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>
54 #include <linux/gfp.h>
56 #include <asm/processor.h>
59 struct futex_pi_state;
60 struct robust_list_head;
63 struct perf_event_context;
67 * List of flags we want to share for kernel threads,
68 * if only because they are not used by them anyway.
70 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
73 * These are the constant used to fake the fixed-point load-average
74 * counting. Some notes:
75 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
76 * a load-average precision of 10 bits integer + 11 bits fractional
77 * - if you want to count load-averages more often, you need more
78 * precision, or rounding will get you. With 2-second counting freq,
79 * the EXP_n values would be 1981, 2034 and 2043 if still using only
82 extern unsigned long avenrun[]; /* Load averages */
83 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
85 #define FSHIFT 11 /* nr of bits of precision */
86 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
87 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
88 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
89 #define EXP_5 2014 /* 1/exp(5sec/5min) */
90 #define EXP_15 2037 /* 1/exp(5sec/15min) */
92 #define CALC_LOAD(load,exp,n) \
94 load += n*(FIXED_1-exp); \
97 extern unsigned long total_forks;
98 extern int nr_threads;
99 DECLARE_PER_CPU(unsigned long, process_counts);
100 extern int nr_processes(void);
101 extern unsigned long nr_running(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);
125 * Task state bitmask. NOTE! These bits are also
126 * encoded in fs/proc/array.c: get_task_state().
128 * We have two separate sets of flags: task->state
129 * is about runnability, while task->exit_state are
130 * about the task exiting. Confusing, but this way
131 * modifying one set can't modify the other one by
134 #define TASK_RUNNING 0
135 #define TASK_INTERRUPTIBLE 1
136 #define TASK_UNINTERRUPTIBLE 2
137 #define __TASK_STOPPED 4
138 #define __TASK_TRACED 8
139 /* in tsk->exit_state */
140 #define EXIT_ZOMBIE 16
142 /* in tsk->state again */
144 #define TASK_WAKEKILL 128
145 #define TASK_WAKING 256
146 #define TASK_PARKED 512
147 #define TASK_STATE_MAX 1024
149 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
151 extern char ___assert_task_state[1 - 2*!!(
152 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
154 /* Convenience macros for the sake of set_task_state */
155 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
156 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
157 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
159 /* Convenience macros for the sake of wake_up */
160 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
161 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
163 /* get_task_state() */
164 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
165 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
168 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
169 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
170 #define task_is_dead(task) ((task)->exit_state != 0)
171 #define task_is_stopped_or_traced(task) \
172 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
173 #define task_contributes_to_load(task) \
174 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
175 (task->flags & PF_FROZEN) == 0)
177 #define __set_task_state(tsk, state_value) \
178 do { (tsk)->state = (state_value); } while (0)
179 #define set_task_state(tsk, state_value) \
180 set_mb((tsk)->state, (state_value))
183 * set_current_state() includes a barrier so that the write of current->state
184 * is correctly serialised wrt the caller's subsequent test of whether to
187 * set_current_state(TASK_UNINTERRUPTIBLE);
188 * if (do_i_need_to_sleep())
191 * If the caller does not need such serialisation then use __set_current_state()
193 #define __set_current_state(state_value) \
194 do { current->state = (state_value); } while (0)
195 #define set_current_state(state_value) \
196 set_mb(current->state, (state_value))
198 /* Task command name length */
199 #define TASK_COMM_LEN 16
201 #include <linux/spinlock.h>
204 * This serializes "schedule()" and also protects
205 * the run-queue from deletions/modifications (but
206 * _adding_ to the beginning of the run-queue has
209 extern rwlock_t tasklist_lock;
210 extern spinlock_t mmlist_lock;
214 #ifdef CONFIG_PROVE_RCU
215 extern int lockdep_tasklist_lock_is_held(void);
216 #endif /* #ifdef CONFIG_PROVE_RCU */
218 extern void sched_init(void);
219 extern void sched_init_smp(void);
220 extern asmlinkage void schedule_tail(struct task_struct *prev);
221 extern void init_idle(struct task_struct *idle, int cpu);
222 extern void init_idle_bootup_task(struct task_struct *idle);
224 extern int runqueue_is_locked(int cpu);
226 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
227 extern void nohz_balance_enter_idle(int cpu);
228 extern void set_cpu_sd_state_idle(void);
229 extern int get_nohz_timer_target(void);
231 static inline void nohz_balance_enter_idle(int cpu) { }
232 static inline void set_cpu_sd_state_idle(void) { }
236 * Only dump TASK_* tasks. (0 for all tasks)
238 extern void show_state_filter(unsigned long state_filter);
240 static inline void show_state(void)
242 show_state_filter(0);
245 extern void show_regs(struct pt_regs *);
248 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
249 * task), SP is the stack pointer of the first frame that should be shown in the back
250 * trace (or NULL if the entire call-chain of the task should be shown).
252 extern void show_stack(struct task_struct *task, unsigned long *sp);
254 void io_schedule(void);
255 long io_schedule_timeout(long timeout);
257 extern void cpu_init (void);
258 extern void trap_init(void);
259 extern void update_process_times(int user);
260 extern void scheduler_tick(void);
262 extern void sched_show_task(struct task_struct *p);
264 #ifdef CONFIG_LOCKUP_DETECTOR
265 extern void touch_softlockup_watchdog(void);
266 extern void touch_softlockup_watchdog_sync(void);
267 extern void touch_all_softlockup_watchdogs(void);
268 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
270 size_t *lenp, loff_t *ppos);
271 extern unsigned int softlockup_panic;
272 void lockup_detector_init(void);
274 static inline void touch_softlockup_watchdog(void)
277 static inline void touch_softlockup_watchdog_sync(void)
280 static inline void touch_all_softlockup_watchdogs(void)
283 static inline void lockup_detector_init(void)
288 /* Attach to any functions which should be ignored in wchan output. */
289 #define __sched __attribute__((__section__(".sched.text")))
291 /* Linker adds these: start and end of __sched functions */
292 extern char __sched_text_start[], __sched_text_end[];
294 /* Is this address in the __sched functions? */
295 extern int in_sched_functions(unsigned long addr);
297 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
298 extern signed long schedule_timeout(signed long timeout);
299 extern signed long schedule_timeout_interruptible(signed long timeout);
300 extern signed long schedule_timeout_killable(signed long timeout);
301 extern signed long schedule_timeout_uninterruptible(signed long timeout);
302 asmlinkage void schedule(void);
303 extern void schedule_preempt_disabled(void);
306 struct user_namespace;
309 extern void arch_pick_mmap_layout(struct mm_struct *mm);
311 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
312 unsigned long, unsigned long);
314 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
315 unsigned long len, unsigned long pgoff,
316 unsigned long flags);
318 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
322 extern void set_dumpable(struct mm_struct *mm, int value);
323 extern int get_dumpable(struct mm_struct *mm);
327 #define MMF_DUMPABLE 0 /* core dump is permitted */
328 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
330 #define MMF_DUMPABLE_BITS 2
331 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
333 /* coredump filter bits */
334 #define MMF_DUMP_ANON_PRIVATE 2
335 #define MMF_DUMP_ANON_SHARED 3
336 #define MMF_DUMP_MAPPED_PRIVATE 4
337 #define MMF_DUMP_MAPPED_SHARED 5
338 #define MMF_DUMP_ELF_HEADERS 6
339 #define MMF_DUMP_HUGETLB_PRIVATE 7
340 #define MMF_DUMP_HUGETLB_SHARED 8
342 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
343 #define MMF_DUMP_FILTER_BITS 7
344 #define MMF_DUMP_FILTER_MASK \
345 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
346 #define MMF_DUMP_FILTER_DEFAULT \
347 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
348 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
350 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
351 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
353 # define MMF_DUMP_MASK_DEFAULT_ELF 0
355 /* leave room for more dump flags */
356 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
357 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
358 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
360 #define MMF_HAS_UPROBES 19 /* has uprobes */
361 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
363 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
365 struct sighand_struct {
367 struct k_sigaction action[_NSIG];
369 wait_queue_head_t signalfd_wqh;
372 struct pacct_struct {
375 unsigned long ac_mem;
376 cputime_t ac_utime, ac_stime;
377 unsigned long ac_minflt, ac_majflt;
388 * struct cputime - snaphsot of system and user cputime
389 * @utime: time spent in user mode
390 * @stime: time spent in system mode
392 * Gathers a generic snapshot of user and system time.
