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>
66 #include <asm/system.h>
68 #include <asm/ptrace.h>
69 #include <asm/cputime.h>
71 #include <linux/smp.h>
72 #include <linux/sem.h>
73 #include <linux/signal.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rculist.h>
83 #include <linux/rtmutex.h>
85 #include <linux/time.h>
86 #include <linux/param.h>
87 #include <linux/resource.h>
88 #include <linux/timer.h>
89 #include <linux/hrtimer.h>
90 #include <linux/task_io_accounting.h>
91 #include <linux/latencytop.h>
92 #include <linux/cred.h>
93 #include <linux/llist.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_FREEZING) == 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 int get_nohz_timer_target(void);
279 static inline void select_nohz_load_balancer(int stop_tick) { }
283 * Only dump TASK_* tasks. (0 for all tasks)
285 extern void show_state_filter(unsigned long state_filter);
287 static inline void show_state(void)
289 show_state_filter(0);
292 extern void show_regs(struct pt_regs *);
295 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
296 * task), SP is the stack pointer of the first frame that should be shown in the back
297 * trace (or NULL if the entire call-chain of the task should be shown).
299 extern void show_stack(struct task_struct *task, unsigned long *sp);
301 void io_schedule(void);
302 long io_schedule_timeout(long timeout);
304 extern void cpu_init (void);
305 extern void trap_init(void);
306 extern void update_process_times(int user);
307 extern void scheduler_tick(void);
309 extern void sched_show_task(struct task_struct *p);
311 #ifdef CONFIG_LOCKUP_DETECTOR
312 extern void touch_softlockup_watchdog(void);
313 extern void touch_softlockup_watchdog_sync(void);
314 extern void touch_all_softlockup_watchdogs(void);
315 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
317 size_t *lenp, loff_t *ppos);
318 extern unsigned int softlockup_panic;
319 void lockup_detector_init(void);
321 static inline void touch_softlockup_watchdog(void)
324 static inline void touch_softlockup_watchdog_sync(void)
327 static inline void touch_all_softlockup_watchdogs(void)
330 static inline void lockup_detector_init(void)
335 #ifdef CONFIG_DETECT_HUNG_TASK
336 extern unsigned int sysctl_hung_task_panic;
337 extern unsigned long sysctl_hung_task_check_count;
338 extern unsigned long sysctl_hung_task_timeout_secs;
339 extern unsigned long sysctl_hung_task_warnings;
340 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
342 size_t *lenp, loff_t *ppos);
344 /* Avoid need for ifdefs elsewhere in the code */
345 enum { sysctl_hung_task_timeout_secs = 0 };
348 /* Attach to any functions which should be ignored in wchan output. */
349 #define __sched __attribute__((__section__(".sched.text")))
351 /* Linker adds these: start and end of __sched functions */
352 extern char __sched_text_start[], __sched_text_end[];
354 /* Is this address in the __sched functions? */
355 extern int in_sched_functions(unsigned long addr);
357 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
358 extern signed long schedule_timeout(signed long timeout);
359 extern signed long schedule_timeout_interruptible(signed long timeout);
360 extern signed long schedule_timeout_killable(signed long timeout);
361 extern signed long schedule_timeout_uninterruptible(signed long timeout);
362 asmlinkage void schedule(void);
363 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
366 struct user_namespace;
369 * Default maximum number of active map areas, this limits the number of vmas
370 * per mm struct. Users can overwrite this number by sysctl but there is a
373 * When a program's coredump is generated as ELF format, a section is created
374 * per a vma. In ELF, the number of sections is represented in unsigned short.
375 * This means the number of sections should be smaller than 65535 at coredump.
376 * Because the kernel adds some informative sections to a image of program at
377 * generating coredump, we need some margin. The number of extra sections is
378 * 1-3 now and depends on arch. We use "5" as safe margin, here.
380 #define MAPCOUNT_ELF_CORE_MARGIN (5)
381 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
383 extern int sysctl_max_map_count;
385 #include <linux/aio.h>
388 extern void arch_pick_mmap_layout(struct mm_struct *mm);
390 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
391 unsigned long, unsigned long);
393 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
394 unsigned long len, unsigned long pgoff,
395 unsigned long flags);
396 extern void arch_unmap_area(struct mm_struct *, unsigned long);
397 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
399 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
403 extern void set_dumpable(struct mm_struct *mm, int value);
404 extern int get_dumpable(struct mm_struct *mm);
406 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
407 #define SUID_DUMP_USER 1 /* Dump as user of process */
408 #define SUID_DUMP_ROOT 2 /* Dump as root */
412 #define MMF_DUMPABLE 0 /* core dump is permitted */
413 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
415 #define MMF_DUMPABLE_BITS 2
416 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
418 /* coredump filter bits */
419 #define MMF_DUMP_ANON_PRIVATE 2
420 #define MMF_DUMP_ANON_SHARED 3
421 #define MMF_DUMP_MAPPED_PRIVATE 4
422 #define MMF_DUMP_MAPPED_SHARED 5
423 #define MMF_DUMP_ELF_HEADERS 6
424 #define MMF_DUMP_HUGETLB_PRIVATE 7
425 #define MMF_DUMP_HUGETLB_SHARED 8
427 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
428 #define MMF_DUMP_FILTER_BITS 7
429 #define MMF_DUMP_FILTER_MASK \
430 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
431 #define MMF_DUMP_FILTER_DEFAULT \
432 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
433 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
435 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
436 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
438 # define MMF_DUMP_MASK_DEFAULT_ELF 0
440 /* leave room for more dump flags */
441 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
442 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
444 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
446 struct sighand_struct {
448 struct k_sigaction action[_NSIG];
450 wait_queue_head_t signalfd_wqh;
453 struct pacct_struct {
456 unsigned long ac_mem;
457 cputime_t ac_utime, ac_stime;
458 unsigned long ac_minflt, ac_majflt;
469 * struct task_cputime - collected CPU time counts
470 * @utime: time spent in user mode, in &cputime_t units
471 * @stime: time spent in kernel mode, in &cputime_t units
472 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
474 * This structure groups together three kinds of CPU time that are
475 * tracked for threads and thread groups. Most things considering
476 * CPU time want to group these counts together and treat all three
477 * of them in parallel.
