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>
25 #include <linux/preempt.h>
28 #include <asm/ptrace.h>
29 #include <asm/cputime.h>
31 #include <linux/smp.h>
32 #include <linux/sem.h>
33 #include <linux/signal.h>
34 #include <linux/compiler.h>
35 #include <linux/completion.h>
36 #include <linux/pid.h>
37 #include <linux/percpu.h>
38 #include <linux/topology.h>
39 #include <linux/proportions.h>
40 #include <linux/seccomp.h>
41 #include <linux/rcupdate.h>
42 #include <linux/rculist.h>
43 #include <linux/rtmutex.h>
45 #include <linux/time.h>
46 #include <linux/param.h>
47 #include <linux/resource.h>
48 #include <linux/timer.h>
49 #include <linux/hrtimer.h>
50 #include <linux/task_io_accounting.h>
51 #include <linux/latencytop.h>
52 #include <linux/cred.h>
53 #include <linux/llist.h>
54 #include <linux/uidgid.h>
55 #include <linux/gfp.h>
57 #include <asm/processor.h>
60 struct futex_pi_state
;
61 struct robust_list_head
;
64 struct perf_event_context
;
68 * List of flags we want to share for kernel threads,
69 * if only because they are not used by them anyway.
71 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
74 * These are the constant used to fake the fixed-point load-average
75 * counting. Some notes:
76 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
77 * a load-average precision of 10 bits integer + 11 bits fractional
78 * - if you want to count load-averages more often, you need more
79 * precision, or rounding will get you. With 2-second counting freq,
80 * the EXP_n values would be 1981, 2034 and 2043 if still using only
83 extern unsigned long avenrun
[]; /* Load averages */
84 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
86 #define FSHIFT 11 /* nr of bits of precision */
87 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
88 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
89 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
90 #define EXP_5 2014 /* 1/exp(5sec/5min) */
91 #define EXP_15 2037 /* 1/exp(5sec/15min) */
93 #define CALC_LOAD(load,exp,n) \
95 load += n*(FIXED_1-exp); \
98 extern unsigned long total_forks
;
99 extern int nr_threads
;
100 DECLARE_PER_CPU(unsigned long, process_counts
);
101 extern int nr_processes(void);
102 extern unsigned long nr_running(void);
103 extern unsigned long nr_iowait(void);
104 extern unsigned long nr_iowait_cpu(int cpu
);
105 extern unsigned long this_cpu_load(void);
108 extern void calc_global_load(unsigned long ticks
);
109 extern void update_cpu_load_nohz(void);
111 extern unsigned long get_parent_ip(unsigned long addr
);
113 extern void dump_cpu_task(int cpu
);
118 #ifdef CONFIG_SCHED_DEBUG
119 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
120 extern void proc_sched_set_task(struct task_struct
*p
);
122 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
);
126 * Task state bitmask. NOTE! These bits are also
127 * encoded in fs/proc/array.c: get_task_state().
129 * We have two separate sets of flags: task->state
130 * is about runnability, while task->exit_state are
131 * about the task exiting. Confusing, but this way
132 * modifying one set can't modify the other one by
135 #define TASK_RUNNING 0
136 #define TASK_INTERRUPTIBLE 1
137 #define TASK_UNINTERRUPTIBLE 2
138 #define __TASK_STOPPED 4
139 #define __TASK_TRACED 8
140 /* in tsk->exit_state */
141 #define EXIT_ZOMBIE 16
143 /* in tsk->state again */
145 #define TASK_WAKEKILL 128
146 #define TASK_WAKING 256
147 #define TASK_PARKED 512
148 #define TASK_STATE_MAX 1024
150 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
152 extern char ___assert_task_state
[1 - 2*!!(
153 sizeof(TASK_STATE_TO_CHAR_STR
)-1 != ilog2(TASK_STATE_MAX
)+1)];
155 /* Convenience macros for the sake of set_task_state */
156 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
157 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
158 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
160 /* Convenience macros for the sake of wake_up */
161 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
162 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
164 /* get_task_state() */
165 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
166 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
169 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
170 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
171 #define task_is_dead(task) ((task)->exit_state != 0)
172 #define task_is_stopped_or_traced(task) \
173 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
174 #define task_contributes_to_load(task) \
175 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
176 (task->flags & PF_FROZEN) == 0)
178 #define __set_task_state(tsk, state_value) \
179 do { (tsk)->state = (state_value); } while (0)
180 #define set_task_state(tsk, state_value) \
181 set_mb((tsk)->state, (state_value))
184 * set_current_state() includes a barrier so that the write of current->state
185 * is correctly serialised wrt the caller's subsequent test of whether to
188 * set_current_state(TASK_UNINTERRUPTIBLE);
189 * if (do_i_need_to_sleep())
192 * If the caller does not need such serialisation then use __set_current_state()
194 #define __set_current_state(state_value) \
195 do { current->state = (state_value); } while (0)
196 #define set_current_state(state_value) \
197 set_mb(current->state, (state_value))
199 /* Task command name length */
200 #define TASK_COMM_LEN 16
202 #include <linux/spinlock.h>
205 * This serializes "schedule()" and also protects
206 * the run-queue from deletions/modifications (but
207 * _adding_ to the beginning of the run-queue has
210 extern rwlock_t tasklist_lock
;
211 extern spinlock_t mmlist_lock
;
215 #ifdef CONFIG_PROVE_RCU
216 extern int lockdep_tasklist_lock_is_held(void);
217 #endif /* #ifdef CONFIG_PROVE_RCU */
219 extern void sched_init(void);
220 extern void sched_init_smp(void);
221 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
222 extern void init_idle(struct task_struct
*idle
, int cpu
);
223 extern void init_idle_bootup_task(struct task_struct
*idle
);
225 extern int runqueue_is_locked(int cpu
);
227 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
228 extern void nohz_balance_enter_idle(int cpu
);
229 extern void set_cpu_sd_state_idle(void);
230 extern int get_nohz_timer_target(void);
232 static inline void nohz_balance_enter_idle(int cpu
) { }
233 static inline void set_cpu_sd_state_idle(void) { }
237 * Only dump TASK_* tasks. (0 for all tasks)
239 extern void show_state_filter(unsigned long state_filter
);
241 static inline void show_state(void)
243 show_state_filter(0);
246 extern void show_regs(struct pt_regs
*);
249 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
250 * task), SP is the stack pointer of the first frame that should be shown in the back
251 * trace (or NULL if the entire call-chain of the task should be shown).
253 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
255 void io_schedule(void);
256 long io_schedule_timeout(long timeout
);
258 extern void cpu_init (void);
259 extern void trap_init(void);
260 extern void update_process_times(int user
);
261 extern void scheduler_tick(void);
263 extern void sched_show_task(struct task_struct
*p
);
265 #ifdef CONFIG_LOCKUP_DETECTOR
266 extern void touch_softlockup_watchdog(void);
267 extern void touch_softlockup_watchdog_sync(void);
268 extern void touch_all_softlockup_watchdogs(void);
269 extern int proc_dowatchdog_thresh(struct ctl_table
*table
, int write
,
271 size_t *lenp
, loff_t
*ppos
);
272 extern unsigned int softlockup_panic
;
273 void lockup_detector_init(void);
275 static inline void touch_softlockup_watchdog(void)
278 static inline void touch_softlockup_watchdog_sync(void)
281 static inline void touch_all_softlockup_watchdogs(void)
284 static inline void lockup_detector_init(void)
289 /* Attach to any functions which should be ignored in wchan output. */
290 #define __sched __attribute__((__section__(".sched.text")))
292 /* Linker adds these: start and end of __sched functions */
293 extern char __sched_text_start
[], __sched_text_end
[];
295 /* Is this address in the __sched functions? */
296 extern int in_sched_functions(unsigned long addr
);
298 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
299 extern signed long schedule_timeout(signed long timeout
);
300 extern signed long schedule_timeout_interruptible(signed long timeout
);
301 extern signed long schedule_timeout_killable(signed long timeout
);
302 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
303 asmlinkage
void schedule(void);
304 extern void schedule_preempt_disabled(void);
307 struct user_namespace
;
310 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
312 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
313 unsigned long, unsigned long);
315 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
316 unsigned long len
, unsigned long pgoff
,
317 unsigned long flags
);
319 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
) {}
323 extern void set_dumpable(struct mm_struct
*mm
, int value
);
324 extern int get_dumpable(struct mm_struct
*mm
);
328 #define MMF_DUMPABLE 0 /* core dump is permitted */
329 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
331 #define MMF_DUMPABLE_BITS 2
332 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
334 /* coredump filter bits */
335 #define MMF_DUMP_ANON_PRIVATE 2
336 #define MMF_DUMP_ANON_SHARED 3
337 #define MMF_DUMP_MAPPED_PRIVATE 4
338 #define MMF_DUMP_MAPPED_SHARED 5
339 #define MMF_DUMP_ELF_HEADERS 6
340 #define MMF_DUMP_HUGETLB_PRIVATE 7
341 #define MMF_DUMP_HUGETLB_SHARED 8
343 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
344 #define MMF_DUMP_FILTER_BITS 7
345 #define MMF_DUMP_FILTER_MASK \
346 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
347 #define MMF_DUMP_FILTER_DEFAULT \
348 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
349 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
351 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
352 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
354 # define MMF_DUMP_MASK_DEFAULT_ELF 0
356 /* leave room for more dump flags */
357 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
358 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
359 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
361 #define MMF_HAS_UPROBES 19 /* has uprobes */
362 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
364 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
366 struct sighand_struct
{
368 struct k_sigaction action
[_NSIG
];
370 wait_queue_head_t signalfd_wqh
;
373 struct pacct_struct
{
376 unsigned long ac_mem
;
377 cputime_t ac_utime
, ac_stime
;
378 unsigned long ac_minflt
, ac_majflt
;
389 * struct cputime - snaphsot of system and user cputime
390 * @utime: time spent in user mode
391 * @stime: time spent in system mode
393 * Gathers a generic snapshot of user and system time.