400 * struct task_cputime - collected CPU time counts
401 * @utime: time spent in user mode, in &cputime_t units
402 * @stime: time spent in kernel mode, in &cputime_t units
403 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
405 * This is an extension of struct cputime that includes the total runtime
406 * spent by the task from the scheduler point of view.
408 * As a result, this structure groups together three kinds of CPU time
409 * that are tracked for threads and thread groups. Most things considering
410 * CPU time want to group these counts together and treat all three
411 * of them in parallel.
413 struct task_cputime {
416 unsigned long long sum_exec_runtime;
418 /* Alternate field names when used to cache expirations. */
419 #define prof_exp stime
420 #define virt_exp utime
421 #define sched_exp sum_exec_runtime
423 #define INIT_CPUTIME \
424 (struct task_cputime) { \
427 .sum_exec_runtime = 0, \
431 * Disable preemption until the scheduler is running.
432 * Reset by start_kernel()->sched_init()->init_idle().
434 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
435 * before the scheduler is active -- see should_resched().
437 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
440 * struct thread_group_cputimer - thread group interval timer counts
441 * @cputime: thread group interval timers.
442 * @running: non-zero when there are timers running and
443 * @cputime receives updates.
444 * @lock: lock for fields in this struct.
446 * This structure contains the version of task_cputime, above, that is
447 * used for thread group CPU timer calculations.
449 struct thread_group_cputimer {
450 struct task_cputime cputime;
455 #include <linux/rwsem.h>
459 * NOTE! "signal_struct" does not have its own
460 * locking, because a shared signal_struct always
461 * implies a shared sighand_struct, so locking
462 * sighand_struct is always a proper superset of
463 * the locking of signal_struct.
465 struct signal_struct {
470 wait_queue_head_t wait_chldexit; /* for wait4() */
472 /* current thread group signal load-balancing target: */
473 struct task_struct *curr_target;
475 /* shared signal handling: */
476 struct sigpending shared_pending;
478 /* thread group exit support */
481 * - notify group_exit_task when ->count is equal to notify_count
482 * - everyone except group_exit_task is stopped during signal delivery
483 * of fatal signals, group_exit_task processes the signal.
486 struct task_struct *group_exit_task;
488 /* thread group stop support, overloads group_exit_code too */
489 int group_stop_count;
490 unsigned int flags; /* see SIGNAL_* flags below */
493 * PR_SET_CHILD_SUBREAPER marks a process, like a service
494 * manager, to re-parent orphan (double-forking) child processes
495 * to this process instead of 'init'. The service manager is
496 * able to receive SIGCHLD signals and is able to investigate
497 * the process until it calls wait(). All children of this
498 * process will inherit a flag if they should look for a
499 * child_subreaper process at exit.
501 unsigned int is_child_subreaper:1;
502 unsigned int has_child_subreaper:1;
504 /* POSIX.1b Interval Timers */
506 struct list_head posix_timers;
508 /* ITIMER_REAL timer for the process */
509 struct hrtimer real_timer;
510 struct pid *leader_pid;
511 ktime_t it_real_incr;
514 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
515 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
516 * values are defined to 0 and 1 respectively
518 struct cpu_itimer it[2];
521 * Thread group totals for process CPU timers.
522 * See thread_group_cputimer(), et al, for details.
524 struct thread_group_cputimer cputimer;
526 /* Earliest-expiration cache. */
527 struct task_cputime cputime_expires;
529 struct list_head cpu_timers[3];
531 struct pid *tty_old_pgrp;
533 /* boolean value for session group leader */
536 struct tty_struct *tty; /* NULL if no tty */
538 #ifdef CONFIG_SCHED_AUTOGROUP
539 struct autogroup *autogroup;
542 * Cumulative resource counters for dead threads in the group,
543 * and for reaped dead child processes forked by this group.
544 * Live threads maintain their own counters and add to these
545 * in __exit_signal, except for the group leader.
547 cputime_t utime, stime, cutime, cstime;
550 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
551 struct cputime prev_cputime;
553 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
554 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
555 unsigned long inblock, oublock, cinblock, coublock;
556 unsigned long maxrss, cmaxrss;
557 struct task_io_accounting ioac;
560 * Cumulative ns of schedule CPU time fo dead threads in the
561 * group, not including a zombie group leader, (This only differs
562 * from jiffies_to_ns(utime + stime) if sched_clock uses something
563 * other than jiffies.)
565 unsigned long long sum_sched_runtime;
568 * We don't bother to synchronize most readers of this at all,
569 * because there is no reader checking a limit that actually needs
570 * to get both rlim_cur and rlim_max atomically, and either one
571 * alone is a single word that can safely be read normally.
572 * getrlimit/setrlimit use task_lock(current->group_leader) to
573 * protect this instead of the siglock, because they really
574 * have no need to disable irqs.
576 struct rlimit rlim[RLIM_NLIMITS];
578 #ifdef CONFIG_BSD_PROCESS_ACCT
579 struct pacct_struct pacct; /* per-process accounting information */
581 #ifdef CONFIG_TASKSTATS
582 struct taskstats *stats;
586 unsigned audit_tty_log_passwd;
587 struct tty_audit_buf *tty_audit_buf;
589 #ifdef CONFIG_CGROUPS
591 * group_rwsem prevents new tasks from entering the threadgroup and
592 * member tasks from exiting,a more specifically, setting of
593 * PF_EXITING. fork and exit paths are protected with this rwsem
594 * using threadgroup_change_begin/end(). Users which require
595 * threadgroup to remain stable should use threadgroup_[un]lock()
596 * which also takes care of exec path. Currently, cgroup is the
599 struct rw_semaphore group_rwsem;
602 oom_flags_t oom_flags;
603 short oom_score_adj; /* OOM kill score adjustment */
604 short oom_score_adj_min; /* OOM kill score adjustment min value.
605 * Only settable by CAP_SYS_RESOURCE. */
607 struct mutex cred_guard_mutex; /* guard against foreign influences on
608 * credential calculations
609 * (notably. ptrace) */
613 * Bits in flags field of signal_struct.
615 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
616 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
617 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
618 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
620 * Pending notifications to parent.
622 #define SIGNAL_CLD_STOPPED 0x00000010
623 #define SIGNAL_CLD_CONTINUED 0x00000020
624 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
626 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
628 /* If true, all threads except ->group_exit_task have pending SIGKILL */
629 static inline int signal_group_exit(const struct signal_struct *sig)
631 return (sig->flags & SIGNAL_GROUP_EXIT) ||
632 (sig->group_exit_task != NULL);
636 * Some day this will be a full-fledged user tracking system..
