7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
25 and is now available for re-use. */
26 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
27 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
28 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
29 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
30 #define CLONE_NEWNET 0x40000000 /* New network namespace */
31 #define CLONE_IO 0x80000000 /* Clone io context */
36 #define SCHED_NORMAL 0
40 /* SCHED_ISO: reserved but not implemented yet */
42 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
43 #define SCHED_RESET_ON_FORK 0x40000000
51 #include <asm/param.h> /* for HZ */
53 #include <linux/capability.h>
54 #include <linux/threads.h>
55 #include <linux/kernel.h>
56 #include <linux/types.h>
57 #include <linux/timex.h>
58 #include <linux/jiffies.h>
59 #include <linux/rbtree.h>
60 #include <linux/thread_info.h>
61 #include <linux/cpumask.h>
62 #include <linux/errno.h>
63 #include <linux/nodemask.h>
64 #include <linux/mm_types.h>
67 #include <asm/ptrace.h>
68 #include <asm/cputime.h>
70 #include <linux/smp.h>
71 #include <linux/sem.h>
72 #include <linux/signal.h>
73 #include <linux/compiler.h>
74 #include <linux/completion.h>
75 #include <linux/pid.h>
76 #include <linux/percpu.h>
77 #include <linux/topology.h>
78 #include <linux/proportions.h>
79 #include <linux/seccomp.h>
80 #include <linux/rcupdate.h>
81 #include <linux/rculist.h>
82 #include <linux/rtmutex.h>
84 #include <linux/time.h>
85 #include <linux/param.h>
86 #include <linux/resource.h>
87 #include <linux/timer.h>
88 #include <linux/hrtimer.h>
89 #include <linux/task_io_accounting.h>
90 #include <linux/latencytop.h>
91 #include <linux/cred.h>
92 #include <linux/llist.h>
93 #include <linux/uidgid.h>
95 #include <asm/processor.h>
98 struct futex_pi_state;
99 struct robust_list_head;
102 struct perf_event_context;
106 * List of flags we want to share for kernel threads,
107 * if only because they are not used by them anyway.
109 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
112 * These are the constant used to fake the fixed-point load-average
113 * counting. Some notes:
114 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
115 * a load-average precision of 10 bits integer + 11 bits fractional
116 * - if you want to count load-averages more often, you need more
117 * precision, or rounding will get you. With 2-second counting freq,
118 * the EXP_n values would be 1981, 2034 and 2043 if still using only
121 extern unsigned long avenrun[]; /* Load averages */
122 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
124 #define FSHIFT 11 /* nr of bits of precision */
125 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
126 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
127 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
128 #define EXP_5 2014 /* 1/exp(5sec/5min) */
129 #define EXP_15 2037 /* 1/exp(5sec/15min) */
131 #define CALC_LOAD(load,exp,n) \
133 load += n*(FIXED_1-exp); \
136 extern unsigned long total_forks;
137 extern int nr_threads;
138 DECLARE_PER_CPU(unsigned long, process_counts);
139 extern int nr_processes(void);
140 extern unsigned long nr_running(void);
141 extern unsigned long nr_uninterruptible(void);
142 extern unsigned long nr_iowait(void);
143 extern unsigned long nr_iowait_cpu(int cpu);
144 extern unsigned long this_cpu_load(void);
147 extern void calc_global_load(unsigned long ticks);
148 extern void update_cpu_load_nohz(void);
150 extern unsigned long get_parent_ip(unsigned long addr);
155 #ifdef CONFIG_SCHED_DEBUG
156 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
157 extern void proc_sched_set_task(struct task_struct *p);
159 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
162 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
165 static inline void proc_sched_set_task(struct task_struct *p)
169 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
175 * Task state bitmask. NOTE! These bits are also
176 * encoded in fs/proc/array.c: get_task_state().
178 * We have two separate sets of flags: task->state
179 * is about runnability, while task->exit_state are
180 * about the task exiting. Confusing, but this way
181 * modifying one set can't modify the other one by
184 #define TASK_RUNNING 0
185 #define TASK_INTERRUPTIBLE 1
186 #define TASK_UNINTERRUPTIBLE 2
187 #define __TASK_STOPPED 4
188 #define __TASK_TRACED 8
189 /* in tsk->exit_state */
190 #define EXIT_ZOMBIE 16
192 /* in tsk->state again */
194 #define TASK_WAKEKILL 128
195 #define TASK_WAKING 256
196 #define TASK_STATE_MAX 512
198 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
200 extern char ___assert_task_state[1 - 2*!!(
201 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
203 /* Convenience macros for the sake of set_task_state */
204 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
205 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
206 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
208 /* Convenience macros for the sake of wake_up */
209 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
210 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
212 /* get_task_state() */
213 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
214 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
217 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
218 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
219 #define task_is_dead(task) ((task)->exit_state != 0)
220 #define task_is_stopped_or_traced(task) \
221 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
222 #define task_contributes_to_load(task) \
223 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
224 (task->flags & PF_FROZEN) == 0)
226 #define __set_task_state(tsk, state_value) \
227 do { (tsk)->state = (state_value); } while (0)
228 #define set_task_state(tsk, state_value) \
229 set_mb((tsk)->state, (state_value))
232 * set_current_state() includes a barrier so that the write of current->state
233 * is correctly serialised wrt the caller's subsequent test of whether to
236 * set_current_state(TASK_UNINTERRUPTIBLE);
237 * if (do_i_need_to_sleep())
240 * If the caller does not need such serialisation then use __set_current_state()
242 #define __set_current_state(state_value) \
243 do { current->state = (state_value); } while (0)
244 #define set_current_state(state_value) \
245 set_mb(current->state, (state_value))
247 /* Task command name length */
248 #define TASK_COMM_LEN 16
250 #include <linux/spinlock.h>
253 * This serializes "schedule()" and also protects
254 * the run-queue from deletions/modifications (but
255 * _adding_ to the beginning of the run-queue has
258 extern rwlock_t tasklist_lock;
259 extern spinlock_t mmlist_lock;
263 #ifdef CONFIG_PROVE_RCU
264 extern int lockdep_tasklist_lock_is_held(void);
265 #endif /* #ifdef CONFIG_PROVE_RCU */
267 extern void sched_init(void);
268 extern void sched_init_smp(void);
269 extern asmlinkage void schedule_tail(struct task_struct *prev);
270 extern void init_idle(struct task_struct *idle, int cpu);
271 extern void init_idle_bootup_task(struct task_struct *idle);
273 extern int runqueue_is_locked(int cpu);
275 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
276 extern void select_nohz_load_balancer(int stop_tick);
277 extern void set_cpu_sd_state_idle(void);
278 extern int get_nohz_timer_target(void);
280 static inline void select_nohz_load_balancer(int stop_tick) { }
281 static inline void set_cpu_sd_state_idle(void) { }
285 * Only dump TASK_* tasks. (0 for all tasks)
287 extern void show_state_filter(unsigned long state_filter);
289 static inline void show_state(void)
291 show_state_filter(0);
294 extern void show_regs(struct pt_regs *);
297 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
298 * task), SP is the stack pointer of the first frame that should be shown in the back
299 * trace (or NULL if the entire call-chain of the task should be shown).
301 extern void show_stack(struct task_struct *task, unsigned long *sp);
303 void io_schedule(void);
304 long io_schedule_timeout(long timeout);
306 extern void cpu_init (void);
307 extern void trap_init(void);
308 extern void update_process_times(int user);
309 extern void scheduler_tick(void);
311 extern void sched_show_task(struct task_struct *p);
313 #ifdef CONFIG_LOCKUP_DETECTOR
314 extern void touch_softlockup_watchdog(void);
315 extern void touch_softlockup_watchdog_sync(void);
316 extern void touch_all_softlockup_watchdogs(void);
317 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
319 size_t *lenp, loff_t *ppos);
320 extern unsigned int softlockup_panic;
321 void lockup_detector_init(void);
323 static inline void touch_softlockup_watchdog(void)
326 static inline void touch_softlockup_watchdog_sync(void)
329 static inline void touch_all_softlockup_watchdogs(void)
332 static inline void lockup_detector_init(void)
337 #ifdef CONFIG_DETECT_HUNG_TASK
338 extern unsigned int sysctl_hung_task_panic;
339 extern unsigned long sysctl_hung_task_check_count;
340 extern unsigned long sysctl_hung_task_timeout_secs;
341 extern unsigned long sysctl_hung_task_warnings;
342 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
344 size_t *lenp, loff_t *ppos);
346 /* Avoid need for ifdefs elsewhere in the code */
347 enum { sysctl_hung_task_timeout_secs = 0 };
350 /* Attach to any functions which should be ignored in wchan output. */
351 #define __sched __attribute__((__section__(".sched.text")))
353 /* Linker adds these: start and end of __sched functions */
354 extern char __sched_text_start[], __sched_text_end[];
356 /* Is this address in the __sched functions? */
357 extern int in_sched_functions(unsigned long addr);
359 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
360 extern signed long schedule_timeout(signed long timeout);
361 extern signed long schedule_timeout_interruptible(signed long timeout);
362 extern signed long schedule_timeout_killable(signed long timeout);
363 extern signed long schedule_timeout_uninterruptible(signed long timeout);
364 asmlinkage void schedule(void);
365 extern void schedule_preempt_disabled(void);
366 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
369 struct user_namespace;
372 * Default maximum number of active map areas, this limits the number of vmas
373 * per mm struct. Users can overwrite this number by sysctl but there is a
376 * When a program's coredump is generated as ELF format, a section is created
377 * per a vma. In ELF, the number of sections is represented in unsigned short.