401 * struct task_cputime - collected CPU time counts
402 * @utime: time spent in user mode, in &cputime_t units
403 * @stime: time spent in kernel mode, in &cputime_t units
404 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
406 * This is an extension of struct cputime that includes the total runtime
407 * spent by the task from the scheduler point of view.
409 * As a result, this structure groups together three kinds of CPU time
410 * that are tracked for threads and thread groups. Most things considering
411 * CPU time want to group these counts together and treat all three
412 * of them in parallel.
414 struct task_cputime
{
417 unsigned long long sum_exec_runtime
;
419 /* Alternate field names when used to cache expirations. */
420 #define prof_exp stime
421 #define virt_exp utime
422 #define sched_exp sum_exec_runtime
424 #define INIT_CPUTIME \
425 (struct task_cputime) { \
428 .sum_exec_runtime = 0, \
431 #define PREEMPT_ENABLED (PREEMPT_NEED_RESCHED)
433 #ifdef CONFIG_PREEMPT_COUNT
434 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
436 #define PREEMPT_DISABLED PREEMPT_ENABLED
440 * Disable preemption until the scheduler is running.
441 * Reset by start_kernel()->sched_init()->init_idle().
443 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
444 * before the scheduler is active -- see should_resched().
446 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
449 * struct thread_group_cputimer - thread group interval timer counts
450 * @cputime: thread group interval timers.
451 * @running: non-zero when there are timers running and
452 * @cputime receives updates.
453 * @lock: lock for fields in this struct.
455 * This structure contains the version of task_cputime, above, that is
456 * used for thread group CPU timer calculations.
458 struct thread_group_cputimer
{
459 struct task_cputime cputime
;
464 #include <linux/rwsem.h>
468 * NOTE! "signal_struct" does not have its own
469 * locking, because a shared signal_struct always
470 * implies a shared sighand_struct, so locking
471 * sighand_struct is always a proper superset of
472 * the locking of signal_struct.
474 struct signal_struct
{
479 wait_queue_head_t wait_chldexit
; /* for wait4() */
481 /* current thread group signal load-balancing target: */
482 struct task_struct
*curr_target
;
484 /* shared signal handling: */
485 struct sigpending shared_pending
;
487 /* thread group exit support */
490 * - notify group_exit_task when ->count is equal to notify_count
491 * - everyone except group_exit_task is stopped during signal delivery
492 * of fatal signals, group_exit_task processes the signal.
495 struct task_struct
*group_exit_task
;
497 /* thread group stop support, overloads group_exit_code too */
498 int group_stop_count
;
499 unsigned int flags
; /* see SIGNAL_* flags below */
502 * PR_SET_CHILD_SUBREAPER marks a process, like a service
503 * manager, to re-parent orphan (double-forking) child processes
504 * to this process instead of 'init'. The service manager is
505 * able to receive SIGCHLD signals and is able to investigate
506 * the process until it calls wait(). All children of this
507 * process will inherit a flag if they should look for a
508 * child_subreaper process at exit.
510 unsigned int is_child_subreaper
:1;
511 unsigned int has_child_subreaper
:1;
513 /* POSIX.1b Interval Timers */
515 struct list_head posix_timers
;
517 /* ITIMER_REAL timer for the process */
518 struct hrtimer real_timer
;
519 struct pid
*leader_pid
;
520 ktime_t it_real_incr
;
523 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
524 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
525 * values are defined to 0 and 1 respectively
527 struct cpu_itimer it
[2];
530 * Thread group totals for process CPU timers.
531 * See thread_group_cputimer(), et al, for details.
533 struct thread_group_cputimer cputimer
;
535 /* Earliest-expiration cache. */
536 struct task_cputime cputime_expires
;
538 struct list_head cpu_timers
[3];
540 struct pid
*tty_old_pgrp
;
542 /* boolean value for session group leader */
545 struct tty_struct
*tty
; /* NULL if no tty */
547 #ifdef CONFIG_SCHED_AUTOGROUP
548 struct autogroup
*autogroup
;
551 * Cumulative resource counters for dead threads in the group,
552 * and for reaped dead child processes forked by this group.
553 * Live threads maintain their own counters and add to these
554 * in __exit_signal, except for the group leader.
556 cputime_t utime
, stime
, cutime
, cstime
;
559 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
560 struct cputime prev_cputime
;
562 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
563 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
564 unsigned long inblock
, oublock
, cinblock
, coublock
;
565 unsigned long maxrss
, cmaxrss
;
566 struct task_io_accounting ioac
;
569 * Cumulative ns of schedule CPU time fo dead threads in the
570 * group, not including a zombie group leader, (This only differs
571 * from jiffies_to_ns(utime + stime) if sched_clock uses something
572 * other than jiffies.)
574 unsigned long long sum_sched_runtime
;
577 * We don't bother to synchronize most readers of this at all,
578 * because there is no reader checking a limit that actually needs
579 * to get both rlim_cur and rlim_max atomically, and either one
580 * alone is a single word that can safely be read normally.
581 * getrlimit/setrlimit use task_lock(current->group_leader) to
582 * protect this instead of the siglock, because they really
583 * have no need to disable irqs.
585 struct rlimit rlim
[RLIM_NLIMITS
];
587 #ifdef CONFIG_BSD_PROCESS_ACCT
588 struct pacct_struct pacct
; /* per-process accounting information */
590 #ifdef CONFIG_TASKSTATS
591 struct taskstats
*stats
;
595 unsigned audit_tty_log_passwd
;
596 struct tty_audit_buf
*tty_audit_buf
;
598 #ifdef CONFIG_CGROUPS
600 * group_rwsem prevents new tasks from entering the threadgroup and
601 * member tasks from exiting,a more specifically, setting of
602 * PF_EXITING. fork and exit paths are protected with this rwsem
603 * using threadgroup_change_begin/end(). Users which require
604 * threadgroup to remain stable should use threadgroup_[un]lock()
605 * which also takes care of exec path. Currently, cgroup is the
608 struct rw_semaphore group_rwsem
;
611 oom_flags_t oom_flags
;
612 short oom_score_adj
; /* OOM kill score adjustment */
613 short oom_score_adj_min
; /* OOM kill score adjustment min value.
614 * Only settable by CAP_SYS_RESOURCE. */
616 struct mutex cred_guard_mutex
; /* guard against foreign influences on
617 * credential calculations
618 * (notably. ptrace) */
622 * Bits in flags field of signal_struct.
624 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
625 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
626 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
627 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
629 * Pending notifications to parent.
631 #define SIGNAL_CLD_STOPPED 0x00000010
632 #define SIGNAL_CLD_CONTINUED 0x00000020
633 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
635 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
637 /* If true, all threads except ->group_exit_task have pending SIGKILL */
638 static inline int signal_group_exit(const struct signal_struct
*sig
)
640 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
641 (sig
->group_exit_task
!= NULL
);
645 * Some day this will be a full-fledged user tracking system..