479 struct task_cputime {
482 unsigned long long sum_exec_runtime;
484 /* Alternate field names when used to cache expirations. */
485 #define prof_exp stime
486 #define virt_exp utime
487 #define sched_exp sum_exec_runtime
489 #define INIT_CPUTIME \
490 (struct task_cputime) { \
491 .utime = cputime_zero, \
492 .stime = cputime_zero, \
493 .sum_exec_runtime = 0, \
497 * Disable preemption until the scheduler is running.
498 * Reset by start_kernel()->sched_init()->init_idle().
500 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
501 * before the scheduler is active -- see should_resched().
503 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
506 * struct thread_group_cputimer - thread group interval timer counts
507 * @cputime: thread group interval timers.
508 * @running: non-zero when there are timers running and
509 * @cputime receives updates.
510 * @lock: lock for fields in this struct.
512 * This structure contains the version of task_cputime, above, that is
513 * used for thread group CPU timer calculations.
515 struct thread_group_cputimer {
516 struct task_cputime cputime;
521 #include <linux/rwsem.h>
525 * NOTE! "signal_struct" does not have its own
526 * locking, because a shared signal_struct always
527 * implies a shared sighand_struct, so locking
528 * sighand_struct is always a proper superset of
529 * the locking of signal_struct.
531 struct signal_struct {
536 wait_queue_head_t wait_chldexit; /* for wait4() */
538 /* current thread group signal load-balancing target: */
539 struct task_struct *curr_target;
541 /* shared signal handling: */
542 struct sigpending shared_pending;
544 /* thread group exit support */
547 * - notify group_exit_task when ->count is equal to notify_count
548 * - everyone except group_exit_task is stopped during signal delivery
549 * of fatal signals, group_exit_task processes the signal.
552 struct task_struct *group_exit_task;
554 /* thread group stop support, overloads group_exit_code too */
555 int group_stop_count;
556 unsigned int flags; /* see SIGNAL_* flags below */
558 /* POSIX.1b Interval Timers */
559 struct list_head posix_timers;
561 /* ITIMER_REAL timer for the process */
562 struct hrtimer real_timer;
563 struct pid *leader_pid;
564 ktime_t it_real_incr;
567 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
568 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
569 * values are defined to 0 and 1 respectively
571 struct cpu_itimer it[2];
574 * Thread group totals for process CPU timers.
575 * See thread_group_cputimer(), et al, for details.
577 struct thread_group_cputimer cputimer;
579 /* Earliest-expiration cache. */
580 struct task_cputime cputime_expires;
582 struct list_head cpu_timers[3];
584 struct pid *tty_old_pgrp;
586 /* boolean value for session group leader */
589 struct tty_struct *tty; /* NULL if no tty */
591 #ifdef CONFIG_SCHED_AUTOGROUP
592 struct autogroup *autogroup;
595 * Cumulative resource counters for dead threads in the group,
596 * and for reaped dead child processes forked by this group.
597 * Live threads maintain their own counters and add to these
598 * in __exit_signal, except for the group leader.
600 cputime_t utime, stime, cutime, cstime;
603 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
604 cputime_t prev_utime, prev_stime;
606 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
607 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
608 unsigned long inblock, oublock, cinblock, coublock;
609 unsigned long maxrss, cmaxrss;
610 struct task_io_accounting ioac;
613 * Cumulative ns of schedule CPU time fo dead threads in the
614 * group, not including a zombie group leader, (This only differs
615 * from jiffies_to_ns(utime + stime) if sched_clock uses something
616 * other than jiffies.)
618 unsigned long long sum_sched_runtime;
621 * We don't bother to synchronize most readers of this at all,
622 * because there is no reader checking a limit that actually needs
623 * to get both rlim_cur and rlim_max atomically, and either one
624 * alone is a single word that can safely be read normally.
625 * getrlimit/setrlimit use task_lock(current->group_leader) to
626 * protect this instead of the siglock, because they really
627 * have no need to disable irqs.