378 * This means the number of sections should be smaller than 65535 at coredump.
379 * Because the kernel adds some informative sections to a image of program at
380 * generating coredump, we need some margin. The number of extra sections is
381 * 1-3 now and depends on arch. We use "5" as safe margin, here.
383 #define MAPCOUNT_ELF_CORE_MARGIN (5)
384 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
386 extern int sysctl_max_map_count;
388 #include <linux/aio.h>
391 extern void arch_pick_mmap_layout(struct mm_struct *mm);
393 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
394 unsigned long, unsigned long);
396 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
397 unsigned long len, unsigned long pgoff,
398 unsigned long flags);
399 extern void arch_unmap_area(struct mm_struct *, unsigned long);
400 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
402 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
406 extern void set_dumpable(struct mm_struct *mm, int value);
407 extern int get_dumpable(struct mm_struct *mm);
409 /* get/set_dumpable() values */
410 #define SUID_DUMPABLE_DISABLED 0
411 #define SUID_DUMPABLE_ENABLED 1
412 #define SUID_DUMPABLE_SAFE 2
416 #define MMF_DUMPABLE 0 /* core dump is permitted */
417 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
419 #define MMF_DUMPABLE_BITS 2
420 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
422 /* coredump filter bits */
423 #define MMF_DUMP_ANON_PRIVATE 2
424 #define MMF_DUMP_ANON_SHARED 3
425 #define MMF_DUMP_MAPPED_PRIVATE 4
426 #define MMF_DUMP_MAPPED_SHARED 5
427 #define MMF_DUMP_ELF_HEADERS 6
428 #define MMF_DUMP_HUGETLB_PRIVATE 7
429 #define MMF_DUMP_HUGETLB_SHARED 8
431 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
432 #define MMF_DUMP_FILTER_BITS 7
433 #define MMF_DUMP_FILTER_MASK \
434 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
435 #define MMF_DUMP_FILTER_DEFAULT \
436 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
437 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
439 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
440 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
442 # define MMF_DUMP_MASK_DEFAULT_ELF 0
444 /* leave room for more dump flags */
445 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
446 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
447 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
449 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
451 struct sighand_struct {
453 struct k_sigaction action[_NSIG];
455 wait_queue_head_t signalfd_wqh;
458 struct pacct_struct {
461 unsigned long ac_mem;
462 cputime_t ac_utime, ac_stime;
463 unsigned long ac_minflt, ac_majflt;
474 * struct task_cputime - collected CPU time counts
475 * @utime: time spent in user mode, in &cputime_t units
476 * @stime: time spent in kernel mode, in &cputime_t units
477 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
479 * This structure groups together three kinds of CPU time that are
480 * tracked for threads and thread groups. Most things considering
481 * CPU time want to group these counts together and treat all three
482 * of them in parallel.
484 struct task_cputime {
487 unsigned long long sum_exec_runtime;
489 /* Alternate field names when used to cache expirations. */
490 #define prof_exp stime
491 #define virt_exp utime
492 #define sched_exp sum_exec_runtime
494 #define INIT_CPUTIME \
495 (struct task_cputime) { \
498 .sum_exec_runtime = 0, \
502 * Disable preemption until the scheduler is running.
503 * Reset by start_kernel()->sched_init()->init_idle().
505 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
506 * before the scheduler is active -- see should_resched().
508 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
511 * struct thread_group_cputimer - thread group interval timer counts
512 * @cputime: thread group interval timers.
513 * @running: non-zero when there are timers running and
514 * @cputime receives updates.
515 * @lock: lock for fields in this struct.
517 * This structure contains the version of task_cputime, above, that is
518 * used for thread group CPU timer calculations.
520 struct thread_group_cputimer {
521 struct task_cputime cputime;
526 #include <linux/rwsem.h>
530 * NOTE! "signal_struct" does not have its own
531 * locking, because a shared signal_struct always
532 * implies a shared sighand_struct, so locking
533 * sighand_struct is always a proper superset of
534 * the locking of signal_struct.
536 struct signal_struct {
541 wait_queue_head_t wait_chldexit; /* for wait4() */
543 /* current thread group signal load-balancing target: */
544 struct task_struct *curr_target;
546 /* shared signal handling: */
547 struct sigpending shared_pending;
549 /* thread group exit support */
552 * - notify group_exit_task when ->count is equal to notify_count
553 * - everyone except group_exit_task is stopped during signal delivery
554 * of fatal signals, group_exit_task processes the signal.
557 struct task_struct *group_exit_task;
559 /* thread group stop support, overloads group_exit_code too */
560 int group_stop_count;
561 unsigned int flags; /* see SIGNAL_* flags below */
564 * PR_SET_CHILD_SUBREAPER marks a process, like a service
565 * manager, to re-parent orphan (double-forking) child processes
566 * to this process instead of 'init'. The service manager is
567 * able to receive SIGCHLD signals and is able to investigate
568 * the process until it calls wait(). All children of this
569 * process will inherit a flag if they should look for a
570 * child_subreaper process at exit.
572 unsigned int is_child_subreaper:1;
573 unsigned int has_child_subreaper:1;
575 /* POSIX.1b Interval Timers */
576 struct list_head posix_timers;
578 /* ITIMER_REAL timer for the process */
579 struct hrtimer real_timer;
580 struct pid *leader_pid;
581 ktime_t it_real_incr;
584 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
585 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
586 * values are defined to 0 and 1 respectively
588 struct cpu_itimer it[2];
591 * Thread group totals for process CPU timers.
592 * See thread_group_cputimer(), et al, for details.
594 struct thread_group_cputimer cputimer;
596 /* Earliest-expiration cache. */
597 struct task_cputime cputime_expires;
599 struct list_head cpu_timers[3];
601 struct pid *tty_old_pgrp;
603 /* boolean value for session group leader */
606 struct tty_struct *tty; /* NULL if no tty */
608 #ifdef CONFIG_SCHED_AUTOGROUP
609 struct autogroup *autogroup;
612 * Cumulative resource counters for dead threads in the group,
613 * and for reaped dead child processes forked by this group.
614 * Live threads maintain their own counters and add to these
615 * in __exit_signal, except for the group leader.
617 cputime_t utime, stime, cutime, cstime;
620 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
621 cputime_t prev_utime, prev_stime;
623 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
624 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
625 unsigned long inblock, oublock, cinblock, coublock;
626 unsigned long maxrss, cmaxrss;
627 struct task_io_accounting ioac;
630 * Cumulative ns of schedule CPU time fo dead threads in the
631 * group, not including a zombie group leader, (This only differs
632 * from jiffies_to_ns(utime + stime) if sched_clock uses something
633 * other than jiffies.)
635 unsigned long long sum_sched_runtime;
638 * We don't bother to synchronize most readers of this at all,
639 * because there is no reader checking a limit that actually needs
640 * to get both rlim_cur and rlim_max atomically, and either one
641 * alone is a single word that can safely be read normally.
642 * getrlimit/setrlimit use task_lock(current->group_leader) to
643 * protect this instead of the siglock, because they really
644 * have no need to disable irqs.
646 struct rlimit rlim[RLIM_NLIMITS];
648 #ifdef CONFIG_BSD_PROCESS_ACCT
649 struct pacct_struct pacct; /* per-process accounting information */
651 #ifdef CONFIG_TASKSTATS
652 struct taskstats *stats;
656 struct tty_audit_buf *tty_audit_buf;
658 #ifdef CONFIG_CGROUPS
660 * group_rwsem prevents new tasks from entering the threadgroup and
661 * member tasks from exiting,a more specifically, setting of
662 * PF_EXITING. fork and exit paths are protected with this rwsem
663 * using threadgroup_change_begin/end(). Users which require
664 * threadgroup to remain stable should use threadgroup_[un]lock()
665 * which also takes care of exec path. Currently, cgroup is the
668 struct rw_semaphore group_rwsem;
671 int oom_adj; /* OOM kill score adjustment (bit shift) */
672 int oom_score_adj; /* OOM kill score adjustment */
673 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
674 * Only settable by CAP_SYS_RESOURCE. */
676 struct mutex cred_guard_mutex; /* guard against foreign influences on
677 * credential calculations
678 * (notably. ptrace) */
681 /* Context switch must be unlocked if interrupts are to be enabled */
682 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
683 # define __ARCH_WANT_UNLOCKED_CTXSW
687 * Bits in flags field of signal_struct.
