4 #include <uapi/linux/sched.h>
11 #include <asm/param.h> /* for HZ */
13 #include <linux/capability.h>
14 #include <linux/threads.h>
15 #include <linux/kernel.h>
16 #include <linux/types.h>
17 #include <linux/timex.h>
18 #include <linux/jiffies.h>
19 #include <linux/rbtree.h>
20 #include <linux/thread_info.h>
21 #include <linux/cpumask.h>
22 #include <linux/errno.h>
23 #include <linux/nodemask.h>
24 #include <linux/mm_types.h>
27 #include <asm/ptrace.h>
28 #include <asm/cputime.h>
30 #include <linux/smp.h>
31 #include <linux/sem.h>
32 #include <linux/signal.h>
33 #include <linux/compiler.h>
34 #include <linux/completion.h>
35 #include <linux/pid.h>
36 #include <linux/percpu.h>
37 #include <linux/topology.h>
38 #include <linux/proportions.h>
39 #include <linux/seccomp.h>
40 #include <linux/rcupdate.h>
41 #include <linux/rculist.h>
42 #include <linux/rtmutex.h>
44 #include <linux/time.h>
45 #include <linux/param.h>
46 #include <linux/resource.h>
47 #include <linux/timer.h>
48 #include <linux/hrtimer.h>
49 #include <linux/task_io_accounting.h>
50 #include <linux/latencytop.h>
51 #include <linux/cred.h>
52 #include <linux/llist.h>
53 #include <linux/uidgid.h>
55 #include <asm/processor.h>
58 struct futex_pi_state
;
59 struct robust_list_head
;
62 struct perf_event_context
;
66 * List of flags we want to share for kernel threads,
67 * if only because they are not used by them anyway.
69 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
72 * These are the constant used to fake the fixed-point load-average
73 * counting. Some notes:
74 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
75 * a load-average precision of 10 bits integer + 11 bits fractional
76 * - if you want to count load-averages more often, you need more
77 * precision, or rounding will get you. With 2-second counting freq,
78 * the EXP_n values would be 1981, 2034 and 2043 if still using only
81 extern unsigned long avenrun
[]; /* Load averages */
82 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
84 #define FSHIFT 11 /* nr of bits of precision */
85 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
86 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
87 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
88 #define EXP_5 2014 /* 1/exp(5sec/5min) */
89 #define EXP_15 2037 /* 1/exp(5sec/15min) */
91 #define CALC_LOAD(load,exp,n) \
93 load += n*(FIXED_1-exp); \
96 extern unsigned long total_forks
;
97 extern int nr_threads
;
98 DECLARE_PER_CPU(unsigned long, process_counts
);
99 extern int nr_processes(void);
100 extern unsigned long nr_running(void);
101 extern unsigned long nr_uninterruptible(void);
102 extern unsigned long nr_iowait(void);
103 extern unsigned long nr_iowait_cpu(int cpu
);
104 extern unsigned long this_cpu_load(void);
107 extern void calc_global_load(unsigned long ticks
);
108 extern void update_cpu_load_nohz(void);
110 extern unsigned long get_parent_ip(unsigned long addr
);
115 #ifdef CONFIG_SCHED_DEBUG
116 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
117 extern void proc_sched_set_task(struct task_struct
*p
);
119 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
);
122 proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
)
125 static inline void proc_sched_set_task(struct task_struct
*p
)
129 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
)
135 * Task state bitmask. NOTE! These bits are also
136 * encoded in fs/proc/array.c: get_task_state().
138 * We have two separate sets of flags: task->state
139 * is about runnability, while task->exit_state are
140 * about the task exiting. Confusing, but this way
141 * modifying one set can't modify the other one by
144 #define TASK_RUNNING 0
145 #define TASK_INTERRUPTIBLE 1
146 #define TASK_UNINTERRUPTIBLE 2
147 #define __TASK_STOPPED 4
148 #define __TASK_TRACED 8
149 /* in tsk->exit_state */
150 #define EXIT_ZOMBIE 16
152 /* in tsk->state again */
154 #define TASK_WAKEKILL 128
155 #define TASK_WAKING 256
156 #define TASK_STATE_MAX 512
158 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
160 extern char ___assert_task_state
[1 - 2*!!(
161 sizeof(TASK_STATE_TO_CHAR_STR
)-1 != ilog2(TASK_STATE_MAX
)+1)];
163 /* Convenience macros for the sake of set_task_state */
164 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
165 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
166 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
168 /* Convenience macros for the sake of wake_up */
169 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
170 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
172 /* get_task_state() */
173 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
174 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
177 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
178 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
179 #define task_is_dead(task) ((task)->exit_state != 0)
180 #define task_is_stopped_or_traced(task) \
181 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
182 #define task_contributes_to_load(task) \
183 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
184 (task->flags & PF_FROZEN) == 0)
186 #define __set_task_state(tsk, state_value) \
187 do { (tsk)->state = (state_value); } while (0)
188 #define set_task_state(tsk, state_value) \
189 set_mb((tsk)->state, (state_value))
192 * set_current_state() includes a barrier so that the write of current->state
193 * is correctly serialised wrt the caller's subsequent test of whether to
196 * set_current_state(TASK_UNINTERRUPTIBLE);
197 * if (do_i_need_to_sleep())
200 * If the caller does not need such serialisation then use __set_current_state()
202 #define __set_current_state(state_value) \
203 do { current->state = (state_value); } while (0)
204 #define set_current_state(state_value) \
205 set_mb(current->state, (state_value))
207 /* Task command name length */
208 #define TASK_COMM_LEN 16
210 #include <linux/spinlock.h>
213 * This serializes "schedule()" and also protects
214 * the run-queue from deletions/modifications (but
215 * _adding_ to the beginning of the run-queue has
218 extern rwlock_t tasklist_lock
;
219 extern spinlock_t mmlist_lock
;
223 #ifdef CONFIG_PROVE_RCU
224 extern int lockdep_tasklist_lock_is_held(void);
225 #endif /* #ifdef CONFIG_PROVE_RCU */
227 extern void sched_init(void);
228 extern void sched_init_smp(void);
229 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
230 extern void init_idle(struct task_struct
*idle
, int cpu
);
231 extern void init_idle_bootup_task(struct task_struct
*idle
);
233 extern int runqueue_is_locked(int cpu
);
235 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
236 extern void nohz_balance_enter_idle(int cpu
);
237 extern void set_cpu_sd_state_idle(void);
238 extern int get_nohz_timer_target(void);
240 static inline void nohz_balance_enter_idle(int cpu
) { }
241 static inline void set_cpu_sd_state_idle(void) { }
245 * Only dump TASK_* tasks. (0 for all tasks)
247 extern void show_state_filter(unsigned long state_filter
);
249 static inline void show_state(void)
251 show_state_filter(0);
254 extern void show_regs(struct pt_regs
*);
257 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
258 * task), SP is the stack pointer of the first frame that should be shown in the back
259 * trace (or NULL if the entire call-chain of the task should be shown).
261 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
263 void io_schedule(void);
264 long io_schedule_timeout(long timeout
);
266 extern void cpu_init (void);
267 extern void trap_init(void);
268 extern void update_process_times(int user
);
269 extern void scheduler_tick(void);
271 extern void sched_show_task(struct task_struct
*p
);
273 #ifdef CONFIG_LOCKUP_DETECTOR
274 extern void touch_softlockup_watchdog(void);
275 extern void touch_softlockup_watchdog_sync(void);
276 extern void touch_all_softlockup_watchdogs(void);
277 extern int proc_dowatchdog_thresh(struct ctl_table
*table
, int write
,
279 size_t *lenp
, loff_t
*ppos
);
280 extern unsigned int softlockup_panic
;
281 void lockup_detector_init(void);
283 static inline void touch_softlockup_watchdog(void)
286 static inline void touch_softlockup_watchdog_sync(void)
289 static inline void touch_all_softlockup_watchdogs(void)
292 static inline void lockup_detector_init(void)
297 #ifdef CONFIG_DETECT_HUNG_TASK
298 extern unsigned int sysctl_hung_task_panic
;
299 extern unsigned long sysctl_hung_task_check_count
;
300 extern unsigned long sysctl_hung_task_timeout_secs
;
301 extern unsigned long sysctl_hung_task_warnings
;
302 extern int proc_dohung_task_timeout_secs(struct ctl_table
*table
, int write
,
304 size_t *lenp
, loff_t
*ppos
);
306 /* Avoid need for ifdefs elsewhere in the code */
307 enum { sysctl_hung_task_timeout_secs
= 0 };
310 /* Attach to any functions which should be ignored in wchan output. */
311 #define __sched __attribute__((__section__(".sched.text")))
313 /* Linker adds these: start and end of __sched functions */
314 extern char __sched_text_start
[], __sched_text_end
[];
316 /* Is this address in the __sched functions? */
317 extern int in_sched_functions(unsigned long addr
);
319 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
320 extern signed long schedule_timeout(signed long timeout
);
321 extern signed long schedule_timeout_interruptible(signed long timeout
);
322 extern signed long schedule_timeout_killable(signed long timeout
);
323 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
324 asmlinkage
void schedule(void);
325 extern void schedule_preempt_disabled(void);
326 extern int mutex_spin_on_owner(struct mutex
*lock
, struct task_struct
*owner
);
329 struct user_namespace
;
332 * Default maximum number of active map areas, this limits the number of vmas
333 * per mm struct. Users can overwrite this number by sysctl but there is a
336 * When a program's coredump is generated as ELF format, a section is created
337 * per a vma. In ELF, the number of sections is represented in unsigned short.