648 atomic_t __count
; /* reference count */
649 atomic_t processes
; /* How many processes does this user have? */
650 atomic_t files
; /* How many open files does this user have? */
651 atomic_t sigpending
; /* How many pending signals does this user have? */
652 #ifdef CONFIG_INOTIFY_USER
653 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
654 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
656 #ifdef CONFIG_FANOTIFY
657 atomic_t fanotify_listeners
;
660 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
662 #ifdef CONFIG_POSIX_MQUEUE
663 /* protected by mq_lock */
664 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
666 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
669 struct key
*uid_keyring
; /* UID specific keyring */
670 struct key
*session_keyring
; /* UID's default session keyring */
673 /* Hash table maintenance information */
674 struct hlist_node uidhash_node
;
677 #ifdef CONFIG_PERF_EVENTS
678 atomic_long_t locked_vm
;
682 extern int uids_sysfs_init(void);
684 extern struct user_struct
*find_user(kuid_t
);
686 extern struct user_struct root_user
;
687 #define INIT_USER (&root_user)
690 struct backing_dev_info
;
691 struct reclaim_state
;
693 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
695 /* cumulative counters */
696 unsigned long pcount
; /* # of times run on this cpu */
697 unsigned long long run_delay
; /* time spent waiting on a runqueue */
700 unsigned long long last_arrival
,/* when we last ran on a cpu */
701 last_queued
; /* when we were last queued to run */
703 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
705 #ifdef CONFIG_TASK_DELAY_ACCT
706 struct task_delay_info
{
708 unsigned int flags
; /* Private per-task flags */
710 /* For each stat XXX, add following, aligned appropriately
712 * struct timespec XXX_start, XXX_end;
716 * Atomicity of updates to XXX_delay, XXX_count protected by
717 * single lock above (split into XXX_lock if contention is an issue).
721 * XXX_count is incremented on every XXX operation, the delay
722 * associated with the operation is added to XXX_delay.
723 * XXX_delay contains the accumulated delay time in nanoseconds.
725 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
726 u64 blkio_delay
; /* wait for sync block io completion */
727 u64 swapin_delay
; /* wait for swapin block io completion */
728 u32 blkio_count
; /* total count of the number of sync block */
729 /* io operations performed */
730 u32 swapin_count
; /* total count of the number of swapin block */
731 /* io operations performed */
733 struct timespec freepages_start
, freepages_end
;
734 u64 freepages_delay
; /* wait for memory reclaim */
735 u32 freepages_count
; /* total count of memory reclaim */
737 #endif /* CONFIG_TASK_DELAY_ACCT */
739 static inline int sched_info_on(void)
741 #ifdef CONFIG_SCHEDSTATS
743 #elif defined(CONFIG_TASK_DELAY_ACCT)
744 extern int delayacct_on
;
759 * Increase resolution of cpu_power calculations
761 #define SCHED_POWER_SHIFT 10
762 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
765 * sched-domains (multiprocessor balancing) declarations:
768 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
769 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
770 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
771 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
772 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
773 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
774 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
775 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
776 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
777 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
778 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
779 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
781 extern int __weak
arch_sd_sibiling_asym_packing(void);
783 struct sched_domain_attr
{
784 int relax_domain_level
;
787 #define SD_ATTR_INIT (struct sched_domain_attr) { \
788 .relax_domain_level = -1, \
791 extern int sched_domain_level_max
;
795 struct sched_domain
{
796 /* These fields must be setup */
797 struct sched_domain
*parent
; /* top domain must be null terminated */
798 struct sched_domain
*child
; /* bottom domain must be null terminated */
799 struct sched_group
*groups
; /* the balancing groups of the domain */
800 unsigned long min_interval
; /* Minimum balance interval ms */
801 unsigned long max_interval
; /* Maximum balance interval ms */
802 unsigned int busy_factor
; /* less balancing by factor if busy */
803 unsigned int imbalance_pct
; /* No balance until over watermark */
804 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
805 unsigned int busy_idx
;
806 unsigned int idle_idx
;
807 unsigned int newidle_idx
;
808 unsigned int wake_idx
;
809 unsigned int forkexec_idx
;
810 unsigned int smt_gain
;
812 int nohz_idle
; /* NOHZ IDLE status */
813 int flags
; /* See SD_* */
816 /* Runtime fields. */
817 unsigned long last_balance
; /* init to jiffies. units in jiffies */
818 unsigned int balance_interval
; /* initialise to 1. units in ms. */
819 unsigned int nr_balance_failed
; /* initialise to 0 */
823 /* idle_balance() stats */
824 u64 max_newidle_lb_cost
;
825 unsigned long next_decay_max_lb_cost
;
827 #ifdef CONFIG_SCHEDSTATS
828 /* load_balance() stats */
829 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
830 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
831 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
832 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
833 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
834 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
835 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
836 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
838 /* Active load balancing */
839 unsigned int alb_count
;
840 unsigned int alb_failed
;
841 unsigned int alb_pushed
;
843 /* SD_BALANCE_EXEC stats */
844 unsigned int sbe_count
;
845 unsigned int sbe_balanced
;
846 unsigned int sbe_pushed
;
848 /* SD_BALANCE_FORK stats */
849 unsigned int sbf_count
;
850 unsigned int sbf_balanced
;
851 unsigned int sbf_pushed
;
853 /* try_to_wake_up() stats */
854 unsigned int ttwu_wake_remote
;
855 unsigned int ttwu_move_affine
;
856 unsigned int ttwu_move_balance
;
858 #ifdef CONFIG_SCHED_DEBUG
862 void *private; /* used during construction */
863 struct rcu_head rcu
; /* used during destruction */
866 unsigned int span_weight
;
868 * Span of all CPUs in this domain.
870 * NOTE: this field is variable length. (Allocated dynamically
871 * by attaching extra space to the end of the structure,
872 * depending on how many CPUs the kernel has booted up with)
874 unsigned long span
[0];
877 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
879 return to_cpumask(sd
->span
);
882 extern void partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
883 struct sched_domain_attr
*dattr_new
);
885 /* Allocate an array of sched domains, for partition_sched_domains(). */
886 cpumask_var_t
*alloc_sched_domains(unsigned int ndoms
);
887 void free_sched_domains(cpumask_var_t doms
[], unsigned int ndoms
);
889 bool cpus_share_cache(int this_cpu
, int that_cpu
);
891 #else /* CONFIG_SMP */
893 struct sched_domain_attr
;
896 partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
897 struct sched_domain_attr
*dattr_new
)
901 static inline bool cpus_share_cache(int this_cpu
, int that_cpu
)
906 #endif /* !CONFIG_SMP */
909 struct io_context
; /* See blkdev.h */
912 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
913 extern void prefetch_stack(struct task_struct
*t
);
915 static inline void prefetch_stack(struct task_struct
*t
) { }
918 struct audit_context
; /* See audit.c */
920 struct pipe_inode_info
;
921 struct uts_namespace
;
924 unsigned long weight
, inv_weight
;
929 * These sums represent an infinite geometric series and so are bound
930 * above by 1024/(1-y). Thus we only need a u32 to store them for all
931 * choices of y < 1-2^(-32)*1024.