629 struct rlimit rlim[RLIM_NLIMITS];
631 #ifdef CONFIG_BSD_PROCESS_ACCT
632 struct pacct_struct pacct; /* per-process accounting information */
634 #ifdef CONFIG_TASKSTATS
635 struct taskstats *stats;
639 struct tty_audit_buf *tty_audit_buf;
641 #ifdef CONFIG_CGROUPS
643 * The threadgroup_fork_lock prevents threads from forking with
644 * CLONE_THREAD while held for writing. Use this for fork-sensitive
645 * threadgroup-wide operations. It's taken for reading in fork.c in
647 * Currently only needed write-side by cgroups.
649 struct rw_semaphore threadgroup_fork_lock;
652 int oom_adj; /* OOM kill score adjustment (bit shift) */
653 int oom_score_adj; /* OOM kill score adjustment */
654 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
655 * Only settable by CAP_SYS_RESOURCE. */
657 struct mutex cred_guard_mutex; /* guard against foreign influences on
658 * credential calculations
659 * (notably. ptrace) */
662 /* Context switch must be unlocked if interrupts are to be enabled */
663 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
664 # define __ARCH_WANT_UNLOCKED_CTXSW
668 * Bits in flags field of signal_struct.
670 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
671 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
672 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
674 * Pending notifications to parent.
676 #define SIGNAL_CLD_STOPPED 0x00000010
677 #define SIGNAL_CLD_CONTINUED 0x00000020
678 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
680 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
682 /* If true, all threads except ->group_exit_task have pending SIGKILL */
683 static inline int signal_group_exit(const struct signal_struct *sig)
685 return (sig->flags & SIGNAL_GROUP_EXIT) ||
686 (sig->group_exit_task != NULL);
690 * Some day this will be a full-fledged user tracking system..
693 atomic_t __count; /* reference count */
694 atomic_t processes; /* How many processes does this user have? */
695 atomic_t files; /* How many open files does this user have? */
696 atomic_t sigpending; /* How many pending signals does this user have? */
697 #ifdef CONFIG_INOTIFY_USER
698 atomic_t inotify_watches; /* How many inotify watches does this user have? */
699 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
701 #ifdef CONFIG_FANOTIFY
702 atomic_t fanotify_listeners;
705 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
707 #ifdef CONFIG_POSIX_MQUEUE
708 /* protected by mq_lock */
709 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
711 unsigned long locked_shm; /* How many pages of mlocked shm ? */
714 struct key *uid_keyring; /* UID specific keyring */
715 struct key *session_keyring; /* UID's default session keyring */
718 /* Hash table maintenance information */
719 struct hlist_node uidhash_node;
721 struct user_namespace *user_ns;
723 #ifdef CONFIG_PERF_EVENTS
724 atomic_long_t locked_vm;
728 extern int uids_sysfs_init(void);
730 extern struct user_struct *find_user(uid_t);
732 extern struct user_struct root_user;
733 #define INIT_USER (&root_user)
736 struct backing_dev_info;
737 struct reclaim_state;
739 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
741 /* cumulative counters */
742 unsigned long pcount; /* # of times run on this cpu */
743 unsigned long long run_delay; /* time spent waiting on a runqueue */
746 unsigned long long last_arrival,/* when we last ran on a cpu */
747 last_queued; /* when we were last queued to run */
749 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
751 #ifdef CONFIG_TASK_DELAY_ACCT
752 struct task_delay_info {
754 unsigned int flags; /* Private per-task flags */
756 /* For each stat XXX, add following, aligned appropriately
758 * struct timespec XXX_start, XXX_end;
762 * Atomicity of updates to XXX_delay, XXX_count protected by
763 * single lock above (split into XXX_lock if contention is an issue).
767 * XXX_count is incremented on every XXX operation, the delay
768 * associated with the operation is added to XXX_delay.
769 * XXX_delay contains the accumulated delay time in nanoseconds.
771 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
772 u64 blkio_delay; /* wait for sync block io completion */
773 u64 swapin_delay; /* wait for swapin block io completion */
774 u32 blkio_count; /* total count of the number of sync block */
775 /* io operations performed */
776 u32 swapin_count; /* total count of the number of swapin block */
777 /* io operations performed */
779 struct timespec freepages_start, freepages_end;
780 u64 freepages_delay; /* wait for memory reclaim */
781 u32 freepages_count; /* total count of memory reclaim */
783 #endif /* CONFIG_TASK_DELAY_ACCT */
785 static inline int sched_info_on(void)
787 #ifdef CONFIG_SCHEDSTATS
789 #elif defined(CONFIG_TASK_DELAY_ACCT)
790 extern int delayacct_on;
805 * Increase resolution of nice-level calculations for 64-bit architectures.
806 * The extra resolution improves shares distribution and load balancing of
807 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
808 * hierarchies, especially on larger systems. This is not a user-visible change
809 * and does not change the user-interface for setting shares/weights.
