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);
112 extern void dump_cpu_task(int cpu);
117 #ifdef CONFIG_SCHED_DEBUG
118 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
119 extern void proc_sched_set_task(struct task_struct *p);
121 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
124 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
127 static inline void proc_sched_set_task(struct task_struct *p)
131 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
137 * Task state bitmask. NOTE! These bits are also
138 * encoded in fs/proc/array.c: get_task_state().
140 * We have two separate sets of flags: task->state
141 * is about runnability, while task->exit_state are
142 * about the task exiting. Confusing, but this way
143 * modifying one set can't modify the other one by
146 #define TASK_RUNNING 0
147 #define TASK_INTERRUPTIBLE 1
148 #define TASK_UNINTERRUPTIBLE 2
149 #define __TASK_STOPPED 4
150 #define __TASK_TRACED 8
151 /* in tsk->exit_state */
152 #define EXIT_ZOMBIE 16
154 /* in tsk->state again */
156 #define TASK_WAKEKILL 128
157 #define TASK_WAKING 256
158 #define TASK_STATE_MAX 512
160 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
162 extern char ___assert_task_state[1 - 2*!!(
163 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
165 /* Convenience macros for the sake of set_task_state */
166 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
167 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
168 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
170 /* Convenience macros for the sake of wake_up */
171 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
172 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
174 /* get_task_state() */
175 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
176 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
179 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
180 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
181 #define task_is_dead(task) ((task)->exit_state != 0)
182 #define task_is_stopped_or_traced(task) \
183 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
184 #define task_contributes_to_load(task) \
185 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
186 (task->flags & PF_FROZEN) == 0)
188 #define __set_task_state(tsk, state_value) \
189 do { (tsk)->state = (state_value); } while (0)
190 #define set_task_state(tsk, state_value) \
191 set_mb((tsk)->state, (state_value))
194 * set_current_state() includes a barrier so that the write of current->state
195 * is correctly serialised wrt the caller's subsequent test of whether to
198 * set_current_state(TASK_UNINTERRUPTIBLE);
199 * if (do_i_need_to_sleep())
202 * If the caller does not need such serialisation then use __set_current_state()
204 #define __set_current_state(state_value) \
205 do { current->state = (state_value); } while (0)
206 #define set_current_state(state_value) \
207 set_mb(current->state, (state_value))
209 /* Task command name length */
210 #define TASK_COMM_LEN 16
212 #include <linux/spinlock.h>
215 * This serializes "schedule()" and also protects
216 * the run-queue from deletions/modifications (but
217 * _adding_ to the beginning of the run-queue has
220 extern rwlock_t tasklist_lock;
221 extern spinlock_t mmlist_lock;
225 #ifdef CONFIG_PROVE_RCU
226 extern int lockdep_tasklist_lock_is_held(void);
227 #endif /* #ifdef CONFIG_PROVE_RCU */
229 extern void sched_init(void);
230 extern void sched_init_smp(void);
231 extern asmlinkage void schedule_tail(struct task_struct *prev);
232 extern void init_idle(struct task_struct *idle, int cpu);
233 extern void init_idle_bootup_task(struct task_struct *idle);
235 extern int runqueue_is_locked(int cpu);
237 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
238 extern void nohz_balance_enter_idle(int cpu);
239 extern void set_cpu_sd_state_idle(void);
240 extern int get_nohz_timer_target(void);
242 static inline void nohz_balance_enter_idle(int cpu) { }
243 static inline void set_cpu_sd_state_idle(void) { }
247 * Only dump TASK_* tasks. (0 for all tasks)
249 extern void show_state_filter(unsigned long state_filter);
251 static inline void show_state(void)
253 show_state_filter(0);
256 extern void show_regs(struct pt_regs *);
259 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
260 * task), SP is the stack pointer of the first frame that should be shown in the back
261 * trace (or NULL if the entire call-chain of the task should be shown).
263 extern void show_stack(struct task_struct *task, unsigned long *sp);
265 void io_schedule(void);
266 long io_schedule_timeout(long timeout);
268 extern void cpu_init (void);
269 extern void trap_init(void);
270 extern void update_process_times(int user);
271 extern void scheduler_tick(void);
273 extern void sched_show_task(struct task_struct *p);
275 #ifdef CONFIG_LOCKUP_DETECTOR
276 extern void touch_softlockup_watchdog(void);
277 extern void touch_softlockup_watchdog_sync(void);
278 extern void touch_all_softlockup_watchdogs(void);
279 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
281 size_t *lenp, loff_t *ppos);
282 extern unsigned int softlockup_panic;
283 void lockup_detector_init(void);
285 static inline void touch_softlockup_watchdog(void)
288 static inline void touch_softlockup_watchdog_sync(void)
291 static inline void touch_all_softlockup_watchdogs(void)
294 static inline void lockup_detector_init(void)
299 #ifdef CONFIG_DETECT_HUNG_TASK
300 extern unsigned int sysctl_hung_task_panic;
301 extern unsigned long sysctl_hung_task_check_count;
302 extern unsigned long sysctl_hung_task_timeout_secs;
303 extern unsigned long sysctl_hung_task_warnings;
304 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
306 size_t *lenp, loff_t *ppos);
308 /* Avoid need for ifdefs elsewhere in the code */
309 enum { sysctl_hung_task_timeout_secs = 0 };
312 /* Attach to any functions which should be ignored in wchan output. */
313 #define __sched __attribute__((__section__(".sched.text")))
315 /* Linker adds these: start and end of __sched functions */
316 extern char __sched_text_start[], __sched_text_end[];
318 /* Is this address in the __sched functions? */
319 extern int in_sched_functions(unsigned long addr);
321 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
322 extern signed long schedule_timeout(signed long timeout);
323 extern signed long schedule_timeout_interruptible(signed long timeout);
324 extern signed long schedule_timeout_killable(signed long timeout);
325 extern signed long schedule_timeout_uninterruptible(signed long timeout);
326 asmlinkage void schedule(void);
327 extern void schedule_preempt_disabled(void);
328 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
331 struct user_namespace;
334 * Default maximum number of active map areas, this limits the number of vmas
335 * per mm struct. Users can overwrite this number by sysctl but there is a
338 * When a program's coredump is generated as ELF format, a section is created
339 * per a vma. In ELF, the number of sections is represented in unsigned short.
340 * This means the number of sections should be smaller than 65535 at coredump.
341 * Because the kernel adds some informative sections to a image of program at
342 * generating coredump, we need some margin. The number of extra sections is
343 * 1-3 now and depends on arch. We use "5" as safe margin, here.
345 #define MAPCOUNT_ELF_CORE_MARGIN (5)
346 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
348 extern int sysctl_max_map_count;
350 #include <linux/aio.h>
353 extern void arch_pick_mmap_layout(struct mm_struct *mm);
355 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
356 unsigned long, unsigned long);
358 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
359 unsigned long len, unsigned long pgoff,
360 unsigned long flags);
361 extern void arch_unmap_area(struct mm_struct *, unsigned long);
362 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
364 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
368 extern void set_dumpable(struct mm_struct *mm, int value);
369 extern int get_dumpable(struct mm_struct *mm);
371 /* get/set_dumpable() values */
372 #define SUID_DUMPABLE_DISABLED 0
373 #define SUID_DUMPABLE_ENABLED 1
374 #define SUID_DUMPABLE_SAFE 2
378 #define MMF_DUMPABLE 0 /* core dump is permitted */
379 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
381 #define MMF_DUMPABLE_BITS 2
382 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
384 /* coredump filter bits */
385 #define MMF_DUMP_ANON_PRIVATE 2
386 #define MMF_DUMP_ANON_SHARED 3
387 #define MMF_DUMP_MAPPED_PRIVATE 4
388 #define MMF_DUMP_MAPPED_SHARED 5
389 #define MMF_DUMP_ELF_HEADERS 6
390 #define MMF_DUMP_HUGETLB_PRIVATE 7
391 #define MMF_DUMP_HUGETLB_SHARED 8
393 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
394 #define MMF_DUMP_FILTER_BITS 7
395 #define MMF_DUMP_FILTER_MASK \
396 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
397 #define MMF_DUMP_FILTER_DEFAULT \
398 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
399 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
401 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
402 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
404 # define MMF_DUMP_MASK_DEFAULT_ELF 0
406 /* leave room for more dump flags */
407 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
408 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
409 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
411 #define MMF_HAS_UPROBES 19 /* has uprobes */
412 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
414 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
416 struct sighand_struct {
418 struct k_sigaction action[_NSIG];
420 wait_queue_head_t signalfd_wqh;
423 struct pacct_struct {
426 unsigned long ac_mem;
427 cputime_t ac_utime, ac_stime;
428 unsigned long ac_minflt, ac_majflt;
439 * struct task_cputime - collected CPU time counts
440 * @utime: time spent in user mode, in &cputime_t units
441 * @stime: time spent in kernel mode, in &cputime_t units
442 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
444 * This structure groups together three kinds of CPU time that are
445 * tracked for threads and thread groups. Most things considering
446 * CPU time want to group these counts together and treat all three
447 * of them in parallel.