639 atomic_t __count; /* reference count */
640 atomic_t processes; /* How many processes does this user have? */
641 atomic_t files; /* How many open files does this user have? */
642 atomic_t sigpending; /* How many pending signals does this user have? */
643 #ifdef CONFIG_INOTIFY_USER
644 atomic_t inotify_watches; /* How many inotify watches does this user have? */
645 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
647 #ifdef CONFIG_FANOTIFY
648 atomic_t fanotify_listeners;
651 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
653 #ifdef CONFIG_POSIX_MQUEUE
654 /* protected by mq_lock */
655 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
657 unsigned long locked_shm; /* How many pages of mlocked shm ? */
660 struct key *uid_keyring; /* UID specific keyring */
661 struct key *session_keyring; /* UID's default session keyring */
664 /* Hash table maintenance information */
665 struct hlist_node uidhash_node;
668 #ifdef CONFIG_PERF_EVENTS
669 atomic_long_t locked_vm;
673 extern int uids_sysfs_init(void);
675 extern struct user_struct *find_user(kuid_t);
677 extern struct user_struct root_user;
678 #define INIT_USER (&root_user)
681 struct backing_dev_info;
682 struct reclaim_state;
684 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
686 /* cumulative counters */
687 unsigned long pcount; /* # of times run on this cpu */
688 unsigned long long run_delay; /* time spent waiting on a runqueue */
691 unsigned long long last_arrival,/* when we last ran on a cpu */
692 last_queued; /* when we were last queued to run */
694 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
696 #ifdef CONFIG_TASK_DELAY_ACCT
697 struct task_delay_info {
699 unsigned int flags; /* Private per-task flags */
701 /* For each stat XXX, add following, aligned appropriately
703 * struct timespec XXX_start, XXX_end;
707 * Atomicity of updates to XXX_delay, XXX_count protected by
708 * single lock above (split into XXX_lock if contention is an issue).
712 * XXX_count is incremented on every XXX operation, the delay
713 * associated with the operation is added to XXX_delay.
714 * XXX_delay contains the accumulated delay time in nanoseconds.
716 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
717 u64 blkio_delay; /* wait for sync block io completion */
718 u64 swapin_delay; /* wait for swapin block io completion */
719 u32 blkio_count; /* total count of the number of sync block */
720 /* io operations performed */
721 u32 swapin_count; /* total count of the number of swapin block */
722 /* io operations performed */
724 struct timespec freepages_start, freepages_end;
725 u64 freepages_delay; /* wait for memory reclaim */
726 u32 freepages_count; /* total count of memory reclaim */
728 #endif /* CONFIG_TASK_DELAY_ACCT */
730 static inline int sched_info_on(void)
732 #ifdef CONFIG_SCHEDSTATS
734 #elif defined(CONFIG_TASK_DELAY_ACCT)
735 extern int delayacct_on;
750 * Increase resolution of cpu_power calculations
752 #define SCHED_POWER_SHIFT 10
753 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
756 * sched-domains (multiprocessor balancing) declarations:
759 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
760 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
761 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
762 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
763 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
764 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
765 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
766 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
767 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
768 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
769 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
770 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
772 extern int __weak arch_sd_sibiling_asym_packing(void);
774 struct sched_domain_attr {
775 int relax_domain_level;
778 #define SD_ATTR_INIT (struct sched_domain_attr) { \
779 .relax_domain_level = -1, \
782 extern int sched_domain_level_max;
786 struct sched_domain {
787 /* These fields must be setup */
788 struct sched_domain *parent; /* top domain must be null terminated */
789 struct sched_domain *child; /* bottom domain must be null terminated */
790 struct sched_group *groups; /* the balancing groups of the domain */
791 unsigned long min_interval; /* Minimum balance interval ms */
792 unsigned long max_interval; /* Maximum balance interval ms */
793 unsigned int busy_factor; /* less balancing by factor if busy */
794 unsigned int imbalance_pct; /* No balance until over watermark */
795 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
796 unsigned int busy_idx;
797 unsigned int idle_idx;
798 unsigned int newidle_idx;
799 unsigned int wake_idx;
800 unsigned int forkexec_idx;
801 unsigned int smt_gain;
803 int nohz_idle; /* NOHZ IDLE status */
804 int flags; /* See SD_* */
807 /* Runtime fields. */
808 unsigned long last_balance; /* init to jiffies. units in jiffies */
809 unsigned int balance_interval; /* initialise to 1. units in ms. */
810 unsigned int nr_balance_failed; /* initialise to 0 */
814 #ifdef CONFIG_SCHEDSTATS
815 /* load_balance() stats */
816 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
817 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
818 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
819 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
820 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
821 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
822 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
823 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
825 /* Active load balancing */
826 unsigned int alb_count;
827 unsigned int alb_failed;
828 unsigned int alb_pushed;
830 /* SD_BALANCE_EXEC stats */
831 unsigned int sbe_count;
832 unsigned int sbe_balanced;
833 unsigned int sbe_pushed;
835 /* SD_BALANCE_FORK stats */
836 unsigned int sbf_count;
837 unsigned int sbf_balanced;
838 unsigned int sbf_pushed;
840 /* try_to_wake_up() stats */
841 unsigned int ttwu_wake_remote;
842 unsigned int ttwu_move_affine;
843 unsigned int ttwu_move_balance;
845 #ifdef CONFIG_SCHED_DEBUG
849 void *private; /* used during construction */
850 struct rcu_head rcu; /* used during destruction */
853 unsigned int span_weight;
855 * Span of all CPUs in this domain.
857 * NOTE: this field is variable length. (Allocated dynamically
858 * by attaching extra space to the end of the structure,
859 * depending on how many CPUs the kernel has booted up with)
861 unsigned long span[0];
864 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
866 return to_cpumask(sd->span);
869 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
870 struct sched_domain_attr *dattr_new);
872 /* Allocate an array of sched domains, for partition_sched_domains(). */
873 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
874 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
876 bool cpus_share_cache(int this_cpu, int that_cpu);
878 #else /* CONFIG_SMP */
880 struct sched_domain_attr;
883 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
884 struct sched_domain_attr *dattr_new)
888 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
893 #endif /* !CONFIG_SMP */
896 struct io_context; /* See blkdev.h */
899 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
900 extern void prefetch_stack(struct task_struct *t);
902 static inline void prefetch_stack(struct task_struct *t) { }
905 struct audit_context; /* See audit.c */
907 struct pipe_inode_info;
908 struct uts_namespace;
911 unsigned long weight, inv_weight;
916 * These sums represent an infinite geometric series and so are bound
917 * above by 1024/(1-y). Thus we only need a u32 to store them for all
918 * choices of y < 1-2^(-32)*1024.