933 u32 runnable_avg_sum
, runnable_avg_period
;
934 u64 last_runnable_update
;
936 unsigned long load_avg_contrib
;
939 #ifdef CONFIG_SCHEDSTATS
940 struct sched_statistics
{
950 s64 sum_sleep_runtime
;
957 u64 nr_migrations_cold
;
958 u64 nr_failed_migrations_affine
;
959 u64 nr_failed_migrations_running
;
960 u64 nr_failed_migrations_hot
;
961 u64 nr_forced_migrations
;
965 u64 nr_wakeups_migrate
;
966 u64 nr_wakeups_local
;
967 u64 nr_wakeups_remote
;
968 u64 nr_wakeups_affine
;
969 u64 nr_wakeups_affine_attempts
;
970 u64 nr_wakeups_passive
;
975 struct sched_entity
{
976 struct load_weight load
; /* for load-balancing */
977 struct rb_node run_node
;
978 struct list_head group_node
;
982 u64 sum_exec_runtime
;
984 u64 prev_sum_exec_runtime
;
988 #ifdef CONFIG_SCHEDSTATS
989 struct sched_statistics statistics
;
992 #ifdef CONFIG_FAIR_GROUP_SCHED
993 struct sched_entity
*parent
;
994 /* rq on which this entity is (to be) queued: */
995 struct cfs_rq
*cfs_rq
;
996 /* rq "owned" by this entity/group: */
1001 /* Per-entity load-tracking */
1002 struct sched_avg avg
;
1006 struct sched_rt_entity
{
1007 struct list_head run_list
;
1008 unsigned long timeout
;
1009 unsigned long watchdog_stamp
;
1010 unsigned int time_slice
;
1012 struct sched_rt_entity
*back
;
1013 #ifdef CONFIG_RT_GROUP_SCHED
1014 struct sched_rt_entity
*parent
;
1015 /* rq on which this entity is (to be) queued: */
1016 struct rt_rq
*rt_rq
;
1017 /* rq "owned" by this entity/group: */
1025 enum perf_event_task_context
{
1026 perf_invalid_context
= -1,
1027 perf_hw_context
= 0,
1029 perf_nr_task_contexts
,
1032 struct task_struct
{
1033 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1036 unsigned int flags
; /* per process flags, defined below */
1037 unsigned int ptrace
;
1040 struct llist_node wake_entry
;
1042 struct task_struct
*last_wakee
;
1043 unsigned long wakee_flips
;
1044 unsigned long wakee_flip_decay_ts
;
1048 int prio
, static_prio
, normal_prio
;
1049 unsigned int rt_priority
;
1050 const struct sched_class
*sched_class
;
1051 struct sched_entity se
;
1052 struct sched_rt_entity rt
;
1053 #ifdef CONFIG_CGROUP_SCHED
1054 struct task_group
*sched_task_group
;
1057 #ifdef CONFIG_PREEMPT_NOTIFIERS
1058 /* list of struct preempt_notifier: */
1059 struct hlist_head preempt_notifiers
;
1063 * fpu_counter contains the number of consecutive context switches
1064 * that the FPU is used. If this is over a threshold, the lazy fpu
1065 * saving becomes unlazy to save the trap. This is an unsigned char
1066 * so that after 256 times the counter wraps and the behavior turns
1067 * lazy again; this to deal with bursty apps that only use FPU for
1070 unsigned char fpu_counter
;
1071 #ifdef CONFIG_BLK_DEV_IO_TRACE
1072 unsigned int btrace_seq
;
1075 unsigned int policy
;
1076 int nr_cpus_allowed
;
1077 cpumask_t cpus_allowed
;
1079 #ifdef CONFIG_PREEMPT_RCU
1080 int rcu_read_lock_nesting
;
1081 char rcu_read_unlock_special
;
1082 struct list_head rcu_node_entry
;
1083 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1084 #ifdef CONFIG_TREE_PREEMPT_RCU
1085 struct rcu_node
*rcu_blocked_node
;
1086 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1087 #ifdef CONFIG_RCU_BOOST
1088 struct rt_mutex
*rcu_boost_mutex
;
1089 #endif /* #ifdef CONFIG_RCU_BOOST */
1091 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1092 struct sched_info sched_info
;
1095 struct list_head tasks
;
1097 struct plist_node pushable_tasks
;
1100 struct mm_struct
*mm
, *active_mm
;
1101 #ifdef CONFIG_COMPAT_BRK
1102 unsigned brk_randomized
:1;
1104 #if defined(SPLIT_RSS_COUNTING)
1105 struct task_rss_stat rss_stat
;
1109 int exit_code
, exit_signal
;
1110 int pdeath_signal
; /* The signal sent when the parent dies */
1111 unsigned int jobctl
; /* JOBCTL_*, siglock protected */
1113 /* Used for emulating ABI behavior of previous Linux versions */
1114 unsigned int personality
;
1116 unsigned did_exec
:1;
1117 unsigned in_execve
:1; /* Tell the LSMs that the process is doing an
1119 unsigned in_iowait
:1;
1121 /* task may not gain privileges */
1122 unsigned no_new_privs
:1;
1124 /* Revert to default priority/policy when forking */
1125 unsigned sched_reset_on_fork
:1;
1126 unsigned sched_contributes_to_load
:1;
1131 #ifdef CONFIG_CC_STACKPROTECTOR
1132 /* Canary value for the -fstack-protector gcc feature */
1133 unsigned long stack_canary
;
1136 * pointers to (original) parent process, youngest child, younger sibling,
1137 * older sibling, respectively. (p->father can be replaced with
1138 * p->real_parent->pid)
1140 struct task_struct __rcu
*real_parent
; /* real parent process */
1141 struct task_struct __rcu
*parent
; /* recipient of SIGCHLD, wait4() reports */
1143 * children/sibling forms the list of my natural children
1145 struct list_head children
; /* list of my children */
1146 struct list_head sibling
; /* linkage in my parent's children list */
1147 struct task_struct
*group_leader
; /* threadgroup leader */
1150 * ptraced is the list of tasks this task is using ptrace on.
1151 * This includes both natural children and PTRACE_ATTACH targets.
1152 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1154 struct list_head ptraced
;
1155 struct list_head ptrace_entry
;
1157 /* PID/PID hash table linkage. */
1158 struct pid_link pids
[PIDTYPE_MAX
];
1159 struct list_head thread_group
;
1161 struct completion
*vfork_done
; /* for vfork() */
1162 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1163 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1165 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1167 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1168 struct cputime prev_cputime
;
1170 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1171 seqlock_t vtime_seqlock
;
1172 unsigned long long vtime_snap
;
1177 } vtime_snap_whence
;
1179 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1180 struct timespec start_time
; /* monotonic time */
1181 struct timespec real_start_time
; /* boot based time */
1182 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1183 unsigned long min_flt
, maj_flt
;
1185 struct task_cputime cputime_expires
;
1186 struct list_head cpu_timers
[3];
1188 /* process credentials */
1189 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1190 * credentials (COW) */
1191 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1192 * credentials (COW) */
1193 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1194 - access with [gs]et_task_comm (which lock
1195 it with task_lock())
1196 - initialized normally by setup_new_exec */
1197 /* file system info */
1198 int link_count
, total_link_count
;
1199 #ifdef CONFIG_SYSVIPC
1201 struct sysv_sem sysvsem
;
1203 #ifdef CONFIG_DETECT_HUNG_TASK
1204 /* hung task detection */
1205 unsigned long last_switch_count
;
1207 /* CPU-specific state of this task */
1208 struct thread_struct thread
;
1209 /* filesystem information */
1210 struct fs_struct
*fs
;
1211 /* open file information */
1212 struct files_struct
*files
;
1214 struct nsproxy
*nsproxy
;
1215 /* signal handlers */
1216 struct signal_struct
*signal
;
1217 struct sighand_struct
*sighand
;
1219 sigset_t blocked
, real_blocked
;
1220 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1221 struct sigpending pending
;
1223 unsigned long sas_ss_sp
;
1225 int (*notifier
)(void *priv
);
1226 void *notifier_data
;
1227 sigset_t
*notifier_mask
;
1228 struct callback_head
*task_works
;
1230 struct audit_context
*audit_context
;
1231 #ifdef CONFIG_AUDITSYSCALL
1233 unsigned int sessionid
;
1235 struct seccomp seccomp
;
1237 /* Thread group tracking */
1240 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1242 spinlock_t alloc_lock
;
1244 /* Protection of the PI data structures: */
1245 raw_spinlock_t pi_lock
;
1247 #ifdef CONFIG_RT_MUTEXES
1248 /* PI waiters blocked on a rt_mutex held by this task */
1249 struct plist_head pi_waiters
;
1250 /* Deadlock detection and priority inheritance handling */
1251 struct rt_mutex_waiter
*pi_blocked_on
;
1254 #ifdef CONFIG_DEBUG_MUTEXES
1255 /* mutex deadlock detection */
1256 struct mutex_waiter
*blocked_on
;
1258 #ifdef CONFIG_TRACE_IRQFLAGS
1259 unsigned int irq_events
;
1260 unsigned long hardirq_enable_ip
;
1261 unsigned long hardirq_disable_ip
;
1262 unsigned int hardirq_enable_event
;
1263 unsigned int hardirq_disable_event
;
1264 int hardirqs_enabled
;
1265 int hardirq_context
;
1266 unsigned long softirq_disable_ip
;
1267 unsigned long softirq_enable_ip
;
1268 unsigned int softirq_disable_event
;
1269 unsigned int softirq_enable_event
;
1270 int softirqs_enabled
;
1271 int softirq_context
;
1273 #ifdef CONFIG_LOCKDEP
1274 # define MAX_LOCK_DEPTH 48UL
1277 unsigned int lockdep_recursion
;
1278 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1279 gfp_t lockdep_reclaim_gfp
;
1282 /* journalling filesystem info */
1285 /* stacked block device info */
1286 struct bio_list
*bio_list
;
1289 /* stack plugging */
1290 struct blk_plug
*plug
;
1294 struct reclaim_state
*reclaim_state
;
1296 struct backing_dev_info
*backing_dev_info
;
1298 struct io_context
*io_context
;
1300 unsigned long ptrace_message
;
1301 siginfo_t
*last_siginfo
; /* For ptrace use. */
1302 struct task_io_accounting ioac
;
1303 #if defined(CONFIG_TASK_XACCT)
1304 u64 acct_rss_mem1
; /* accumulated rss usage */
1305 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1306 cputime_t acct_timexpd
; /* stime + utime since last update */
1308 #ifdef CONFIG_CPUSETS
1309 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1310 seqcount_t mems_allowed_seq
; /* Seqence no to catch updates */
1311 int cpuset_mem_spread_rotor
;
1312 int cpuset_slab_spread_rotor
;
1314 #ifdef CONFIG_CGROUPS
1315 /* Control Group info protected by css_set_lock */
1316 struct css_set __rcu
*cgroups
;
1317 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1318 struct list_head cg_list
;
1321 struct robust_list_head __user
*robust_list
;
1322 #ifdef CONFIG_COMPAT
1323 struct compat_robust_list_head __user
*compat_robust_list
;
1325 struct list_head pi_state_list
;
1326 struct futex_pi_state
*pi_state_cache
;
1328 #ifdef CONFIG_PERF_EVENTS
1329 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1330 struct mutex perf_event_mutex
;
1331 struct list_head perf_event_list
;
1334 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1336 short pref_node_fork
;
1338 #ifdef CONFIG_NUMA_BALANCING
1340 int numa_migrate_seq
;
1341 unsigned int numa_scan_period
;
1342 unsigned int numa_scan_period_max
;
1343 u64 node_stamp
; /* migration stamp */
1344 struct callback_head numa_work
;
1347 * Exponential decaying average of faults on a per-node basis.