338 * This means the number of sections should be smaller than 65535 at coredump.
339 * Because the kernel adds some informative sections to a image of program at
340 * generating coredump, we need some margin. The number of extra sections is
341 * 1-3 now and depends on arch. We use "5" as safe margin, here.
343 #define MAPCOUNT_ELF_CORE_MARGIN (5)
344 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
346 extern int sysctl_max_map_count
;
348 #include <linux/aio.h>
351 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
353 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
354 unsigned long, unsigned long);
356 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
357 unsigned long len
, unsigned long pgoff
,
358 unsigned long flags
);
359 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
360 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
362 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
) {}
366 extern void set_dumpable(struct mm_struct
*mm
, int value
);
367 extern int get_dumpable(struct mm_struct
*mm
);
369 /* get/set_dumpable() values */
370 #define SUID_DUMPABLE_DISABLED 0
371 #define SUID_DUMPABLE_ENABLED 1
372 #define SUID_DUMPABLE_SAFE 2
376 #define MMF_DUMPABLE 0 /* core dump is permitted */
377 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
379 #define MMF_DUMPABLE_BITS 2
380 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
382 /* coredump filter bits */
383 #define MMF_DUMP_ANON_PRIVATE 2
384 #define MMF_DUMP_ANON_SHARED 3
385 #define MMF_DUMP_MAPPED_PRIVATE 4
386 #define MMF_DUMP_MAPPED_SHARED 5
387 #define MMF_DUMP_ELF_HEADERS 6
388 #define MMF_DUMP_HUGETLB_PRIVATE 7
389 #define MMF_DUMP_HUGETLB_SHARED 8
391 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
392 #define MMF_DUMP_FILTER_BITS 7
393 #define MMF_DUMP_FILTER_MASK \
394 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
395 #define MMF_DUMP_FILTER_DEFAULT \
396 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
397 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
399 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
400 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
402 # define MMF_DUMP_MASK_DEFAULT_ELF 0
404 /* leave room for more dump flags */
405 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
406 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
407 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
409 #define MMF_HAS_UPROBES 19 /* has uprobes */
410 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
412 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
414 struct sighand_struct
{
416 struct k_sigaction action
[_NSIG
];
418 wait_queue_head_t signalfd_wqh
;
421 struct pacct_struct
{
424 unsigned long ac_mem
;
425 cputime_t ac_utime
, ac_stime
;
426 unsigned long ac_minflt
, ac_majflt
;
437 * struct task_cputime - collected CPU time counts
438 * @utime: time spent in user mode, in &cputime_t units
439 * @stime: time spent in kernel mode, in &cputime_t units
440 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
442 * This structure groups together three kinds of CPU time that are
443 * tracked for threads and thread groups. Most things considering
444 * CPU time want to group these counts together and treat all three
445 * of them in parallel.
447 struct task_cputime
{
450 unsigned long long sum_exec_runtime
;
452 /* Alternate field names when used to cache expirations. */
453 #define prof_exp stime
454 #define virt_exp utime
455 #define sched_exp sum_exec_runtime
457 #define INIT_CPUTIME \
458 (struct task_cputime) { \
461 .sum_exec_runtime = 0, \
465 * Disable preemption until the scheduler is running.
466 * Reset by start_kernel()->sched_init()->init_idle().
468 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
469 * before the scheduler is active -- see should_resched().
471 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
474 * struct thread_group_cputimer - thread group interval timer counts
475 * @cputime: thread group interval timers.
476 * @running: non-zero when there are timers running and
477 * @cputime receives updates.
478 * @lock: lock for fields in this struct.
480 * This structure contains the version of task_cputime, above, that is
481 * used for thread group CPU timer calculations.
483 struct thread_group_cputimer
{
484 struct task_cputime cputime
;
489 #include <linux/rwsem.h>
493 * NOTE! "signal_struct" does not have its own
494 * locking, because a shared signal_struct always
495 * implies a shared sighand_struct, so locking
496 * sighand_struct is always a proper superset of
497 * the locking of signal_struct.
499 struct signal_struct
{
504 wait_queue_head_t wait_chldexit
; /* for wait4() */
506 /* current thread group signal load-balancing target: */
507 struct task_struct
*curr_target
;
509 /* shared signal handling: */
510 struct sigpending shared_pending
;
512 /* thread group exit support */
515 * - notify group_exit_task when ->count is equal to notify_count
516 * - everyone except group_exit_task is stopped during signal delivery
517 * of fatal signals, group_exit_task processes the signal.
520 struct task_struct
*group_exit_task
;
522 /* thread group stop support, overloads group_exit_code too */
523 int group_stop_count
;
524 unsigned int flags
; /* see SIGNAL_* flags below */
527 * PR_SET_CHILD_SUBREAPER marks a process, like a service
528 * manager, to re-parent orphan (double-forking) child processes
529 * to this process instead of 'init'. The service manager is
530 * able to receive SIGCHLD signals and is able to investigate
531 * the process until it calls wait(). All children of this
532 * process will inherit a flag if they should look for a
533 * child_subreaper process at exit.
535 unsigned int is_child_subreaper
:1;
536 unsigned int has_child_subreaper
:1;
538 /* POSIX.1b Interval Timers */
539 struct list_head posix_timers
;
541 /* ITIMER_REAL timer for the process */
542 struct hrtimer real_timer
;
543 struct pid
*leader_pid
;
544 ktime_t it_real_incr
;
547 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
548 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
549 * values are defined to 0 and 1 respectively
551 struct cpu_itimer it
[2];
554 * Thread group totals for process CPU timers.
555 * See thread_group_cputimer(), et al, for details.
557 struct thread_group_cputimer cputimer
;
559 /* Earliest-expiration cache. */
560 struct task_cputime cputime_expires
;
562 struct list_head cpu_timers
[3];
564 struct pid
*tty_old_pgrp
;
566 /* boolean value for session group leader */
569 struct tty_struct
*tty
; /* NULL if no tty */
571 #ifdef CONFIG_SCHED_AUTOGROUP
572 struct autogroup
*autogroup
;
575 * Cumulative resource counters for dead threads in the group,
576 * and for reaped dead child processes forked by this group.
577 * Live threads maintain their own counters and add to these
578 * in __exit_signal, except for the group leader.
580 cputime_t utime
, stime
, cutime
, cstime
;
583 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
584 cputime_t prev_utime
, prev_stime
;
586 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
587 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
588 unsigned long inblock
, oublock
, cinblock
, coublock
;
589 unsigned long maxrss
, cmaxrss
;
590 struct task_io_accounting ioac
;
593 * Cumulative ns of schedule CPU time fo dead threads in the
594 * group, not including a zombie group leader, (This only differs
595 * from jiffies_to_ns(utime + stime) if sched_clock uses something
596 * other than jiffies.)
598 unsigned long long sum_sched_runtime
;
601 * We don't bother to synchronize most readers of this at all,
602 * because there is no reader checking a limit that actually needs
603 * to get both rlim_cur and rlim_max atomically, and either one
604 * alone is a single word that can safely be read normally.
605 * getrlimit/setrlimit use task_lock(current->group_leader) to
606 * protect this instead of the siglock, because they really
607 * have no need to disable irqs.
609 struct rlimit rlim
[RLIM_NLIMITS
];
611 #ifdef CONFIG_BSD_PROCESS_ACCT
612 struct pacct_struct pacct
; /* per-process accounting information */
614 #ifdef CONFIG_TASKSTATS
615 struct taskstats
*stats
;
619 struct tty_audit_buf
*tty_audit_buf
;
621 #ifdef CONFIG_CGROUPS
623 * group_rwsem prevents new tasks from entering the threadgroup and
624 * member tasks from exiting,a more specifically, setting of
625 * PF_EXITING. fork and exit paths are protected with this rwsem
626 * using threadgroup_change_begin/end(). Users which require
627 * threadgroup to remain stable should use threadgroup_[un]lock()
628 * which also takes care of exec path. Currently, cgroup is the
631 struct rw_semaphore group_rwsem
;
634 int oom_score_adj
; /* OOM kill score adjustment */
635 int oom_score_adj_min
; /* OOM kill score adjustment minimum value.
636 * Only settable by CAP_SYS_RESOURCE. */
638 struct mutex cred_guard_mutex
; /* guard against foreign influences on
639 * credential calculations
640 * (notably. ptrace) */
644 * Bits in flags field of signal_struct.
646 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
647 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
648 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
650 * Pending notifications to parent.