920 u32 runnable_avg_sum, runnable_avg_period;
921 u64 last_runnable_update;
923 unsigned long load_avg_contrib;
926 #ifdef CONFIG_SCHEDSTATS
927 struct sched_statistics {
937 s64 sum_sleep_runtime;
944 u64 nr_migrations_cold;
945 u64 nr_failed_migrations_affine;
946 u64 nr_failed_migrations_running;
947 u64 nr_failed_migrations_hot;
948 u64 nr_forced_migrations;
952 u64 nr_wakeups_migrate;
953 u64 nr_wakeups_local;
954 u64 nr_wakeups_remote;
955 u64 nr_wakeups_affine;
956 u64 nr_wakeups_affine_attempts;
957 u64 nr_wakeups_passive;
962 struct sched_entity {
963 struct load_weight load; /* for load-balancing */
964 struct rb_node run_node;
965 struct list_head group_node;
969 u64 sum_exec_runtime;
971 u64 prev_sum_exec_runtime;
975 #ifdef CONFIG_SCHEDSTATS
976 struct sched_statistics statistics;
979 #ifdef CONFIG_FAIR_GROUP_SCHED
980 struct sched_entity *parent;
981 /* rq on which this entity is (to be) queued: */
982 struct cfs_rq *cfs_rq;
983 /* rq "owned" by this entity/group: */
988 /* Per-entity load-tracking */
989 struct sched_avg avg;
993 struct sched_rt_entity {
994 struct list_head run_list;
995 unsigned long timeout;
996 unsigned long watchdog_stamp;
997 unsigned int time_slice;
999 struct sched_rt_entity *back;
1000 #ifdef CONFIG_RT_GROUP_SCHED
1001 struct sched_rt_entity *parent;
1002 /* rq on which this entity is (to be) queued: */
1003 struct rt_rq *rt_rq;
1004 /* rq "owned" by this entity/group: */
1012 enum perf_event_task_context {
1013 perf_invalid_context = -1,
1014 perf_hw_context = 0,
1016 perf_nr_task_contexts,
1019 struct task_struct {
1020 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1023 unsigned int flags; /* per process flags, defined below */
1024 unsigned int ptrace;
1027 struct llist_node wake_entry;
1029 struct task_struct *last_wakee;
1030 unsigned long wakee_flips;
1031 unsigned long wakee_flip_decay_ts;
1035 int prio, static_prio, normal_prio;
1036 unsigned int rt_priority;
1037 const struct sched_class *sched_class;
1038 struct sched_entity se;
1039 struct sched_rt_entity rt;
1040 #ifdef CONFIG_CGROUP_SCHED
1041 struct task_group *sched_task_group;
1044 #ifdef CONFIG_PREEMPT_NOTIFIERS
1045 /* list of struct preempt_notifier: */
1046 struct hlist_head preempt_notifiers;
1050 * fpu_counter contains the number of consecutive context switches
1051 * that the FPU is used. If this is over a threshold, the lazy fpu
1052 * saving becomes unlazy to save the trap. This is an unsigned char
1053 * so that after 256 times the counter wraps and the behavior turns
1054 * lazy again; this to deal with bursty apps that only use FPU for
1057 unsigned char fpu_counter;
1058 #ifdef CONFIG_BLK_DEV_IO_TRACE
1059 unsigned int btrace_seq;
1062 unsigned int policy;
1063 int nr_cpus_allowed;
1064 cpumask_t cpus_allowed;
1066 #ifdef CONFIG_PREEMPT_RCU
1067 int rcu_read_lock_nesting;
1068 char rcu_read_unlock_special;
1069 struct list_head rcu_node_entry;
1070 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1071 #ifdef CONFIG_TREE_PREEMPT_RCU
1072 struct rcu_node *rcu_blocked_node;
1073 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1074 #ifdef CONFIG_RCU_BOOST
1075 struct rt_mutex *rcu_boost_mutex;
1076 #endif /* #ifdef CONFIG_RCU_BOOST */
1078 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1079 struct sched_info sched_info;
1082 struct list_head tasks;
1084 struct plist_node pushable_tasks;
1087 struct mm_struct *mm, *active_mm;
1088 #ifdef CONFIG_COMPAT_BRK
1089 unsigned brk_randomized:1;
1091 #if defined(SPLIT_RSS_COUNTING)
1092 struct task_rss_stat rss_stat;
1096 int exit_code, exit_signal;
1097 int pdeath_signal; /* The signal sent when the parent dies */
1098 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1100 /* Used for emulating ABI behavior of previous Linux versions */
1101 unsigned int personality;
1103 unsigned did_exec:1;
1104 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1106 unsigned in_iowait:1;
1108 /* task may not gain privileges */
1109 unsigned no_new_privs:1;
1111 /* Revert to default priority/policy when forking */
1112 unsigned sched_reset_on_fork:1;
1113 unsigned sched_contributes_to_load:1;
1118 #ifdef CONFIG_CC_STACKPROTECTOR
1119 /* Canary value for the -fstack-protector gcc feature */
1120 unsigned long stack_canary;
1123 * pointers to (original) parent process, youngest child, younger sibling,
1124 * older sibling, respectively. (p->father can be replaced with
1125 * p->real_parent->pid)
1127 struct task_struct __rcu *real_parent; /* real parent process */
1128 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1130 * children/sibling forms the list of my natural children
1132 struct list_head children; /* list of my children */
1133 struct list_head sibling; /* linkage in my parent's children list */
1134 struct task_struct *group_leader; /* threadgroup leader */
1137 * ptraced is the list of tasks this task is using ptrace on.
1138 * This includes both natural children and PTRACE_ATTACH targets.
1139 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1141 struct list_head ptraced;
1142 struct list_head ptrace_entry;
1144 /* PID/PID hash table linkage. */
1145 struct pid_link pids[PIDTYPE_MAX];
1146 struct list_head thread_group;
1148 struct completion *vfork_done; /* for vfork() */
1149 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1150 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1152 cputime_t utime, stime, utimescaled, stimescaled;
1154 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1155 struct cputime prev_cputime;
1157 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1158 seqlock_t vtime_seqlock;
1159 unsigned long long vtime_snap;
1164 } vtime_snap_whence;
1166 unsigned long nvcsw, nivcsw; /* context switch counts */
1167 struct timespec start_time; /* monotonic time */
1168 struct timespec real_start_time; /* boot based time */
1169 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1170 unsigned long min_flt, maj_flt;
1172 struct task_cputime cputime_expires;
1173 struct list_head cpu_timers[3];
1175 /* process credentials */
1176 const struct cred __rcu *real_cred; /* objective and real subjective task
1177 * credentials (COW) */
1178 const struct cred __rcu *cred; /* effective (overridable) subjective task
1179 * credentials (COW) */
1180 char comm[TASK_COMM_LEN]; /* executable name excluding path
1181 - access with [gs]et_task_comm (which lock
1182 it with task_lock())
1183 - initialized normally by setup_new_exec */
1184 /* file system info */
1185 int link_count, total_link_count;
1186 #ifdef CONFIG_SYSVIPC
1188 struct sysv_sem sysvsem;
1190 #ifdef CONFIG_DETECT_HUNG_TASK
1191 /* hung task detection */
1192 unsigned long last_switch_count;
1194 /* CPU-specific state of this task */
1195 struct thread_struct thread;
1196 /* filesystem information */
1197 struct fs_struct *fs;
1198 /* open file information */
1199 struct files_struct *files;
1201 struct nsproxy *nsproxy;
1202 /* signal handlers */
1203 struct signal_struct *signal;
1204 struct sighand_struct *sighand;
1206 sigset_t blocked, real_blocked;
1207 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1208 struct sigpending pending;
1210 unsigned long sas_ss_sp;
1212 int (*notifier)(void *priv);
1213 void *notifier_data;
1214 sigset_t *notifier_mask;
1215 struct callback_head *task_works;
1217 struct audit_context *audit_context;
1218 #ifdef CONFIG_AUDITSYSCALL
1220 unsigned int sessionid;
1222 struct seccomp seccomp;
1224 /* Thread group tracking */
1227 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1229 spinlock_t alloc_lock;
1231 /* Protection of the PI data structures: */
1232 raw_spinlock_t pi_lock;
1234 #ifdef CONFIG_RT_MUTEXES
1235 /* PI waiters blocked on a rt_mutex held by this task */
1236 struct plist_head pi_waiters;
1237 /* Deadlock detection and priority inheritance handling */
1238 struct rt_mutex_waiter *pi_blocked_on;
1241 #ifdef CONFIG_DEBUG_MUTEXES
1242 /* mutex deadlock detection */
1243 struct mutex_waiter *blocked_on;
1245 #ifdef CONFIG_TRACE_IRQFLAGS
1246 unsigned int irq_events;
1247 unsigned long hardirq_enable_ip;
1248 unsigned long hardirq_disable_ip;
1249 unsigned int hardirq_enable_event;
1250 unsigned int hardirq_disable_event;
1251 int hardirqs_enabled;
1252 int hardirq_context;
1253 unsigned long softirq_disable_ip;
1254 unsigned long softirq_enable_ip;
1255 unsigned int softirq_disable_event;
1256 unsigned int softirq_enable_event;
1257 int softirqs_enabled;
1258 int softirq_context;
1260 #ifdef CONFIG_LOCKDEP
1261 # define MAX_LOCK_DEPTH 48UL
1264 unsigned int lockdep_recursion;
1265 struct held_lock held_locks[MAX_LOCK_DEPTH];
1266 gfp_t lockdep_reclaim_gfp;
1269 /* journalling filesystem info */
1272 /* stacked block device info */
1273 struct bio_list *bio_list;
1276 /* stack plugging */
1277 struct blk_plug *plug;
1281 struct reclaim_state *reclaim_state;
1283 struct backing_dev_info *backing_dev_info;
1285 struct io_context *io_context;
1287 unsigned long ptrace_message;
1288 siginfo_t *last_siginfo; /* For ptrace use. */
1289 struct task_io_accounting ioac;
1290 #if defined(CONFIG_TASK_XACCT)
1291 u64 acct_rss_mem1; /* accumulated rss usage */
1292 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1293 cputime_t acct_timexpd; /* stime + utime since last update */
1295 #ifdef CONFIG_CPUSETS
1296 nodemask_t mems_allowed; /* Protected by alloc_lock */
1297 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1298 int cpuset_mem_spread_rotor;
1299 int cpuset_slab_spread_rotor;
1301 #ifdef CONFIG_CGROUPS
1302 /* Control Group info protected by css_set_lock */
1303 struct css_set __rcu *cgroups;
1304 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1305 struct list_head cg_list;
1308 struct robust_list_head __user *robust_list;
1309 #ifdef CONFIG_COMPAT
1310 struct compat_robust_list_head __user *compat_robust_list;
1312 struct list_head pi_state_list;
1313 struct futex_pi_state *pi_state_cache;
1315 #ifdef CONFIG_PERF_EVENTS
1316 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1317 struct mutex perf_event_mutex;
1318 struct list_head perf_event_list;
1321 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1323 short pref_node_fork;
1325 #ifdef CONFIG_NUMA_BALANCING
1327 int numa_migrate_seq;
1328 unsigned int numa_scan_period;
1329 u64 node_stamp; /* migration stamp */
1330 struct callback_head numa_work;
1331 #endif /* CONFIG_NUMA_BALANCING */
1333 struct rcu_head rcu;
1336 * cache last used pipe for splice
1338 struct pipe_inode_info *splice_pipe;
1340 struct page_frag task_frag;
1342 #ifdef CONFIG_TASK_DELAY_ACCT
1343 struct task_delay_info *delays;
1345 #ifdef CONFIG_FAULT_INJECTION
1349 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1350 * balance_dirty_pages() for some dirty throttling pause
1353 int nr_dirtied_pause;
1354 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1356 #ifdef CONFIG_LATENCYTOP
1357 int latency_record_count;
1358 struct latency_record latency_record[LT_SAVECOUNT];
1361 * time slack values; these are used to round up poll() and
1362 * select() etc timeout values. These are in nanoseconds.