689 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
690 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
691 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
693 * Pending notifications to parent.
695 #define SIGNAL_CLD_STOPPED 0x00000010
696 #define SIGNAL_CLD_CONTINUED 0x00000020
697 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
699 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
701 /* If true, all threads except ->group_exit_task have pending SIGKILL */
702 static inline int signal_group_exit(const struct signal_struct *sig)
704 return (sig->flags & SIGNAL_GROUP_EXIT) ||
705 (sig->group_exit_task != NULL);
709 * Some day this will be a full-fledged user tracking system..
712 atomic_t __count; /* reference count */
713 atomic_t processes; /* How many processes does this user have? */
714 atomic_t files; /* How many open files does this user have? */
715 atomic_t sigpending; /* How many pending signals does this user have? */
716 #ifdef CONFIG_INOTIFY_USER
717 atomic_t inotify_watches; /* How many inotify watches does this user have? */
718 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
720 #ifdef CONFIG_FANOTIFY
721 atomic_t fanotify_listeners;
724 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
726 #ifdef CONFIG_POSIX_MQUEUE
727 /* protected by mq_lock */
728 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
730 unsigned long locked_shm; /* How many pages of mlocked shm ? */
733 struct key *uid_keyring; /* UID specific keyring */
734 struct key *session_keyring; /* UID's default session keyring */
737 /* Hash table maintenance information */
738 struct hlist_node uidhash_node;
741 #ifdef CONFIG_PERF_EVENTS
742 atomic_long_t locked_vm;
746 extern int uids_sysfs_init(void);
748 extern struct user_struct *find_user(kuid_t);
750 extern struct user_struct root_user;
751 #define INIT_USER (&root_user)
754 struct backing_dev_info;
755 struct reclaim_state;
757 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
759 /* cumulative counters */
760 unsigned long pcount; /* # of times run on this cpu */
761 unsigned long long run_delay; /* time spent waiting on a runqueue */
764 unsigned long long last_arrival,/* when we last ran on a cpu */
765 last_queued; /* when we were last queued to run */
767 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
769 #ifdef CONFIG_TASK_DELAY_ACCT
770 struct task_delay_info {
772 unsigned int flags; /* Private per-task flags */
774 /* For each stat XXX, add following, aligned appropriately
776 * struct timespec XXX_start, XXX_end;
780 * Atomicity of updates to XXX_delay, XXX_count protected by
781 * single lock above (split into XXX_lock if contention is an issue).
785 * XXX_count is incremented on every XXX operation, the delay
786 * associated with the operation is added to XXX_delay.
787 * XXX_delay contains the accumulated delay time in nanoseconds.
789 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
790 u64 blkio_delay; /* wait for sync block io completion */
791 u64 swapin_delay; /* wait for swapin block io completion */
792 u32 blkio_count; /* total count of the number of sync block */
793 /* io operations performed */
794 u32 swapin_count; /* total count of the number of swapin block */
795 /* io operations performed */
797 struct timespec freepages_start, freepages_end;
798 u64 freepages_delay; /* wait for memory reclaim */
799 u32 freepages_count; /* total count of memory reclaim */
801 #endif /* CONFIG_TASK_DELAY_ACCT */
803 static inline int sched_info_on(void)
805 #ifdef CONFIG_SCHEDSTATS
807 #elif defined(CONFIG_TASK_DELAY_ACCT)
808 extern int delayacct_on;
823 * Increase resolution of nice-level calculations for 64-bit architectures.
824 * The extra resolution improves shares distribution and load balancing of
825 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
826 * hierarchies, especially on larger systems. This is not a user-visible change
827 * and does not change the user-interface for setting shares/weights.
829 * We increase resolution only if we have enough bits to allow this increased
830 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
831 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
834 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
835 # define SCHED_LOAD_RESOLUTION 10
836 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
837 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
839 # define SCHED_LOAD_RESOLUTION 0
840 # define scale_load(w) (w)
841 # define scale_load_down(w) (w)
844 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
845 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
848 * Increase resolution of cpu_power calculations
850 #define SCHED_POWER_SHIFT 10
851 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
854 * sched-domains (multiprocessor balancing) declarations:
857 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
858 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
859 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
860 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
861 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
862 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
863 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
864 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
865 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
866 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
867 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
868 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
869 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
871 extern int __weak arch_sd_sibiling_asym_packing(void);
873 struct sched_group_power {
876 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
879 unsigned int power, power_orig;
880 unsigned long next_update;
882 * Number of busy cpus in this group.
884 atomic_t nr_busy_cpus;
886 unsigned long cpumask[0]; /* iteration mask */
890 struct sched_group *next; /* Must be a circular list */
893 unsigned int group_weight;
894 struct sched_group_power *sgp;
897 * The CPUs this group covers.
899 * NOTE: this field is variable length. (Allocated dynamically
900 * by attaching extra space to the end of the structure,
901 * depending on how many CPUs the kernel has booted up with)
903 unsigned long cpumask[0];
906 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
908 return to_cpumask(sg->cpumask);
912 * cpumask masking which cpus in the group are allowed to iterate up the domain
915 static inline struct cpumask *sched_group_mask(struct sched_group *sg)
917 return to_cpumask(sg->sgp->cpumask);
921 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
922 * @group: The group whose first cpu is to be returned.
924 static inline unsigned int group_first_cpu(struct sched_group *group)
926 return cpumask_first(sched_group_cpus(group));
929 struct sched_domain_attr {
930 int relax_domain_level;
933 #define SD_ATTR_INIT (struct sched_domain_attr) { \
934 .relax_domain_level = -1, \
937 extern int sched_domain_level_max;
939 struct sched_domain {
940 /* These fields must be setup */
941 struct sched_domain *parent; /* top domain must be null terminated */
942 struct sched_domain *child; /* bottom domain must be null terminated */
943 struct sched_group *groups; /* the balancing groups of the domain */
944 unsigned long min_interval; /* Minimum balance interval ms */
945 unsigned long max_interval; /* Maximum balance interval ms */
946 unsigned int busy_factor; /* less balancing by factor if busy */
947 unsigned int imbalance_pct; /* No balance until over watermark */
948 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
949 unsigned int busy_idx;
950 unsigned int idle_idx;
951 unsigned int newidle_idx;
952 unsigned int wake_idx;
953 unsigned int forkexec_idx;
954 unsigned int smt_gain;
955 int flags; /* See SD_* */
958 /* Runtime fields. */
959 unsigned long last_balance; /* init to jiffies. units in jiffies */
960 unsigned int balance_interval; /* initialise to 1. units in ms. */
961 unsigned int nr_balance_failed; /* initialise to 0 */
965 #ifdef CONFIG_SCHEDSTATS
966 /* load_balance() stats */
967 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
968 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
969 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
970 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
971 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
972 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
973 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
974 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
976 /* Active load balancing */
977 unsigned int alb_count;
978 unsigned int alb_failed;
979 unsigned int alb_pushed;
981 /* SD_BALANCE_EXEC stats */
982 unsigned int sbe_count;
983 unsigned int sbe_balanced;
984 unsigned int sbe_pushed;
986 /* SD_BALANCE_FORK stats */
987 unsigned int sbf_count;
988 unsigned int sbf_balanced;
989 unsigned int sbf_pushed;
991 /* try_to_wake_up() stats */
992 unsigned int ttwu_wake_remote;
993 unsigned int ttwu_move_affine;
994 unsigned int ttwu_move_balance;
996 #ifdef CONFIG_SCHED_DEBUG
1000 void *private; /* used during construction */
1001 struct rcu_head rcu; /* used during destruction */
1004 unsigned int span_weight;
1006 * Span of all CPUs in this domain.