811 * We increase resolution only if we have enough bits to allow this increased
812 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
813 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
816 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
817 # define SCHED_LOAD_RESOLUTION 10
818 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
819 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
821 # define SCHED_LOAD_RESOLUTION 0
822 # define scale_load(w) (w)
823 # define scale_load_down(w) (w)
826 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
827 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
830 * Increase resolution of cpu_power calculations
832 #define SCHED_POWER_SHIFT 10
833 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
836 * sched-domains (multiprocessor balancing) declarations:
839 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
840 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
841 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
842 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
843 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
844 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
845 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
846 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
847 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
848 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
849 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
850 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
851 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
852 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
854 enum powersavings_balance_level {
855 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
856 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
857 * first for long running threads
859 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
860 * cpu package for power savings
862 MAX_POWERSAVINGS_BALANCE_LEVELS
865 extern int sched_mc_power_savings, sched_smt_power_savings;
867 static inline int sd_balance_for_mc_power(void)
869 if (sched_smt_power_savings)
870 return SD_POWERSAVINGS_BALANCE;
872 if (!sched_mc_power_savings)
873 return SD_PREFER_SIBLING;
878 static inline int sd_balance_for_package_power(void)
880 if (sched_mc_power_savings | sched_smt_power_savings)
881 return SD_POWERSAVINGS_BALANCE;
883 return SD_PREFER_SIBLING;
886 extern int __weak arch_sd_sibiling_asym_packing(void);
889 * Optimise SD flags for power savings:
890 * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings.
891 * Keep default SD flags if sched_{smt,mc}_power_saving=0
894 static inline int sd_power_saving_flags(void)
896 if (sched_mc_power_savings | sched_smt_power_savings)
897 return SD_BALANCE_NEWIDLE;
902 struct sched_group_power {
905 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
908 unsigned int power, power_orig;
912 struct sched_group *next; /* Must be a circular list */
915 unsigned int group_weight;
916 struct sched_group_power *sgp;
919 * The CPUs this group covers.
921 * NOTE: this field is variable length. (Allocated dynamically
922 * by attaching extra space to the end of the structure,
923 * depending on how many CPUs the kernel has booted up with)
925 unsigned long cpumask[0];
928 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
930 return to_cpumask(sg->cpumask);
933 struct sched_domain_attr {
934 int relax_domain_level;
937 #define SD_ATTR_INIT (struct sched_domain_attr) { \
938 .relax_domain_level = -1, \
941 extern int sched_domain_level_max;
943 struct sched_domain {
944 /* These fields must be setup */
945 struct sched_domain *parent; /* top domain must be null terminated */
946 struct sched_domain *child; /* bottom domain must be null terminated */
947 struct sched_group *groups; /* the balancing groups of the domain */
948 unsigned long min_interval; /* Minimum balance interval ms */
949 unsigned long max_interval; /* Maximum balance interval ms */
950 unsigned int busy_factor; /* less balancing by factor if busy */
951 unsigned int imbalance_pct; /* No balance until over watermark */
952 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
953 unsigned int busy_idx;
954 unsigned int idle_idx;
955 unsigned int newidle_idx;
956 unsigned int wake_idx;
957 unsigned int forkexec_idx;
958 unsigned int smt_gain;
959 int flags; /* See SD_* */
962 /* Runtime fields. */
963 unsigned long last_balance; /* init to jiffies. units in jiffies */
964 unsigned int balance_interval; /* initialise to 1. units in ms. */
965 unsigned int nr_balance_failed; /* initialise to 0 */
969 #ifdef CONFIG_SCHEDSTATS
970 /* load_balance() stats */
971 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
972 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
973 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
974 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
975 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
976 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
977 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
978 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
980 /* Active load balancing */
981 unsigned int alb_count;
982 unsigned int alb_failed;
983 unsigned int alb_pushed;
985 /* SD_BALANCE_EXEC stats */
986 unsigned int sbe_count;
987 unsigned int sbe_balanced;
988 unsigned int sbe_pushed;
990 /* SD_BALANCE_FORK stats */
991 unsigned int sbf_count;
992 unsigned int sbf_balanced;
993 unsigned int sbf_pushed;
995 /* try_to_wake_up() stats */
996 unsigned int ttwu_wake_remote;
997 unsigned int ttwu_move_affine;
998 unsigned int ttwu_move_balance;
1000 #ifdef CONFIG_SCHED_DEBUG
1004 void *private; /* used during construction */
1005 struct rcu_head rcu; /* used during destruction */
1008 unsigned int span_weight;
1010 * Span of all CPUs in this domain.
