1634 int sched_rt_handler(struct ctl_table *table, int write,
1635 struct file *filp, void __user *buffer, size_t *lenp,
1638 extern unsigned int sysctl_sched_compat_yield;
1640 #ifdef CONFIG_RT_MUTEXES
1641 extern int rt_mutex_getprio(struct task_struct *p);
1642 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1643 extern void rt_mutex_adjust_pi(struct task_struct *p);
1645 static inline int rt_mutex_getprio(struct task_struct *p)
1647 return p->normal_prio;
1649 # define rt_mutex_adjust_pi(p) do { } while (0)
1652 extern void set_user_nice(struct task_struct *p, long nice);
1653 extern int task_prio(const struct task_struct *p);
1654 extern int task_nice(const struct task_struct *p);
1655 extern int can_nice(const struct task_struct *p, const int nice);
1656 extern int task_curr(const struct task_struct *p);
1657 extern int idle_cpu(int cpu);
1658 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1659 extern struct task_struct *idle_task(int cpu);
1660 extern struct task_struct *curr_task(int cpu);
1661 extern void set_curr_task(int cpu, struct task_struct *p);
1666 * The default (Linux) execution domain.
1668 extern struct exec_domain default_exec_domain;
1670 union thread_union {
1671 struct thread_info thread_info;
1672 unsigned long stack[THREAD_SIZE/sizeof(long)];
1675 #ifndef __HAVE_ARCH_KSTACK_END
1676 static inline int kstack_end(void *addr)
1678 /* Reliable end of stack detection:
1679 * Some APM bios versions misalign the stack
1681 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1685 extern union thread_union init_thread_union;
1686 extern struct task_struct init_task;
1688 extern struct mm_struct init_mm;
1690 extern struct pid_namespace init_pid_ns;
1693 * find a task by one of its numerical ids
1695 * find_task_by_pid_type_ns():
1696 * it is the most generic call - it finds a task by all id,
1697 * type and namespace specified
1698 * find_task_by_pid_ns():
1699 * finds a task by its pid in the specified namespace
1700 * find_task_by_vpid():
1701 * finds a task by its virtual pid
1702 * find_task_by_pid():
1703 * finds a task by its global pid
1705 * see also find_pid() etc in include/linux/pid.h
1708 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1709 struct pid_namespace *ns);
1711 static inline struct task_struct *__deprecated find_task_by_pid(pid_t nr)
1713 return find_task_by_pid_type_ns(PIDTYPE_PID, nr, &init_pid_ns);
1715 extern struct task_struct *find_task_by_vpid(pid_t nr);
1716 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1717 struct pid_namespace *ns);
1719 extern void __set_special_pids(struct pid *pid);
1721 /* per-UID process charging. */
1722 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1723 static inline struct user_struct *get_uid(struct user_struct *u)
1725 atomic_inc(&u->__count);
1728 extern void free_uid(struct user_struct *);
1729 extern void switch_uid(struct user_struct *);
1730 extern void release_uids(struct user_namespace *ns);
1732 #include <asm/current.h>
1734 extern void do_timer(unsigned long ticks);
1736 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1737 extern int wake_up_process(struct task_struct *tsk);
1738 extern void wake_up_new_task(struct task_struct *tsk,
1739 unsigned long clone_flags);
1741 extern void kick_process(struct task_struct *tsk);
1743 static inline void kick_process(struct task_struct *tsk) { }
1745 extern void sched_fork(struct task_struct *p, int clone_flags);
1746 extern void sched_dead(struct task_struct *p);
1748 extern int in_group_p(gid_t);
1749 extern int in_egroup_p(gid_t);
1751 extern void proc_caches_init(void);
1752 extern void flush_signals(struct task_struct *);
1753 extern void ignore_signals(struct task_struct *);
1754 extern void flush_signal_handlers(struct task_struct *, int force_default);
1755 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1757 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1759 unsigned long flags;
1762 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1763 ret = dequeue_signal(tsk, mask, info);
1764 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1769 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1771 extern void unblock_all_signals(void);
1772 extern void release_task(struct task_struct * p);
1773 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1774 extern int force_sigsegv(int, struct task_struct *);
1775 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1776 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1777 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1778 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1779 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1780 extern int kill_pid(struct pid *pid, int sig, int priv);
1781 extern int kill_proc_info(int, struct siginfo *, pid_t);
1782 extern void do_notify_parent(struct task_struct *, int);
1783 extern void force_sig(int, struct task_struct *);
1784 extern void force_sig_specific(int, struct task_struct *);