652 #define SIGNAL_CLD_STOPPED 0x00000010
653 #define SIGNAL_CLD_CONTINUED 0x00000020
654 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
656 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
658 /* If true, all threads except ->group_exit_task have pending SIGKILL */
659 static inline int signal_group_exit(const struct signal_struct
*sig
)
661 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
662 (sig
->group_exit_task
!= NULL
);
666 * Some day this will be a full-fledged user tracking system..
669 atomic_t __count
; /* reference count */
670 atomic_t processes
; /* How many processes does this user have? */
671 atomic_t files
; /* How many open files does this user have? */
672 atomic_t sigpending
; /* How many pending signals does this user have? */
673 #ifdef CONFIG_INOTIFY_USER
674 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
675 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
677 #ifdef CONFIG_FANOTIFY
678 atomic_t fanotify_listeners
;
681 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
683 #ifdef CONFIG_POSIX_MQUEUE
684 /* protected by mq_lock */
685 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
687 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
690 struct key
*uid_keyring
; /* UID specific keyring */
691 struct key
*session_keyring
; /* UID's default session keyring */
694 /* Hash table maintenance information */
695 struct hlist_node uidhash_node
;
698 #ifdef CONFIG_PERF_EVENTS
699 atomic_long_t locked_vm
;
703 extern int uids_sysfs_init(void);
705 extern struct user_struct
*find_user(kuid_t
);
707 extern struct user_struct root_user
;
708 #define INIT_USER (&root_user)
711 struct backing_dev_info
;
712 struct reclaim_state
;
714 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
716 /* cumulative counters */
717 unsigned long pcount
; /* # of times run on this cpu */
718 unsigned long long run_delay
; /* time spent waiting on a runqueue */
721 unsigned long long last_arrival
,/* when we last ran on a cpu */
722 last_queued
; /* when we were last queued to run */
724 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
726 #ifdef CONFIG_TASK_DELAY_ACCT
727 struct task_delay_info
{
729 unsigned int flags
; /* Private per-task flags */
731 /* For each stat XXX, add following, aligned appropriately
733 * struct timespec XXX_start, XXX_end;
737 * Atomicity of updates to XXX_delay, XXX_count protected by
738 * single lock above (split into XXX_lock if contention is an issue).
742 * XXX_count is incremented on every XXX operation, the delay
743 * associated with the operation is added to XXX_delay.
744 * XXX_delay contains the accumulated delay time in nanoseconds.
746 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
747 u64 blkio_delay
; /* wait for sync block io completion */
748 u64 swapin_delay
; /* wait for swapin block io completion */
749 u32 blkio_count
; /* total count of the number of sync block */
750 /* io operations performed */
751 u32 swapin_count
; /* total count of the number of swapin block */
752 /* io operations performed */
754 struct timespec freepages_start
, freepages_end
;
755 u64 freepages_delay
; /* wait for memory reclaim */
756 u32 freepages_count
; /* total count of memory reclaim */
758 #endif /* CONFIG_TASK_DELAY_ACCT */
760 static inline int sched_info_on(void)
762 #ifdef CONFIG_SCHEDSTATS
764 #elif defined(CONFIG_TASK_DELAY_ACCT)
765 extern int delayacct_on
;
780 * Increase resolution of nice-level calculations for 64-bit architectures.
781 * The extra resolution improves shares distribution and load balancing of
782 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
783 * hierarchies, especially on larger systems. This is not a user-visible change
784 * and does not change the user-interface for setting shares/weights.
786 * We increase resolution only if we have enough bits to allow this increased
787 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
788 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
791 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
792 # define SCHED_LOAD_RESOLUTION 10
793 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
794 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
796 # define SCHED_LOAD_RESOLUTION 0
797 # define scale_load(w) (w)
798 # define scale_load_down(w) (w)
801 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
802 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
805 * Increase resolution of cpu_power calculations
807 #define SCHED_POWER_SHIFT 10
808 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
811 * sched-domains (multiprocessor balancing) declarations:
814 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
815 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
816 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
817 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
818 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
819 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
820 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
821 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
822 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
823 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
824 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
825 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
827 extern int __weak
arch_sd_sibiling_asym_packing(void);
829 struct sched_group_power
{
832 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
835 unsigned int power
, power_orig
;
836 unsigned long next_update
;
838 * Number of busy cpus in this group.
840 atomic_t nr_busy_cpus
;
842 unsigned long cpumask
[0]; /* iteration mask */
846 struct sched_group
*next
; /* Must be a circular list */
849 unsigned int group_weight
;
850 struct sched_group_power
*sgp
;
853 * The CPUs this group covers.
855 * NOTE: this field is variable length. (Allocated dynamically
856 * by attaching extra space to the end of the structure,
857 * depending on how many CPUs the kernel has booted up with)
859 unsigned long cpumask
[0];
862 static inline struct cpumask
*sched_group_cpus(struct sched_group
*sg
)
864 return to_cpumask(sg
->cpumask
);
868 * cpumask masking which cpus in the group are allowed to iterate up the domain
871 static inline struct cpumask
*sched_group_mask(struct sched_group
*sg
)
873 return to_cpumask(sg
->sgp
->cpumask
);
877 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
878 * @group: The group whose first cpu is to be returned.
880 static inline unsigned int group_first_cpu(struct sched_group
*group
)
882 return cpumask_first(sched_group_cpus(group
));
885 struct sched_domain_attr
{
886 int relax_domain_level
;
889 #define SD_ATTR_INIT (struct sched_domain_attr) { \
890 .relax_domain_level = -1, \
893 extern int sched_domain_level_max
;
895 struct sched_domain
{
896 /* These fields must be setup */
897 struct sched_domain
*parent
; /* top domain must be null terminated */
898 struct sched_domain
*child
; /* bottom domain must be null terminated */
899 struct sched_group
*groups
; /* the balancing groups of the domain */
900 unsigned long min_interval
; /* Minimum balance interval ms */
901 unsigned long max_interval
; /* Maximum balance interval ms */
902 unsigned int busy_factor
; /* less balancing by factor if busy */
903 unsigned int imbalance_pct
; /* No balance until over watermark */
904 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
905 unsigned int busy_idx
;
906 unsigned int idle_idx
;
907 unsigned int newidle_idx
;
908 unsigned int wake_idx
;
909 unsigned int forkexec_idx
;
910 unsigned int smt_gain
;
911 int flags
; /* See SD_* */
914 /* Runtime fields. */
915 unsigned long last_balance
; /* init to jiffies. units in jiffies */
916 unsigned int balance_interval
; /* initialise to 1. units in ms. */
917 unsigned int nr_balance_failed
; /* initialise to 0 */
921 #ifdef CONFIG_SCHEDSTATS
922 /* load_balance() stats */
923 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
924 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
925 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
926 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
927 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
928 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
929 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
930 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
932 /* Active load balancing */
933 unsigned int alb_count
;
934 unsigned int alb_failed
;
935 unsigned int alb_pushed
;
937 /* SD_BALANCE_EXEC stats */
938 unsigned int sbe_count
;
939 unsigned int sbe_balanced
;
940 unsigned int sbe_pushed
;
942 /* SD_BALANCE_FORK stats */
943 unsigned int sbf_count
;
944 unsigned int sbf_balanced
;
945 unsigned int sbf_pushed
;
947 /* try_to_wake_up() stats */
948 unsigned int ttwu_wake_remote
;
949 unsigned int ttwu_move_affine
;
950 unsigned int ttwu_move_balance
;
952 #ifdef CONFIG_SCHED_DEBUG
956 void *private; /* used during construction */
957 struct rcu_head rcu
; /* used during destruction */
960 unsigned int span_weight
;
962 * Span of all CPUs in this domain.