449 struct task_cputime {
452 unsigned long long sum_exec_runtime;
454 /* Alternate field names when used to cache expirations. */
455 #define prof_exp stime
456 #define virt_exp utime
457 #define sched_exp sum_exec_runtime
459 #define INIT_CPUTIME \
460 (struct task_cputime) { \
463 .sum_exec_runtime = 0, \
467 * Disable preemption until the scheduler is running.
468 * Reset by start_kernel()->sched_init()->init_idle().
470 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
471 * before the scheduler is active -- see should_resched().
473 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
476 * struct thread_group_cputimer - thread group interval timer counts
477 * @cputime: thread group interval timers.
478 * @running: non-zero when there are timers running and
479 * @cputime receives updates.
480 * @lock: lock for fields in this struct.
482 * This structure contains the version of task_cputime, above, that is
483 * used for thread group CPU timer calculations.
485 struct thread_group_cputimer {
486 struct task_cputime cputime;
491 #include <linux/rwsem.h>
495 * NOTE! "signal_struct" does not have its own
496 * locking, because a shared signal_struct always
497 * implies a shared sighand_struct, so locking
498 * sighand_struct is always a proper superset of
499 * the locking of signal_struct.
501 struct signal_struct {
506 wait_queue_head_t wait_chldexit; /* for wait4() */
508 /* current thread group signal load-balancing target: */
509 struct task_struct *curr_target;
511 /* shared signal handling: */
512 struct sigpending shared_pending;
514 /* thread group exit support */
517 * - notify group_exit_task when ->count is equal to notify_count
518 * - everyone except group_exit_task is stopped during signal delivery
519 * of fatal signals, group_exit_task processes the signal.
522 struct task_struct *group_exit_task;
524 /* thread group stop support, overloads group_exit_code too */
525 int group_stop_count;
526 unsigned int flags; /* see SIGNAL_* flags below */
529 * PR_SET_CHILD_SUBREAPER marks a process, like a service
530 * manager, to re-parent orphan (double-forking) child processes
531 * to this process instead of 'init'. The service manager is
532 * able to receive SIGCHLD signals and is able to investigate
533 * the process until it calls wait(). All children of this
534 * process will inherit a flag if they should look for a
535 * child_subreaper process at exit.
537 unsigned int is_child_subreaper:1;
538 unsigned int has_child_subreaper:1;
540 /* POSIX.1b Interval Timers */
541 struct list_head posix_timers;
543 /* ITIMER_REAL timer for the process */
544 struct hrtimer real_timer;
545 struct pid *leader_pid;
546 ktime_t it_real_incr;
549 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
550 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
551 * values are defined to 0 and 1 respectively
553 struct cpu_itimer it[2];
556 * Thread group totals for process CPU timers.
557 * See thread_group_cputimer(), et al, for details.
559 struct thread_group_cputimer cputimer;
561 /* Earliest-expiration cache. */
562 struct task_cputime cputime_expires;
564 struct list_head cpu_timers[3];
566 struct pid *tty_old_pgrp;
568 /* boolean value for session group leader */
571 struct tty_struct *tty; /* NULL if no tty */
573 #ifdef CONFIG_SCHED_AUTOGROUP
574 struct autogroup *autogroup;
577 * Cumulative resource counters for dead threads in the group,
578 * and for reaped dead child processes forked by this group.
579 * Live threads maintain their own counters and add to these
580 * in __exit_signal, except for the group leader.
582 cputime_t utime, stime, cutime, cstime;
585 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
586 cputime_t prev_utime, prev_stime;
588 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
589 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
590 unsigned long inblock, oublock, cinblock, coublock;
591 unsigned long maxrss, cmaxrss;
592 struct task_io_accounting ioac;
595 * Cumulative ns of schedule CPU time fo dead threads in the
596 * group, not including a zombie group leader, (This only differs
597 * from jiffies_to_ns(utime + stime) if sched_clock uses something
598 * other than jiffies.)
600 unsigned long long sum_sched_runtime;
603 * We don't bother to synchronize most readers of this at all,
604 * because there is no reader checking a limit that actually needs
605 * to get both rlim_cur and rlim_max atomically, and either one
606 * alone is a single word that can safely be read normally.
607 * getrlimit/setrlimit use task_lock(current->group_leader) to
608 * protect this instead of the siglock, because they really
609 * have no need to disable irqs.
611 struct rlimit rlim[RLIM_NLIMITS];
613 #ifdef CONFIG_BSD_PROCESS_ACCT
614 struct pacct_struct pacct; /* per-process accounting information */
616 #ifdef CONFIG_TASKSTATS
617 struct taskstats *stats;
621 struct tty_audit_buf *tty_audit_buf;
623 #ifdef CONFIG_CGROUPS
625 * group_rwsem prevents new tasks from entering the threadgroup and
626 * member tasks from exiting,a more specifically, setting of
627 * PF_EXITING. fork and exit paths are protected with this rwsem
628 * using threadgroup_change_begin/end(). Users which require
629 * threadgroup to remain stable should use threadgroup_[un]lock()
630 * which also takes care of exec path. Currently, cgroup is the
633 struct rw_semaphore group_rwsem;
636 oom_flags_t oom_flags;
637 short oom_score_adj; /* OOM kill score adjustment */
638 short oom_score_adj_min; /* OOM kill score adjustment min value.
639 * Only settable by CAP_SYS_RESOURCE. */
641 struct mutex cred_guard_mutex; /* guard against foreign influences on
642 * credential calculations
643 * (notably. ptrace) */
647 * Bits in flags field of signal_struct.
649 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
650 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
651 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
653 * Pending notifications to parent.
655 #define SIGNAL_CLD_STOPPED 0x00000010
656 #define SIGNAL_CLD_CONTINUED 0x00000020
657 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
659 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
661 /* If true, all threads except ->group_exit_task have pending SIGKILL */
662 static inline int signal_group_exit(const struct signal_struct *sig)
664 return (sig->flags & SIGNAL_GROUP_EXIT) ||
665 (sig->group_exit_task != NULL);
669 * Some day this will be a full-fledged user tracking system..
672 atomic_t __count; /* reference count */
673 atomic_t processes; /* How many processes does this user have? */
674 atomic_t files; /* How many open files does this user have? */
675 atomic_t sigpending; /* How many pending signals does this user have? */
676 #ifdef CONFIG_INOTIFY_USER
677 atomic_t inotify_watches; /* How many inotify watches does this user have? */
678 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
680 #ifdef CONFIG_FANOTIFY
681 atomic_t fanotify_listeners;
684 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
686 #ifdef CONFIG_POSIX_MQUEUE
687 /* protected by mq_lock */
688 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
690 unsigned long locked_shm; /* How many pages of mlocked shm ? */
693 struct key *uid_keyring; /* UID specific keyring */
694 struct key *session_keyring; /* UID's default session keyring */
697 /* Hash table maintenance information */
698 struct hlist_node uidhash_node;
701 #ifdef CONFIG_PERF_EVENTS
702 atomic_long_t locked_vm;
706 extern int uids_sysfs_init(void);
708 extern struct user_struct *find_user(kuid_t);
710 extern struct user_struct root_user;
711 #define INIT_USER (&root_user)
714 struct backing_dev_info;
715 struct reclaim_state;
717 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
719 /* cumulative counters */
720 unsigned long pcount; /* # of times run on this cpu */
721 unsigned long long run_delay; /* time spent waiting on a runqueue */
724 unsigned long long last_arrival,/* when we last ran on a cpu */
725 last_queued; /* when we were last queued to run */
727 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
729 #ifdef CONFIG_TASK_DELAY_ACCT
730 struct task_delay_info {
732 unsigned int flags; /* Private per-task flags */
734 /* For each stat XXX, add following, aligned appropriately
736 * struct timespec XXX_start, XXX_end;
740 * Atomicity of updates to XXX_delay, XXX_count protected by
741 * single lock above (split into XXX_lock if contention is an issue).