1364 unsigned long timer_slack_ns;
1365 unsigned long default_timer_slack_ns;
1367 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1368 /* Index of current stored address in ret_stack */
1370 /* Stack of return addresses for return function tracing */
1371 struct ftrace_ret_stack *ret_stack;
1372 /* time stamp for last schedule */
1373 unsigned long long ftrace_timestamp;
1375 * Number of functions that haven't been traced
1376 * because of depth overrun.
1378 atomic_t trace_overrun;
1379 /* Pause for the tracing */
1380 atomic_t tracing_graph_pause;
1382 #ifdef CONFIG_TRACING
1383 /* state flags for use by tracers */
1384 unsigned long trace;
1385 /* bitmask and counter of trace recursion */
1386 unsigned long trace_recursion;
1387 #endif /* CONFIG_TRACING */
1388 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1389 struct memcg_batch_info {
1390 int do_batch; /* incremented when batch uncharge started */
1391 struct mem_cgroup *memcg; /* target memcg of uncharge */
1392 unsigned long nr_pages; /* uncharged usage */
1393 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1395 unsigned int memcg_kmem_skip_account;
1396 struct memcg_oom_info {
1397 unsigned int may_oom:1;
1398 unsigned int in_memcg_oom:1;
1399 unsigned int oom_locked:1;
1401 struct mem_cgroup *wait_on_memcg;
1404 #ifdef CONFIG_UPROBES
1405 struct uprobe_task *utask;
1407 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1408 unsigned int sequential_io;
1409 unsigned int sequential_io_avg;
1413 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1414 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1416 #ifdef CONFIG_NUMA_BALANCING
1417 extern void task_numa_fault(int node, int pages, bool migrated);
1418 extern void set_numabalancing_state(bool enabled);
1420 static inline void task_numa_fault(int node, int pages, bool migrated)
1423 static inline void set_numabalancing_state(bool enabled)
1428 static inline struct pid *task_pid(struct task_struct *task)
1430 return task->pids[PIDTYPE_PID].pid;
1433 static inline struct pid *task_tgid(struct task_struct *task)
1435 return task->group_leader->pids[PIDTYPE_PID].pid;
1439 * Without tasklist or rcu lock it is not safe to dereference
1440 * the result of task_pgrp/task_session even if task == current,
1441 * we can race with another thread doing sys_setsid/sys_setpgid.
1443 static inline struct pid *task_pgrp(struct task_struct *task)
1445 return task->group_leader->pids[PIDTYPE_PGID].pid;
1448 static inline struct pid *task_session(struct task_struct *task)
1450 return task->group_leader->pids[PIDTYPE_SID].pid;
1453 struct pid_namespace;
1456 * the helpers to get the task's different pids as they are seen
1457 * from various namespaces
1459 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1460 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1462 * task_xid_nr_ns() : id seen from the ns specified;
1464 * set_task_vxid() : assigns a virtual id to a task;
1466 * see also pid_nr() etc in include/linux/pid.h
1468 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1469 struct pid_namespace *ns);
1471 static inline pid_t task_pid_nr(struct task_struct *tsk)
1476 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1477 struct pid_namespace *ns)
1479 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1482 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1484 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1488 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1493 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1495 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1497 return pid_vnr(task_tgid(tsk));
1501 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1502 struct pid_namespace *ns)
1504 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1507 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1509 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1513 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1514 struct pid_namespace *ns)
1516 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1519 static inline pid_t task_session_vnr(struct task_struct *tsk)
1521 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1524 /* obsolete, do not use */
1525 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1527 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1531 * pid_alive - check that a task structure is not stale
1532 * @p: Task structure to be checked.
1534 * Test if a process is not yet dead (at most zombie state)
1535 * If pid_alive fails, then pointers within the task structure
1536 * can be stale and must not be dereferenced.
1538 * Return: 1 if the process is alive. 0 otherwise.
1540 static inline int pid_alive(struct task_struct *p)
1542 return p->pids[PIDTYPE_PID].pid != NULL;
1546 * is_global_init - check if a task structure is init
1547 * @tsk: Task structure to be checked.
1549 * Check if a task structure is the first user space task the kernel created.
1551 * Return: 1 if the task structure is init. 0 otherwise.
1553 static inline int is_global_init(struct task_struct *tsk)
1555 return tsk->pid == 1;
1558 extern struct pid *cad_pid;
1560 extern void free_task(struct task_struct *tsk);
1561 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1563 extern void __put_task_struct(struct task_struct *t);
1565 static inline void put_task_struct(struct task_struct *t)
1567 if (atomic_dec_and_test(&t->usage))
1568 __put_task_struct(t);
1571 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1572 extern void task_cputime(struct task_struct *t,
1573 cputime_t *utime, cputime_t *stime);
1574 extern void task_cputime_scaled(struct task_struct *t,
1575 cputime_t *utimescaled, cputime_t *stimescaled);
1576 extern cputime_t task_gtime(struct task_struct *t);
1578 static inline void task_cputime(struct task_struct *t,
1579 cputime_t *utime, cputime_t *stime)
1587 static inline void task_cputime_scaled(struct task_struct *t,
1588 cputime_t *utimescaled,
1589 cputime_t *stimescaled)
1592 *utimescaled = t->utimescaled;
1594 *stimescaled = t->stimescaled;
1597 static inline cputime_t task_gtime(struct task_struct *t)
1602 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1603 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1608 #define PF_EXITING 0x00000004 /* getting shut down */
1609 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1610 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1611 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1612 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1613 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1614 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1615 #define PF_DUMPCORE 0x00000200 /* dumped core */
1616 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1617 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1618 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1619 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1620 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1621 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1622 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1623 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1624 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1625 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1626 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1627 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1628 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1629 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1630 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1631 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1632 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1633 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1634 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1635 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1636 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1637 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1640 * Only the _current_ task can read/write to tsk->flags, but other
1641 * tasks can access tsk->flags in readonly mode for example
1642 * with tsk_used_math (like during threaded core dumping).