1785 extern int send_sig(int, struct task_struct *, int);
1786 extern void zap_other_threads(struct task_struct *p);
1787 extern int kill_proc(pid_t, int, int);
1788 extern struct sigqueue *sigqueue_alloc(void);
1789 extern void sigqueue_free(struct sigqueue *);
1790 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1791 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1792 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1794 static inline int kill_cad_pid(int sig, int priv)
1796 return kill_pid(cad_pid, sig, priv);
1799 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1800 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1801 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1802 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1804 static inline int is_si_special(const struct siginfo *info)
1806 return info <= SEND_SIG_FORCED;
1809 /* True if we are on the alternate signal stack. */
1811 static inline int on_sig_stack(unsigned long sp)
1813 return (sp - current->sas_ss_sp < current->sas_ss_size);
1816 static inline int sas_ss_flags(unsigned long sp)
1818 return (current->sas_ss_size == 0 ? SS_DISABLE
1819 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1823 * Routines for handling mm_structs
1825 extern struct mm_struct * mm_alloc(void);
1827 /* mmdrop drops the mm and the page tables */
1828 extern void __mmdrop(struct mm_struct *);
1829 static inline void mmdrop(struct mm_struct * mm)
1831 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1835 /* mmput gets rid of the mappings and all user-space */
1836 extern void mmput(struct mm_struct *);
1837 /* Grab a reference to a task's mm, if it is not already going away */
1838 extern struct mm_struct *get_task_mm(struct task_struct *task);
1839 /* Remove the current tasks stale references to the old mm_struct */
1840 extern void mm_release(struct task_struct *, struct mm_struct *);
1841 /* Allocate a new mm structure and copy contents from tsk->mm */
1842 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1844 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1845 extern void flush_thread(void);
1846 extern void exit_thread(void);
1848 extern void exit_files(struct task_struct *);
1849 extern void __cleanup_signal(struct signal_struct *);
1850 extern void __cleanup_sighand(struct sighand_struct *);
1851 extern void exit_itimers(struct signal_struct *);
1853 extern NORET_TYPE void do_group_exit(int);
1855 extern void daemonize(const char *, ...);
1856 extern int allow_signal(int);
1857 extern int disallow_signal(int);
1859 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1860 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1861 struct task_struct *fork_idle(int);
1863 extern void set_task_comm(struct task_struct *tsk, char *from);
1864 extern char *get_task_comm(char *to, struct task_struct *tsk);
1867 extern void wait_task_inactive(struct task_struct * p);
1869 #define wait_task_inactive(p) do { } while (0)
1872 #define remove_parent(p) list_del_init(&(p)->sibling)
1873 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1875 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1877 #define for_each_process(p) \
1878 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1881 * Careful: do_each_thread/while_each_thread is a double loop so
1882 * 'break' will not work as expected - use goto instead.
1884 #define do_each_thread(g, t) \
1885 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1887 #define while_each_thread(g, t) \
1888 while ((t = next_thread(t)) != g)
1890 /* de_thread depends on thread_group_leader not being a pid based check */
1891 #define thread_group_leader(p) (p == p->group_leader)
1893 /* Do to the insanities of de_thread it is possible for a process
1894 * to have the pid of the thread group leader without actually being
1895 * the thread group leader. For iteration through the pids in proc
1896 * all we care about is that we have a task with the appropriate
1897 * pid, we don't actually care if we have the right task.
1899 static inline int has_group_leader_pid(struct task_struct *p)
1901 return p->pid == p->tgid;
1905 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1907 return p1->tgid == p2->tgid;
1910 static inline struct task_struct *next_thread(const struct task_struct *p)
1912 return list_entry(rcu_dereference(p->thread_group.next),
1913 struct task_struct, thread_group);
1916 static inline int thread_group_empty(struct task_struct *p)
1918 return list_empty(&p->thread_group);
1921 #define delay_group_leader(p) \
1922 (thread_group_leader(p) && !thread_group_empty(p))
1925 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1926 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1927 * pins the final release of task.io_context. Also protects ->cpuset and
1928 * ->cgroup.subsys[].
1930 * Nests both inside and outside of read_lock(&tasklist_lock).
1931 * It must not be nested with write_lock_irq(&tasklist_lock),
1932 * neither inside nor outside.