964 * NOTE: this field is variable length. (Allocated dynamically
965 * by attaching extra space to the end of the structure,
966 * depending on how many CPUs the kernel has booted up with)
968 unsigned long span
[0];
971 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
973 return to_cpumask(sd
->span
);
976 extern void partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
977 struct sched_domain_attr
*dattr_new
);
979 /* Allocate an array of sched domains, for partition_sched_domains(). */
980 cpumask_var_t
*alloc_sched_domains(unsigned int ndoms
);
981 void free_sched_domains(cpumask_var_t doms
[], unsigned int ndoms
);
983 /* Test a flag in parent sched domain */
984 static inline int test_sd_parent(struct sched_domain
*sd
, int flag
)
986 if (sd
->parent
&& (sd
->parent
->flags
& flag
))
992 unsigned long default_scale_freq_power(struct sched_domain
*sd
, int cpu
);
993 unsigned long default_scale_smt_power(struct sched_domain
*sd
, int cpu
);
995 bool cpus_share_cache(int this_cpu
, int that_cpu
);
997 #else /* CONFIG_SMP */
999 struct sched_domain_attr
;
1002 partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1003 struct sched_domain_attr
*dattr_new
)
1007 static inline bool cpus_share_cache(int this_cpu
, int that_cpu
)
1012 #endif /* !CONFIG_SMP */
1015 struct io_context
; /* See blkdev.h */
1018 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1019 extern void prefetch_stack(struct task_struct
*t
);
1021 static inline void prefetch_stack(struct task_struct
*t
) { }
1024 struct audit_context
; /* See audit.c */
1026 struct pipe_inode_info
;
1027 struct uts_namespace
;
1030 struct sched_domain
;
1035 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1036 #define WF_FORK 0x02 /* child wakeup after fork */
1037 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1039 #define ENQUEUE_WAKEUP 1
1040 #define ENQUEUE_HEAD 2
1042 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1044 #define ENQUEUE_WAKING 0
1047 #define DEQUEUE_SLEEP 1
1049 struct sched_class
{
1050 const struct sched_class
*next
;
1052 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int flags
);
1053 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int flags
);
1054 void (*yield_task
) (struct rq
*rq
);
1055 bool (*yield_to_task
) (struct rq
*rq
, struct task_struct
*p
, bool preempt
);
1057 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
, int flags
);
1059 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
1060 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
1063 int (*select_task_rq
)(struct task_struct
*p
, int sd_flag
, int flags
);
1065 void (*pre_schedule
) (struct rq
*this_rq
, struct task_struct
*task
);
1066 void (*post_schedule
) (struct rq
*this_rq
);
1067 void (*task_waking
) (struct task_struct
*task
);
1068 void (*task_woken
) (struct rq
*this_rq
, struct task_struct
*task
);
1070 void (*set_cpus_allowed
)(struct task_struct
*p
,
1071 const struct cpumask
*newmask
);
1073 void (*rq_online
)(struct rq
*rq
);
1074 void (*rq_offline
)(struct rq
*rq
);
1077 void (*set_curr_task
) (struct rq
*rq
);
1078 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
, int queued
);
1079 void (*task_fork
) (struct task_struct
*p
);
1081 void (*switched_from
) (struct rq
*this_rq
, struct task_struct
*task
);
1082 void (*switched_to
) (struct rq
*this_rq
, struct task_struct
*task
);
1083 void (*prio_changed
) (struct rq
*this_rq
, struct task_struct
*task
,
1086 unsigned int (*get_rr_interval
) (struct rq
*rq
,
1087 struct task_struct
*task
);
1089 #ifdef CONFIG_FAIR_GROUP_SCHED
1090 void (*task_move_group
) (struct task_struct
*p
, int on_rq
);
1094 struct load_weight
{
1095 unsigned long weight
, inv_weight
;
1100 * These sums represent an infinite geometric series and so are bound
1101 * above by 1024/(1-y). Thus we only need a u32 to store them for for all
1102 * choices of y < 1-2^(-32)*1024.
1104 u32 runnable_avg_sum
, runnable_avg_period
;
1105 u64 last_runnable_update
;
1108 #ifdef CONFIG_SCHEDSTATS
1109 struct sched_statistics
{
1119 s64 sum_sleep_runtime
;
1126 u64 nr_migrations_cold
;
1127 u64 nr_failed_migrations_affine
;
1128 u64 nr_failed_migrations_running
;
1129 u64 nr_failed_migrations_hot
;
1130 u64 nr_forced_migrations
;
1133 u64 nr_wakeups_sync
;
1134 u64 nr_wakeups_migrate
;
1135 u64 nr_wakeups_local
;
1136 u64 nr_wakeups_remote
;
1137 u64 nr_wakeups_affine
;
1138 u64 nr_wakeups_affine_attempts
;
1139 u64 nr_wakeups_passive
;
1140 u64 nr_wakeups_idle
;
1144 struct sched_entity
{
1145 struct load_weight load
; /* for load-balancing */
1146 struct rb_node run_node
;
1147 struct list_head group_node
;
1151 u64 sum_exec_runtime
;
1153 u64 prev_sum_exec_runtime
;
1157 #ifdef CONFIG_SCHEDSTATS
1158 struct sched_statistics statistics
;
1161 #ifdef CONFIG_FAIR_GROUP_SCHED
1162 struct sched_entity
*parent
;
1163 /* rq on which this entity is (to be) queued: */
1164 struct cfs_rq
*cfs_rq
;
1165 /* rq "owned" by this entity/group: */
1166 struct cfs_rq
*my_q
;
1169 struct sched_avg avg
;
1173 struct sched_rt_entity
{
1174 struct list_head run_list
;
1175 unsigned long timeout
;
1176 unsigned int time_slice
;
1178 struct sched_rt_entity
*back
;
1179 #ifdef CONFIG_RT_GROUP_SCHED
1180 struct sched_rt_entity
*parent
;
1181 /* rq on which this entity is (to be) queued: */
1182 struct rt_rq
*rt_rq
;
1183 /* rq "owned" by this entity/group: */
1189 * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1190 * Timeslices get refilled after they expire.
1192 #define RR_TIMESLICE (100 * HZ / 1000)
1196 enum perf_event_task_context
{
1197 perf_invalid_context
= -1,
1198 perf_hw_context
= 0,
1200 perf_nr_task_contexts
,
1203 struct task_struct
{
1204 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1207 unsigned int flags
; /* per process flags, defined below */
1208 unsigned int ptrace
;
1211 struct llist_node wake_entry
;
1216 int prio
, static_prio
, normal_prio
;
1217 unsigned int rt_priority
;
1218 const struct sched_class
*sched_class
;
1219 struct sched_entity se
;
1220 struct sched_rt_entity rt
;
1221 #ifdef CONFIG_CGROUP_SCHED
1222 struct task_group
*sched_task_group
;
1225 #ifdef CONFIG_PREEMPT_NOTIFIERS
1226 /* list of struct preempt_notifier: */
1227 struct hlist_head preempt_notifiers
;
1231 * fpu_counter contains the number of consecutive context switches
1232 * that the FPU is used. If this is over a threshold, the lazy fpu
1233 * saving becomes unlazy to save the trap. This is an unsigned char
1234 * so that after 256 times the counter wraps and the behavior turns
1235 * lazy again; this to deal with bursty apps that only use FPU for
1238 unsigned char fpu_counter
;
1239 #ifdef CONFIG_BLK_DEV_IO_TRACE
1240 unsigned int btrace_seq
;
1243 unsigned int policy
;
1244 int nr_cpus_allowed
;
1245 cpumask_t cpus_allowed
;
1247 #ifdef CONFIG_PREEMPT_RCU
1248 int rcu_read_lock_nesting
;
1249 char rcu_read_unlock_special
;
1250 struct list_head rcu_node_entry
;
1251 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1252 #ifdef CONFIG_TREE_PREEMPT_RCU
1253 struct rcu_node
*rcu_blocked_node
;
1254 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1255 #ifdef CONFIG_RCU_BOOST
1256 struct rt_mutex
*rcu_boost_mutex
;
1257 #endif /* #ifdef CONFIG_RCU_BOOST */
1259 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1260 struct sched_info sched_info
;
1263 struct list_head tasks
;
1265 struct plist_node pushable_tasks
;
1268 struct mm_struct
*mm
, *active_mm
;
1269 #ifdef CONFIG_COMPAT_BRK
1270 unsigned brk_randomized
:1;
1272 #if defined(SPLIT_RSS_COUNTING)
1273 struct task_rss_stat rss_stat
;
1277 int exit_code
, exit_signal
;
1278 int pdeath_signal
; /* The signal sent when the parent dies */
1279 unsigned int jobctl
; /* JOBCTL_*, siglock protected */
1281 unsigned int personality
;
1282 unsigned did_exec
:1;
1283 unsigned in_execve
:1; /* Tell the LSMs that the process is doing an
1285 unsigned in_iowait
:1;
1287 /* task may not gain privileges */
1288 unsigned no_new_privs
:1;
1290 /* Revert to default priority/policy when forking */
1291 unsigned sched_reset_on_fork
:1;
1292 unsigned sched_contributes_to_load
:1;
1297 #ifdef CONFIG_CC_STACKPROTECTOR
1298 /* Canary value for the -fstack-protector gcc feature */
1299 unsigned long stack_canary
;
1302 * pointers to (original) parent process, youngest child, younger sibling,
1303 * older sibling, respectively. (p->father can be replaced with
1304 * p->real_parent->pid)
1306 struct task_struct __rcu
*real_parent
; /* real parent process */
1307 struct task_struct __rcu
*parent
; /* recipient of SIGCHLD, wait4() reports */
1309 * children/sibling forms the list of my natural children
1311 struct list_head children
; /* list of my children */
1312 struct list_head sibling
; /* linkage in my parent's children list */
1313 struct task_struct
*group_leader
; /* threadgroup leader */
1316 * ptraced is the list of tasks this task is using ptrace on.
1317 * This includes both natural children and PTRACE_ATTACH targets.