1348 * Scheduling placement decisions are made based on the these counts.
1349 * The values remain static for the duration of a PTE scan
1351 unsigned long *numa_faults
;
1354 * numa_faults_buffer records faults per node during the current
1355 * scan window. When the scan completes, the counts in numa_faults
1356 * decay and these values are copied.
1358 unsigned long *numa_faults_buffer
;
1360 int numa_preferred_nid
;
1361 #endif /* CONFIG_NUMA_BALANCING */
1363 struct rcu_head rcu
;
1366 * cache last used pipe for splice
1368 struct pipe_inode_info
*splice_pipe
;
1370 struct page_frag task_frag
;
1372 #ifdef CONFIG_TASK_DELAY_ACCT
1373 struct task_delay_info
*delays
;
1375 #ifdef CONFIG_FAULT_INJECTION
1379 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1380 * balance_dirty_pages() for some dirty throttling pause
1383 int nr_dirtied_pause
;
1384 unsigned long dirty_paused_when
; /* start of a write-and-pause period */
1386 #ifdef CONFIG_LATENCYTOP
1387 int latency_record_count
;
1388 struct latency_record latency_record
[LT_SAVECOUNT
];
1391 * time slack values; these are used to round up poll() and
1392 * select() etc timeout values. These are in nanoseconds.
1394 unsigned long timer_slack_ns
;
1395 unsigned long default_timer_slack_ns
;
1397 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1398 /* Index of current stored address in ret_stack */
1400 /* Stack of return addresses for return function tracing */
1401 struct ftrace_ret_stack
*ret_stack
;
1402 /* time stamp for last schedule */
1403 unsigned long long ftrace_timestamp
;
1405 * Number of functions that haven't been traced
1406 * because of depth overrun.
1408 atomic_t trace_overrun
;
1409 /* Pause for the tracing */
1410 atomic_t tracing_graph_pause
;
1412 #ifdef CONFIG_TRACING
1413 /* state flags for use by tracers */
1414 unsigned long trace
;
1415 /* bitmask and counter of trace recursion */
1416 unsigned long trace_recursion
;
1417 #endif /* CONFIG_TRACING */
1418 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1419 struct memcg_batch_info
{
1420 int do_batch
; /* incremented when batch uncharge started */
1421 struct mem_cgroup
*memcg
; /* target memcg of uncharge */
1422 unsigned long nr_pages
; /* uncharged usage */
1423 unsigned long memsw_nr_pages
; /* uncharged mem+swap usage */
1425 unsigned int memcg_kmem_skip_account
;
1426 struct memcg_oom_info
{
1427 unsigned int may_oom
:1;
1428 unsigned int in_memcg_oom
:1;
1429 unsigned int oom_locked
:1;
1431 struct mem_cgroup
*wait_on_memcg
;
1434 #ifdef CONFIG_UPROBES
1435 struct uprobe_task
*utask
;
1437 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1438 unsigned int sequential_io
;
1439 unsigned int sequential_io_avg
;
1443 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1444 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1446 #ifdef CONFIG_NUMA_BALANCING
1447 extern void task_numa_fault(int node
, int pages
, bool migrated
);
1448 extern void set_numabalancing_state(bool enabled
);
1450 static inline void task_numa_fault(int node
, int pages
, bool migrated
)
1453 static inline void set_numabalancing_state(bool enabled
)
1458 static inline struct pid
*task_pid(struct task_struct
*task
)
1460 return task
->pids
[PIDTYPE_PID
].pid
;
1463 static inline struct pid
*task_tgid(struct task_struct
*task
)
1465 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1469 * Without tasklist or rcu lock it is not safe to dereference
1470 * the result of task_pgrp/task_session even if task == current,
1471 * we can race with another thread doing sys_setsid/sys_setpgid.
1473 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1475 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1478 static inline struct pid
*task_session(struct task_struct
*task
)
1480 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1483 struct pid_namespace
;
1486 * the helpers to get the task's different pids as they are seen
1487 * from various namespaces
1489 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1490 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1492 * task_xid_nr_ns() : id seen from the ns specified;
1494 * set_task_vxid() : assigns a virtual id to a task;
1496 * see also pid_nr() etc in include/linux/pid.h
1498 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
1499 struct pid_namespace
*ns
);
1501 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1506 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
1507 struct pid_namespace
*ns
)
1509 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1512 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1514 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1518 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1523 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1525 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1527 return pid_vnr(task_tgid(tsk
));
1531 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
1532 struct pid_namespace
*ns
)
1534 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1537 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1539 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1543 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
1544 struct pid_namespace
*ns
)
1546 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1549 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1551 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1554 /* obsolete, do not use */
1555 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1557 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1561 * pid_alive - check that a task structure is not stale
1562 * @p: Task structure to be checked.
1564 * Test if a process is not yet dead (at most zombie state)
1565 * If pid_alive fails, then pointers within the task structure
1566 * can be stale and must not be dereferenced.
1568 * Return: 1 if the process is alive. 0 otherwise.
1570 static inline int pid_alive(struct task_struct
*p
)
1572 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1576 * is_global_init - check if a task structure is init
1577 * @tsk: Task structure to be checked.
1579 * Check if a task structure is the first user space task the kernel created.
1581 * Return: 1 if the task structure is init. 0 otherwise.
1583 static inline int is_global_init(struct task_struct
*tsk
)
1585 return tsk
->pid
== 1;
1588 extern struct pid
*cad_pid
;
1590 extern void free_task(struct task_struct
*tsk
);
1591 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1593 extern void __put_task_struct(struct task_struct
*t
);
1595 static inline void put_task_struct(struct task_struct
*t
)
1597 if (atomic_dec_and_test(&t
->usage
))
1598 __put_task_struct(t
);
1601 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1602 extern void task_cputime(struct task_struct
*t
,
1603 cputime_t
*utime
, cputime_t
*stime
);
1604 extern void task_cputime_scaled(struct task_struct
*t
,
1605 cputime_t
*utimescaled
, cputime_t
*stimescaled
);
1606 extern cputime_t
task_gtime(struct task_struct
*t
);
1608 static inline void task_cputime(struct task_struct
*t
,
1609 cputime_t
*utime
, cputime_t
*stime
)
1617 static inline void task_cputime_scaled(struct task_struct
*t
,
1618 cputime_t
*utimescaled
,
1619 cputime_t
*stimescaled
)
1622 *utimescaled
= t
->utimescaled
;
1624 *stimescaled
= t
->stimescaled
;
1627 static inline cputime_t
task_gtime(struct task_struct
*t
)
1632 extern void task_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
1633 extern void thread_group_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
1638 #define PF_EXITING 0x00000004 /* getting shut down */
1639 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1640 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1641 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1642 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1643 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1644 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1645 #define PF_DUMPCORE 0x00000200 /* dumped core */
1646 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1647 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1648 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1649 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1650 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1651 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1652 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1653 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1654 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1655 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1656 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1657 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1658 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1659 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1660 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1661 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1662 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1663 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1664 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1665 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1666 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1667 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1670 * Only the _current_ task can read/write to tsk->flags, but other
1671 * tasks can access tsk->flags in readonly mode for example
1672 * with tsk_used_math (like during threaded core dumping).