1643 * There is however an exception to this rule during ptrace
1644 * or during fork: the ptracer task is allowed to write to the
1645 * child->flags of its traced child (same goes for fork, the parent
1646 * can write to the child->flags), because we're guaranteed the
1647 * child is not running and in turn not changing child->flags
1648 * at the same time the parent does it.
1650 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1651 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1652 #define clear_used_math() clear_stopped_child_used_math(current)
1653 #define set_used_math() set_stopped_child_used_math(current)
1654 #define conditional_stopped_child_used_math(condition, child) \
1655 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1656 #define conditional_used_math(condition) \
1657 conditional_stopped_child_used_math(condition, current)
1658 #define copy_to_stopped_child_used_math(child) \
1659 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1660 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1661 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1662 #define used_math() tsk_used_math(current)
1664 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1665 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1667 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1672 static inline unsigned int memalloc_noio_save(void)
1674 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1675 current->flags |= PF_MEMALLOC_NOIO;
1679 static inline void memalloc_noio_restore(unsigned int flags)
1681 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1685 * task->jobctl flags
1687 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1689 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1690 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1691 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1692 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1693 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1694 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1695 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1697 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1698 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1699 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1700 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1701 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1702 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1703 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1705 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1706 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1708 extern bool task_set_jobctl_pending(struct task_struct *task,
1710 extern void task_clear_jobctl_trapping(struct task_struct *task);
1711 extern void task_clear_jobctl_pending(struct task_struct *task,
1714 #ifdef CONFIG_PREEMPT_RCU
1716 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1717 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1719 static inline void rcu_copy_process(struct task_struct *p)
1721 p->rcu_read_lock_nesting = 0;
1722 p->rcu_read_unlock_special = 0;
1723 #ifdef CONFIG_TREE_PREEMPT_RCU
1724 p->rcu_blocked_node = NULL;
1725 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1726 #ifdef CONFIG_RCU_BOOST
1727 p->rcu_boost_mutex = NULL;
1728 #endif /* #ifdef CONFIG_RCU_BOOST */
1729 INIT_LIST_HEAD(&p->rcu_node_entry);
1734 static inline void rcu_copy_process(struct task_struct *p)
1740 static inline void tsk_restore_flags(struct task_struct *task,
1741 unsigned long orig_flags, unsigned long flags)
1743 task->flags &= ~flags;
1744 task->flags |= orig_flags & flags;
1748 extern void do_set_cpus_allowed(struct task_struct *p,
1749 const struct cpumask *new_mask);
1751 extern int set_cpus_allowed_ptr(struct task_struct *p,
1752 const struct cpumask *new_mask);
1754 static inline void do_set_cpus_allowed(struct task_struct *p,
1755 const struct cpumask *new_mask)
1758 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1759 const struct cpumask *new_mask)
1761 if (!cpumask_test_cpu(0, new_mask))
1767 #ifdef CONFIG_NO_HZ_COMMON
1768 void calc_load_enter_idle(void);
1769 void calc_load_exit_idle(void);
1771 static inline void calc_load_enter_idle(void) { }
1772 static inline void calc_load_exit_idle(void) { }
1773 #endif /* CONFIG_NO_HZ_COMMON */
1775 #ifndef CONFIG_CPUMASK_OFFSTACK
1776 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1778 return set_cpus_allowed_ptr(p, &new_mask);
1783 * Do not use outside of architecture code which knows its limitations.
1785 * sched_clock() has no promise of monotonicity or bounded drift between
1786 * CPUs, use (which you should not) requires disabling IRQs.
1788 * Please use one of the three interfaces below.
1790 extern unsigned long long notrace sched_clock(void);
1792 * See the comment in kernel/sched/clock.c
1794 extern u64 cpu_clock(int cpu);
1795 extern u64 local_clock(void);
1796 extern u64 sched_clock_cpu(int cpu);
1799 extern void sched_clock_init(void);
1801 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1802 static inline void sched_clock_tick(void)
1806 static inline void sched_clock_idle_sleep_event(void)
1810 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1815 * Architectures can set this to 1 if they have specified
1816 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1817 * but then during bootup it turns out that sched_clock()
1818 * is reliable after all:
1820 extern int sched_clock_stable;
1822 extern void sched_clock_tick(void);
1823 extern void sched_clock_idle_sleep_event(void);
1824 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1827 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1829 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1830 * The reason for this explicit opt-in is not to have perf penalty with
1831 * slow sched_clocks.
1833 extern void enable_sched_clock_irqtime(void);
1834 extern void disable_sched_clock_irqtime(void);
1836 static inline void enable_sched_clock_irqtime(void) {}
1837 static inline void disable_sched_clock_irqtime(void) {}
1840 extern unsigned long long
1841 task_sched_runtime(struct task_struct *task);
1843 /* sched_exec is called by processes performing an exec */
1845 extern void sched_exec(void);
1847 #define sched_exec() {}
1850 extern void sched_clock_idle_sleep_event(void);
1851 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1853 #ifdef CONFIG_HOTPLUG_CPU
1854 extern void idle_task_exit(void);
1856 static inline void idle_task_exit(void) {}
1859 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1860 extern void wake_up_nohz_cpu(int cpu);
1862 static inline void wake_up_nohz_cpu(int cpu) { }
1865 #ifdef CONFIG_NO_HZ_FULL
1866 extern bool sched_can_stop_tick(void);
1867 extern u64 scheduler_tick_max_deferment(void);
1869 static inline bool sched_can_stop_tick(void) { return false; }
1872 #ifdef CONFIG_SCHED_AUTOGROUP
1873 extern void sched_autogroup_create_attach(struct task_struct *p);
1874 extern void sched_autogroup_detach(struct task_struct *p);
1875 extern void sched_autogroup_fork(struct signal_struct *sig);
1876 extern void sched_autogroup_exit(struct signal_struct *sig);
1877 #ifdef CONFIG_PROC_FS
1878 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1879 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
1882 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1883 static inline void sched_autogroup_detach(struct task_struct *p) { }
1884 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1885 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1888 extern bool yield_to(struct task_struct *p, bool preempt);
1889 extern void set_user_nice(struct task_struct *p, long nice);
1890 extern int task_prio(const struct task_struct *p);
1891 extern int task_nice(const struct task_struct *p);
1892 extern int can_nice(const struct task_struct *p, const int nice);
1893 extern int task_curr(const struct task_struct *p);
1894 extern int idle_cpu(int cpu);
1895 extern int sched_setscheduler(struct task_struct *, int,
1896 const struct sched_param *);
1897 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1898 const struct sched_param *);
1899 extern struct task_struct *idle_task(int cpu);
1901 * is_idle_task - is the specified task an idle task?
1902 * @p: the task in question.
1904 * Return: 1 if @p is an idle task. 0 otherwise.
1906 static inline bool is_idle_task(const struct task_struct *p)
1910 extern struct task_struct *curr_task(int cpu);
1911 extern void set_curr_task(int cpu, struct task_struct *p);
1916 * The default (Linux) execution domain.