745 * XXX_count is incremented on every XXX operation, the delay
746 * associated with the operation is added to XXX_delay.
747 * XXX_delay contains the accumulated delay time in nanoseconds.
749 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
750 u64 blkio_delay; /* wait for sync block io completion */
751 u64 swapin_delay; /* wait for swapin block io completion */
752 u32 blkio_count; /* total count of the number of sync block */
753 /* io operations performed */
754 u32 swapin_count; /* total count of the number of swapin block */
755 /* io operations performed */
757 struct timespec freepages_start, freepages_end;
758 u64 freepages_delay; /* wait for memory reclaim */
759 u32 freepages_count; /* total count of memory reclaim */
761 #endif /* CONFIG_TASK_DELAY_ACCT */
763 static inline int sched_info_on(void)
765 #ifdef CONFIG_SCHEDSTATS
767 #elif defined(CONFIG_TASK_DELAY_ACCT)
768 extern int delayacct_on;
783 * Increase resolution of nice-level calculations for 64-bit architectures.
784 * The extra resolution improves shares distribution and load balancing of
785 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
786 * hierarchies, especially on larger systems. This is not a user-visible change
787 * and does not change the user-interface for setting shares/weights.
789 * We increase resolution only if we have enough bits to allow this increased
790 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
791 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
794 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
795 # define SCHED_LOAD_RESOLUTION 10
796 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
797 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
799 # define SCHED_LOAD_RESOLUTION 0
800 # define scale_load(w) (w)
801 # define scale_load_down(w) (w)
804 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
805 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
808 * Increase resolution of cpu_power calculations
810 #define SCHED_POWER_SHIFT 10
811 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
814 * sched-domains (multiprocessor balancing) declarations:
817 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
818 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
819 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
820 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
821 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
822 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
823 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
824 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
825 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
826 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
827 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
828 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
830 extern int __weak arch_sd_sibiling_asym_packing(void);
832 struct sched_group_power {
835 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
838 unsigned int power, power_orig;
839 unsigned long next_update;
841 * Number of busy cpus in this group.
843 atomic_t nr_busy_cpus;
845 unsigned long cpumask[0]; /* iteration mask */
849 struct sched_group *next; /* Must be a circular list */
852 unsigned int group_weight;
853 struct sched_group_power *sgp;
856 * The CPUs this group covers.
858 * NOTE: this field is variable length. (Allocated dynamically
859 * by attaching extra space to the end of the structure,
860 * depending on how many CPUs the kernel has booted up with)
862 unsigned long cpumask[0];
865 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
867 return to_cpumask(sg->cpumask);
871 * cpumask masking which cpus in the group are allowed to iterate up the domain
874 static inline struct cpumask *sched_group_mask(struct sched_group *sg)
876 return to_cpumask(sg->sgp->cpumask);
880 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
881 * @group: The group whose first cpu is to be returned.
883 static inline unsigned int group_first_cpu(struct sched_group *group)
885 return cpumask_first(sched_group_cpus(group));
888 struct sched_domain_attr {
889 int relax_domain_level;
892 #define SD_ATTR_INIT (struct sched_domain_attr) { \
893 .relax_domain_level = -1, \
896 extern int sched_domain_level_max;
898 struct sched_domain {
899 /* These fields must be setup */
900 struct sched_domain *parent; /* top domain must be null terminated */
901 struct sched_domain *child; /* bottom domain must be null terminated */
902 struct sched_group *groups; /* the balancing groups of the domain */
903 unsigned long min_interval; /* Minimum balance interval ms */
904 unsigned long max_interval; /* Maximum balance interval ms */
905 unsigned int busy_factor; /* less balancing by factor if busy */
906 unsigned int imbalance_pct; /* No balance until over watermark */
907 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
908 unsigned int busy_idx;
909 unsigned int idle_idx;
910 unsigned int newidle_idx;
911 unsigned int wake_idx;
912 unsigned int forkexec_idx;
913 unsigned int smt_gain;
914 int flags; /* See SD_* */
917 /* Runtime fields. */
918 unsigned long last_balance; /* init to jiffies. units in jiffies */
919 unsigned int balance_interval; /* initialise to 1. units in ms. */
920 unsigned int nr_balance_failed; /* initialise to 0 */
924 #ifdef CONFIG_SCHEDSTATS
925 /* load_balance() stats */
926 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
927 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
928 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
929 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
931 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
932 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
933 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
935 /* Active load balancing */
936 unsigned int alb_count;
937 unsigned int alb_failed;
938 unsigned int alb_pushed;
940 /* SD_BALANCE_EXEC stats */
941 unsigned int sbe_count;
942 unsigned int sbe_balanced;
943 unsigned int sbe_pushed;
945 /* SD_BALANCE_FORK stats */
946 unsigned int sbf_count;
947 unsigned int sbf_balanced;
948 unsigned int sbf_pushed;
950 /* try_to_wake_up() stats */
951 unsigned int ttwu_wake_remote;
952 unsigned int ttwu_move_affine;
953 unsigned int ttwu_move_balance;
955 #ifdef CONFIG_SCHED_DEBUG
959 void *private; /* used during construction */
960 struct rcu_head rcu; /* used during destruction */
963 unsigned int span_weight;
965 * Span of all CPUs in this domain.
967 * NOTE: this field is variable length. (Allocated dynamically
968 * by attaching extra space to the end of the structure,
969 * depending on how many CPUs the kernel has booted up with)
971 unsigned long span[0];
974 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
976 return to_cpumask(sd->span);
979 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
980 struct sched_domain_attr *dattr_new);
982 /* Allocate an array of sched domains, for partition_sched_domains(). */
983 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
984 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
986 /* Test a flag in parent sched domain */
987 static inline int test_sd_parent(struct sched_domain *sd, int flag)
989 if (sd->parent && (sd->parent->flags & flag))
995 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
996 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
998 bool cpus_share_cache(int this_cpu, int that_cpu);
1000 #else /* CONFIG_SMP */
1002 struct sched_domain_attr;
1005 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1006 struct sched_domain_attr *dattr_new)
1010 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1015 #endif /* !CONFIG_SMP */
1018 struct io_context; /* See blkdev.h */
1021 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1022 extern void prefetch_stack(struct task_struct *t);
1024 static inline void prefetch_stack(struct task_struct *t) { }
1027 struct audit_context; /* See audit.c */
1029 struct pipe_inode_info;
1030 struct uts_namespace;
1033 struct sched_domain;
1038 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1039 #define WF_FORK 0x02 /* child wakeup after fork */
1040 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1042 #define ENQUEUE_WAKEUP 1
1043 #define ENQUEUE_HEAD 2
1045 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1047 #define ENQUEUE_WAKING 0
1050 #define DEQUEUE_SLEEP 1
1052 struct sched_class {
1053 const struct sched_class *next;
1055 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1056 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1057 void (*yield_task) (struct rq *rq);
1058 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1060 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1062 struct task_struct * (*pick_next_task) (struct rq *rq);
1063 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1066 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1068 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1069 void (*post_schedule) (struct rq *this_rq);
1070 void (*task_waking) (struct task_struct *task);
1071 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1073 void (*set_cpus_allowed)(struct task_struct *p,
1074 const struct cpumask *newmask);
1076 void (*rq_online)(struct rq *rq);
1077 void (*rq_offline)(struct rq *rq);
1080 void (*set_curr_task) (struct rq *rq);
1081 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1082 void (*task_fork) (struct task_struct *p);
1084 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1085 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1086 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1089 unsigned int (*get_rr_interval) (struct rq *rq,
1090 struct task_struct *task);
1092 #ifdef CONFIG_FAIR_GROUP_SCHED
1093 void (*task_move_group) (struct task_struct *p, int on_rq);
1097 struct load_weight {
1098 unsigned long weight, inv_weight;
1101 #ifdef CONFIG_SCHEDSTATS
1102 struct sched_statistics {
1112 s64 sum_sleep_runtime;
1119 u64 nr_migrations_cold;
1120 u64 nr_failed_migrations_affine;
1121 u64 nr_failed_migrations_running;
1122 u64 nr_failed_migrations_hot;
1123 u64 nr_forced_migrations;
1126 u64 nr_wakeups_sync;
1127 u64 nr_wakeups_migrate;
1128 u64 nr_wakeups_local;
1129 u64 nr_wakeups_remote;
1130 u64 nr_wakeups_affine;
1131 u64 nr_wakeups_affine_attempts;
1132 u64 nr_wakeups_passive;
1133 u64 nr_wakeups_idle;
1137 struct sched_entity {
1138 struct load_weight load; /* for load-balancing */
1139 struct rb_node run_node;
1140 struct list_head group_node;
1144 u64 sum_exec_runtime;
1146 u64 prev_sum_exec_runtime;
1150 #ifdef CONFIG_SCHEDSTATS
1151 struct sched_statistics statistics;
1154 #ifdef CONFIG_FAIR_GROUP_SCHED
1155 struct sched_entity *parent;
1156 /* rq on which this entity is (to be) queued: */
1157 struct cfs_rq *cfs_rq;
1158 /* rq "owned" by this entity/group: */
1159 struct cfs_rq *my_q;
1163 struct sched_rt_entity {
1164 struct list_head run_list;
1165 unsigned long timeout;
1166 unsigned int time_slice;
1168 struct sched_rt_entity *back;
1169 #ifdef CONFIG_RT_GROUP_SCHED
1170 struct sched_rt_entity *parent;
1171 /* rq on which this entity is (to be) queued: */
1172 struct rt_rq *rt_rq;
1173 /* rq "owned" by this entity/group: */
1179 * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1180 * Timeslices get refilled after they expire.