1012 * NOTE: this field is variable length. (Allocated dynamically
1013 * by attaching extra space to the end of the structure,
1014 * depending on how many CPUs the kernel has booted up with)
1016 unsigned long span[0];
1019 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1021 return to_cpumask(sd->span);
1024 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1025 struct sched_domain_attr *dattr_new);
1027 /* Allocate an array of sched domains, for partition_sched_domains(). */
1028 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1029 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1031 /* Test a flag in parent sched domain */
1032 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1034 if (sd->parent && (sd->parent->flags & flag))
1040 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1041 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1043 #else /* CONFIG_SMP */
1045 struct sched_domain_attr;
1048 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1049 struct sched_domain_attr *dattr_new)
1052 #endif /* !CONFIG_SMP */
1055 struct io_context; /* See blkdev.h */
1058 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1059 extern void prefetch_stack(struct task_struct *t);
1061 static inline void prefetch_stack(struct task_struct *t) { }
1064 struct audit_context; /* See audit.c */
1066 struct pipe_inode_info;
1067 struct uts_namespace;
1070 struct sched_domain;
1075 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1076 #define WF_FORK 0x02 /* child wakeup after fork */
1077 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1079 #define ENQUEUE_WAKEUP 1
1080 #define ENQUEUE_HEAD 2
1082 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1084 #define ENQUEUE_WAKING 0
1087 #define DEQUEUE_SLEEP 1
1089 struct sched_class {
1090 const struct sched_class *next;
1092 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1093 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1094 void (*yield_task) (struct rq *rq);
1095 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1097 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1099 struct task_struct * (*pick_next_task) (struct rq *rq);
1100 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1103 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1105 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1106 void (*post_schedule) (struct rq *this_rq);
1107 void (*task_waking) (struct task_struct *task);
1108 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1110 void (*set_cpus_allowed)(struct task_struct *p,
1111 const struct cpumask *newmask);
1113 void (*rq_online)(struct rq *rq);
1114 void (*rq_offline)(struct rq *rq);
1117 void (*set_curr_task) (struct rq *rq);
1118 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1119 void (*task_fork) (struct task_struct *p);
1121 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1122 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1123 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1126 unsigned int (*get_rr_interval) (struct rq *rq,
1127 struct task_struct *task);
1129 #ifdef CONFIG_FAIR_GROUP_SCHED
1130 void (*task_move_group) (struct task_struct *p, int on_rq);
1134 struct load_weight {
1135 unsigned long weight, inv_weight;
1138 #ifdef CONFIG_SCHEDSTATS
1139 struct sched_statistics {
1149 s64 sum_sleep_runtime;
1156 u64 nr_migrations_cold;
1157 u64 nr_failed_migrations_affine;
1158 u64 nr_failed_migrations_running;
1159 u64 nr_failed_migrations_hot;
1160 u64 nr_forced_migrations;
1163 u64 nr_wakeups_sync;
1164 u64 nr_wakeups_migrate;
1165 u64 nr_wakeups_local;
1166 u64 nr_wakeups_remote;
1167 u64 nr_wakeups_affine;
1168 u64 nr_wakeups_affine_attempts;
1169 u64 nr_wakeups_passive;
1170 u64 nr_wakeups_idle;
1174 struct sched_entity {
1175 struct load_weight load; /* for load-balancing */
1176 struct rb_node run_node;
1177 struct list_head group_node;
1181 u64 sum_exec_runtime;
1183 u64 prev_sum_exec_runtime;
1187 #ifdef CONFIG_SCHEDSTATS
1188 struct sched_statistics statistics;
1191 #ifdef CONFIG_FAIR_GROUP_SCHED
1192 struct sched_entity *parent;
1193 /* rq on which this entity is (to be) queued: */
1194 struct cfs_rq *cfs_rq;
1195 /* rq "owned" by this entity/group: */
1196 struct cfs_rq *my_q;
1200 struct sched_rt_entity {
1201 struct list_head run_list;
1202 unsigned long timeout;
1203 unsigned long watchdog_stamp;
1204 unsigned int time_slice;
1205 int nr_cpus_allowed;
1207 struct sched_rt_entity *back;
1208 #ifdef CONFIG_RT_GROUP_SCHED
1209 struct sched_rt_entity *parent;
1210 /* rq on which this entity is (to be) queued: */
1211 struct rt_rq *rt_rq;
1212 /* rq "owned" by this entity/group: */
1219 enum perf_event_task_context {
1220 perf_invalid_context = -1,
1221 perf_hw_context = 0,
1223 perf_nr_task_contexts,
1226 struct task_struct {
1227 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1230 unsigned int flags; /* per process flags, defined below */
1231 unsigned int ptrace;
1234 struct llist_node wake_entry;
1239 int prio, static_prio, normal_prio;
1240 unsigned int rt_priority;
1241 const struct sched_class *sched_class;
1242 struct sched_entity se;
1243 struct sched_rt_entity rt;
1244 #ifdef CONFIG_CGROUP_SCHED
1245 struct task_group *sched_task_group;
1248 #ifdef CONFIG_PREEMPT_NOTIFIERS
1249 /* list of struct preempt_notifier: */
1250 struct hlist_head preempt_notifiers;
1254 * fpu_counter contains the number of consecutive context switches
1255 * that the FPU is used. If this is over a threshold, the lazy fpu
1256 * saving becomes unlazy to save the trap. This is an unsigned char
1257 * so that after 256 times the counter wraps and the behavior turns
1258 * lazy again; this to deal with bursty apps that only use FPU for
1261 unsigned char fpu_counter;
1262 #ifdef CONFIG_BLK_DEV_IO_TRACE
1263 unsigned int btrace_seq;
1266 unsigned int policy;
1267 cpumask_t cpus_allowed;
1269 #ifdef CONFIG_PREEMPT_RCU
1270 int rcu_read_lock_nesting;
1271 char rcu_read_unlock_special;
1272 struct list_head rcu_node_entry;
1273 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1274 #ifdef CONFIG_TREE_PREEMPT_RCU
1275 struct rcu_node *rcu_blocked_node;
1276 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1277 #ifdef CONFIG_RCU_BOOST
1278 struct rt_mutex *rcu_boost_mutex;
1279 #endif /* #ifdef CONFIG_RCU_BOOST */
1281 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1282 struct sched_info sched_info;
1285 struct list_head tasks;
1287 struct plist_node pushable_tasks;
1290 struct mm_struct *mm, *active_mm;
1291 #ifdef CONFIG_COMPAT_BRK
1292 unsigned brk_randomized:1;
1294 #if defined(SPLIT_RSS_COUNTING)
1295 struct task_rss_stat rss_stat;
1299 int exit_code, exit_signal;
1300 int pdeath_signal; /* The signal sent when the parent dies */
1301 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1303 unsigned int personality;
1304 unsigned did_exec:1;
1305 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1307 unsigned in_iowait:1;
1310 /* Revert to default priority/policy when forking */
1311 unsigned sched_reset_on_fork:1;
1312 unsigned sched_contributes_to_load:1;
1317 #ifdef CONFIG_CC_STACKPROTECTOR
1318 /* Canary value for the -fstack-protector gcc feature */
1319 unsigned long stack_canary;
1323 * pointers to (original) parent process, youngest child, younger sibling,
1324 * older sibling, respectively. (p->father can be replaced with
1325 * p->real_parent->pid)
1327 struct task_struct *real_parent; /* real parent process */
1328 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1330 * children/sibling forms the list of my natural children
1332 struct list_head children; /* list of my children */
1333 struct list_head sibling; /* linkage in my parent's children list */
1334 struct task_struct *group_leader; /* threadgroup leader */
1337 * ptraced is the list of tasks this task is using ptrace on.