1673 * There is however an exception to this rule during ptrace
1674 * or during fork: the ptracer task is allowed to write to the
1675 * child->flags of its traced child (same goes for fork, the parent
1676 * can write to the child->flags), because we're guaranteed the
1677 * child is not running and in turn not changing child->flags
1678 * at the same time the parent does it.
1680 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1681 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1682 #define clear_used_math() clear_stopped_child_used_math(current)
1683 #define set_used_math() set_stopped_child_used_math(current)
1684 #define conditional_stopped_child_used_math(condition, child) \
1685 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1686 #define conditional_used_math(condition) \
1687 conditional_stopped_child_used_math(condition, current)
1688 #define copy_to_stopped_child_used_math(child) \
1689 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1690 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1691 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1692 #define used_math() tsk_used_math(current)
1694 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1695 static inline gfp_t
memalloc_noio_flags(gfp_t flags
)
1697 if (unlikely(current
->flags
& PF_MEMALLOC_NOIO
))
1702 static inline unsigned int memalloc_noio_save(void)
1704 unsigned int flags
= current
->flags
& PF_MEMALLOC_NOIO
;
1705 current
->flags
|= PF_MEMALLOC_NOIO
;
1709 static inline void memalloc_noio_restore(unsigned int flags
)
1711 current
->flags
= (current
->flags
& ~PF_MEMALLOC_NOIO
) | flags
;
1715 * task->jobctl flags
1717 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1719 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1720 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1721 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1722 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1723 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1724 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1725 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1727 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1728 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1729 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1730 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1731 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1732 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1733 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1735 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1736 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1738 extern bool task_set_jobctl_pending(struct task_struct
*task
,
1740 extern void task_clear_jobctl_trapping(struct task_struct
*task
);
1741 extern void task_clear_jobctl_pending(struct task_struct
*task
,
1744 #ifdef CONFIG_PREEMPT_RCU
1746 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1747 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1749 static inline void rcu_copy_process(struct task_struct
*p
)
1751 p
->rcu_read_lock_nesting
= 0;
1752 p
->rcu_read_unlock_special
= 0;
1753 #ifdef CONFIG_TREE_PREEMPT_RCU
1754 p
->rcu_blocked_node
= NULL
;
1755 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1756 #ifdef CONFIG_RCU_BOOST
1757 p
->rcu_boost_mutex
= NULL
;
1758 #endif /* #ifdef CONFIG_RCU_BOOST */
1759 INIT_LIST_HEAD(&p
->rcu_node_entry
);
1764 static inline void rcu_copy_process(struct task_struct
*p
)
1770 static inline void tsk_restore_flags(struct task_struct
*task
,
1771 unsigned long orig_flags
, unsigned long flags
)
1773 task
->flags
&= ~flags
;
1774 task
->flags
|= orig_flags
& flags
;
1778 extern void do_set_cpus_allowed(struct task_struct
*p
,
1779 const struct cpumask
*new_mask
);
1781 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
1782 const struct cpumask
*new_mask
);
1784 static inline void do_set_cpus_allowed(struct task_struct
*p
,
1785 const struct cpumask
*new_mask
)
1788 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
1789 const struct cpumask
*new_mask
)
1791 if (!cpumask_test_cpu(0, new_mask
))
1797 #ifdef CONFIG_NO_HZ_COMMON
1798 void calc_load_enter_idle(void);
1799 void calc_load_exit_idle(void);
1801 static inline void calc_load_enter_idle(void) { }
1802 static inline void calc_load_exit_idle(void) { }
1803 #endif /* CONFIG_NO_HZ_COMMON */
1805 #ifndef CONFIG_CPUMASK_OFFSTACK
1806 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1808 return set_cpus_allowed_ptr(p
, &new_mask
);
1813 * Do not use outside of architecture code which knows its limitations.
1815 * sched_clock() has no promise of monotonicity or bounded drift between
1816 * CPUs, use (which you should not) requires disabling IRQs.
1818 * Please use one of the three interfaces below.
1820 extern unsigned long long notrace
sched_clock(void);
1822 * See the comment in kernel/sched/clock.c
1824 extern u64
cpu_clock(int cpu
);
1825 extern u64
local_clock(void);
1826 extern u64
sched_clock_cpu(int cpu
);
1829 extern void sched_clock_init(void);
1831 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1832 static inline void sched_clock_tick(void)
1836 static inline void sched_clock_idle_sleep_event(void)
1840 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
1845 * Architectures can set this to 1 if they have specified
1846 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1847 * but then during bootup it turns out that sched_clock()
1848 * is reliable after all:
1850 extern int sched_clock_stable
;
1852 extern void sched_clock_tick(void);
1853 extern void sched_clock_idle_sleep_event(void);
1854 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1857 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1859 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1860 * The reason for this explicit opt-in is not to have perf penalty with
1861 * slow sched_clocks.
1863 extern void enable_sched_clock_irqtime(void);
1864 extern void disable_sched_clock_irqtime(void);
1866 static inline void enable_sched_clock_irqtime(void) {}
1867 static inline void disable_sched_clock_irqtime(void) {}
1870 extern unsigned long long
1871 task_sched_runtime(struct task_struct
*task
);
1873 /* sched_exec is called by processes performing an exec */
1875 extern void sched_exec(void);
1877 #define sched_exec() {}
1880 extern void sched_clock_idle_sleep_event(void);
1881 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1883 #ifdef CONFIG_HOTPLUG_CPU
1884 extern void idle_task_exit(void);
1886 static inline void idle_task_exit(void) {}
1889 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1890 extern void wake_up_nohz_cpu(int cpu
);
1892 static inline void wake_up_nohz_cpu(int cpu
) { }
1895 #ifdef CONFIG_NO_HZ_FULL
1896 extern bool sched_can_stop_tick(void);
1897 extern u64
scheduler_tick_max_deferment(void);
1899 static inline bool sched_can_stop_tick(void) { return false; }
1902 #ifdef CONFIG_SCHED_AUTOGROUP
1903 extern void sched_autogroup_create_attach(struct task_struct
*p
);
1904 extern void sched_autogroup_detach(struct task_struct
*p
);
1905 extern void sched_autogroup_fork(struct signal_struct
*sig
);
1906 extern void sched_autogroup_exit(struct signal_struct
*sig
);
1907 #ifdef CONFIG_PROC_FS
1908 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
1909 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int nice
);
1912 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
1913 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
1914 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
1915 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
1918 extern bool yield_to(struct task_struct
*p
, bool preempt
);
1919 extern void set_user_nice(struct task_struct
*p
, long nice
);
1920 extern int task_prio(const struct task_struct
*p
);
1921 extern int task_nice(const struct task_struct
*p
);
1922 extern int can_nice(const struct task_struct
*p
, const int nice
);
1923 extern int task_curr(const struct task_struct
*p
);
1924 extern int idle_cpu(int cpu
);
1925 extern int sched_setscheduler(struct task_struct
*, int,
1926 const struct sched_param
*);
1927 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
1928 const struct sched_param
*);
1929 extern struct task_struct
*idle_task(int cpu
);
1931 * is_idle_task - is the specified task an idle task?
1932 * @p: the task in question.
1934 * Return: 1 if @p is an idle task. 0 otherwise.
1936 static inline bool is_idle_task(const struct task_struct
*p
)
1940 extern struct task_struct
*curr_task(int cpu
);
1941 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1946 * The default (Linux) execution domain.