1182 #define RR_TIMESLICE (100 * HZ / 1000)
1186 enum perf_event_task_context {
1187 perf_invalid_context = -1,
1188 perf_hw_context = 0,
1190 perf_nr_task_contexts,
1193 struct task_struct {
1194 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1197 unsigned int flags; /* per process flags, defined below */
1198 unsigned int ptrace;
1201 struct llist_node wake_entry;
1206 int prio, static_prio, normal_prio;
1207 unsigned int rt_priority;
1208 const struct sched_class *sched_class;
1209 struct sched_entity se;
1210 struct sched_rt_entity rt;
1211 #ifdef CONFIG_CGROUP_SCHED
1212 struct task_group *sched_task_group;
1215 #ifdef CONFIG_PREEMPT_NOTIFIERS
1216 /* list of struct preempt_notifier: */
1217 struct hlist_head preempt_notifiers;
1221 * fpu_counter contains the number of consecutive context switches
1222 * that the FPU is used. If this is over a threshold, the lazy fpu
1223 * saving becomes unlazy to save the trap. This is an unsigned char
1224 * so that after 256 times the counter wraps and the behavior turns
1225 * lazy again; this to deal with bursty apps that only use FPU for
1228 unsigned char fpu_counter;
1229 #ifdef CONFIG_BLK_DEV_IO_TRACE
1230 unsigned int btrace_seq;
1233 unsigned int policy;
1234 int nr_cpus_allowed;
1235 cpumask_t cpus_allowed;
1237 #ifdef CONFIG_PREEMPT_RCU
1238 int rcu_read_lock_nesting;
1239 char rcu_read_unlock_special;
1240 struct list_head rcu_node_entry;
1241 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1242 #ifdef CONFIG_TREE_PREEMPT_RCU
1243 struct rcu_node *rcu_blocked_node;
1244 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1245 #ifdef CONFIG_RCU_BOOST
1246 struct rt_mutex *rcu_boost_mutex;
1247 #endif /* #ifdef CONFIG_RCU_BOOST */
1249 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1250 struct sched_info sched_info;
1253 struct list_head tasks;
1255 struct plist_node pushable_tasks;
1258 struct mm_struct *mm, *active_mm;
1259 #ifdef CONFIG_COMPAT_BRK
1260 unsigned brk_randomized:1;
1262 #if defined(SPLIT_RSS_COUNTING)
1263 struct task_rss_stat rss_stat;
1267 int exit_code, exit_signal;
1268 int pdeath_signal; /* The signal sent when the parent dies */
1269 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1271 unsigned int personality;
1272 unsigned did_exec:1;
1273 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1275 unsigned in_iowait:1;
1277 /* task may not gain privileges */
1278 unsigned no_new_privs:1;
1280 /* Revert to default priority/policy when forking */
1281 unsigned sched_reset_on_fork:1;
1282 unsigned sched_contributes_to_load:1;
1287 #ifdef CONFIG_CC_STACKPROTECTOR
1288 /* Canary value for the -fstack-protector gcc feature */
1289 unsigned long stack_canary;
1292 * pointers to (original) parent process, youngest child, younger sibling,
1293 * older sibling, respectively. (p->father can be replaced with
1294 * p->real_parent->pid)
1296 struct task_struct __rcu *real_parent; /* real parent process */
1297 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1299 * children/sibling forms the list of my natural children
1301 struct list_head children; /* list of my children */
1302 struct list_head sibling; /* linkage in my parent's children list */
1303 struct task_struct *group_leader; /* threadgroup leader */
1306 * ptraced is the list of tasks this task is using ptrace on.
1307 * This includes both natural children and PTRACE_ATTACH targets.
1308 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1310 struct list_head ptraced;
1311 struct list_head ptrace_entry;
1313 /* PID/PID hash table linkage. */
1314 struct pid_link pids[PIDTYPE_MAX];
1315 struct list_head thread_group;
1317 struct completion *vfork_done; /* for vfork() */
1318 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1319 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1321 cputime_t utime, stime, utimescaled, stimescaled;
1323 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1324 cputime_t prev_utime, prev_stime;
1326 unsigned long nvcsw, nivcsw; /* context switch counts */
1327 struct timespec start_time; /* monotonic time */
1328 struct timespec real_start_time; /* boot based time */
1329 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1330 unsigned long min_flt, maj_flt;
1332 struct task_cputime cputime_expires;
1333 struct list_head cpu_timers[3];
1335 /* process credentials */
1336 const struct cred __rcu *real_cred; /* objective and real subjective task
1337 * credentials (COW) */
1338 const struct cred __rcu *cred; /* effective (overridable) subjective task
1339 * credentials (COW) */
1340 char comm[TASK_COMM_LEN]; /* executable name excluding path
1341 - access with [gs]et_task_comm (which lock
1342 it with task_lock())
1343 - initialized normally by setup_new_exec */
1344 /* file system info */
1345 int link_count, total_link_count;
1346 #ifdef CONFIG_SYSVIPC
1348 struct sysv_sem sysvsem;
1350 #ifdef CONFIG_DETECT_HUNG_TASK
1351 /* hung task detection */
1352 unsigned long last_switch_count;
1354 /* CPU-specific state of this task */
1355 struct thread_struct thread;
1356 /* filesystem information */
1357 struct fs_struct *fs;
1358 /* open file information */
1359 struct files_struct *files;
1361 struct nsproxy *nsproxy;
1362 /* signal handlers */
1363 struct signal_struct *signal;
1364 struct sighand_struct *sighand;
1366 sigset_t blocked, real_blocked;
1367 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1368 struct sigpending pending;
1370 unsigned long sas_ss_sp;
1372 int (*notifier)(void *priv);
1373 void *notifier_data;
1374 sigset_t *notifier_mask;
1375 struct callback_head *task_works;
1377 struct audit_context *audit_context;
1378 #ifdef CONFIG_AUDITSYSCALL
1380 unsigned int sessionid;
1382 struct seccomp seccomp;
1384 /* Thread group tracking */
1387 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1389 spinlock_t alloc_lock;
1391 /* Protection of the PI data structures: */
1392 raw_spinlock_t pi_lock;
1394 #ifdef CONFIG_RT_MUTEXES
1395 /* PI waiters blocked on a rt_mutex held by this task */
1396 struct plist_head pi_waiters;
1397 /* Deadlock detection and priority inheritance handling */
1398 struct rt_mutex_waiter *pi_blocked_on;
1401 #ifdef CONFIG_DEBUG_MUTEXES
1402 /* mutex deadlock detection */
1403 struct mutex_waiter *blocked_on;
1405 #ifdef CONFIG_TRACE_IRQFLAGS
1406 unsigned int irq_events;
1407 unsigned long hardirq_enable_ip;
1408 unsigned long hardirq_disable_ip;
1409 unsigned int hardirq_enable_event;
1410 unsigned int hardirq_disable_event;
1411 int hardirqs_enabled;
1412 int hardirq_context;
1413 unsigned long softirq_disable_ip;
1414 unsigned long softirq_enable_ip;
1415 unsigned int softirq_disable_event;
1416 unsigned int softirq_enable_event;
1417 int softirqs_enabled;
1418 int softirq_context;
1420 #ifdef CONFIG_LOCKDEP
1421 # define MAX_LOCK_DEPTH 48UL
1424 unsigned int lockdep_recursion;
1425 struct held_lock held_locks[MAX_LOCK_DEPTH];
1426 gfp_t lockdep_reclaim_gfp;
1429 /* journalling filesystem info */
1432 /* stacked block device info */
1433 struct bio_list *bio_list;
1436 /* stack plugging */
1437 struct blk_plug *plug;
1441 struct reclaim_state *reclaim_state;
1443 struct backing_dev_info *backing_dev_info;
1445 struct io_context *io_context;
1447 unsigned long ptrace_message;
1448 siginfo_t *last_siginfo; /* For ptrace use. */
1449 struct task_io_accounting ioac;
1450 #if defined(CONFIG_TASK_XACCT)
1451 u64 acct_rss_mem1; /* accumulated rss usage */
1452 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1453 cputime_t acct_timexpd; /* stime + utime since last update */
1455 #ifdef CONFIG_CPUSETS
1456 nodemask_t mems_allowed; /* Protected by alloc_lock */
1457 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1458 int cpuset_mem_spread_rotor;
1459 int cpuset_slab_spread_rotor;
1461 #ifdef CONFIG_CGROUPS
1462 /* Control Group info protected by css_set_lock */
1463 struct css_set __rcu *cgroups;
1464 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1465 struct list_head cg_list;
1468 struct robust_list_head __user *robust_list;
1469 #ifdef CONFIG_COMPAT
1470 struct compat_robust_list_head __user *compat_robust_list;
1472 struct list_head pi_state_list;
1473 struct futex_pi_state *pi_state_cache;
1475 #ifdef CONFIG_PERF_EVENTS
1476 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1477 struct mutex perf_event_mutex;
1478 struct list_head perf_event_list;
1481 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1483 short pref_node_fork;