1008 * NOTE: this field is variable length. (Allocated dynamically
1009 * by attaching extra space to the end of the structure,
1010 * depending on how many CPUs the kernel has booted up with)
1012 unsigned long span[0];
1015 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1017 return to_cpumask(sd->span);
1020 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1021 struct sched_domain_attr *dattr_new);
1023 /* Allocate an array of sched domains, for partition_sched_domains(). */
1024 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1025 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1027 /* Test a flag in parent sched domain */
1028 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1030 if (sd->parent && (sd->parent->flags & flag))
1036 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1037 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1039 bool cpus_share_cache(int this_cpu, int that_cpu);
1041 #else /* CONFIG_SMP */
1043 struct sched_domain_attr;
1046 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1047 struct sched_domain_attr *dattr_new)
1051 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1056 #endif /* !CONFIG_SMP */
1059 struct io_context; /* See blkdev.h */
1062 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1063 extern void prefetch_stack(struct task_struct *t);
1065 static inline void prefetch_stack(struct task_struct *t) { }
1068 struct audit_context; /* See audit.c */
1070 struct pipe_inode_info;
1071 struct uts_namespace;
1074 struct sched_domain;
1079 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1080 #define WF_FORK 0x02 /* child wakeup after fork */
1081 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1083 #define ENQUEUE_WAKEUP 1
1084 #define ENQUEUE_HEAD 2
1086 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1088 #define ENQUEUE_WAKING 0
1091 #define DEQUEUE_SLEEP 1
1093 struct sched_class {
1094 const struct sched_class *next;
1096 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1097 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1098 void (*yield_task) (struct rq *rq);
1099 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1101 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1103 struct task_struct * (*pick_next_task) (struct rq *rq);
1104 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1107 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1109 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1110 void (*post_schedule) (struct rq *this_rq);
1111 void (*task_waking) (struct task_struct *task);
1112 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1114 void (*set_cpus_allowed)(struct task_struct *p,
1115 const struct cpumask *newmask);
1117 void (*rq_online)(struct rq *rq);
1118 void (*rq_offline)(struct rq *rq);
1121 void (*set_curr_task) (struct rq *rq);
1122 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1123 void (*task_fork) (struct task_struct *p);
1125 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1126 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1127 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1130 unsigned int (*get_rr_interval) (struct rq *rq,
1131 struct task_struct *task);
1133 #ifdef CONFIG_FAIR_GROUP_SCHED
1134 void (*task_move_group) (struct task_struct *p, int on_rq);
1138 struct load_weight {
1139 unsigned long weight, inv_weight;
1142 #ifdef CONFIG_SCHEDSTATS
1143 struct sched_statistics {
1153 s64 sum_sleep_runtime;
1160 u64 nr_migrations_cold;
1161 u64 nr_failed_migrations_affine;
1162 u64 nr_failed_migrations_running;
1163 u64 nr_failed_migrations_hot;
1164 u64 nr_forced_migrations;
1167 u64 nr_wakeups_sync;
1168 u64 nr_wakeups_migrate;
1169 u64 nr_wakeups_local;
1170 u64 nr_wakeups_remote;
1171 u64 nr_wakeups_affine;
1172 u64 nr_wakeups_affine_attempts;
1173 u64 nr_wakeups_passive;
1174 u64 nr_wakeups_idle;
1178 struct sched_entity {
1179 struct load_weight load; /* for load-balancing */
1180 struct rb_node run_node;
1181 struct list_head group_node;
1185 u64 sum_exec_runtime;
1187 u64 prev_sum_exec_runtime;
1191 #ifdef CONFIG_SCHEDSTATS
1192 struct sched_statistics statistics;
1195 #ifdef CONFIG_FAIR_GROUP_SCHED
1196 struct sched_entity *parent;
1197 /* rq on which this entity is (to be) queued: */
1198 struct cfs_rq *cfs_rq;
1199 /* rq "owned" by this entity/group: */
1200 struct cfs_rq *my_q;
1204 struct sched_rt_entity {
1205 struct list_head run_list;
1206 unsigned long timeout;
1207 unsigned int time_slice;
1209 struct sched_rt_entity *back;
1210 #ifdef CONFIG_RT_GROUP_SCHED
1211 struct sched_rt_entity *parent;
1212 /* rq on which this entity is (to be) queued: */
1213 struct rt_rq *rt_rq;
1214 /* rq "owned" by this entity/group: */
1220 * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1221 * Timeslices get refilled after they expire.
1223 #define RR_TIMESLICE (100 * HZ / 1000)
1227 enum perf_event_task_context {
1228 perf_invalid_context = -1,
1229 perf_hw_context = 0,
1231 perf_nr_task_contexts,
1234 struct task_struct {
1235 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1238 unsigned int flags; /* per process flags, defined below */
1239 unsigned int ptrace;
1242 struct llist_node wake_entry;
1247 int prio, static_prio, normal_prio;
1248 unsigned int rt_priority;
1249 const struct sched_class *sched_class;
1250 struct sched_entity se;
1251 struct sched_rt_entity rt;
1252 #ifdef CONFIG_CGROUP_SCHED
1253 struct task_group *sched_task_group;
1256 #ifdef CONFIG_PREEMPT_NOTIFIERS
1257 /* list of struct preempt_notifier: */
1258 struct hlist_head preempt_notifiers;
1262 * fpu_counter contains the number of consecutive context switches
1263 * that the FPU is used. If this is over a threshold, the lazy fpu
1264 * saving becomes unlazy to save the trap. This is an unsigned char
1265 * so that after 256 times the counter wraps and the behavior turns
1266 * lazy again; this to deal with bursty apps that only use FPU for
1269 unsigned char fpu_counter;
1270 #ifdef CONFIG_BLK_DEV_IO_TRACE
1271 unsigned int btrace_seq;
1274 unsigned int policy;
1275 int nr_cpus_allowed;
1276 cpumask_t cpus_allowed;
1278 #ifdef CONFIG_PREEMPT_RCU
1279 int rcu_read_lock_nesting;
1280 char rcu_read_unlock_special;
1281 struct list_head rcu_node_entry;
1282 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1283 #ifdef CONFIG_TREE_PREEMPT_RCU
1284 struct rcu_node *rcu_blocked_node;
1285 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1286 #ifdef CONFIG_RCU_BOOST
1287 struct rt_mutex *rcu_boost_mutex;
1288 #endif /* #ifdef CONFIG_RCU_BOOST */
1290 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1291 struct sched_info sched_info;
1294 struct list_head tasks;
1296 struct plist_node pushable_tasks;
1299 struct mm_struct *mm, *active_mm;
1300 #ifdef CONFIG_COMPAT_BRK
1301 unsigned brk_randomized:1;
1303 #if defined(SPLIT_RSS_COUNTING)
1304 struct task_rss_stat rss_stat;
1308 int exit_code, exit_signal;
1309 int pdeath_signal; /* The signal sent when the parent dies */
1310 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1312 unsigned int personality;
1313 unsigned did_exec:1;
1314 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1316 unsigned in_iowait:1;
1318 /* task may not gain privileges */
1319 unsigned no_new_privs:1;
1321 /* Revert to default priority/policy when forking */
1322 unsigned sched_reset_on_fork:1;
1323 unsigned sched_contributes_to_load:1;
1328 #ifdef CONFIG_CC_STACKPROTECTOR
1329 /* Canary value for the -fstack-protector gcc feature */
1330 unsigned long stack_canary;
1333 * pointers to (original) parent process, youngest child, younger sibling,
1334 * older sibling, respectively. (p->father can be replaced with
1335 * p->real_parent->pid)
1337 struct task_struct __rcu *real_parent; /* real parent process */
1338 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1340 * children/sibling forms the list of my natural children
1342 struct list_head children; /* list of my children */
1343 struct list_head sibling; /* linkage in my parent's children list */
1344 struct task_struct *group_leader; /* threadgroup leader */
1347 * ptraced is the list of tasks this task is using ptrace on.
1348 * This includes both natural children and PTRACE_ATTACH targets.