1318 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1320 struct list_head ptraced
;
1321 struct list_head ptrace_entry
;
1323 /* PID/PID hash table linkage. */
1324 struct pid_link pids
[PIDTYPE_MAX
];
1325 struct list_head thread_group
;
1327 struct completion
*vfork_done
; /* for vfork() */
1328 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1329 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1331 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1333 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1334 cputime_t prev_utime
, prev_stime
;
1336 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1337 struct timespec start_time
; /* monotonic time */
1338 struct timespec real_start_time
; /* boot based time */
1339 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1340 unsigned long min_flt
, maj_flt
;
1342 struct task_cputime cputime_expires
;
1343 struct list_head cpu_timers
[3];
1345 /* process credentials */
1346 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1347 * credentials (COW) */
1348 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1349 * credentials (COW) */
1350 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1351 - access with [gs]et_task_comm (which lock
1352 it with task_lock())
1353 - initialized normally by setup_new_exec */
1354 /* file system info */
1355 int link_count
, total_link_count
;
1356 #ifdef CONFIG_SYSVIPC
1358 struct sysv_sem sysvsem
;
1360 #ifdef CONFIG_DETECT_HUNG_TASK
1361 /* hung task detection */
1362 unsigned long last_switch_count
;
1364 /* CPU-specific state of this task */
1365 struct thread_struct thread
;
1366 /* filesystem information */
1367 struct fs_struct
*fs
;
1368 /* open file information */
1369 struct files_struct
*files
;
1371 struct nsproxy
*nsproxy
;
1372 /* signal handlers */
1373 struct signal_struct
*signal
;
1374 struct sighand_struct
*sighand
;
1376 sigset_t blocked
, real_blocked
;
1377 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1378 struct sigpending pending
;
1380 unsigned long sas_ss_sp
;
1382 int (*notifier
)(void *priv
);
1383 void *notifier_data
;
1384 sigset_t
*notifier_mask
;
1385 struct callback_head
*task_works
;
1387 struct audit_context
*audit_context
;
1388 #ifdef CONFIG_AUDITSYSCALL
1390 unsigned int sessionid
;
1392 struct seccomp seccomp
;
1394 /* Thread group tracking */
1397 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1399 spinlock_t alloc_lock
;
1401 /* Protection of the PI data structures: */
1402 raw_spinlock_t pi_lock
;
1404 #ifdef CONFIG_RT_MUTEXES
1405 /* PI waiters blocked on a rt_mutex held by this task */
1406 struct plist_head pi_waiters
;
1407 /* Deadlock detection and priority inheritance handling */
1408 struct rt_mutex_waiter
*pi_blocked_on
;
1411 #ifdef CONFIG_DEBUG_MUTEXES
1412 /* mutex deadlock detection */
1413 struct mutex_waiter
*blocked_on
;
1415 #ifdef CONFIG_TRACE_IRQFLAGS
1416 unsigned int irq_events
;
1417 unsigned long hardirq_enable_ip
;
1418 unsigned long hardirq_disable_ip
;
1419 unsigned int hardirq_enable_event
;
1420 unsigned int hardirq_disable_event
;
1421 int hardirqs_enabled
;
1422 int hardirq_context
;
1423 unsigned long softirq_disable_ip
;
1424 unsigned long softirq_enable_ip
;
1425 unsigned int softirq_disable_event
;
1426 unsigned int softirq_enable_event
;
1427 int softirqs_enabled
;
1428 int softirq_context
;
1430 #ifdef CONFIG_LOCKDEP
1431 # define MAX_LOCK_DEPTH 48UL
1434 unsigned int lockdep_recursion
;
1435 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1436 gfp_t lockdep_reclaim_gfp
;
1439 /* journalling filesystem info */
1442 /* stacked block device info */
1443 struct bio_list
*bio_list
;
1446 /* stack plugging */
1447 struct blk_plug
*plug
;
1451 struct reclaim_state
*reclaim_state
;
1453 struct backing_dev_info
*backing_dev_info
;
1455 struct io_context
*io_context
;
1457 unsigned long ptrace_message
;
1458 siginfo_t
*last_siginfo
; /* For ptrace use. */
1459 struct task_io_accounting ioac
;
1460 #if defined(CONFIG_TASK_XACCT)
1461 u64 acct_rss_mem1
; /* accumulated rss usage */
1462 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1463 cputime_t acct_timexpd
; /* stime + utime since last update */
1465 #ifdef CONFIG_CPUSETS
1466 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1467 seqcount_t mems_allowed_seq
; /* Seqence no to catch updates */
1468 int cpuset_mem_spread_rotor
;
1469 int cpuset_slab_spread_rotor
;
1471 #ifdef CONFIG_CGROUPS
1472 /* Control Group info protected by css_set_lock */
1473 struct css_set __rcu
*cgroups
;
1474 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1475 struct list_head cg_list
;
1478 struct robust_list_head __user
*robust_list
;
1479 #ifdef CONFIG_COMPAT
1480 struct compat_robust_list_head __user
*compat_robust_list
;
1482 struct list_head pi_state_list
;
1483 struct futex_pi_state
*pi_state_cache
;
1485 #ifdef CONFIG_PERF_EVENTS
1486 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1487 struct mutex perf_event_mutex
;
1488 struct list_head perf_event_list
;
1491 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1493 short pref_node_fork
;
1495 struct rcu_head rcu
;
1498 * cache last used pipe for splice
1500 struct pipe_inode_info
*splice_pipe
;
1502 struct page_frag task_frag
;
1504 #ifdef CONFIG_TASK_DELAY_ACCT
1505 struct task_delay_info
*delays
;
1507 #ifdef CONFIG_FAULT_INJECTION
1511 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1512 * balance_dirty_pages() for some dirty throttling pause
1515 int nr_dirtied_pause
;
1516 unsigned long dirty_paused_when
; /* start of a write-and-pause period */
1518 #ifdef CONFIG_LATENCYTOP
1519 int latency_record_count
;
1520 struct latency_record latency_record
[LT_SAVECOUNT
];
1523 * time slack values; these are used to round up poll() and
1524 * select() etc timeout values. These are in nanoseconds.
1526 unsigned long timer_slack_ns
;
1527 unsigned long default_timer_slack_ns
;
1529 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1530 /* Index of current stored address in ret_stack */
1532 /* Stack of return addresses for return function tracing */
1533 struct ftrace_ret_stack
*ret_stack
;
1534 /* time stamp for last schedule */
1535 unsigned long long ftrace_timestamp
;
1537 * Number of functions that haven't been traced
1538 * because of depth overrun.
1540 atomic_t trace_overrun
;
1541 /* Pause for the tracing */
1542 atomic_t tracing_graph_pause
;
1544 #ifdef CONFIG_TRACING
1545 /* state flags for use by tracers */
1546 unsigned long trace
;
1547 /* bitmask and counter of trace recursion */
1548 unsigned long trace_recursion
;
1549 #endif /* CONFIG_TRACING */
1550 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1551 struct memcg_batch_info
{
1552 int do_batch
; /* incremented when batch uncharge started */
1553 struct mem_cgroup
*memcg
; /* target memcg of uncharge */
1554 unsigned long nr_pages
; /* uncharged usage */
1555 unsigned long memsw_nr_pages
; /* uncharged mem+swap usage */
1558 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1559 atomic_t ptrace_bp_refcnt
;
1561 #ifdef CONFIG_UPROBES
1562 struct uprobe_task
*utask
;
1566 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1567 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1570 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1571 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1572 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1573 * values are inverted: lower p->prio value means higher priority.
1575 * The MAX_USER_RT_PRIO value allows the actual maximum
1576 * RT priority to be separate from the value exported to
1577 * user-space. This allows kernel threads to set their
1578 * priority to a value higher than any user task. Note:
1579 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1582 #define MAX_USER_RT_PRIO 100
1583 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1585 #define MAX_PRIO (MAX_RT_PRIO + 40)
1586 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1588 static inline int rt_prio(int prio
)
1590 if (unlikely(prio
< MAX_RT_PRIO
))
1595 static inline int rt_task(struct task_struct
*p
)
1597 return rt_prio(p
->prio
);
1600 static inline struct pid
*task_pid(struct task_struct
*task
)
1602 return task
->pids
[PIDTYPE_PID
].pid
;
1605 static inline struct pid
*task_tgid(struct task_struct
*task
)
1607 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1611 * Without tasklist or rcu lock it is not safe to dereference
1612 * the result of task_pgrp/task_session even if task == current,
1613 * we can race with another thread doing sys_setsid/sys_setpgid.
1615 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1617 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1620 static inline struct pid
*task_session(struct task_struct
*task
)
1622 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1625 struct pid_namespace
;
1628 * the helpers to get the task's different pids as they are seen
1629 * from various namespaces
1631 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1632 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1634 * task_xid_nr_ns() : id seen from the ns specified;
1636 * set_task_vxid() : assigns a virtual id to a task;
1638 * see also pid_nr() etc in include/linux/pid.h
1640 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
1641 struct pid_namespace
*ns
);
1643 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1648 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
1649 struct pid_namespace
*ns
)
1651 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1654 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1656 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1660 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1665 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1667 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1669 return pid_vnr(task_tgid(tsk
));
1673 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
1674 struct pid_namespace
*ns
)
1676 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1679 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1681 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1685 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
1686 struct pid_namespace
*ns
)
1688 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1691 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1693 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1696 /* obsolete, do not use */
1697 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1699 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1703 * pid_alive - check that a task structure is not stale
1704 * @p: Task structure to be checked.
1706 * Test if a process is not yet dead (at most zombie state)
1707 * If pid_alive fails, then pointers within the task structure
1708 * can be stale and must not be dereferenced.