1338 * This includes both natural children and PTRACE_ATTACH targets.
1339 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1341 struct list_head ptraced;
1342 struct list_head ptrace_entry;
1344 /* PID/PID hash table linkage. */
1345 struct pid_link pids[PIDTYPE_MAX];
1346 struct list_head thread_group;
1348 struct completion *vfork_done; /* for vfork() */
1349 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1350 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1352 cputime_t utime, stime, utimescaled, stimescaled;
1354 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1355 cputime_t prev_utime, prev_stime;
1357 unsigned long nvcsw, nivcsw; /* context switch counts */
1358 struct timespec start_time; /* monotonic time */
1359 struct timespec real_start_time; /* boot based time */
1360 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1361 unsigned long min_flt, maj_flt;
1363 struct task_cputime cputime_expires;
1364 struct list_head cpu_timers[3];
1366 /* process credentials */
1367 const struct cred __rcu *real_cred; /* objective and real subjective task
1368 * credentials (COW) */
1369 const struct cred __rcu *cred; /* effective (overridable) subjective task
1370 * credentials (COW) */
1371 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1373 char comm[TASK_COMM_LEN]; /* executable name excluding path
1374 - access with [gs]et_task_comm (which lock
1375 it with task_lock())
1376 - initialized normally by setup_new_exec */
1377 /* file system info */
1378 int link_count, total_link_count;
1379 #ifdef CONFIG_SYSVIPC
1381 struct sysv_sem sysvsem;
1383 #ifdef CONFIG_DETECT_HUNG_TASK
1384 /* hung task detection */
1385 unsigned long last_switch_count;
1387 /* CPU-specific state of this task */
1388 struct thread_struct thread;
1389 /* filesystem information */
1390 struct fs_struct *fs;
1391 /* open file information */
1392 struct files_struct *files;
1394 struct nsproxy *nsproxy;
1395 /* signal handlers */
1396 struct signal_struct *signal;
1397 struct sighand_struct *sighand;
1399 sigset_t blocked, real_blocked;
1400 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1401 struct sigpending pending;
1403 unsigned long sas_ss_sp;
1405 int (*notifier)(void *priv);
1406 void *notifier_data;
1407 sigset_t *notifier_mask;
1408 struct audit_context *audit_context;
1409 #ifdef CONFIG_AUDITSYSCALL
1411 unsigned int sessionid;
1415 /* Thread group tracking */
1418 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1420 spinlock_t alloc_lock;
1422 #ifdef CONFIG_GENERIC_HARDIRQS
1423 /* IRQ handler threads */
1424 struct irqaction *irqaction;
1427 /* Protection of the PI data structures: */
1428 raw_spinlock_t pi_lock;
1430 #ifdef CONFIG_RT_MUTEXES
1431 /* PI waiters blocked on a rt_mutex held by this task */
1432 struct plist_head pi_waiters;
1433 /* Deadlock detection and priority inheritance handling */
1434 struct rt_mutex_waiter *pi_blocked_on;
1437 #ifdef CONFIG_DEBUG_MUTEXES
1438 /* mutex deadlock detection */
1439 struct mutex_waiter *blocked_on;
1441 #ifdef CONFIG_TRACE_IRQFLAGS
1442 unsigned int irq_events;
1443 unsigned long hardirq_enable_ip;
1444 unsigned long hardirq_disable_ip;
1445 unsigned int hardirq_enable_event;
1446 unsigned int hardirq_disable_event;
1447 int hardirqs_enabled;
1448 int hardirq_context;
1449 unsigned long softirq_disable_ip;
1450 unsigned long softirq_enable_ip;
1451 unsigned int softirq_disable_event;
1452 unsigned int softirq_enable_event;
1453 int softirqs_enabled;
1454 int softirq_context;
1456 #ifdef CONFIG_LOCKDEP
1457 # define MAX_LOCK_DEPTH 48UL
1460 unsigned int lockdep_recursion;
1461 struct held_lock held_locks[MAX_LOCK_DEPTH];
1462 gfp_t lockdep_reclaim_gfp;
1465 /* journalling filesystem info */
1468 /* stacked block device info */
1469 struct bio_list *bio_list;
1472 /* stack plugging */
1473 struct blk_plug *plug;
1477 struct reclaim_state *reclaim_state;
1479 struct backing_dev_info *backing_dev_info;
1481 struct io_context *io_context;
1483 unsigned long ptrace_message;
1484 siginfo_t *last_siginfo; /* For ptrace use. */
1485 struct task_io_accounting ioac;
1486 #if defined(CONFIG_TASK_XACCT)
1487 u64 acct_rss_mem1; /* accumulated rss usage */
1488 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1489 cputime_t acct_timexpd; /* stime + utime since last update */
1491 #ifdef CONFIG_CPUSETS
1492 nodemask_t mems_allowed; /* Protected by alloc_lock */
1493 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1494 int cpuset_mem_spread_rotor;
1495 int cpuset_slab_spread_rotor;
1497 #ifdef CONFIG_CGROUPS
1498 /* Control Group info protected by css_set_lock */
1499 struct css_set __rcu *cgroups;
1500 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1501 struct list_head cg_list;
1504 struct robust_list_head __user *robust_list;
1505 #ifdef CONFIG_COMPAT
1506 struct compat_robust_list_head __user *compat_robust_list;
1508 struct list_head pi_state_list;
1509 struct futex_pi_state *pi_state_cache;
1511 #ifdef CONFIG_PERF_EVENTS
1512 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1513 struct mutex perf_event_mutex;
1514 struct list_head perf_event_list;
1517 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1519 short pref_node_fork;
1521 struct rcu_head rcu;
1524 * cache last used pipe for splice