1918 extern struct exec_domain default_exec_domain;
1920 union thread_union {
1921 struct thread_info thread_info;
1922 unsigned long stack[THREAD_SIZE/sizeof(long)];
1925 #ifndef __HAVE_ARCH_KSTACK_END
1926 static inline int kstack_end(void *addr)
1928 /* Reliable end of stack detection:
1929 * Some APM bios versions misalign the stack
1931 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1935 extern union thread_union init_thread_union;
1936 extern struct task_struct init_task;
1938 extern struct mm_struct init_mm;
1940 extern struct pid_namespace init_pid_ns;
1943 * find a task by one of its numerical ids
1945 * find_task_by_pid_ns():
1946 * finds a task by its pid in the specified namespace
1947 * find_task_by_vpid():
1948 * finds a task by its virtual pid
1950 * see also find_vpid() etc in include/linux/pid.h
1953 extern struct task_struct *find_task_by_vpid(pid_t nr);
1954 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1955 struct pid_namespace *ns);
1957 /* per-UID process charging. */
1958 extern struct user_struct * alloc_uid(kuid_t);
1959 static inline struct user_struct *get_uid(struct user_struct *u)
1961 atomic_inc(&u->__count);
1964 extern void free_uid(struct user_struct *);
1966 #include <asm/current.h>
1968 extern void xtime_update(unsigned long ticks);
1970 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1971 extern int wake_up_process(struct task_struct *tsk);
1972 extern void wake_up_new_task(struct task_struct *tsk);
1974 extern void kick_process(struct task_struct *tsk);
1976 static inline void kick_process(struct task_struct *tsk) { }
1978 extern void sched_fork(struct task_struct *p);
1979 extern void sched_dead(struct task_struct *p);
1981 extern void proc_caches_init(void);
1982 extern void flush_signals(struct task_struct *);
1983 extern void __flush_signals(struct task_struct *);
1984 extern void ignore_signals(struct task_struct *);
1985 extern void flush_signal_handlers(struct task_struct *, int force_default);
1986 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1988 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1990 unsigned long flags;
1993 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1994 ret = dequeue_signal(tsk, mask, info);
1995 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2000 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2002 extern void unblock_all_signals(void);
2003 extern void release_task(struct task_struct * p);
2004 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2005 extern int force_sigsegv(int, struct task_struct *);
2006 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2007 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2008 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2009 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2010 const struct cred *, u32);
2011 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2012 extern int kill_pid(struct pid *pid, int sig, int priv);
2013 extern int kill_proc_info(int, struct siginfo *, pid_t);
2014 extern __must_check bool do_notify_parent(struct task_struct *, int);
2015 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2016 extern void force_sig(int, struct task_struct *);
2017 extern int send_sig(int, struct task_struct *, int);
2018 extern int zap_other_threads(struct task_struct *p);
2019 extern struct sigqueue *sigqueue_alloc(void);
2020 extern void sigqueue_free(struct sigqueue *);
2021 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2022 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2024 static inline void restore_saved_sigmask(void)
2026 if (test_and_clear_restore_sigmask())
2027 __set_current_blocked(¤t->saved_sigmask);
2030 static inline sigset_t *sigmask_to_save(void)
2032 sigset_t *res = ¤t->blocked;
2033 if (unlikely(test_restore_sigmask()))
2034 res = ¤t->saved_sigmask;
2038 static inline int kill_cad_pid(int sig, int priv)
2040 return kill_pid(cad_pid, sig, priv);
2043 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2044 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2045 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2046 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2049 * True if we are on the alternate signal stack.
2051 static inline int on_sig_stack(unsigned long sp)
2053 #ifdef CONFIG_STACK_GROWSUP
2054 return sp >= current->sas_ss_sp &&
2055 sp - current->sas_ss_sp < current->sas_ss_size;
2057 return sp > current->sas_ss_sp &&
2058 sp - current->sas_ss_sp <= current->sas_ss_size;
2062 static inline int sas_ss_flags(unsigned long sp)
2064 return (current->sas_ss_size == 0 ? SS_DISABLE
2065 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2068 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2070 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2071 #ifdef CONFIG_STACK_GROWSUP
2072 return current->sas_ss_sp;
2074 return current->sas_ss_sp + current->sas_ss_size;
2080 * Routines for handling mm_structs
2082 extern struct mm_struct * mm_alloc(void);
2084 /* mmdrop drops the mm and the page tables */
2085 extern void __mmdrop(struct mm_struct *);
2086 static inline void mmdrop(struct mm_struct * mm)
2088 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2092 /* mmput gets rid of the mappings and all user-space */
2093 extern void mmput(struct mm_struct *);
2094 /* Grab a reference to a task's mm, if it is not already going away */
2095 extern struct mm_struct *get_task_mm(struct task_struct *task);
2097 * Grab a reference to a task's mm, if it is not already going away
2098 * and ptrace_may_access with the mode parameter passed to it
2101 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2102 /* Remove the current tasks stale references to the old mm_struct */
2103 extern void mm_release(struct task_struct *, struct mm_struct *);
2104 /* Allocate a new mm structure and copy contents from tsk->mm */
2105 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2107 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2108 struct task_struct *);
2109 extern void flush_thread(void);
2110 extern void exit_thread(void);
2112 extern void exit_files(struct task_struct *);
2113 extern void __cleanup_sighand(struct sighand_struct *);
2115 extern void exit_itimers(struct signal_struct *);
2116 extern void flush_itimer_signals(void);
2118 extern void do_group_exit(int);
2120 extern int allow_signal(int);
2121 extern int disallow_signal(int);
2123 extern int do_execve(const char *,
2124 const char __user * const __user *,
2125 const char __user * const __user *);
2126 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2127 struct task_struct *fork_idle(int);
2128 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2130 extern void set_task_comm(struct task_struct *tsk, char *from);
2131 extern char *get_task_comm(char *to, struct task_struct *tsk);
2134 void scheduler_ipi(void);
2135 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2137 static inline void scheduler_ipi(void) { }
2138 static inline unsigned long wait_task_inactive(struct task_struct *p,
2145 #define next_task(p) \
2146 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2148 #define for_each_process(p) \
2149 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2151 extern bool current_is_single_threaded(void);
2154 * Careful: do_each_thread/while_each_thread is a double loop so
2155 * 'break' will not work as expected - use goto instead.
2157 #define do_each_thread(g, t) \
2158 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2160 #define while_each_thread(g, t) \
2161 while ((t = next_thread(t)) != g)
2163 static inline int get_nr_threads(struct task_struct *tsk)
2165 return tsk->signal->nr_threads;
2168 static inline bool thread_group_leader(struct task_struct *p)
2170 return p->exit_signal >= 0;
2173 /* Do to the insanities of de_thread it is possible for a process
2174 * to have the pid of the thread group leader without actually being
2175 * the thread group leader. For iteration through the pids in proc
2176 * all we care about is that we have a task with the appropriate
2177 * pid, we don't actually care if we have the right task.
2179 static inline bool has_group_leader_pid(struct task_struct *p)
2181 return task_pid(p) == p->signal->leader_pid;
2185 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2187 return p1->signal == p2->signal;
2190 static inline struct task_struct *next_thread(const struct task_struct *p)
2192 return list_entry_rcu(p->thread_group.next,
2193 struct task_struct, thread_group);
2196 static inline int thread_group_empty(struct task_struct *p)
2198 return list_empty(&p->thread_group);
2201 #define delay_group_leader(p) \
2202 (thread_group_leader(p) && !thread_group_empty(p))
2205 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2206 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2207 * pins the final release of task.io_context. Also protects ->cpuset and
2208 * ->cgroup.subsys[]. And ->vfork_done.
2210 * Nests both inside and outside of read_lock(&tasklist_lock).
2211 * It must not be nested with write_lock_irq(&tasklist_lock),
2212 * neither inside nor outside.
2214 static inline void task_lock(struct task_struct *p)
2216 spin_lock(&p->alloc_lock);
2219 static inline void task_unlock(struct task_struct *p)
2221 spin_unlock(&p->alloc_lock);
2224 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2225 unsigned long *flags);
2227 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2228 unsigned long *flags)
2230 struct sighand_struct *ret;
2232 ret = __lock_task_sighand(tsk, flags);
2233 (void)__cond_lock(&tsk->sighand->siglock, ret);
2237 static inline void unlock_task_sighand(struct task_struct *tsk,
2238 unsigned long *flags)
2240 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2243 #ifdef CONFIG_CGROUPS
2244 static inline void threadgroup_change_begin(struct task_struct *tsk)
2246 down_read(&tsk->signal->group_rwsem);
2248 static inline void threadgroup_change_end(struct task_struct *tsk)
2250 up_read(&tsk->signal->group_rwsem);
2254 * threadgroup_lock - lock threadgroup
2255 * @tsk: member task of the threadgroup to lock
2257 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2258 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2259 * change ->group_leader/pid. This is useful for cases where the threadgroup
2260 * needs to stay stable across blockable operations.