1710 static inline int pid_alive(struct task_struct
*p
)
1712 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1716 * is_global_init - check if a task structure is init
1717 * @tsk: Task structure to be checked.
1719 * Check if a task structure is the first user space task the kernel created.
1721 static inline int is_global_init(struct task_struct
*tsk
)
1723 return tsk
->pid
== 1;
1727 * is_container_init:
1728 * check whether in the task is init in its own pid namespace.
1730 extern int is_container_init(struct task_struct
*tsk
);
1732 extern struct pid
*cad_pid
;
1734 extern void free_task(struct task_struct
*tsk
);
1735 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1737 extern void __put_task_struct(struct task_struct
*t
);
1739 static inline void put_task_struct(struct task_struct
*t
)
1741 if (atomic_dec_and_test(&t
->usage
))
1742 __put_task_struct(t
);
1745 extern void task_times(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
1746 extern void thread_group_times(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
1751 #define PF_EXITING 0x00000004 /* getting shut down */
1752 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1753 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1754 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1755 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1756 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1757 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1758 #define PF_DUMPCORE 0x00000200 /* dumped core */
1759 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1760 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1761 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1762 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1763 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1764 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1765 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1766 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1767 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1768 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1769 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1770 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1771 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1772 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1773 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1774 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1775 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1776 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1777 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1780 * Only the _current_ task can read/write to tsk->flags, but other
1781 * tasks can access tsk->flags in readonly mode for example
1782 * with tsk_used_math (like during threaded core dumping).
1783 * There is however an exception to this rule during ptrace
1784 * or during fork: the ptracer task is allowed to write to the
1785 * child->flags of its traced child (same goes for fork, the parent
1786 * can write to the child->flags), because we're guaranteed the
1787 * child is not running and in turn not changing child->flags
1788 * at the same time the parent does it.
1790 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1791 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1792 #define clear_used_math() clear_stopped_child_used_math(current)
1793 #define set_used_math() set_stopped_child_used_math(current)
1794 #define conditional_stopped_child_used_math(condition, child) \
1795 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1796 #define conditional_used_math(condition) \
1797 conditional_stopped_child_used_math(condition, current)
1798 #define copy_to_stopped_child_used_math(child) \
1799 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1800 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1801 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1802 #define used_math() tsk_used_math(current)
1805 * task->jobctl flags
1807 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1809 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1810 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1811 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1812 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1813 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1814 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1815 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1817 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1818 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1819 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1820 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1821 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1822 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1823 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1825 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1826 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1828 extern bool task_set_jobctl_pending(struct task_struct
*task
,
1830 extern void task_clear_jobctl_trapping(struct task_struct
*task
);
1831 extern void task_clear_jobctl_pending(struct task_struct
*task
,
1834 #ifdef CONFIG_PREEMPT_RCU
1836 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1837 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1839 static inline void rcu_copy_process(struct task_struct
*p
)
1841 p
->rcu_read_lock_nesting
= 0;
1842 p
->rcu_read_unlock_special
= 0;
1843 #ifdef CONFIG_TREE_PREEMPT_RCU
1844 p
->rcu_blocked_node
= NULL
;
1845 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1846 #ifdef CONFIG_RCU_BOOST
1847 p
->rcu_boost_mutex
= NULL
;
1848 #endif /* #ifdef CONFIG_RCU_BOOST */
1849 INIT_LIST_HEAD(&p
->rcu_node_entry
);
1854 static inline void rcu_copy_process(struct task_struct
*p
)
1860 static inline void rcu_switch(struct task_struct
*prev
,
1861 struct task_struct
*next
)
1863 #ifdef CONFIG_RCU_USER_QS
1864 rcu_user_hooks_switch(prev
, next
);
1868 static inline void tsk_restore_flags(struct task_struct
*task
,
1869 unsigned long orig_flags
, unsigned long flags
)
1871 task
->flags
&= ~flags
;
1872 task
->flags
|= orig_flags
& flags
;
1876 extern void do_set_cpus_allowed(struct task_struct
*p
,
1877 const struct cpumask
*new_mask
);
1879 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
1880 const struct cpumask
*new_mask
);
1882 static inline void do_set_cpus_allowed(struct task_struct
*p
,
1883 const struct cpumask
*new_mask
)
1886 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
1887 const struct cpumask
*new_mask
)
1889 if (!cpumask_test_cpu(0, new_mask
))
1896 void calc_load_enter_idle(void);
1897 void calc_load_exit_idle(void);
1899 static inline void calc_load_enter_idle(void) { }
1900 static inline void calc_load_exit_idle(void) { }
1901 #endif /* CONFIG_NO_HZ */
1903 #ifndef CONFIG_CPUMASK_OFFSTACK
1904 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1906 return set_cpus_allowed_ptr(p
, &new_mask
);
1911 * Do not use outside of architecture code which knows its limitations.
1913 * sched_clock() has no promise of monotonicity or bounded drift between
1914 * CPUs, use (which you should not) requires disabling IRQs.
1916 * Please use one of the three interfaces below.
1918 extern unsigned long long notrace
sched_clock(void);
1920 * See the comment in kernel/sched/clock.c
1922 extern u64
cpu_clock(int cpu
);
1923 extern u64
local_clock(void);
1924 extern u64
sched_clock_cpu(int cpu
);
1927 extern void sched_clock_init(void);
1929 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1930 static inline void sched_clock_tick(void)
1934 static inline void sched_clock_idle_sleep_event(void)
1938 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
1943 * Architectures can set this to 1 if they have specified
1944 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1945 * but then during bootup it turns out that sched_clock()
1946 * is reliable after all:
1948 extern int sched_clock_stable
;
1950 extern void sched_clock_tick(void);
1951 extern void sched_clock_idle_sleep_event(void);
1952 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1955 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1957 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1958 * The reason for this explicit opt-in is not to have perf penalty with
1959 * slow sched_clocks.
1961 extern void enable_sched_clock_irqtime(void);
1962 extern void disable_sched_clock_irqtime(void);
1964 static inline void enable_sched_clock_irqtime(void) {}
1965 static inline void disable_sched_clock_irqtime(void) {}
1968 extern unsigned long long
1969 task_sched_runtime(struct task_struct
*task
);
1971 /* sched_exec is called by processes performing an exec */
1973 extern void sched_exec(void);
1975 #define sched_exec() {}
1978 extern void sched_clock_idle_sleep_event(void);
1979 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1981 #ifdef CONFIG_HOTPLUG_CPU
1982 extern void idle_task_exit(void);
1984 static inline void idle_task_exit(void) {}
1987 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1988 extern void wake_up_idle_cpu(int cpu
);
1990 static inline void wake_up_idle_cpu(int cpu
) { }
1993 extern unsigned int sysctl_sched_latency
;
1994 extern unsigned int sysctl_sched_min_granularity
;
1995 extern unsigned int sysctl_sched_wakeup_granularity
;
1996 extern unsigned int sysctl_sched_child_runs_first
;
1998 enum sched_tunable_scaling
{
1999 SCHED_TUNABLESCALING_NONE
,
2000 SCHED_TUNABLESCALING_LOG
,
2001 SCHED_TUNABLESCALING_LINEAR
,
2002 SCHED_TUNABLESCALING_END
,
2004 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling
;
2006 #ifdef CONFIG_SCHED_DEBUG
2007 extern unsigned int sysctl_sched_migration_cost
;
2008 extern unsigned int sysctl_sched_nr_migrate
;
2009 extern unsigned int sysctl_sched_time_avg
;
2010 extern unsigned int sysctl_timer_migration
;
2011 extern unsigned int sysctl_sched_shares_window
;
2013 int sched_proc_update_handler(struct ctl_table
*table
, int write
,
2014 void __user
*buffer
, size_t *length
,
2017 #ifdef CONFIG_SCHED_DEBUG
2018 static inline unsigned int get_sysctl_timer_migration(void)
2020 return sysctl_timer_migration
;
2023 static inline unsigned int get_sysctl_timer_migration(void)
2028 extern unsigned int sysctl_sched_rt_period
;
2029 extern int sysctl_sched_rt_runtime
;
2031 int sched_rt_handler(struct ctl_table
*table
, int write
,
2032 void __user
*buffer
, size_t *lenp
,
2035 #ifdef CONFIG_SCHED_AUTOGROUP
2036 extern unsigned int sysctl_sched_autogroup_enabled
;
2038 extern void sched_autogroup_create_attach(struct task_struct
*p
);
2039 extern void sched_autogroup_detach(struct task_struct
*p
);
2040 extern void sched_autogroup_fork(struct signal_struct
*sig
);
2041 extern void sched_autogroup_exit(struct signal_struct
*sig
);
2042 #ifdef CONFIG_PROC_FS
2043 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
2044 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int nice
);
2047 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
2048 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
2049 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
2050 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
2053 #ifdef CONFIG_CFS_BANDWIDTH
2054 extern unsigned int sysctl_sched_cfs_bandwidth_slice
;
2057 #ifdef CONFIG_RT_MUTEXES
2058 extern int rt_mutex_getprio(struct task_struct
*p
);
2059 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
2060 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
2061 static inline bool tsk_is_pi_blocked(struct task_struct
*tsk
)
2063 return tsk
->pi_blocked_on
!= NULL
;
2066 static inline int rt_mutex_getprio(struct task_struct
*p
)
2068 return p
->normal_prio
;
2070 # define rt_mutex_adjust_pi(p) do { } while (0)
2071 static inline bool tsk_is_pi_blocked(struct task_struct
*tsk
)
2077 extern bool yield_to(struct task_struct
*p
, bool preempt
);
2078 extern void set_user_nice(struct task_struct
*p
, long nice
);
2079 extern int task_prio(const struct task_struct
*p
);
2080 extern int task_nice(const struct task_struct
*p
);
2081 extern int can_nice(const struct task_struct
*p
, const int nice
);
2082 extern int task_curr(const struct task_struct
*p
);
2083 extern int idle_cpu(int cpu
);
2084 extern int sched_setscheduler(struct task_struct
*, int,
2085 const struct sched_param
*);
2086 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
2087 const struct sched_param
*);
2088 extern struct task_struct
*idle_task(int cpu
);
2090 * is_idle_task - is the specified task an idle task?