1526 struct pipe_inode_info *splice_pipe;
1527 #ifdef CONFIG_TASK_DELAY_ACCT
1528 struct task_delay_info *delays;
1530 #ifdef CONFIG_FAULT_INJECTION
1534 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1535 * balance_dirty_pages() for some dirty throttling pause
1538 int nr_dirtied_pause;
1540 #ifdef CONFIG_LATENCYTOP
1541 int latency_record_count;
1542 struct latency_record latency_record[LT_SAVECOUNT];
1545 * time slack values; these are used to round up poll() and
1546 * select() etc timeout values. These are in nanoseconds.
1548 unsigned long timer_slack_ns;
1549 unsigned long default_timer_slack_ns;
1551 struct list_head *scm_work_list;
1552 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1553 /* Index of current stored address in ret_stack */
1555 /* Stack of return addresses for return function tracing */
1556 struct ftrace_ret_stack *ret_stack;
1557 /* time stamp for last schedule */
1558 unsigned long long ftrace_timestamp;
1560 * Number of functions that haven't been traced
1561 * because of depth overrun.
1563 atomic_t trace_overrun;
1564 /* Pause for the tracing */
1565 atomic_t tracing_graph_pause;
1567 #ifdef CONFIG_TRACING
1568 /* state flags for use by tracers */
1569 unsigned long trace;
1570 /* bitmask and counter of trace recursion */
1571 unsigned long trace_recursion;
1572 #endif /* CONFIG_TRACING */
1573 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1574 struct memcg_batch_info {
1575 int do_batch; /* incremented when batch uncharge started */
1576 struct mem_cgroup *memcg; /* target memcg of uncharge */
1577 unsigned long nr_pages; /* uncharged usage */
1578 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1581 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1582 atomic_t ptrace_bp_refcnt;
1586 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1587 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1590 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1591 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1592 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1593 * values are inverted: lower p->prio value means higher priority.
1595 * The MAX_USER_RT_PRIO value allows the actual maximum
1596 * RT priority to be separate from the value exported to
1597 * user-space. This allows kernel threads to set their
1598 * priority to a value higher than any user task. Note:
1599 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1602 #define MAX_USER_RT_PRIO 100
1603 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1605 #define MAX_PRIO (MAX_RT_PRIO + 40)
1606 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1608 static inline int rt_prio(int prio)
1610 if (unlikely(prio < MAX_RT_PRIO))
1615 static inline int rt_task(struct task_struct *p)
1617 return rt_prio(p->prio);
1620 static inline struct pid *task_pid(struct task_struct *task)
1622 return task->pids[PIDTYPE_PID].pid;
1625 static inline struct pid *task_tgid(struct task_struct *task)
1627 return task->group_leader->pids[PIDTYPE_PID].pid;
1631 * Without tasklist or rcu lock it is not safe to dereference
1632 * the result of task_pgrp/task_session even if task == current,
1633 * we can race with another thread doing sys_setsid/sys_setpgid.
1635 static inline struct pid *task_pgrp(struct task_struct *task)
1637 return task->group_leader->pids[PIDTYPE_PGID].pid;
1640 static inline struct pid *task_session(struct task_struct *task)
1642 return task->group_leader->pids[PIDTYPE_SID].pid;
1645 struct pid_namespace;
1648 * the helpers to get the task's different pids as they are seen
1649 * from various namespaces
1651 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1652 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1654 * task_xid_nr_ns() : id seen from the ns specified;
1656 * set_task_vxid() : assigns a virtual id to a task;
1658 * see also pid_nr() etc in include/linux/pid.h
1660 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1661 struct pid_namespace *ns);
1663 static inline pid_t task_pid_nr(struct task_struct *tsk)
1668 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1669 struct pid_namespace *ns)
1671 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1674 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1676 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1680 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1685 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1687 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1689 return pid_vnr(task_tgid(tsk));
1693 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1694 struct pid_namespace *ns)
1696 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1699 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1701 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1705 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1706 struct pid_namespace *ns)
1708 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1711 static inline pid_t task_session_vnr(struct task_struct *tsk)
1713 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1716 /* obsolete, do not use */
1717 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1719 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1723 * pid_alive - check that a task structure is not stale
1724 * @p: Task structure to be checked.