1934 static inline void task_lock(struct task_struct *p)
1936 spin_lock(&p->alloc_lock);
1939 static inline void task_unlock(struct task_struct *p)
1941 spin_unlock(&p->alloc_lock);
1944 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1945 unsigned long *flags);
1947 static inline void unlock_task_sighand(struct task_struct *tsk,
1948 unsigned long *flags)
1950 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1953 #ifndef __HAVE_THREAD_FUNCTIONS
1955 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1956 #define task_stack_page(task) ((task)->stack)
1958 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1960 *task_thread_info(p) = *task_thread_info(org);
1961 task_thread_info(p)->task = p;
1964 static inline unsigned long *end_of_stack(struct task_struct *p)
1966 return (unsigned long *)(task_thread_info(p) + 1);
1971 extern void thread_info_cache_init(void);
1973 /* set thread flags in other task's structures
1974 * - see asm/thread_info.h for TIF_xxxx flags available
1976 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1978 set_ti_thread_flag(task_thread_info(tsk), flag);
1981 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1983 clear_ti_thread_flag(task_thread_info(tsk), flag);
1986 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1988 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1991 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1993 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1996 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1998 return test_ti_thread_flag(task_thread_info(tsk), flag);
2001 static inline void set_tsk_need_resched(struct task_struct *tsk)
2003 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2006 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2008 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2011 static inline int test_tsk_need_resched(struct task_struct *tsk)
2013 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2016 static inline int signal_pending(struct task_struct *p)
2018 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2021 extern int __fatal_signal_pending(struct task_struct *p);
2023 static inline int fatal_signal_pending(struct task_struct *p)
2025 return signal_pending(p) && __fatal_signal_pending(p);
2028 static inline int need_resched(void)
2030 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2034 * cond_resched() and cond_resched_lock(): latency reduction via
2035 * explicit rescheduling in places that are safe. The return
2036 * value indicates whether a reschedule was done in fact.
2037 * cond_resched_lock() will drop the spinlock before scheduling,
2038 * cond_resched_softirq() will enable bhs before scheduling.
2040 extern int _cond_resched(void);
2041 #ifdef CONFIG_PREEMPT_BKL
2042 static inline int cond_resched(void)
2047 static inline int cond_resched(void)
2049 return _cond_resched();
2052 extern int cond_resched_lock(spinlock_t * lock);
2053 extern int cond_resched_softirq(void);
2054 static inline int cond_resched_bkl(void)
2056 return _cond_resched();
2060 * Does a critical section need to be broken due to another
2061 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2062 * but a general need for low latency)
2064 static inline int spin_needbreak(spinlock_t *lock)
2066 #ifdef CONFIG_PREEMPT
2067 return spin_is_contended(lock);
2074 * Reevaluate whether the task has signals pending delivery.
2075 * Wake the task if so.
2076 * This is required every time the blocked sigset_t changes.
2077 * callers must hold sighand->siglock.
2079 extern void recalc_sigpending_and_wake(struct task_struct *t);
2080 extern void recalc_sigpending(void);
2082 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2085 * Wrappers for p->thread_info->cpu access. No-op on UP.
2089 static inline unsigned int task_cpu(const struct task_struct *p)
2091 return task_thread_info(p)->cpu;
2094 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2098 static inline unsigned int task_cpu(const struct task_struct *p)
2103 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2107 #endif /* CONFIG_SMP */
2109 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2110 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2112 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
2114 mm->mmap_base = TASK_UNMAPPED_BASE;
2115 mm->get_unmapped_area = arch_get_unmapped_area;
2116 mm->unmap_area = arch_unmap_area;
2120 extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask);
2121 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2123 extern int sched_mc_power_savings, sched_smt_power_savings;
2125 extern void normalize_rt_tasks(void);
2127 #ifdef CONFIG_GROUP_SCHED
2129 extern struct task_group init_task_group;
2130 #ifdef CONFIG_USER_SCHED
2131 extern struct task_group root_task_group;
2134 extern struct task_group *sched_create_group(struct task_group *parent);
2135 extern void sched_destroy_group(struct task_group *tg);
2136 extern void sched_move_task(struct task_struct *tsk);
2137 #ifdef CONFIG_FAIR_GROUP_SCHED
2138 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2139 extern unsigned long sched_group_shares(struct task_group *tg);
2141 #ifdef CONFIG_RT_GROUP_SCHED
2142 extern int sched_group_set_rt_runtime(struct task_group *tg,
2143 long rt_runtime_us);
2144 extern long sched_group_rt_runtime(struct task_group *tg);
2145 extern int sched_group_set_rt_period(struct task_group *tg,
2147 extern long sched_group_rt_period(struct task_group *tg);
2151 #ifdef CONFIG_TASK_XACCT
2152 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2157 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2162 static inline void inc_syscr(struct task_struct *tsk)
2167 static inline void inc_syscw(struct task_struct *tsk)
2172 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2176 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2180 static inline void inc_syscr(struct task_struct *tsk)
2184 static inline void inc_syscw(struct task_struct *tsk)
2190 void migration_init(void);
2192 static inline void migration_init(void)
2197 #ifndef TASK_SIZE_OF
2198 #define TASK_SIZE_OF(tsk) TASK_SIZE
2201 #ifdef CONFIG_MM_OWNER
2202 extern void mm_update_next_owner(struct mm_struct *mm);
2203 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2205 static inline void mm_update_next_owner(struct mm_struct *mm)
2209 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2212 #endif /* CONFIG_MM_OWNER */
2214 #endif /* __KERNEL__ */