1485 struct rcu_head rcu;
1488 * cache last used pipe for splice
1490 struct pipe_inode_info *splice_pipe;
1492 struct page_frag task_frag;
1494 #ifdef CONFIG_TASK_DELAY_ACCT
1495 struct task_delay_info *delays;
1497 #ifdef CONFIG_FAULT_INJECTION
1501 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1502 * balance_dirty_pages() for some dirty throttling pause
1505 int nr_dirtied_pause;
1506 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1508 #ifdef CONFIG_LATENCYTOP
1509 int latency_record_count;
1510 struct latency_record latency_record[LT_SAVECOUNT];
1513 * time slack values; these are used to round up poll() and
1514 * select() etc timeout values. These are in nanoseconds.
1516 unsigned long timer_slack_ns;
1517 unsigned long default_timer_slack_ns;
1519 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1520 /* Index of current stored address in ret_stack */
1522 /* Stack of return addresses for return function tracing */
1523 struct ftrace_ret_stack *ret_stack;
1524 /* time stamp for last schedule */
1525 unsigned long long ftrace_timestamp;
1527 * Number of functions that haven't been traced
1528 * because of depth overrun.
1530 atomic_t trace_overrun;
1531 /* Pause for the tracing */
1532 atomic_t tracing_graph_pause;
1534 #ifdef CONFIG_TRACING
1535 /* state flags for use by tracers */
1536 unsigned long trace;
1537 /* bitmask and counter of trace recursion */
1538 unsigned long trace_recursion;
1539 #endif /* CONFIG_TRACING */
1540 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1541 struct memcg_batch_info {
1542 int do_batch; /* incremented when batch uncharge started */
1543 struct mem_cgroup *memcg; /* target memcg of uncharge */
1544 unsigned long nr_pages; /* uncharged usage */
1545 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1548 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1549 atomic_t ptrace_bp_refcnt;
1551 #ifdef CONFIG_UPROBES
1552 struct uprobe_task *utask;
1556 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1557 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1560 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1561 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1562 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1563 * values are inverted: lower p->prio value means higher priority.
1565 * The MAX_USER_RT_PRIO value allows the actual maximum
1566 * RT priority to be separate from the value exported to
1567 * user-space. This allows kernel threads to set their
1568 * priority to a value higher than any user task. Note:
1569 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1572 #define MAX_USER_RT_PRIO 100
1573 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1575 #define MAX_PRIO (MAX_RT_PRIO + 40)
1576 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1578 static inline int rt_prio(int prio)
1580 if (unlikely(prio < MAX_RT_PRIO))
1585 static inline int rt_task(struct task_struct *p)
1587 return rt_prio(p->prio);
1590 static inline struct pid *task_pid(struct task_struct *task)
1592 return task->pids[PIDTYPE_PID].pid;
1595 static inline struct pid *task_tgid(struct task_struct *task)
1597 return task->group_leader->pids[PIDTYPE_PID].pid;
1601 * Without tasklist or rcu lock it is not safe to dereference
1602 * the result of task_pgrp/task_session even if task == current,
1603 * we can race with another thread doing sys_setsid/sys_setpgid.
1605 static inline struct pid *task_pgrp(struct task_struct *task)
1607 return task->group_leader->pids[PIDTYPE_PGID].pid;
1610 static inline struct pid *task_session(struct task_struct *task)
1612 return task->group_leader->pids[PIDTYPE_SID].pid;
1615 struct pid_namespace;
1618 * the helpers to get the task's different pids as they are seen
1619 * from various namespaces
1621 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1622 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1624 * task_xid_nr_ns() : id seen from the ns specified;
1626 * set_task_vxid() : assigns a virtual id to a task;
1628 * see also pid_nr() etc in include/linux/pid.h
1630 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1631 struct pid_namespace *ns);
1633 static inline pid_t task_pid_nr(struct task_struct *tsk)
1638 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1639 struct pid_namespace *ns)
1641 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1644 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1646 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1650 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1655 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1657 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1659 return pid_vnr(task_tgid(tsk));
1663 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1664 struct pid_namespace *ns)
1666 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1669 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1671 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1675 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1676 struct pid_namespace *ns)
1678 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1681 static inline pid_t task_session_vnr(struct task_struct *tsk)
1683 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1686 /* obsolete, do not use */
1687 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1689 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1693 * pid_alive - check that a task structure is not stale
1694 * @p: Task structure to be checked.
1696 * Test if a process is not yet dead (at most zombie state)
1697 * If pid_alive fails, then pointers within the task structure
1698 * can be stale and must not be dereferenced.
1700 static inline int pid_alive(struct task_struct *p)
1702 return p->pids[PIDTYPE_PID].pid != NULL;
1706 * is_global_init - check if a task structure is init
1707 * @tsk: Task structure to be checked.
1709 * Check if a task structure is the first user space task the kernel created.
1711 static inline int is_global_init(struct task_struct *tsk)
1713 return tsk->pid == 1;
1717 * is_container_init:
1718 * check whether in the task is init in its own pid namespace.
1720 extern int is_container_init(struct task_struct *tsk);
1722 extern struct pid *cad_pid;
1724 extern void free_task(struct task_struct *tsk);
1725 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1727 extern void __put_task_struct(struct task_struct *t);
1729 static inline void put_task_struct(struct task_struct *t)
1731 if (atomic_dec_and_test(&t->usage))
1732 __put_task_struct(t);
1735 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1736 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1741 #define PF_EXITING 0x00000004 /* getting shut down */
1742 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1743 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1744 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1745 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1746 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1747 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1748 #define PF_DUMPCORE 0x00000200 /* dumped core */
1749 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1750 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1751 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1752 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1753 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1754 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1755 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1756 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1757 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1758 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1759 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1760 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1761 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1762 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1763 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1764 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1765 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1766 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1767 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1770 * Only the _current_ task can read/write to tsk->flags, but other
1771 * tasks can access tsk->flags in readonly mode for example
1772 * with tsk_used_math (like during threaded core dumping).