2091 * @p: the task in question.
2093 static inline bool is_idle_task(const struct task_struct
*p
)
2097 extern struct task_struct
*curr_task(int cpu
);
2098 extern void set_curr_task(int cpu
, struct task_struct
*p
);
2103 * The default (Linux) execution domain.
2105 extern struct exec_domain default_exec_domain
;
2107 union thread_union
{
2108 struct thread_info thread_info
;
2109 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
2112 #ifndef __HAVE_ARCH_KSTACK_END
2113 static inline int kstack_end(void *addr
)
2115 /* Reliable end of stack detection:
2116 * Some APM bios versions misalign the stack
2118 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
2122 extern union thread_union init_thread_union
;
2123 extern struct task_struct init_task
;
2125 extern struct mm_struct init_mm
;
2127 extern struct pid_namespace init_pid_ns
;
2130 * find a task by one of its numerical ids
2132 * find_task_by_pid_ns():
2133 * finds a task by its pid in the specified namespace
2134 * find_task_by_vpid():
2135 * finds a task by its virtual pid
2137 * see also find_vpid() etc in include/linux/pid.h
2140 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
2141 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
2142 struct pid_namespace
*ns
);
2144 extern void __set_special_pids(struct pid
*pid
);
2146 /* per-UID process charging. */
2147 extern struct user_struct
* alloc_uid(kuid_t
);
2148 static inline struct user_struct
*get_uid(struct user_struct
*u
)
2150 atomic_inc(&u
->__count
);
2153 extern void free_uid(struct user_struct
*);
2155 #include <asm/current.h>
2157 extern void xtime_update(unsigned long ticks
);
2159 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2160 extern int wake_up_process(struct task_struct
*tsk
);
2161 extern void wake_up_new_task(struct task_struct
*tsk
);
2163 extern void kick_process(struct task_struct
*tsk
);
2165 static inline void kick_process(struct task_struct
*tsk
) { }
2167 extern void sched_fork(struct task_struct
*p
);
2168 extern void sched_dead(struct task_struct
*p
);
2170 extern void proc_caches_init(void);
2171 extern void flush_signals(struct task_struct
*);
2172 extern void __flush_signals(struct task_struct
*);
2173 extern void ignore_signals(struct task_struct
*);
2174 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2175 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2177 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
2179 unsigned long flags
;
2182 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
2183 ret
= dequeue_signal(tsk
, mask
, info
);
2184 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
2189 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
2191 extern void unblock_all_signals(void);
2192 extern void release_task(struct task_struct
* p
);
2193 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2194 extern int force_sigsegv(int, struct task_struct
*);
2195 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2196 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2197 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2198 extern int kill_pid_info_as_cred(int, struct siginfo
*, struct pid
*,
2199 const struct cred
*, u32
);
2200 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2201 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2202 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2203 extern __must_check
bool do_notify_parent(struct task_struct
*, int);
2204 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2205 extern void force_sig(int, struct task_struct
*);
2206 extern int send_sig(int, struct task_struct
*, int);
2207 extern int zap_other_threads(struct task_struct
*p
);
2208 extern struct sigqueue
*sigqueue_alloc(void);
2209 extern void sigqueue_free(struct sigqueue
*);
2210 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2211 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2212 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
2214 static inline void restore_saved_sigmask(void)
2216 if (test_and_clear_restore_sigmask())
2217 __set_current_blocked(¤t
->saved_sigmask
);
2220 static inline sigset_t
*sigmask_to_save(void)
2222 sigset_t
*res
= ¤t
->blocked
;
2223 if (unlikely(test_restore_sigmask()))
2224 res
= ¤t
->saved_sigmask
;
2228 static inline int kill_cad_pid(int sig
, int priv
)
2230 return kill_pid(cad_pid
, sig
, priv
);
2233 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2234 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2235 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2236 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2239 * True if we are on the alternate signal stack.
2241 static inline int on_sig_stack(unsigned long sp
)
2243 #ifdef CONFIG_STACK_GROWSUP
2244 return sp
>= current
->sas_ss_sp
&&
2245 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2247 return sp
> current
->sas_ss_sp
&&
2248 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2252 static inline int sas_ss_flags(unsigned long sp
)
2254 return (current
->sas_ss_size
== 0 ? SS_DISABLE
2255 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
2259 * Routines for handling mm_structs
2261 extern struct mm_struct
* mm_alloc(void);
2263 /* mmdrop drops the mm and the page tables */
2264 extern void __mmdrop(struct mm_struct
*);
2265 static inline void mmdrop(struct mm_struct
* mm
)
2267 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2271 /* mmput gets rid of the mappings and all user-space */
2272 extern void mmput(struct mm_struct
*);
2273 /* Grab a reference to a task's mm, if it is not already going away */
2274 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2276 * Grab a reference to a task's mm, if it is not already going away
2277 * and ptrace_may_access with the mode parameter passed to it
2280 extern struct mm_struct
*mm_access(struct task_struct
*task
, unsigned int mode
);
2281 /* Remove the current tasks stale references to the old mm_struct */
2282 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2283 /* Allocate a new mm structure and copy contents from tsk->mm */
2284 extern struct mm_struct
*dup_mm(struct task_struct
*tsk
);
2286 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2287 struct task_struct
*, struct pt_regs
*);
2288 extern void flush_thread(void);
2289 extern void exit_thread(void);
2291 extern void exit_files(struct task_struct
*);
2292 extern void __cleanup_sighand(struct sighand_struct
*);
2294 extern void exit_itimers(struct signal_struct
*);
2295 extern void flush_itimer_signals(void);
2297 extern void do_group_exit(int);
2299 extern void daemonize(const char *, ...);
2300 extern int allow_signal(int);
2301 extern int disallow_signal(int);
2303 extern int do_execve(const char *,
2304 const char __user
* const __user
*,
2305 const char __user
* const __user
*, struct pt_regs
*);
2306 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
2307 struct task_struct
*fork_idle(int);
2308 #ifdef CONFIG_GENERIC_KERNEL_THREAD
2309 extern pid_t
kernel_thread(int (*fn
)(void *), void *arg
, unsigned long flags
);
2312 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
2313 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2316 void scheduler_ipi(void);
2317 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2319 static inline void scheduler_ipi(void) { }
2320 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2327 #define next_task(p) \
2328 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2330 #define for_each_process(p) \
2331 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2333 extern bool current_is_single_threaded(void);
2336 * Careful: do_each_thread/while_each_thread is a double loop so
2337 * 'break' will not work as expected - use goto instead.
2339 #define do_each_thread(g, t) \
2340 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2342 #define while_each_thread(g, t) \
2343 while ((t = next_thread(t)) != g)
2345 static inline int get_nr_threads(struct task_struct
*tsk
)
2347 return tsk
->signal
->nr_threads
;
2350 static inline bool thread_group_leader(struct task_struct
*p
)
2352 return p
->exit_signal
>= 0;
2355 /* Do to the insanities of de_thread it is possible for a process
2356 * to have the pid of the thread group leader without actually being
2357 * the thread group leader. For iteration through the pids in proc
2358 * all we care about is that we have a task with the appropriate
2359 * pid, we don't actually care if we have the right task.
2361 static inline int has_group_leader_pid(struct task_struct
*p
)
2363 return p
->pid
== p
->tgid
;
2367 int same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2369 return p1
->tgid
== p2
->tgid
;
2372 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2374 return list_entry_rcu(p
->thread_group
.next
,
2375 struct task_struct
, thread_group
);
2378 static inline int thread_group_empty(struct task_struct
*p
)
2380 return list_empty(&p
->thread_group
);
2383 #define delay_group_leader(p) \
2384 (thread_group_leader(p) && !thread_group_empty(p))
2387 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2388 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2389 * pins the final release of task.io_context. Also protects ->cpuset and
2390 * ->cgroup.subsys[]. And ->vfork_done.
2392 * Nests both inside and outside of read_lock(&tasklist_lock).