1349 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1351 struct list_head ptraced;
1352 struct list_head ptrace_entry;
1354 /* PID/PID hash table linkage. */
1355 struct pid_link pids[PIDTYPE_MAX];
1356 struct list_head thread_group;
1358 struct completion *vfork_done; /* for vfork() */
1359 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1360 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1362 cputime_t utime, stime, utimescaled, stimescaled;
1364 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1365 cputime_t prev_utime, prev_stime;
1367 unsigned long nvcsw, nivcsw; /* context switch counts */
1368 struct timespec start_time; /* monotonic time */
1369 struct timespec real_start_time; /* boot based time */
1370 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1371 unsigned long min_flt, maj_flt;
1373 struct task_cputime cputime_expires;
1374 struct list_head cpu_timers[3];
1376 /* process credentials */
1377 const struct cred __rcu *real_cred; /* objective and real subjective task
1378 * credentials (COW) */
1379 const struct cred __rcu *cred; /* effective (overridable) subjective task
1380 * credentials (COW) */
1381 char comm[TASK_COMM_LEN]; /* executable name excluding path
1382 - access with [gs]et_task_comm (which lock
1383 it with task_lock())
1384 - initialized normally by setup_new_exec */
1385 /* file system info */
1386 int link_count, total_link_count;
1387 #ifdef CONFIG_SYSVIPC
1389 struct sysv_sem sysvsem;
1391 #ifdef CONFIG_DETECT_HUNG_TASK
1392 /* hung task detection */
1393 unsigned long last_switch_count;
1395 /* CPU-specific state of this task */
1396 struct thread_struct thread;
1397 /* filesystem information */
1398 struct fs_struct *fs;
1399 /* open file information */
1400 struct files_struct *files;
1402 struct nsproxy *nsproxy;
1403 /* signal handlers */
1404 struct signal_struct *signal;
1405 struct sighand_struct *sighand;
1407 sigset_t blocked, real_blocked;
1408 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1409 struct sigpending pending;
1411 unsigned long sas_ss_sp;
1413 int (*notifier)(void *priv);
1414 void *notifier_data;
1415 sigset_t *notifier_mask;
1416 struct callback_head *task_works;
1418 struct audit_context *audit_context;
1419 #ifdef CONFIG_AUDITSYSCALL
1421 unsigned int sessionid;
1423 struct seccomp seccomp;
1425 /* Thread group tracking */
1428 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1430 spinlock_t alloc_lock;
1432 /* Protection of the PI data structures: */
1433 raw_spinlock_t pi_lock;
1435 #ifdef CONFIG_RT_MUTEXES
1436 /* PI waiters blocked on a rt_mutex held by this task */
1437 struct plist_head pi_waiters;
1438 /* Deadlock detection and priority inheritance handling */
1439 struct rt_mutex_waiter *pi_blocked_on;
1442 #ifdef CONFIG_DEBUG_MUTEXES
1443 /* mutex deadlock detection */
1444 struct mutex_waiter *blocked_on;
1446 #ifdef CONFIG_TRACE_IRQFLAGS
1447 unsigned int irq_events;
1448 unsigned long hardirq_enable_ip;
1449 unsigned long hardirq_disable_ip;
1450 unsigned int hardirq_enable_event;
1451 unsigned int hardirq_disable_event;
1452 int hardirqs_enabled;
1453 int hardirq_context;
1454 unsigned long softirq_disable_ip;
1455 unsigned long softirq_enable_ip;
1456 unsigned int softirq_disable_event;
1457 unsigned int softirq_enable_event;
1458 int softirqs_enabled;
1459 int softirq_context;
1461 #ifdef CONFIG_LOCKDEP
1462 # define MAX_LOCK_DEPTH 48UL
1465 unsigned int lockdep_recursion;
1466 struct held_lock held_locks[MAX_LOCK_DEPTH];
1467 gfp_t lockdep_reclaim_gfp;
1470 /* journalling filesystem info */
1473 /* stacked block device info */
1474 struct bio_list *bio_list;
1477 /* stack plugging */
1478 struct blk_plug *plug;
1482 struct reclaim_state *reclaim_state;
1484 struct backing_dev_info *backing_dev_info;
1486 struct io_context *io_context;
1488 unsigned long ptrace_message;
1489 siginfo_t *last_siginfo; /* For ptrace use. */
1490 struct task_io_accounting ioac;
1491 #if defined(CONFIG_TASK_XACCT)
1492 u64 acct_rss_mem1; /* accumulated rss usage */
1493 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1494 cputime_t acct_timexpd; /* stime + utime since last update */
1496 #ifdef CONFIG_CPUSETS
1497 nodemask_t mems_allowed; /* Protected by alloc_lock */
1498 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1499 int cpuset_mem_spread_rotor;
1500 int cpuset_slab_spread_rotor;
1502 #ifdef CONFIG_CGROUPS
1503 /* Control Group info protected by css_set_lock */
1504 struct css_set __rcu *cgroups;
1505 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1506 struct list_head cg_list;
1509 struct robust_list_head __user *robust_list;
1510 #ifdef CONFIG_COMPAT
1511 struct compat_robust_list_head __user *compat_robust_list;
1513 struct list_head pi_state_list;
1514 struct futex_pi_state *pi_state_cache;
1516 #ifdef CONFIG_PERF_EVENTS
1517 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1518 struct mutex perf_event_mutex;
1519 struct list_head perf_event_list;
1522 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1524 short pref_node_fork;
1526 struct rcu_head rcu;
1529 * cache last used pipe for splice
1531 struct pipe_inode_info *splice_pipe;
1533 struct page_frag task_frag;
1535 #ifdef CONFIG_TASK_DELAY_ACCT
1536 struct task_delay_info *delays;
1538 #ifdef CONFIG_FAULT_INJECTION
1542 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1543 * balance_dirty_pages() for some dirty throttling pause
1546 int nr_dirtied_pause;
1547 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1549 #ifdef CONFIG_LATENCYTOP
1550 int latency_record_count;
1551 struct latency_record latency_record[LT_SAVECOUNT];
1554 * time slack values; these are used to round up poll() and
1555 * select() etc timeout values. These are in nanoseconds.
1557 unsigned long timer_slack_ns;
1558 unsigned long default_timer_slack_ns;
1560 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1561 /* Index of current stored address in ret_stack */
1563 /* Stack of return addresses for return function tracing */
1564 struct ftrace_ret_stack *ret_stack;
1565 /* time stamp for last schedule */
1566 unsigned long long ftrace_timestamp;
1568 * Number of functions that haven't been traced
1569 * because of depth overrun.
1571 atomic_t trace_overrun;
1572 /* Pause for the tracing */
1573 atomic_t tracing_graph_pause;
1575 #ifdef CONFIG_TRACING
1576 /* state flags for use by tracers */
1577 unsigned long trace;
1578 /* bitmask and counter of trace recursion */
1579 unsigned long trace_recursion;
1580 #endif /* CONFIG_TRACING */
1581 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1582 struct memcg_batch_info {
1583 int do_batch; /* incremented when batch uncharge started */
1584 struct mem_cgroup *memcg; /* target memcg of uncharge */
1585 unsigned long nr_pages; /* uncharged usage */
1586 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1589 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1590 atomic_t ptrace_bp_refcnt;
1592 #ifdef CONFIG_UPROBES
1593 struct uprobe_task *utask;
1597 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1598 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1601 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1602 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1603 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1604 * values are inverted: lower p->prio value means higher priority.
1606 * The MAX_USER_RT_PRIO value allows the actual maximum
1607 * RT priority to be separate from the value exported to
1608 * user-space. This allows kernel threads to set their
1609 * priority to a value higher than any user task. Note:
1610 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1613 #define MAX_USER_RT_PRIO 100
1614 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1616 #define MAX_PRIO (MAX_RT_PRIO + 40)
1617 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1619 static inline int rt_prio(int prio)
1621 if (unlikely(prio < MAX_RT_PRIO))
1626 static inline int rt_task(struct task_struct *p)
1628 return rt_prio(p->prio);
1631 static inline struct pid *task_pid(struct task_struct *task)
1633 return task->pids[PIDTYPE_PID].pid;
1636 static inline struct pid *task_tgid(struct task_struct *task)
1638 return task->group_leader->pids[PIDTYPE_PID].pid;
1642 * Without tasklist or rcu lock it is not safe to dereference
1643 * the result of task_pgrp/task_session even if task == current,
1644 * we can race with another thread doing sys_setsid/sys_setpgid.
1646 static inline struct pid *task_pgrp(struct task_struct *task)
1648 return task->group_leader->pids[PIDTYPE_PGID].pid;
1651 static inline struct pid *task_session(struct task_struct *task)
1653 return task->group_leader->pids[PIDTYPE_SID].pid;
1656 struct pid_namespace;
1659 * the helpers to get the task's different pids as they are seen
1660 * from various namespaces
1662 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1663 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1665 * task_xid_nr_ns() : id seen from the ns specified;
1667 * set_task_vxid() : assigns a virtual id to a task;
1669 * see also pid_nr() etc in include/linux/pid.h
1671 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1672 struct pid_namespace *ns);
1674 static inline pid_t task_pid_nr(struct task_struct *tsk)
1679 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1680 struct pid_namespace *ns)
1682 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1685 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1687 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1691 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1696 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1698 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1700 return pid_vnr(task_tgid(tsk));
1704 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1705 struct pid_namespace *ns)
1707 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1710 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1712 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1716 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1717 struct pid_namespace *ns)
1719 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1722 static inline pid_t task_session_vnr(struct task_struct *tsk)
1724 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1727 /* obsolete, do not use */
1728 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1730 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1734 * pid_alive - check that a task structure is not stale
1735 * @p: Task structure to be checked.
1737 * Test if a process is not yet dead (at most zombie state)
1738 * If pid_alive fails, then pointers within the task structure
1739 * can be stale and must not be dereferenced.
1741 static inline int pid_alive(struct task_struct *p)
1743 return p->pids[PIDTYPE_PID].pid != NULL;
1747 * is_global_init - check if a task structure is init
1748 * @tsk: Task structure to be checked.