1773 * There is however an exception to this rule during ptrace
1774 * or during fork: the ptracer task is allowed to write to the
1775 * child->flags of its traced child (same goes for fork, the parent
1776 * can write to the child->flags), because we're guaranteed the
1777 * child is not running and in turn not changing child->flags
1778 * at the same time the parent does it.
1780 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1781 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1782 #define clear_used_math() clear_stopped_child_used_math(current)
1783 #define set_used_math() set_stopped_child_used_math(current)
1784 #define conditional_stopped_child_used_math(condition, child) \
1785 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1786 #define conditional_used_math(condition) \
1787 conditional_stopped_child_used_math(condition, current)
1788 #define copy_to_stopped_child_used_math(child) \
1789 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1790 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1791 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1792 #define used_math() tsk_used_math(current)
1795 * task->jobctl flags
1797 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1799 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1800 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1801 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1802 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1803 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1804 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1805 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1807 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1808 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1809 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1810 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1811 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1812 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1813 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1815 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1816 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1818 extern bool task_set_jobctl_pending(struct task_struct *task,
1820 extern void task_clear_jobctl_trapping(struct task_struct *task);
1821 extern void task_clear_jobctl_pending(struct task_struct *task,
1824 #ifdef CONFIG_PREEMPT_RCU
1826 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1827 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1829 static inline void rcu_copy_process(struct task_struct *p)
1831 p->rcu_read_lock_nesting = 0;
1832 p->rcu_read_unlock_special = 0;
1833 #ifdef CONFIG_TREE_PREEMPT_RCU
1834 p->rcu_blocked_node = NULL;
1835 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1836 #ifdef CONFIG_RCU_BOOST
1837 p->rcu_boost_mutex = NULL;
1838 #endif /* #ifdef CONFIG_RCU_BOOST */
1839 INIT_LIST_HEAD(&p->rcu_node_entry);
1844 static inline void rcu_copy_process(struct task_struct *p)
1850 static inline void tsk_restore_flags(struct task_struct *task,
1851 unsigned long orig_flags, unsigned long flags)
1853 task->flags &= ~flags;
1854 task->flags |= orig_flags & flags;
1858 extern void do_set_cpus_allowed(struct task_struct *p,
1859 const struct cpumask *new_mask);
1861 extern int set_cpus_allowed_ptr(struct task_struct *p,
1862 const struct cpumask *new_mask);
1864 static inline void do_set_cpus_allowed(struct task_struct *p,
1865 const struct cpumask *new_mask)
1868 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1869 const struct cpumask *new_mask)
1871 if (!cpumask_test_cpu(0, new_mask))
1878 void calc_load_enter_idle(void);
1879 void calc_load_exit_idle(void);
1881 static inline void calc_load_enter_idle(void) { }
1882 static inline void calc_load_exit_idle(void) { }
1883 #endif /* CONFIG_NO_HZ */
1885 #ifndef CONFIG_CPUMASK_OFFSTACK
1886 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1888 return set_cpus_allowed_ptr(p, &new_mask);
1893 * Do not use outside of architecture code which knows its limitations.
1895 * sched_clock() has no promise of monotonicity or bounded drift between
1896 * CPUs, use (which you should not) requires disabling IRQs.
1898 * Please use one of the three interfaces below.
1900 extern unsigned long long notrace sched_clock(void);
1902 * See the comment in kernel/sched/clock.c
1904 extern u64 cpu_clock(int cpu);
1905 extern u64 local_clock(void);
1906 extern u64 sched_clock_cpu(int cpu);
1909 extern void sched_clock_init(void);
1911 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1912 static inline void sched_clock_tick(void)
1916 static inline void sched_clock_idle_sleep_event(void)
1920 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1925 * Architectures can set this to 1 if they have specified
1926 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1927 * but then during bootup it turns out that sched_clock()
1928 * is reliable after all:
1930 extern int sched_clock_stable;
1932 extern void sched_clock_tick(void);
1933 extern void sched_clock_idle_sleep_event(void);
1934 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1937 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1939 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1940 * The reason for this explicit opt-in is not to have perf penalty with
1941 * slow sched_clocks.
1943 extern void enable_sched_clock_irqtime(void);
1944 extern void disable_sched_clock_irqtime(void);
1946 static inline void enable_sched_clock_irqtime(void) {}
1947 static inline void disable_sched_clock_irqtime(void) {}
1950 extern unsigned long long
1951 task_sched_runtime(struct task_struct *task);
1953 /* sched_exec is called by processes performing an exec */
1955 extern void sched_exec(void);
1957 #define sched_exec() {}
1960 extern void sched_clock_idle_sleep_event(void);
1961 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1963 #ifdef CONFIG_HOTPLUG_CPU
1964 extern void idle_task_exit(void);
1966 static inline void idle_task_exit(void) {}
1969 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1970 extern void wake_up_idle_cpu(int cpu);
1972 static inline void wake_up_idle_cpu(int cpu) { }
1975 extern unsigned int sysctl_sched_latency;
1976 extern unsigned int sysctl_sched_min_granularity;
1977 extern unsigned int sysctl_sched_wakeup_granularity;
1978 extern unsigned int sysctl_sched_child_runs_first;
1980 enum sched_tunable_scaling {
1981 SCHED_TUNABLESCALING_NONE,
1982 SCHED_TUNABLESCALING_LOG,
1983 SCHED_TUNABLESCALING_LINEAR,
1984 SCHED_TUNABLESCALING_END,
1986 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1988 #ifdef CONFIG_SCHED_DEBUG
1989 extern unsigned int sysctl_sched_migration_cost;
1990 extern unsigned int sysctl_sched_nr_migrate;
1991 extern unsigned int sysctl_sched_time_avg;
1992 extern unsigned int sysctl_timer_migration;
1993 extern unsigned int sysctl_sched_shares_window;
1995 int sched_proc_update_handler(struct ctl_table *table, int write,
1996 void __user *buffer, size_t *length,
1999 #ifdef CONFIG_SCHED_DEBUG
2000 static inline unsigned int get_sysctl_timer_migration(void)
2002 return sysctl_timer_migration;
2005 static inline unsigned int get_sysctl_timer_migration(void)
2010 extern unsigned int sysctl_sched_rt_period;
2011 extern int sysctl_sched_rt_runtime;
2013 int sched_rt_handler(struct ctl_table *table, int write,
2014 void __user *buffer, size_t *lenp,
2017 #ifdef CONFIG_SCHED_AUTOGROUP
2018 extern unsigned int sysctl_sched_autogroup_enabled;
2020 extern void sched_autogroup_create_attach(struct task_struct *p);
2021 extern void sched_autogroup_detach(struct task_struct *p);
2022 extern void sched_autogroup_fork(struct signal_struct *sig);
2023 extern void sched_autogroup_exit(struct signal_struct *sig);
2024 #ifdef CONFIG_PROC_FS
2025 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2026 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2029 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2030 static inline void sched_autogroup_detach(struct task_struct *p) { }
2031 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2032 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2035 #ifdef CONFIG_CFS_BANDWIDTH
2036 extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2039 #ifdef CONFIG_RT_MUTEXES
2040 extern int rt_mutex_getprio(struct task_struct *p);
2041 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2042 extern void rt_mutex_adjust_pi(struct task_struct *p);
2043 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2045 return tsk->pi_blocked_on != NULL;
2048 static inline int rt_mutex_getprio(struct task_struct *p)
2050 return p->normal_prio;
2052 # define rt_mutex_adjust_pi(p) do { } while (0)
2053 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2059 extern bool yield_to(struct task_struct *p, bool preempt);
2060 extern void set_user_nice(struct task_struct *p, long nice);
2061 extern int task_prio(const struct task_struct *p);
2062 extern int task_nice(const struct task_struct *p);
2063 extern int can_nice(const struct task_struct *p, const int nice);
2064 extern int task_curr(const struct task_struct *p);
2065 extern int idle_cpu(int cpu);
2066 extern int sched_setscheduler(struct task_struct *, int,
2067 const struct sched_param *);
2068 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2069 const struct sched_param *);
2070 extern struct task_struct *idle_task(int cpu);
2072 * is_idle_task - is the specified task an idle task?
2073 * @p: the task in question.
2075 static inline bool is_idle_task(const struct task_struct *p)
2079 extern struct task_struct *curr_task(int cpu);
2080 extern void set_curr_task(int cpu, struct task_struct *p);
2085 * The default (Linux) execution domain.