2393 * It must not be nested with write_lock_irq(&tasklist_lock),
2394 * neither inside nor outside.
2396 static inline void task_lock(struct task_struct
*p
)
2398 spin_lock(&p
->alloc_lock
);
2401 static inline void task_unlock(struct task_struct
*p
)
2403 spin_unlock(&p
->alloc_lock
);
2406 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
2407 unsigned long *flags
);
2409 static inline struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
2410 unsigned long *flags
)
2412 struct sighand_struct
*ret
;
2414 ret
= __lock_task_sighand(tsk
, flags
);
2415 (void)__cond_lock(&tsk
->sighand
->siglock
, ret
);
2419 static inline void unlock_task_sighand(struct task_struct
*tsk
,
2420 unsigned long *flags
)
2422 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2425 #ifdef CONFIG_CGROUPS
2426 static inline void threadgroup_change_begin(struct task_struct
*tsk
)
2428 down_read(&tsk
->signal
->group_rwsem
);
2430 static inline void threadgroup_change_end(struct task_struct
*tsk
)
2432 up_read(&tsk
->signal
->group_rwsem
);
2436 * threadgroup_lock - lock threadgroup
2437 * @tsk: member task of the threadgroup to lock
2439 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2440 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2441 * perform exec. This is useful for cases where the threadgroup needs to
2442 * stay stable across blockable operations.
2444 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2445 * synchronization. While held, no new task will be added to threadgroup
2446 * and no existing live task will have its PF_EXITING set.
2448 * During exec, a task goes and puts its thread group through unusual
2449 * changes. After de-threading, exclusive access is assumed to resources
2450 * which are usually shared by tasks in the same group - e.g. sighand may
2451 * be replaced with a new one. Also, the exec'ing task takes over group
2452 * leader role including its pid. Exclude these changes while locked by
2453 * grabbing cred_guard_mutex which is used to synchronize exec path.
2455 static inline void threadgroup_lock(struct task_struct
*tsk
)
2458 * exec uses exit for de-threading nesting group_rwsem inside
2459 * cred_guard_mutex. Grab cred_guard_mutex first.
2461 mutex_lock(&tsk
->signal
->cred_guard_mutex
);
2462 down_write(&tsk
->signal
->group_rwsem
);
2466 * threadgroup_unlock - unlock threadgroup
2467 * @tsk: member task of the threadgroup to unlock
2469 * Reverse threadgroup_lock().
2471 static inline void threadgroup_unlock(struct task_struct
*tsk
)
2473 up_write(&tsk
->signal
->group_rwsem
);
2474 mutex_unlock(&tsk
->signal
->cred_guard_mutex
);
2477 static inline void threadgroup_change_begin(struct task_struct
*tsk
) {}
2478 static inline void threadgroup_change_end(struct task_struct
*tsk
) {}
2479 static inline void threadgroup_lock(struct task_struct
*tsk
) {}
2480 static inline void threadgroup_unlock(struct task_struct
*tsk
) {}
2483 #ifndef __HAVE_THREAD_FUNCTIONS
2485 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2486 #define task_stack_page(task) ((task)->stack)
2488 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2490 *task_thread_info(p
) = *task_thread_info(org
);
2491 task_thread_info(p
)->task
= p
;
2494 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2496 return (unsigned long *)(task_thread_info(p
) + 1);
2501 static inline int object_is_on_stack(void *obj
)
2503 void *stack
= task_stack_page(current
);
2505 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2508 extern void thread_info_cache_init(void);
2510 #ifdef CONFIG_DEBUG_STACK_USAGE
2511 static inline unsigned long stack_not_used(struct task_struct
*p
)
2513 unsigned long *n
= end_of_stack(p
);
2515 do { /* Skip over canary */
2519 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
2523 /* set thread flags in other task's structures
2524 * - see asm/thread_info.h for TIF_xxxx flags available
2526 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2528 set_ti_thread_flag(task_thread_info(tsk
), flag
);
2531 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2533 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2536 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2538 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
2541 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2543 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2546 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2548 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
2551 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
2553 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2556 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
2558 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2561 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
2563 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
2566 static inline int restart_syscall(void)
2568 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
2569 return -ERESTARTNOINTR
;
2572 static inline int signal_pending(struct task_struct
*p
)
2574 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
2577 static inline int __fatal_signal_pending(struct task_struct
*p
)
2579 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
2582 static inline int fatal_signal_pending(struct task_struct
*p
)
2584 return signal_pending(p
) && __fatal_signal_pending(p
);
2587 static inline int signal_pending_state(long state
, struct task_struct
*p
)
2589 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
2591 if (!signal_pending(p
))
2594 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
2597 static inline int need_resched(void)
2599 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
2603 * cond_resched() and cond_resched_lock(): latency reduction via
2604 * explicit rescheduling in places that are safe. The return
2605 * value indicates whether a reschedule was done in fact.
2606 * cond_resched_lock() will drop the spinlock before scheduling,
2607 * cond_resched_softirq() will enable bhs before scheduling.
2609 extern int _cond_resched(void);
2611 #define cond_resched() ({ \
2612 __might_sleep(__FILE__, __LINE__, 0); \
2616 extern int __cond_resched_lock(spinlock_t
*lock
);
2618 #ifdef CONFIG_PREEMPT_COUNT
2619 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2621 #define PREEMPT_LOCK_OFFSET 0
2624 #define cond_resched_lock(lock) ({ \
2625 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2626 __cond_resched_lock(lock); \
2629 extern int __cond_resched_softirq(void);
2631 #define cond_resched_softirq() ({ \
2632 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2633 __cond_resched_softirq(); \
2637 * Does a critical section need to be broken due to another
2638 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2639 * but a general need for low latency)
2641 static inline int spin_needbreak(spinlock_t
*lock
)
2643 #ifdef CONFIG_PREEMPT
2644 return spin_is_contended(lock
);
2651 * Thread group CPU time accounting.
2653 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
2654 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
2656 static inline void thread_group_cputime_init(struct signal_struct
*sig
)
2658 raw_spin_lock_init(&sig
->cputimer
.lock
);
2662 * Reevaluate whether the task has signals pending delivery.
2663 * Wake the task if so.
2664 * This is required every time the blocked sigset_t changes.
2665 * callers must hold sighand->siglock.
2667 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
2668 extern void recalc_sigpending(void);
2670 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
2673 * Wrappers for p->thread_info->cpu access. No-op on UP.
2677 static inline unsigned int task_cpu(const struct task_struct
*p
)
2679 return task_thread_info(p
)->cpu
;
2682 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
2686 static inline unsigned int task_cpu(const struct task_struct
*p
)
2691 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
2695 #endif /* CONFIG_SMP */
2697 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
2698 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
2700 extern void normalize_rt_tasks(void);
2702 #ifdef CONFIG_CGROUP_SCHED
2704 extern struct task_group root_task_group
;
2706 extern struct task_group
*sched_create_group(struct task_group
*parent
);
2707 extern void sched_destroy_group(struct task_group
*tg
);
2708 extern void sched_move_task(struct task_struct
*tsk
);
2709 #ifdef CONFIG_FAIR_GROUP_SCHED
2710 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
2711 extern unsigned long sched_group_shares(struct task_group
*tg
);
2713 #ifdef CONFIG_RT_GROUP_SCHED
2714 extern int sched_group_set_rt_runtime(struct task_group
*tg
,
2715 long rt_runtime_us
);
2716 extern long sched_group_rt_runtime(struct task_group
*tg
);
2717 extern int sched_group_set_rt_period(struct task_group
*tg
,
2719 extern long sched_group_rt_period(struct task_group
*tg
);
2720 extern int sched_rt_can_attach(struct task_group
*tg
, struct task_struct
*tsk
);
2722 #endif /* CONFIG_CGROUP_SCHED */
2724 extern int task_can_switch_user(struct user_struct
*up
,
2725 struct task_struct
*tsk
);
2727 #ifdef CONFIG_TASK_XACCT
2728 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2730 tsk
->ioac
.rchar
+= amt
;
2733 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2735 tsk
->ioac
.wchar
+= amt
;
2738 static inline void inc_syscr(struct task_struct
*tsk
)
2743 static inline void inc_syscw(struct task_struct
*tsk
)
2748 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2752 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2756 static inline void inc_syscr(struct task_struct
*tsk
)
2760 static inline void inc_syscw(struct task_struct
*tsk
)
2765 #ifndef TASK_SIZE_OF
2766 #define TASK_SIZE_OF(tsk) TASK_SIZE
2769 #ifdef CONFIG_MM_OWNER
2770 extern void mm_update_next_owner(struct mm_struct
*mm
);
2771 extern void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
);
2773 static inline void mm_update_next_owner(struct mm_struct
*mm
)
2777 static inline void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
)
2780 #endif /* CONFIG_MM_OWNER */
2782 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
2785 return ACCESS_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
2788 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
2791 return ACCESS_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
2794 static inline unsigned long rlimit(unsigned int limit
)
2796 return task_rlimit(current
, limit
);
2799 static inline unsigned long rlimit_max(unsigned int limit
)
2801 return task_rlimit_max(current
, limit
);