1750 * Check if a task structure is the first user space task the kernel created.
1752 static inline int is_global_init(struct task_struct *tsk)
1754 return tsk->pid == 1;
1758 * is_container_init:
1759 * check whether in the task is init in its own pid namespace.
1761 extern int is_container_init(struct task_struct *tsk);
1763 extern struct pid *cad_pid;
1765 extern void free_task(struct task_struct *tsk);
1766 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1768 extern void __put_task_struct(struct task_struct *t);
1770 static inline void put_task_struct(struct task_struct *t)
1772 if (atomic_dec_and_test(&t->usage))
1773 __put_task_struct(t);
1776 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1777 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1782 #define PF_EXITING 0x00000004 /* getting shut down */
1783 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1784 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1785 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1786 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1787 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1788 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1789 #define PF_DUMPCORE 0x00000200 /* dumped core */
1790 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1791 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1792 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1793 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1794 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1795 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1796 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1797 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1798 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1799 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1800 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1801 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1802 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1803 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1804 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1805 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1806 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1807 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1808 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1811 * Only the _current_ task can read/write to tsk->flags, but other
1812 * tasks can access tsk->flags in readonly mode for example
1813 * with tsk_used_math (like during threaded core dumping).
1814 * There is however an exception to this rule during ptrace
1815 * or during fork: the ptracer task is allowed to write to the
1816 * child->flags of its traced child (same goes for fork, the parent
1817 * can write to the child->flags), because we're guaranteed the
1818 * child is not running and in turn not changing child->flags
1819 * at the same time the parent does it.
1821 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1822 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1823 #define clear_used_math() clear_stopped_child_used_math(current)
1824 #define set_used_math() set_stopped_child_used_math(current)
1825 #define conditional_stopped_child_used_math(condition, child) \
1826 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1827 #define conditional_used_math(condition) \
1828 conditional_stopped_child_used_math(condition, current)
1829 #define copy_to_stopped_child_used_math(child) \
1830 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1831 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1832 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1833 #define used_math() tsk_used_math(current)
1836 * task->jobctl flags
1838 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1840 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1841 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1842 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1843 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1844 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1845 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1846 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1848 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1849 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1850 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1851 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1852 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1853 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1854 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1856 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1857 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1859 extern bool task_set_jobctl_pending(struct task_struct *task,
1861 extern void task_clear_jobctl_trapping(struct task_struct *task);
1862 extern void task_clear_jobctl_pending(struct task_struct *task,
1865 #ifdef CONFIG_PREEMPT_RCU
1867 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1868 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1870 static inline void rcu_copy_process(struct task_struct *p)
1872 p->rcu_read_lock_nesting = 0;
1873 p->rcu_read_unlock_special = 0;
1874 #ifdef CONFIG_TREE_PREEMPT_RCU
1875 p->rcu_blocked_node = NULL;
1876 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1877 #ifdef CONFIG_RCU_BOOST
1878 p->rcu_boost_mutex = NULL;
1879 #endif /* #ifdef CONFIG_RCU_BOOST */
1880 INIT_LIST_HEAD(&p->rcu_node_entry);
1885 static inline void rcu_copy_process(struct task_struct *p)
1891 static inline void tsk_restore_flags(struct task_struct *task,
1892 unsigned long orig_flags, unsigned long flags)
1894 task->flags &= ~flags;
1895 task->flags |= orig_flags & flags;
1899 extern void do_set_cpus_allowed(struct task_struct *p,
1900 const struct cpumask *new_mask);
1902 extern int set_cpus_allowed_ptr(struct task_struct *p,
1903 const struct cpumask *new_mask);
1905 static inline void do_set_cpus_allowed(struct task_struct *p,
1906 const struct cpumask *new_mask)
1909 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1910 const struct cpumask *new_mask)
1912 if (!cpumask_test_cpu(0, new_mask))
1919 void calc_load_enter_idle(void);
1920 void calc_load_exit_idle(void);
1922 static inline void calc_load_enter_idle(void) { }
1923 static inline void calc_load_exit_idle(void) { }
1924 #endif /* CONFIG_NO_HZ */
1926 #ifndef CONFIG_CPUMASK_OFFSTACK
1927 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1929 return set_cpus_allowed_ptr(p, &new_mask);
1934 * Do not use outside of architecture code which knows its limitations.
1936 * sched_clock() has no promise of monotonicity or bounded drift between
1937 * CPUs, use (which you should not) requires disabling IRQs.
1939 * Please use one of the three interfaces below.
1941 extern unsigned long long notrace sched_clock(void);
1943 * See the comment in kernel/sched/clock.c
1945 extern u64 cpu_clock(int cpu);
1946 extern u64 local_clock(void);
1947 extern u64 sched_clock_cpu(int cpu);
1950 extern void sched_clock_init(void);
1952 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1953 static inline void sched_clock_tick(void)
1957 static inline void sched_clock_idle_sleep_event(void)
1961 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1966 * Architectures can set this to 1 if they have specified
1967 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1968 * but then during bootup it turns out that sched_clock()
1969 * is reliable after all:
1971 extern int sched_clock_stable;
1973 extern void sched_clock_tick(void);
1974 extern void sched_clock_idle_sleep_event(void);
1975 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1978 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1980 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1981 * The reason for this explicit opt-in is not to have perf penalty with
1982 * slow sched_clocks.
1984 extern void enable_sched_clock_irqtime(void);
1985 extern void disable_sched_clock_irqtime(void);
1987 static inline void enable_sched_clock_irqtime(void) {}
1988 static inline void disable_sched_clock_irqtime(void) {}
1991 extern unsigned long long
1992 task_sched_runtime(struct task_struct *task);
1994 /* sched_exec is called by processes performing an exec */
1996 extern void sched_exec(void);
1998 #define sched_exec() {}
2001 extern void sched_clock_idle_sleep_event(void);
2002 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2004 #ifdef CONFIG_HOTPLUG_CPU
2005 extern void idle_task_exit(void);
2007 static inline void idle_task_exit(void) {}
2010 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2011 extern void wake_up_idle_cpu(int cpu);
2013 static inline void wake_up_idle_cpu(int cpu) { }
2016 extern unsigned int sysctl_sched_latency;
2017 extern unsigned int sysctl_sched_min_granularity;
2018 extern unsigned int sysctl_sched_wakeup_granularity;
2019 extern unsigned int sysctl_sched_child_runs_first;
2021 enum sched_tunable_scaling {
2022 SCHED_TUNABLESCALING_NONE,
2023 SCHED_TUNABLESCALING_LOG,
2024 SCHED_TUNABLESCALING_LINEAR,
2025 SCHED_TUNABLESCALING_END,
2027 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2029 #ifdef CONFIG_SCHED_DEBUG
2030 extern unsigned int sysctl_sched_migration_cost;
2031 extern unsigned int sysctl_sched_nr_migrate;
2032 extern unsigned int sysctl_sched_time_avg;
2033 extern unsigned int sysctl_timer_migration;
2034 extern unsigned int sysctl_sched_shares_window;
2036 int sched_proc_update_handler(struct ctl_table *table, int write,
2037 void __user *buffer, size_t *length,
2040 #ifdef CONFIG_SCHED_DEBUG
2041 static inline unsigned int get_sysctl_timer_migration(void)
2043 return sysctl_timer_migration;
2046 static inline unsigned int get_sysctl_timer_migration(void)
2051 extern unsigned int sysctl_sched_rt_period;
2052 extern int sysctl_sched_rt_runtime;
2054 int sched_rt_handler(struct ctl_table *table, int write,
2055 void __user *buffer, size_t *lenp,
2058 #ifdef CONFIG_SCHED_AUTOGROUP
2059 extern unsigned int sysctl_sched_autogroup_enabled;
2061 extern void sched_autogroup_create_attach(struct task_struct *p);
2062 extern void sched_autogroup_detach(struct task_struct *p);
2063 extern void sched_autogroup_fork(struct signal_struct *sig);
2064 extern void sched_autogroup_exit(struct signal_struct *sig);
2065 #ifdef CONFIG_PROC_FS
2066 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2067 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2070 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2071 static inline void sched_autogroup_detach(struct task_struct *p) { }
2072 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2073 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2076 #ifdef CONFIG_CFS_BANDWIDTH
2077 extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2080 #ifdef CONFIG_RT_MUTEXES
2081 extern int rt_mutex_getprio(struct task_struct *p);
2082 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2083 extern void rt_mutex_adjust_pi(struct task_struct *p);
2084 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2086 return tsk->pi_blocked_on != NULL;
2089 static inline int rt_mutex_getprio(struct task_struct *p)
2091 return p->normal_prio;
2093 # define rt_mutex_adjust_pi(p) do { } while (0)
2094 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2100 extern bool yield_to(struct task_struct *p, bool preempt);
2101 extern void set_user_nice(struct task_struct *p, long nice);
2102 extern int task_prio(const struct task_struct *p);
2103 extern int task_nice(const struct task_struct *p);
2104 extern int can_nice(const struct task_struct *p, const int nice);
2105 extern int task_curr(const struct task_struct *p);
2106 extern int idle_cpu(int cpu);
2107 extern int sched_setscheduler(struct task_struct *, int,
2108 const struct sched_param *);
2109 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2110 const struct sched_param *);
2111 extern struct task_struct *idle_task(int cpu);
2113 * is_idle_task - is the specified task an idle task?