1948 extern struct exec_domain default_exec_domain
;
1950 union thread_union
{
1951 struct thread_info thread_info
;
1952 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1955 #ifndef __HAVE_ARCH_KSTACK_END
1956 static inline int kstack_end(void *addr
)
1958 /* Reliable end of stack detection:
1959 * Some APM bios versions misalign the stack
1961 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1965 extern union thread_union init_thread_union
;
1966 extern struct task_struct init_task
;
1968 extern struct mm_struct init_mm
;
1970 extern struct pid_namespace init_pid_ns
;
1973 * find a task by one of its numerical ids
1975 * find_task_by_pid_ns():
1976 * finds a task by its pid in the specified namespace
1977 * find_task_by_vpid():
1978 * finds a task by its virtual pid
1980 * see also find_vpid() etc in include/linux/pid.h
1983 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1984 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
1985 struct pid_namespace
*ns
);
1987 /* per-UID process charging. */
1988 extern struct user_struct
* alloc_uid(kuid_t
);
1989 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1991 atomic_inc(&u
->__count
);
1994 extern void free_uid(struct user_struct
*);
1996 #include <asm/current.h>
1998 extern void xtime_update(unsigned long ticks
);
2000 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2001 extern int wake_up_process(struct task_struct
*tsk
);
2002 extern void wake_up_new_task(struct task_struct
*tsk
);
2004 extern void kick_process(struct task_struct
*tsk
);
2006 static inline void kick_process(struct task_struct
*tsk
) { }
2008 extern void sched_fork(struct task_struct
*p
);
2009 extern void sched_dead(struct task_struct
*p
);
2011 extern void proc_caches_init(void);
2012 extern void flush_signals(struct task_struct
*);
2013 extern void __flush_signals(struct task_struct
*);
2014 extern void ignore_signals(struct task_struct
*);
2015 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2016 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2018 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
2020 unsigned long flags
;
2023 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
2024 ret
= dequeue_signal(tsk
, mask
, info
);
2025 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
2030 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
2032 extern void unblock_all_signals(void);
2033 extern void release_task(struct task_struct
* p
);
2034 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2035 extern int force_sigsegv(int, struct task_struct
*);
2036 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2037 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2038 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2039 extern int kill_pid_info_as_cred(int, struct siginfo
*, struct pid
*,
2040 const struct cred
*, u32
);
2041 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2042 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2043 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2044 extern __must_check
bool do_notify_parent(struct task_struct
*, int);
2045 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2046 extern void force_sig(int, struct task_struct
*);
2047 extern int send_sig(int, struct task_struct
*, int);
2048 extern int zap_other_threads(struct task_struct
*p
);
2049 extern struct sigqueue
*sigqueue_alloc(void);
2050 extern void sigqueue_free(struct sigqueue
*);
2051 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2052 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2054 static inline void restore_saved_sigmask(void)
2056 if (test_and_clear_restore_sigmask())
2057 __set_current_blocked(¤t
->saved_sigmask
);
2060 static inline sigset_t
*sigmask_to_save(void)
2062 sigset_t
*res
= ¤t
->blocked
;
2063 if (unlikely(test_restore_sigmask()))
2064 res
= ¤t
->saved_sigmask
;
2068 static inline int kill_cad_pid(int sig
, int priv
)
2070 return kill_pid(cad_pid
, sig
, priv
);
2073 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2074 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2075 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2076 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2079 * True if we are on the alternate signal stack.
2081 static inline int on_sig_stack(unsigned long sp
)
2083 #ifdef CONFIG_STACK_GROWSUP
2084 return sp
>= current
->sas_ss_sp
&&
2085 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2087 return sp
> current
->sas_ss_sp
&&
2088 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2092 static inline int sas_ss_flags(unsigned long sp
)
2094 return (current
->sas_ss_size
== 0 ? SS_DISABLE
2095 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
2098 static inline unsigned long sigsp(unsigned long sp
, struct ksignal
*ksig
)
2100 if (unlikely((ksig
->ka
.sa
.sa_flags
& SA_ONSTACK
)) && ! sas_ss_flags(sp
))
2101 #ifdef CONFIG_STACK_GROWSUP
2102 return current
->sas_ss_sp
;
2104 return current
->sas_ss_sp
+ current
->sas_ss_size
;
2110 * Routines for handling mm_structs
2112 extern struct mm_struct
* mm_alloc(void);
2114 /* mmdrop drops the mm and the page tables */
2115 extern void __mmdrop(struct mm_struct
*);
2116 static inline void mmdrop(struct mm_struct
* mm
)
2118 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2122 /* mmput gets rid of the mappings and all user-space */
2123 extern void mmput(struct mm_struct
*);
2124 /* Grab a reference to a task's mm, if it is not already going away */
2125 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2127 * Grab a reference to a task's mm, if it is not already going away
2128 * and ptrace_may_access with the mode parameter passed to it
2131 extern struct mm_struct
*mm_access(struct task_struct
*task
, unsigned int mode
);
2132 /* Remove the current tasks stale references to the old mm_struct */
2133 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2134 /* Allocate a new mm structure and copy contents from tsk->mm */
2135 extern struct mm_struct
*dup_mm(struct task_struct
*tsk
);
2137 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2138 struct task_struct
*);
2139 extern void flush_thread(void);
2140 extern void exit_thread(void);
2142 extern void exit_files(struct task_struct
*);
2143 extern void __cleanup_sighand(struct sighand_struct
*);
2145 extern void exit_itimers(struct signal_struct
*);
2146 extern void flush_itimer_signals(void);
2148 extern void do_group_exit(int);
2150 extern int allow_signal(int);
2151 extern int disallow_signal(int);
2153 extern int do_execve(const char *,
2154 const char __user
* const __user
*,
2155 const char __user
* const __user
*);
2156 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*);
2157 struct task_struct
*fork_idle(int);
2158 extern pid_t
kernel_thread(int (*fn
)(void *), void *arg
, unsigned long flags
);
2160 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
2161 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2164 void scheduler_ipi(void);
2165 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2167 static inline void scheduler_ipi(void) { }
2168 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2175 #define next_task(p) \
2176 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2178 #define for_each_process(p) \
2179 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2181 extern bool current_is_single_threaded(void);
2184 * Careful: do_each_thread/while_each_thread is a double loop so
2185 * 'break' will not work as expected - use goto instead.
2187 #define do_each_thread(g, t) \
2188 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2190 #define while_each_thread(g, t) \
2191 while ((t = next_thread(t)) != g)
2193 static inline int get_nr_threads(struct task_struct
*tsk
)
2195 return tsk
->signal
->nr_threads
;
2198 static inline bool thread_group_leader(struct task_struct
*p
)
2200 return p
->exit_signal
>= 0;
2203 /* Do to the insanities of de_thread it is possible for a process
2204 * to have the pid of the thread group leader without actually being
2205 * the thread group leader. For iteration through the pids in proc
2206 * all we care about is that we have a task with the appropriate
2207 * pid, we don't actually care if we have the right task.
2209 static inline bool has_group_leader_pid(struct task_struct
*p
)
2211 return task_pid(p
) == p
->signal
->leader_pid
;
2215 bool same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2217 return p1
->signal
== p2
->signal
;
2220 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2222 return list_entry_rcu(p
->thread_group
.next
,
2223 struct task_struct
, thread_group
);
2226 static inline int thread_group_empty(struct task_struct
*p
)
2228 return list_empty(&p
->thread_group
);
2231 #define delay_group_leader(p) \
2232 (thread_group_leader(p) && !thread_group_empty(p))
2235 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2236 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2237 * pins the final release of task.io_context. Also protects ->cpuset and
2238 * ->cgroup.subsys[]. And ->vfork_done.
2240 * Nests both inside and outside of read_lock(&tasklist_lock).
2241 * It must not be nested with write_lock_irq(&tasklist_lock),
2242 * neither inside nor outside.
2244 static inline void task_lock(struct task_struct
*p
)
2246 spin_lock(&p
->alloc_lock
);
2249 static inline void task_unlock(struct task_struct
*p
)
2251 spin_unlock(&p
->alloc_lock
);
2254 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
2255 unsigned long *flags
);
2257 static inline struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
2258 unsigned long *flags
)
2260 struct sighand_struct
*ret
;
2262 ret
= __lock_task_sighand(tsk
, flags
);
2263 (void)__cond_lock(&tsk
->sighand
->siglock
, ret
);
2267 static inline void unlock_task_sighand(struct task_struct
*tsk
,
2268 unsigned long *flags
)
2270 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2273 #ifdef CONFIG_CGROUPS
2274 static inline void threadgroup_change_begin(struct task_struct
*tsk
)
2276 down_read(&tsk
->signal
->group_rwsem
);
2278 static inline void threadgroup_change_end(struct task_struct
*tsk
)
2280 up_read(&tsk
->signal
->group_rwsem
);
2284 * threadgroup_lock - lock threadgroup
2285 * @tsk: member task of the threadgroup to lock
2287 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2288 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2289 * change ->group_leader/pid. This is useful for cases where the threadgroup
2290 * needs to stay stable across blockable operations.
2292 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2293 * synchronization. While held, no new task will be added to threadgroup
2294 * and no existing live task will have its PF_EXITING set.
2296 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2297 * sub-thread becomes a new leader.
2299 static inline void threadgroup_lock(struct task_struct
*tsk
)
2301 down_write(&tsk
->signal
->group_rwsem
);
2305 * threadgroup_unlock - unlock threadgroup
2306 * @tsk: member task of the threadgroup to unlock
2308 * Reverse threadgroup_lock().