2087 extern struct exec_domain default_exec_domain;
2089 union thread_union {
2090 struct thread_info thread_info;
2091 unsigned long stack[THREAD_SIZE/sizeof(long)];
2094 #ifndef __HAVE_ARCH_KSTACK_END
2095 static inline int kstack_end(void *addr)
2097 /* Reliable end of stack detection:
2098 * Some APM bios versions misalign the stack
2100 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2104 extern union thread_union init_thread_union;
2105 extern struct task_struct init_task;
2107 extern struct mm_struct init_mm;
2109 extern struct pid_namespace init_pid_ns;
2112 * find a task by one of its numerical ids
2114 * find_task_by_pid_ns():
2115 * finds a task by its pid in the specified namespace
2116 * find_task_by_vpid():
2117 * finds a task by its virtual pid
2119 * see also find_vpid() etc in include/linux/pid.h
2122 extern struct task_struct *find_task_by_vpid(pid_t nr);
2123 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2124 struct pid_namespace *ns);
2126 extern void __set_special_pids(struct pid *pid);
2128 /* per-UID process charging. */
2129 extern struct user_struct * alloc_uid(kuid_t);
2130 static inline struct user_struct *get_uid(struct user_struct *u)
2132 atomic_inc(&u->__count);
2135 extern void free_uid(struct user_struct *);
2137 #include <asm/current.h>
2139 extern void xtime_update(unsigned long ticks);
2141 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2142 extern int wake_up_process(struct task_struct *tsk);
2143 extern void wake_up_new_task(struct task_struct *tsk);
2145 extern void kick_process(struct task_struct *tsk);
2147 static inline void kick_process(struct task_struct *tsk) { }
2149 extern void sched_fork(struct task_struct *p);
2150 extern void sched_dead(struct task_struct *p);
2152 extern void proc_caches_init(void);
2153 extern void flush_signals(struct task_struct *);
2154 extern void __flush_signals(struct task_struct *);
2155 extern void ignore_signals(struct task_struct *);
2156 extern void flush_signal_handlers(struct task_struct *, int force_default);
2157 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2159 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2161 unsigned long flags;
2164 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2165 ret = dequeue_signal(tsk, mask, info);
2166 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2171 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2173 extern void unblock_all_signals(void);
2174 extern void release_task(struct task_struct * p);
2175 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2176 extern int force_sigsegv(int, struct task_struct *);
2177 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2178 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2179 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2180 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2181 const struct cred *, u32);
2182 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2183 extern int kill_pid(struct pid *pid, int sig, int priv);
2184 extern int kill_proc_info(int, struct siginfo *, pid_t);
2185 extern __must_check bool do_notify_parent(struct task_struct *, int);
2186 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2187 extern void force_sig(int, struct task_struct *);
2188 extern int send_sig(int, struct task_struct *, int);
2189 extern int zap_other_threads(struct task_struct *p);
2190 extern struct sigqueue *sigqueue_alloc(void);
2191 extern void sigqueue_free(struct sigqueue *);
2192 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2193 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2194 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2196 static inline void restore_saved_sigmask(void)
2198 if (test_and_clear_restore_sigmask())
2199 __set_current_blocked(¤t->saved_sigmask);
2202 static inline sigset_t *sigmask_to_save(void)
2204 sigset_t *res = ¤t->blocked;
2205 if (unlikely(test_restore_sigmask()))
2206 res = ¤t->saved_sigmask;
2210 static inline int kill_cad_pid(int sig, int priv)
2212 return kill_pid(cad_pid, sig, priv);
2215 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2216 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2217 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2218 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2221 * True if we are on the alternate signal stack.
2223 static inline int on_sig_stack(unsigned long sp)
2225 #ifdef CONFIG_STACK_GROWSUP
2226 return sp >= current->sas_ss_sp &&
2227 sp - current->sas_ss_sp < current->sas_ss_size;
2229 return sp > current->sas_ss_sp &&
2230 sp - current->sas_ss_sp <= current->sas_ss_size;
2234 static inline int sas_ss_flags(unsigned long sp)
2236 return (current->sas_ss_size == 0 ? SS_DISABLE
2237 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2241 * Routines for handling mm_structs
2243 extern struct mm_struct * mm_alloc(void);
2245 /* mmdrop drops the mm and the page tables */
2246 extern void __mmdrop(struct mm_struct *);
2247 static inline void mmdrop(struct mm_struct * mm)
2249 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2253 /* mmput gets rid of the mappings and all user-space */
2254 extern void mmput(struct mm_struct *);
2255 /* Grab a reference to a task's mm, if it is not already going away */
2256 extern struct mm_struct *get_task_mm(struct task_struct *task);
2258 * Grab a reference to a task's mm, if it is not already going away
2259 * and ptrace_may_access with the mode parameter passed to it
2262 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2263 /* Remove the current tasks stale references to the old mm_struct */
2264 extern void mm_release(struct task_struct *, struct mm_struct *);
2265 /* Allocate a new mm structure and copy contents from tsk->mm */
2266 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2268 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2269 struct task_struct *, struct pt_regs *);
2270 extern void flush_thread(void);
2271 extern void exit_thread(void);
2273 extern void exit_files(struct task_struct *);
2274 extern void __cleanup_sighand(struct sighand_struct *);
2276 extern void exit_itimers(struct signal_struct *);
2277 extern void flush_itimer_signals(void);
2279 extern void do_group_exit(int);
2281 extern void daemonize(const char *, ...);
2282 extern int allow_signal(int);
2283 extern int disallow_signal(int);
2285 extern int do_execve(const char *,
2286 const char __user * const __user *,
2287 const char __user * const __user *, struct pt_regs *);
2288 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2289 struct task_struct *fork_idle(int);
2290 #ifdef CONFIG_GENERIC_KERNEL_THREAD
2291 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2294 extern void set_task_comm(struct task_struct *tsk, char *from);
2295 extern char *get_task_comm(char *to, struct task_struct *tsk);
2298 void scheduler_ipi(void);
2299 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2301 static inline void scheduler_ipi(void) { }
2302 static inline unsigned long wait_task_inactive(struct task_struct *p,
2309 #define next_task(p) \
2310 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2312 #define for_each_process(p) \
2313 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2315 extern bool current_is_single_threaded(void);
2318 * Careful: do_each_thread/while_each_thread is a double loop so
2319 * 'break' will not work as expected - use goto instead.
2321 #define do_each_thread(g, t) \
2322 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2324 #define while_each_thread(g, t) \
2325 while ((t = next_thread(t)) != g)
2327 static inline int get_nr_threads(struct task_struct *tsk)
2329 return tsk->signal->nr_threads;
2332 static inline bool thread_group_leader(struct task_struct *p)
2334 return p->exit_signal >= 0;
2337 /* Do to the insanities of de_thread it is possible for a process
2338 * to have the pid of the thread group leader without actually being
2339 * the thread group leader. For iteration through the pids in proc
2340 * all we care about is that we have a task with the appropriate
2341 * pid, we don't actually care if we have the right task.
2343 static inline int has_group_leader_pid(struct task_struct *p)
2345 return p->pid == p->tgid;
2349 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2351 return p1->tgid == p2->tgid;
2354 static inline struct task_struct *next_thread(const struct task_struct *p)
2356 return list_entry_rcu(p->thread_group.next,
2357 struct task_struct, thread_group);
2360 static inline int thread_group_empty(struct task_struct *p)
2362 return list_empty(&p->thread_group);
2365 #define delay_group_leader(p) \
2366 (thread_group_leader(p) && !thread_group_empty(p))
2369 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2370 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2371 * pins the final release of task.io_context. Also protects ->cpuset and
2372 * ->cgroup.subsys[]. And ->vfork_done.
2374 * Nests both inside and outside of read_lock(&tasklist_lock).
2375 * It must not be nested with write_lock_irq(&tasklist_lock),
2376 * neither inside nor outside.