2114 * @p: the task in question.
2116 static inline bool is_idle_task(const struct task_struct *p)
2120 extern struct task_struct *curr_task(int cpu);
2121 extern void set_curr_task(int cpu, struct task_struct *p);
2126 * The default (Linux) execution domain.
2128 extern struct exec_domain default_exec_domain;
2130 union thread_union {
2131 struct thread_info thread_info;
2132 unsigned long stack[THREAD_SIZE/sizeof(long)];
2135 #ifndef __HAVE_ARCH_KSTACK_END
2136 static inline int kstack_end(void *addr)
2138 /* Reliable end of stack detection:
2139 * Some APM bios versions misalign the stack
2141 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2145 extern union thread_union init_thread_union;
2146 extern struct task_struct init_task;
2148 extern struct mm_struct init_mm;
2150 extern struct pid_namespace init_pid_ns;
2153 * find a task by one of its numerical ids
2155 * find_task_by_pid_ns():
2156 * finds a task by its pid in the specified namespace
2157 * find_task_by_vpid():
2158 * finds a task by its virtual pid
2160 * see also find_vpid() etc in include/linux/pid.h
2163 extern struct task_struct *find_task_by_vpid(pid_t nr);
2164 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2165 struct pid_namespace *ns);
2167 extern void __set_special_pids(struct pid *pid);
2169 /* per-UID process charging. */
2170 extern struct user_struct * alloc_uid(kuid_t);
2171 static inline struct user_struct *get_uid(struct user_struct *u)
2173 atomic_inc(&u->__count);
2176 extern void free_uid(struct user_struct *);
2178 #include <asm/current.h>
2180 extern void xtime_update(unsigned long ticks);
2182 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2183 extern int wake_up_process(struct task_struct *tsk);
2184 extern void wake_up_new_task(struct task_struct *tsk);
2186 extern void kick_process(struct task_struct *tsk);
2188 static inline void kick_process(struct task_struct *tsk) { }
2190 extern void sched_fork(struct task_struct *p);
2191 extern void sched_dead(struct task_struct *p);
2193 extern void proc_caches_init(void);
2194 extern void flush_signals(struct task_struct *);
2195 extern void __flush_signals(struct task_struct *);
2196 extern void ignore_signals(struct task_struct *);
2197 extern void flush_signal_handlers(struct task_struct *, int force_default);
2198 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2200 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2202 unsigned long flags;
2205 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2206 ret = dequeue_signal(tsk, mask, info);
2207 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2212 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2214 extern void unblock_all_signals(void);
2215 extern void release_task(struct task_struct * p);
2216 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2217 extern int force_sigsegv(int, struct task_struct *);
2218 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2219 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2220 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2221 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2222 const struct cred *, u32);
2223 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2224 extern int kill_pid(struct pid *pid, int sig, int priv);
2225 extern int kill_proc_info(int, struct siginfo *, pid_t);
2226 extern __must_check bool do_notify_parent(struct task_struct *, int);
2227 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2228 extern void force_sig(int, struct task_struct *);
2229 extern int send_sig(int, struct task_struct *, int);
2230 extern int zap_other_threads(struct task_struct *p);
2231 extern struct sigqueue *sigqueue_alloc(void);
2232 extern void sigqueue_free(struct sigqueue *);
2233 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2234 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2235 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2237 static inline void restore_saved_sigmask(void)
2239 if (test_and_clear_restore_sigmask())
2240 __set_current_blocked(¤t->saved_sigmask);
2243 static inline sigset_t *sigmask_to_save(void)
2245 sigset_t *res = ¤t->blocked;
2246 if (unlikely(test_restore_sigmask()))
2247 res = ¤t->saved_sigmask;
2251 static inline int kill_cad_pid(int sig, int priv)
2253 return kill_pid(cad_pid, sig, priv);
2256 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2257 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2258 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2259 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2262 * True if we are on the alternate signal stack.
2264 static inline int on_sig_stack(unsigned long sp)
2266 #ifdef CONFIG_STACK_GROWSUP
2267 return sp >= current->sas_ss_sp &&
2268 sp - current->sas_ss_sp < current->sas_ss_size;
2270 return sp > current->sas_ss_sp &&
2271 sp - current->sas_ss_sp <= current->sas_ss_size;
2275 static inline int sas_ss_flags(unsigned long sp)
2277 return (current->sas_ss_size == 0 ? SS_DISABLE
2278 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2282 * Routines for handling mm_structs
2284 extern struct mm_struct * mm_alloc(void);
2286 /* mmdrop drops the mm and the page tables */
2287 extern void __mmdrop(struct mm_struct *);
2288 static inline void mmdrop(struct mm_struct * mm)
2290 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2294 /* mmput gets rid of the mappings and all user-space */
2295 extern void mmput(struct mm_struct *);
2296 /* Grab a reference to a task's mm, if it is not already going away */
2297 extern struct mm_struct *get_task_mm(struct task_struct *task);
2299 * Grab a reference to a task's mm, if it is not already going away
2300 * and ptrace_may_access with the mode parameter passed to it
2303 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2304 /* Remove the current tasks stale references to the old mm_struct */
2305 extern void mm_release(struct task_struct *, struct mm_struct *);
2306 /* Allocate a new mm structure and copy contents from tsk->mm */
2307 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2309 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2310 struct task_struct *, struct pt_regs *);
2311 extern void flush_thread(void);
2312 extern void exit_thread(void);
2314 extern void exit_files(struct task_struct *);
2315 extern void __cleanup_sighand(struct sighand_struct *);
2317 extern void exit_itimers(struct signal_struct *);
2318 extern void flush_itimer_signals(void);
2320 extern void do_group_exit(int);
2322 extern void daemonize(const char *, ...);
2323 extern int allow_signal(int);
2324 extern int disallow_signal(int);
2326 extern int do_execve(const char *,
2327 const char __user * const __user *,
2328 const char __user * const __user *, struct pt_regs *);
2329 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2330 struct task_struct *fork_idle(int);
2332 extern void set_task_comm(struct task_struct *tsk, char *from);
2333 extern char *get_task_comm(char *to, struct task_struct *tsk);
2336 void scheduler_ipi(void);
2337 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2339 static inline void scheduler_ipi(void) { }
2340 static inline unsigned long wait_task_inactive(struct task_struct *p,
2347 #define next_task(p) \
2348 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2350 #define for_each_process(p) \
2351 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2353 extern bool current_is_single_threaded(void);
2356 * Careful: do_each_thread/while_each_thread is a double loop so
2357 * 'break' will not work as expected - use goto instead.
2359 #define do_each_thread(g, t) \
2360 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2362 #define while_each_thread(g, t) \
2363 while ((t = next_thread(t)) != g)
2365 static inline int get_nr_threads(struct task_struct *tsk)
2367 return tsk->signal->nr_threads;
2370 static inline bool thread_group_leader(struct task_struct *p)
2372 return p->exit_signal >= 0;
2375 /* Do to the insanities of de_thread it is possible for a process
2376 * to have the pid of the thread group leader without actually being
2377 * the thread group leader. For iteration through the pids in proc
2378 * all we care about is that we have a task with the appropriate
2379 * pid, we don't actually care if we have the right task.
2381 static inline int has_group_leader_pid(struct task_struct *p)
2383 return p->pid == p->tgid;
2387 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2389 return p1->tgid == p2->tgid;
2392 static inline struct task_struct *next_thread(const struct task_struct *p)
2394 return list_entry_rcu(p->thread_group.next,
2395 struct task_struct, thread_group);
2398 static inline int thread_group_empty(struct task_struct *p)
2400 return list_empty(&p->thread_group);
2403 #define delay_group_leader(p) \
2404 (thread_group_leader(p) && !thread_group_empty(p))
2407 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2408 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2409 * pins the final release of task.io_context. Also protects ->cpuset and
2410 * ->cgroup.subsys[]. And ->vfork_done.
2412 * Nests both inside and outside of read_lock(&tasklist_lock).
2413 * It must not be nested with write_lock_irq(&tasklist_lock),
2414 * neither inside nor outside.