2262 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2263 * synchronization. While held, no new task will be added to threadgroup
2264 * and no existing live task will have its PF_EXITING set.
2266 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2267 * sub-thread becomes a new leader.
2269 static inline void threadgroup_lock(struct task_struct *tsk)
2271 down_write(&tsk->signal->group_rwsem);
2275 * threadgroup_unlock - unlock threadgroup
2276 * @tsk: member task of the threadgroup to unlock
2278 * Reverse threadgroup_lock().
2280 static inline void threadgroup_unlock(struct task_struct *tsk)
2282 up_write(&tsk->signal->group_rwsem);
2285 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2286 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2287 static inline void threadgroup_lock(struct task_struct *tsk) {}
2288 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2291 #ifndef __HAVE_THREAD_FUNCTIONS
2293 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2294 #define task_stack_page(task) ((task)->stack)
2296 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2298 *task_thread_info(p) = *task_thread_info(org);
2299 task_thread_info(p)->task = p;
2302 static inline unsigned long *end_of_stack(struct task_struct *p)
2304 return (unsigned long *)(task_thread_info(p) + 1);
2309 static inline int object_is_on_stack(void *obj)
2311 void *stack = task_stack_page(current);
2313 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2316 extern void thread_info_cache_init(void);
2318 #ifdef CONFIG_DEBUG_STACK_USAGE
2319 static inline unsigned long stack_not_used(struct task_struct *p)
2321 unsigned long *n = end_of_stack(p);
2323 do { /* Skip over canary */
2327 return (unsigned long)n - (unsigned long)end_of_stack(p);
2331 /* set thread flags in other task's structures
2332 * - see asm/thread_info.h for TIF_xxxx flags available
2334 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2336 set_ti_thread_flag(task_thread_info(tsk), flag);
2339 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2341 clear_ti_thread_flag(task_thread_info(tsk), flag);
2344 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2346 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2349 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2351 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2354 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2356 return test_ti_thread_flag(task_thread_info(tsk), flag);
2359 static inline void set_tsk_need_resched(struct task_struct *tsk)
2361 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2364 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2366 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2369 static inline int test_tsk_need_resched(struct task_struct *tsk)
2371 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2374 static inline int restart_syscall(void)
2376 set_tsk_thread_flag(current, TIF_SIGPENDING);
2377 return -ERESTARTNOINTR;
2380 static inline int signal_pending(struct task_struct *p)
2382 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2385 static inline int __fatal_signal_pending(struct task_struct *p)
2387 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2390 static inline int fatal_signal_pending(struct task_struct *p)
2392 return signal_pending(p) && __fatal_signal_pending(p);
2395 static inline int signal_pending_state(long state, struct task_struct *p)
2397 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2399 if (!signal_pending(p))
2402 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2405 static inline int need_resched(void)
2407 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2411 * cond_resched() and cond_resched_lock(): latency reduction via
2412 * explicit rescheduling in places that are safe. The return
2413 * value indicates whether a reschedule was done in fact.
2414 * cond_resched_lock() will drop the spinlock before scheduling,
2415 * cond_resched_softirq() will enable bhs before scheduling.
2417 extern int _cond_resched(void);
2419 #define cond_resched() ({ \
2420 __might_sleep(__FILE__, __LINE__, 0); \
2424 extern int __cond_resched_lock(spinlock_t *lock);
2426 #ifdef CONFIG_PREEMPT_COUNT
2427 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2429 #define PREEMPT_LOCK_OFFSET 0
2432 #define cond_resched_lock(lock) ({ \
2433 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2434 __cond_resched_lock(lock); \
2437 extern int __cond_resched_softirq(void);
2439 #define cond_resched_softirq() ({ \
2440 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2441 __cond_resched_softirq(); \
2444 static inline void cond_resched_rcu(void)
2446 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2454 * Does a critical section need to be broken due to another
2455 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2456 * but a general need for low latency)
2458 static inline int spin_needbreak(spinlock_t *lock)
2460 #ifdef CONFIG_PREEMPT
2461 return spin_is_contended(lock);
2468 * Idle thread specific functions to determine the need_resched
2469 * polling state. We have two versions, one based on TS_POLLING in
2470 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2474 static inline int tsk_is_polling(struct task_struct *p)
2476 return task_thread_info(p)->status & TS_POLLING;
2478 static inline void current_set_polling(void)
2480 current_thread_info()->status |= TS_POLLING;
2483 static inline void current_clr_polling(void)
2485 current_thread_info()->status &= ~TS_POLLING;
2486 smp_mb__after_clear_bit();
2488 #elif defined(TIF_POLLING_NRFLAG)
2489 static inline int tsk_is_polling(struct task_struct *p)
2491 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2493 static inline void current_set_polling(void)
2495 set_thread_flag(TIF_POLLING_NRFLAG);
2498 static inline void current_clr_polling(void)
2500 clear_thread_flag(TIF_POLLING_NRFLAG);
2503 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2504 static inline void current_set_polling(void) { }
2505 static inline void current_clr_polling(void) { }
2509 * Thread group CPU time accounting.
2511 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2512 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2514 static inline void thread_group_cputime_init(struct signal_struct *sig)
2516 raw_spin_lock_init(&sig->cputimer.lock);
2520 * Reevaluate whether the task has signals pending delivery.
2521 * Wake the task if so.
2522 * This is required every time the blocked sigset_t changes.
2523 * callers must hold sighand->siglock.
2525 extern void recalc_sigpending_and_wake(struct task_struct *t);
2526 extern void recalc_sigpending(void);
2528 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2530 static inline void signal_wake_up(struct task_struct *t, bool resume)
2532 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2534 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2536 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2540 * Wrappers for p->thread_info->cpu access. No-op on UP.
2544 static inline unsigned int task_cpu(const struct task_struct *p)
2546 return task_thread_info(p)->cpu;
2549 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2553 static inline unsigned int task_cpu(const struct task_struct *p)
2558 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2562 #endif /* CONFIG_SMP */
2564 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2565 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2567 #ifdef CONFIG_CGROUP_SCHED
2568 extern struct task_group root_task_group;
2569 #endif /* CONFIG_CGROUP_SCHED */
2571 extern int task_can_switch_user(struct user_struct *up,
2572 struct task_struct *tsk);
2574 #ifdef CONFIG_TASK_XACCT
2575 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2577 tsk->ioac.rchar += amt;
2580 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2582 tsk->ioac.wchar += amt;
2585 static inline void inc_syscr(struct task_struct *tsk)
2590 static inline void inc_syscw(struct task_struct *tsk)
2595 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2599 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2603 static inline void inc_syscr(struct task_struct *tsk)
2607 static inline void inc_syscw(struct task_struct *tsk)
2612 #ifndef TASK_SIZE_OF
2613 #define TASK_SIZE_OF(tsk) TASK_SIZE
2616 #ifdef CONFIG_MM_OWNER
2617 extern void mm_update_next_owner(struct mm_struct *mm);
2618 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2620 static inline void mm_update_next_owner(struct mm_struct *mm)
2624 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2627 #endif /* CONFIG_MM_OWNER */
2629 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2632 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2635 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2638 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2641 static inline unsigned long rlimit(unsigned int limit)
2643 return task_rlimit(current, limit);
2646 static inline unsigned long rlimit_max(unsigned int limit)
2648 return task_rlimit_max(current, limit);