2378 static inline void task_lock(struct task_struct *p)
2380 spin_lock(&p->alloc_lock);
2383 static inline void task_unlock(struct task_struct *p)
2385 spin_unlock(&p->alloc_lock);
2388 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2389 unsigned long *flags);
2391 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2392 unsigned long *flags)
2394 struct sighand_struct *ret;
2396 ret = __lock_task_sighand(tsk, flags);
2397 (void)__cond_lock(&tsk->sighand->siglock, ret);
2401 static inline void unlock_task_sighand(struct task_struct *tsk,
2402 unsigned long *flags)
2404 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2407 #ifdef CONFIG_CGROUPS
2408 static inline void threadgroup_change_begin(struct task_struct *tsk)
2410 down_read(&tsk->signal->group_rwsem);
2412 static inline void threadgroup_change_end(struct task_struct *tsk)
2414 up_read(&tsk->signal->group_rwsem);
2418 * threadgroup_lock - lock threadgroup
2419 * @tsk: member task of the threadgroup to lock
2421 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2422 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2423 * perform exec. This is useful for cases where the threadgroup needs to
2424 * stay stable across blockable operations.
2426 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2427 * synchronization. While held, no new task will be added to threadgroup
2428 * and no existing live task will have its PF_EXITING set.
2430 * During exec, a task goes and puts its thread group through unusual
2431 * changes. After de-threading, exclusive access is assumed to resources
2432 * which are usually shared by tasks in the same group - e.g. sighand may
2433 * be replaced with a new one. Also, the exec'ing task takes over group
2434 * leader role including its pid. Exclude these changes while locked by
2435 * grabbing cred_guard_mutex which is used to synchronize exec path.
2437 static inline void threadgroup_lock(struct task_struct *tsk)
2440 * exec uses exit for de-threading nesting group_rwsem inside
2441 * cred_guard_mutex. Grab cred_guard_mutex first.
2443 mutex_lock(&tsk->signal->cred_guard_mutex);
2444 down_write(&tsk->signal->group_rwsem);
2448 * threadgroup_unlock - unlock threadgroup
2449 * @tsk: member task of the threadgroup to unlock
2451 * Reverse threadgroup_lock().
2453 static inline void threadgroup_unlock(struct task_struct *tsk)
2455 up_write(&tsk->signal->group_rwsem);
2456 mutex_unlock(&tsk->signal->cred_guard_mutex);
2459 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2460 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2461 static inline void threadgroup_lock(struct task_struct *tsk) {}
2462 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2465 #ifndef __HAVE_THREAD_FUNCTIONS
2467 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2468 #define task_stack_page(task) ((task)->stack)
2470 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2472 *task_thread_info(p) = *task_thread_info(org);
2473 task_thread_info(p)->task = p;
2476 static inline unsigned long *end_of_stack(struct task_struct *p)
2478 return (unsigned long *)(task_thread_info(p) + 1);
2483 static inline int object_is_on_stack(void *obj)
2485 void *stack = task_stack_page(current);
2487 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2490 extern void thread_info_cache_init(void);
2492 #ifdef CONFIG_DEBUG_STACK_USAGE
2493 static inline unsigned long stack_not_used(struct task_struct *p)
2495 unsigned long *n = end_of_stack(p);
2497 do { /* Skip over canary */
2501 return (unsigned long)n - (unsigned long)end_of_stack(p);
2505 /* set thread flags in other task's structures
2506 * - see asm/thread_info.h for TIF_xxxx flags available
2508 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2510 set_ti_thread_flag(task_thread_info(tsk), flag);
2513 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2515 clear_ti_thread_flag(task_thread_info(tsk), flag);
2518 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2520 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2523 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2525 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2528 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2530 return test_ti_thread_flag(task_thread_info(tsk), flag);
2533 static inline void set_tsk_need_resched(struct task_struct *tsk)
2535 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2538 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2540 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2543 static inline int test_tsk_need_resched(struct task_struct *tsk)
2545 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2548 static inline int restart_syscall(void)
2550 set_tsk_thread_flag(current, TIF_SIGPENDING);
2551 return -ERESTARTNOINTR;
2554 static inline int signal_pending(struct task_struct *p)
2556 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2559 static inline int __fatal_signal_pending(struct task_struct *p)
2561 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2564 static inline int fatal_signal_pending(struct task_struct *p)
2566 return signal_pending(p) && __fatal_signal_pending(p);
2569 static inline int signal_pending_state(long state, struct task_struct *p)
2571 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2573 if (!signal_pending(p))
2576 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2579 static inline int need_resched(void)
2581 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2585 * cond_resched() and cond_resched_lock(): latency reduction via
2586 * explicit rescheduling in places that are safe. The return
2587 * value indicates whether a reschedule was done in fact.
2588 * cond_resched_lock() will drop the spinlock before scheduling,
2589 * cond_resched_softirq() will enable bhs before scheduling.
2591 extern int _cond_resched(void);
2593 #define cond_resched() ({ \
2594 __might_sleep(__FILE__, __LINE__, 0); \
2598 extern int __cond_resched_lock(spinlock_t *lock);
2600 #ifdef CONFIG_PREEMPT_COUNT
2601 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2603 #define PREEMPT_LOCK_OFFSET 0
2606 #define cond_resched_lock(lock) ({ \
2607 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2608 __cond_resched_lock(lock); \
2611 extern int __cond_resched_softirq(void);
2613 #define cond_resched_softirq() ({ \
2614 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2615 __cond_resched_softirq(); \
2619 * Does a critical section need to be broken due to another
2620 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2621 * but a general need for low latency)
2623 static inline int spin_needbreak(spinlock_t *lock)
2625 #ifdef CONFIG_PREEMPT
2626 return spin_is_contended(lock);
2633 * Thread group CPU time accounting.
2635 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2636 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2638 static inline void thread_group_cputime_init(struct signal_struct *sig)
2640 raw_spin_lock_init(&sig->cputimer.lock);
2644 * Reevaluate whether the task has signals pending delivery.
2645 * Wake the task if so.
2646 * This is required every time the blocked sigset_t changes.
2647 * callers must hold sighand->siglock.
2649 extern void recalc_sigpending_and_wake(struct task_struct *t);
2650 extern void recalc_sigpending(void);
2652 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2655 * Wrappers for p->thread_info->cpu access. No-op on UP.
2659 static inline unsigned int task_cpu(const struct task_struct *p)
2661 return task_thread_info(p)->cpu;
2664 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2668 static inline unsigned int task_cpu(const struct task_struct *p)
2673 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2677 #endif /* CONFIG_SMP */
2679 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2680 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2682 extern void normalize_rt_tasks(void);
2684 #ifdef CONFIG_CGROUP_SCHED
2686 extern struct task_group root_task_group;
2688 extern struct task_group *sched_create_group(struct task_group *parent);
2689 extern void sched_destroy_group(struct task_group *tg);
2690 extern void sched_move_task(struct task_struct *tsk);
2691 #ifdef CONFIG_FAIR_GROUP_SCHED
2692 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2693 extern unsigned long sched_group_shares(struct task_group *tg);
2695 #ifdef CONFIG_RT_GROUP_SCHED
2696 extern int sched_group_set_rt_runtime(struct task_group *tg,
2697 long rt_runtime_us);
2698 extern long sched_group_rt_runtime(struct task_group *tg);
2699 extern int sched_group_set_rt_period(struct task_group *tg,
2701 extern long sched_group_rt_period(struct task_group *tg);
2702 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2704 #endif /* CONFIG_CGROUP_SCHED */
2706 extern int task_can_switch_user(struct user_struct *up,
2707 struct task_struct *tsk);
2709 #ifdef CONFIG_TASK_XACCT
2710 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2712 tsk->ioac.rchar += amt;
2715 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2717 tsk->ioac.wchar += amt;
2720 static inline void inc_syscr(struct task_struct *tsk)
2725 static inline void inc_syscw(struct task_struct *tsk)
2730 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2734 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2738 static inline void inc_syscr(struct task_struct *tsk)
2742 static inline void inc_syscw(struct task_struct *tsk)
2747 #ifndef TASK_SIZE_OF
2748 #define TASK_SIZE_OF(tsk) TASK_SIZE
2751 #ifdef CONFIG_MM_OWNER
2752 extern void mm_update_next_owner(struct mm_struct *mm);
2753 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2755 static inline void mm_update_next_owner(struct mm_struct *mm)
2759 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2762 #endif /* CONFIG_MM_OWNER */
2764 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2767 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2770 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2773 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2776 static inline unsigned long rlimit(unsigned int limit)
2778 return task_rlimit(current, limit);
2781 static inline unsigned long rlimit_max(unsigned int limit)
2783 return task_rlimit_max(current, limit);