2416 static inline void task_lock(struct task_struct *p)
2418 spin_lock(&p->alloc_lock);
2421 static inline void task_unlock(struct task_struct *p)
2423 spin_unlock(&p->alloc_lock);
2426 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2427 unsigned long *flags);
2429 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2430 unsigned long *flags)
2432 struct sighand_struct *ret;
2434 ret = __lock_task_sighand(tsk, flags);
2435 (void)__cond_lock(&tsk->sighand->siglock, ret);
2439 static inline void unlock_task_sighand(struct task_struct *tsk,
2440 unsigned long *flags)
2442 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2445 #ifdef CONFIG_CGROUPS
2446 static inline void threadgroup_change_begin(struct task_struct *tsk)
2448 down_read(&tsk->signal->group_rwsem);
2450 static inline void threadgroup_change_end(struct task_struct *tsk)
2452 up_read(&tsk->signal->group_rwsem);
2456 * threadgroup_lock - lock threadgroup
2457 * @tsk: member task of the threadgroup to lock
2459 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2460 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2461 * perform exec. This is useful for cases where the threadgroup needs to
2462 * stay stable across blockable operations.
2464 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2465 * synchronization. While held, no new task will be added to threadgroup
2466 * and no existing live task will have its PF_EXITING set.
2468 * During exec, a task goes and puts its thread group through unusual
2469 * changes. After de-threading, exclusive access is assumed to resources
2470 * which are usually shared by tasks in the same group - e.g. sighand may
2471 * be replaced with a new one. Also, the exec'ing task takes over group
2472 * leader role including its pid. Exclude these changes while locked by
2473 * grabbing cred_guard_mutex which is used to synchronize exec path.
2475 static inline void threadgroup_lock(struct task_struct *tsk)
2478 * exec uses exit for de-threading nesting group_rwsem inside
2479 * cred_guard_mutex. Grab cred_guard_mutex first.
2481 mutex_lock(&tsk->signal->cred_guard_mutex);
2482 down_write(&tsk->signal->group_rwsem);
2486 * threadgroup_unlock - unlock threadgroup
2487 * @tsk: member task of the threadgroup to unlock
2489 * Reverse threadgroup_lock().
2491 static inline void threadgroup_unlock(struct task_struct *tsk)
2493 up_write(&tsk->signal->group_rwsem);
2494 mutex_unlock(&tsk->signal->cred_guard_mutex);
2497 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2498 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2499 static inline void threadgroup_lock(struct task_struct *tsk) {}
2500 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2503 #ifndef __HAVE_THREAD_FUNCTIONS
2505 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2506 #define task_stack_page(task) ((task)->stack)
2508 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2510 *task_thread_info(p) = *task_thread_info(org);
2511 task_thread_info(p)->task = p;
2514 static inline unsigned long *end_of_stack(struct task_struct *p)
2516 return (unsigned long *)(task_thread_info(p) + 1);
2521 static inline int object_is_on_stack(void *obj)
2523 void *stack = task_stack_page(current);
2525 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2528 extern void thread_info_cache_init(void);
2530 #ifdef CONFIG_DEBUG_STACK_USAGE
2531 static inline unsigned long stack_not_used(struct task_struct *p)
2533 unsigned long *n = end_of_stack(p);
2535 do { /* Skip over canary */
2539 return (unsigned long)n - (unsigned long)end_of_stack(p);
2543 /* set thread flags in other task's structures
2544 * - see asm/thread_info.h for TIF_xxxx flags available
2546 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2548 set_ti_thread_flag(task_thread_info(tsk), flag);
2551 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2553 clear_ti_thread_flag(task_thread_info(tsk), flag);
2556 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2558 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2561 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2563 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2566 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2568 return test_ti_thread_flag(task_thread_info(tsk), flag);
2571 static inline void set_tsk_need_resched(struct task_struct *tsk)
2573 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2576 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2578 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2581 static inline int test_tsk_need_resched(struct task_struct *tsk)
2583 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2586 static inline int restart_syscall(void)
2588 set_tsk_thread_flag(current, TIF_SIGPENDING);
2589 return -ERESTARTNOINTR;
2592 static inline int signal_pending(struct task_struct *p)
2594 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2597 static inline int __fatal_signal_pending(struct task_struct *p)
2599 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2602 static inline int fatal_signal_pending(struct task_struct *p)
2604 return signal_pending(p) && __fatal_signal_pending(p);
2607 static inline int signal_pending_state(long state, struct task_struct *p)
2609 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2611 if (!signal_pending(p))
2614 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2617 static inline int need_resched(void)
2619 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2623 * cond_resched() and cond_resched_lock(): latency reduction via
2624 * explicit rescheduling in places that are safe. The return
2625 * value indicates whether a reschedule was done in fact.
2626 * cond_resched_lock() will drop the spinlock before scheduling,
2627 * cond_resched_softirq() will enable bhs before scheduling.
2629 extern int _cond_resched(void);
2631 #define cond_resched() ({ \
2632 __might_sleep(__FILE__, __LINE__, 0); \
2636 extern int __cond_resched_lock(spinlock_t *lock);
2638 #ifdef CONFIG_PREEMPT_COUNT
2639 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2641 #define PREEMPT_LOCK_OFFSET 0
2644 #define cond_resched_lock(lock) ({ \
2645 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2646 __cond_resched_lock(lock); \
2649 extern int __cond_resched_softirq(void);
2651 #define cond_resched_softirq() ({ \
2652 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2653 __cond_resched_softirq(); \
2657 * Does a critical section need to be broken due to another
2658 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2659 * but a general need for low latency)
2661 static inline int spin_needbreak(spinlock_t *lock)
2663 #ifdef CONFIG_PREEMPT
2664 return spin_is_contended(lock);
2671 * Thread group CPU time accounting.
2673 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2674 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2676 static inline void thread_group_cputime_init(struct signal_struct *sig)
2678 raw_spin_lock_init(&sig->cputimer.lock);
2682 * Reevaluate whether the task has signals pending delivery.
2683 * Wake the task if so.
2684 * This is required every time the blocked sigset_t changes.
2685 * callers must hold sighand->siglock.
2687 extern void recalc_sigpending_and_wake(struct task_struct *t);
2688 extern void recalc_sigpending(void);
2690 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2693 * Wrappers for p->thread_info->cpu access. No-op on UP.
2697 static inline unsigned int task_cpu(const struct task_struct *p)
2699 return task_thread_info(p)->cpu;
2702 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2706 static inline unsigned int task_cpu(const struct task_struct *p)
2711 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2715 #endif /* CONFIG_SMP */
2717 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2718 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2720 extern void normalize_rt_tasks(void);
2722 #ifdef CONFIG_CGROUP_SCHED
2724 extern struct task_group root_task_group;
2726 extern struct task_group *sched_create_group(struct task_group *parent);
2727 extern void sched_destroy_group(struct task_group *tg);
2728 extern void sched_move_task(struct task_struct *tsk);
2729 #ifdef CONFIG_FAIR_GROUP_SCHED
2730 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2731 extern unsigned long sched_group_shares(struct task_group *tg);
2733 #ifdef CONFIG_RT_GROUP_SCHED
2734 extern int sched_group_set_rt_runtime(struct task_group *tg,
2735 long rt_runtime_us);
2736 extern long sched_group_rt_runtime(struct task_group *tg);
2737 extern int sched_group_set_rt_period(struct task_group *tg,
2739 extern long sched_group_rt_period(struct task_group *tg);
2740 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2742 #endif /* CONFIG_CGROUP_SCHED */
2744 extern int task_can_switch_user(struct user_struct *up,
2745 struct task_struct *tsk);
2747 #ifdef CONFIG_TASK_XACCT
2748 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2750 tsk->ioac.rchar += amt;
2753 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2755 tsk->ioac.wchar += amt;
2758 static inline void inc_syscr(struct task_struct *tsk)
2763 static inline void inc_syscw(struct task_struct *tsk)
2768 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2772 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2776 static inline void inc_syscr(struct task_struct *tsk)
2780 static inline void inc_syscw(struct task_struct *tsk)
2785 #ifndef TASK_SIZE_OF
2786 #define TASK_SIZE_OF(tsk) TASK_SIZE
2789 #ifdef CONFIG_MM_OWNER
2790 extern void mm_update_next_owner(struct mm_struct *mm);
2791 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2793 static inline void mm_update_next_owner(struct mm_struct *mm)
2797 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2800 #endif /* CONFIG_MM_OWNER */
2802 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2805 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2808 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2811 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2814 static inline unsigned long rlimit(unsigned int limit)
2816 return task_rlimit(current, limit);
2819 static inline unsigned long rlimit_max(unsigned int limit)
2821 return task_rlimit_max(current, limit);
2824 #endif /* __KERNEL__ */