2310 static inline void threadgroup_unlock(struct task_struct
*tsk
)
2312 up_write(&tsk
->signal
->group_rwsem
);
2315 static inline void threadgroup_change_begin(struct task_struct
*tsk
) {}
2316 static inline void threadgroup_change_end(struct task_struct
*tsk
) {}
2317 static inline void threadgroup_lock(struct task_struct
*tsk
) {}
2318 static inline void threadgroup_unlock(struct task_struct
*tsk
) {}
2321 #ifndef __HAVE_THREAD_FUNCTIONS
2323 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2324 #define task_stack_page(task) ((task)->stack)
2326 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2328 *task_thread_info(p
) = *task_thread_info(org
);
2329 task_thread_info(p
)->task
= p
;
2332 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2334 return (unsigned long *)(task_thread_info(p
) + 1);
2339 static inline int object_is_on_stack(void *obj
)
2341 void *stack
= task_stack_page(current
);
2343 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2346 extern void thread_info_cache_init(void);
2348 #ifdef CONFIG_DEBUG_STACK_USAGE
2349 static inline unsigned long stack_not_used(struct task_struct
*p
)
2351 unsigned long *n
= end_of_stack(p
);
2353 do { /* Skip over canary */
2357 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
2361 /* set thread flags in other task's structures
2362 * - see asm/thread_info.h for TIF_xxxx flags available
2364 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2366 set_ti_thread_flag(task_thread_info(tsk
), flag
);
2369 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2371 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2374 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2376 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
2379 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2381 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2384 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2386 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
2389 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
2391 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2394 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
2396 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2399 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
2401 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
2404 static inline int restart_syscall(void)
2406 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
2407 return -ERESTARTNOINTR
;
2410 static inline int signal_pending(struct task_struct
*p
)
2412 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
2415 static inline int __fatal_signal_pending(struct task_struct
*p
)
2417 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
2420 static inline int fatal_signal_pending(struct task_struct
*p
)
2422 return signal_pending(p
) && __fatal_signal_pending(p
);
2425 static inline int signal_pending_state(long state
, struct task_struct
*p
)
2427 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
2429 if (!signal_pending(p
))
2432 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
2436 * cond_resched() and cond_resched_lock(): latency reduction via
2437 * explicit rescheduling in places that are safe. The return
2438 * value indicates whether a reschedule was done in fact.
2439 * cond_resched_lock() will drop the spinlock before scheduling,
2440 * cond_resched_softirq() will enable bhs before scheduling.
2442 extern int _cond_resched(void);
2444 #define cond_resched() ({ \
2445 __might_sleep(__FILE__, __LINE__, 0); \
2449 extern int __cond_resched_lock(spinlock_t
*lock
);
2451 #ifdef CONFIG_PREEMPT_COUNT
2452 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2454 #define PREEMPT_LOCK_OFFSET 0
2457 #define cond_resched_lock(lock) ({ \
2458 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2459 __cond_resched_lock(lock); \
2462 extern int __cond_resched_softirq(void);
2464 #define cond_resched_softirq() ({ \
2465 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2466 __cond_resched_softirq(); \
2469 static inline void cond_resched_rcu(void)
2471 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2479 * Does a critical section need to be broken due to another
2480 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2481 * but a general need for low latency)
2483 static inline int spin_needbreak(spinlock_t
*lock
)
2485 #ifdef CONFIG_PREEMPT
2486 return spin_is_contended(lock
);
2493 * Idle thread specific functions to determine the need_resched
2494 * polling state. We have two versions, one based on TS_POLLING in
2495 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2499 static inline int tsk_is_polling(struct task_struct
*p
)
2501 return task_thread_info(p
)->status
& TS_POLLING
;
2503 static inline void __current_set_polling(void)
2505 current_thread_info()->status
|= TS_POLLING
;
2508 static inline bool __must_check
current_set_polling_and_test(void)
2510 __current_set_polling();
2513 * Polling state must be visible before we test NEED_RESCHED,
2514 * paired by resched_task()
2518 return unlikely(tif_need_resched());
2521 static inline void __current_clr_polling(void)
2523 current_thread_info()->status
&= ~TS_POLLING
;
2526 static inline bool __must_check
current_clr_polling_and_test(void)
2528 __current_clr_polling();
2531 * Polling state must be visible before we test NEED_RESCHED,
2532 * paired by resched_task()
2536 return unlikely(tif_need_resched());
2538 #elif defined(TIF_POLLING_NRFLAG)
2539 static inline int tsk_is_polling(struct task_struct
*p
)
2541 return test_tsk_thread_flag(p
, TIF_POLLING_NRFLAG
);
2544 static inline void __current_set_polling(void)
2546 set_thread_flag(TIF_POLLING_NRFLAG
);
2549 static inline bool __must_check
current_set_polling_and_test(void)
2551 __current_set_polling();
2554 * Polling state must be visible before we test NEED_RESCHED,
2555 * paired by resched_task()
2557 * XXX: assumes set/clear bit are identical barrier wise.
2559 smp_mb__after_clear_bit();
2561 return unlikely(tif_need_resched());
2564 static inline void __current_clr_polling(void)
2566 clear_thread_flag(TIF_POLLING_NRFLAG
);
2569 static inline bool __must_check
current_clr_polling_and_test(void)
2571 __current_clr_polling();
2574 * Polling state must be visible before we test NEED_RESCHED,
2575 * paired by resched_task()
2577 smp_mb__after_clear_bit();
2579 return unlikely(tif_need_resched());
2583 static inline int tsk_is_polling(struct task_struct
*p
) { return 0; }
2584 static inline void __current_set_polling(void) { }
2585 static inline void __current_clr_polling(void) { }
2587 static inline bool __must_check
current_set_polling_and_test(void)
2589 return unlikely(tif_need_resched());
2591 static inline bool __must_check
current_clr_polling_and_test(void)
2593 return unlikely(tif_need_resched());
2597 static __always_inline
bool need_resched(void)
2599 return unlikely(tif_need_resched());
2603 * Thread group CPU time accounting.
2605 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
2606 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
2608 static inline void thread_group_cputime_init(struct signal_struct
*sig
)
2610 raw_spin_lock_init(&sig
->cputimer
.lock
);
2614 * Reevaluate whether the task has signals pending delivery.
2615 * Wake the task if so.
2616 * This is required every time the blocked sigset_t changes.
2617 * callers must hold sighand->siglock.
2619 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
2620 extern void recalc_sigpending(void);
2622 extern void signal_wake_up_state(struct task_struct
*t
, unsigned int state
);
2624 static inline void signal_wake_up(struct task_struct
*t
, bool resume
)
2626 signal_wake_up_state(t
, resume
? TASK_WAKEKILL
: 0);
2628 static inline void ptrace_signal_wake_up(struct task_struct
*t
, bool resume
)
2630 signal_wake_up_state(t
, resume
? __TASK_TRACED
: 0);
2634 * Wrappers for p->thread_info->cpu access. No-op on UP.
2638 static inline unsigned int task_cpu(const struct task_struct
*p
)
2640 return task_thread_info(p
)->cpu
;
2643 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
2647 static inline unsigned int task_cpu(const struct task_struct
*p
)
2652 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
2656 #endif /* CONFIG_SMP */
2658 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
2659 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
2661 #ifdef CONFIG_CGROUP_SCHED
2662 extern struct task_group root_task_group
;
2663 #endif /* CONFIG_CGROUP_SCHED */
2665 extern int task_can_switch_user(struct user_struct
*up
,
2666 struct task_struct
*tsk
);
2668 #ifdef CONFIG_TASK_XACCT
2669 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2671 tsk
->ioac
.rchar
+= amt
;
2674 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2676 tsk
->ioac
.wchar
+= amt
;
2679 static inline void inc_syscr(struct task_struct
*tsk
)
2684 static inline void inc_syscw(struct task_struct
*tsk
)
2689 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2693 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2697 static inline void inc_syscr(struct task_struct
*tsk
)
2701 static inline void inc_syscw(struct task_struct
*tsk
)
2706 #ifndef TASK_SIZE_OF
2707 #define TASK_SIZE_OF(tsk) TASK_SIZE
2710 #ifdef CONFIG_MM_OWNER
2711 extern void mm_update_next_owner(struct mm_struct
*mm
);
2712 extern void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
);
2714 static inline void mm_update_next_owner(struct mm_struct
*mm
)
2718 static inline void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
)
2721 #endif /* CONFIG_MM_OWNER */
2723 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
2726 return ACCESS_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
2729 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
2732 return ACCESS_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
2735 static inline unsigned long rlimit(unsigned int limit
)
2737 return task_rlimit(current
, limit
);
2740 static inline unsigned long rlimit_max(unsigned int limit
)
2742 return task_rlimit_max(current
, limit
);