1726 * Test if a process is not yet dead (at most zombie state)
1727 * If pid_alive fails, then pointers within the task structure
1728 * can be stale and must not be dereferenced.
1730 static inline int pid_alive(struct task_struct *p)
1732 return p->pids[PIDTYPE_PID].pid != NULL;
1736 * is_global_init - check if a task structure is init
1737 * @tsk: Task structure to be checked.
1739 * Check if a task structure is the first user space task the kernel created.
1741 static inline int is_global_init(struct task_struct *tsk)
1743 return tsk->pid == 1;
1747 * is_container_init:
1748 * check whether in the task is init in its own pid namespace.
1750 extern int is_container_init(struct task_struct *tsk);
1752 extern struct pid *cad_pid;
1754 extern void free_task(struct task_struct *tsk);
1755 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1757 extern void __put_task_struct(struct task_struct *t);
1759 static inline void put_task_struct(struct task_struct *t)
1761 if (atomic_dec_and_test(&t->usage))
1762 __put_task_struct(t);
1765 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1766 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1771 #define PF_STARTING 0x00000002 /* being created */
1772 #define PF_EXITING 0x00000004 /* getting shut down */
1773 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1774 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1775 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1776 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1777 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1778 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1779 #define PF_DUMPCORE 0x00000200 /* dumped core */
1780 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1781 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1782 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1783 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1784 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1785 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1786 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1787 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1788 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1789 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1790 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1791 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1792 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1793 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1794 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1795 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1796 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1797 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1798 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1799 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1800 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1803 * Only the _current_ task can read/write to tsk->flags, but other
1804 * tasks can access tsk->flags in readonly mode for example
1805 * with tsk_used_math (like during threaded core dumping).
1806 * There is however an exception to this rule during ptrace
1807 * or during fork: the ptracer task is allowed to write to the
1808 * child->flags of its traced child (same goes for fork, the parent
1809 * can write to the child->flags), because we're guaranteed the
1810 * child is not running and in turn not changing child->flags
1811 * at the same time the parent does it.
1813 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1814 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1815 #define clear_used_math() clear_stopped_child_used_math(current)
1816 #define set_used_math() set_stopped_child_used_math(current)
1817 #define conditional_stopped_child_used_math(condition, child) \
1818 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1819 #define conditional_used_math(condition) \
1820 conditional_stopped_child_used_math(condition, current)
1821 #define copy_to_stopped_child_used_math(child) \
1822 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1823 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1824 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1825 #define used_math() tsk_used_math(current)
1828 * task->jobctl flags
1830 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1832 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1833 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1834 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1835 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1836 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1837 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1838 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1840 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1841 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1842 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1843 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1844 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1845 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1846 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1848 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1849 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1851 extern bool task_set_jobctl_pending(struct task_struct *task,
1853 extern void task_clear_jobctl_trapping(struct task_struct *task);
1854 extern void task_clear_jobctl_pending(struct task_struct *task,
1857 #ifdef CONFIG_PREEMPT_RCU
1859 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1860 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1861 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1863 static inline void rcu_copy_process(struct task_struct *p)
1865 p->rcu_read_lock_nesting = 0;
1866 p->rcu_read_unlock_special = 0;
1867 #ifdef CONFIG_TREE_PREEMPT_RCU
1868 p->rcu_blocked_node = NULL;
1869 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1870 #ifdef CONFIG_RCU_BOOST
1871 p->rcu_boost_mutex = NULL;
1872 #endif /* #ifdef CONFIG_RCU_BOOST */
1873 INIT_LIST_HEAD(&p->rcu_node_entry);
1878 static inline void rcu_copy_process(struct task_struct *p)
1885 extern void do_set_cpus_allowed(struct task_struct *p,
1886 const struct cpumask *new_mask);
1888 extern int set_cpus_allowed_ptr(struct task_struct *p,
1889 const struct cpumask *new_mask);
1891 static inline void do_set_cpus_allowed(struct task_struct *p,
1892 const struct cpumask *new_mask)
1895 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1896 const struct cpumask *new_mask)
1898 if (!cpumask_test_cpu(0, new_mask))
1905 void calc_load_enter_idle(void);
1906 void calc_load_exit_idle(void);
1908 static inline void calc_load_enter_idle(void) { }
1909 static inline void calc_load_exit_idle(void) { }
1910 #endif /* CONFIG_NO_HZ */
1912 #ifndef CONFIG_CPUMASK_OFFSTACK
1913 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1915 return set_cpus_allowed_ptr(p, &new_mask);
1920 * Do not use outside of architecture code which knows its limitations.
1922 * sched_clock() has no promise of monotonicity or bounded drift between
1923 * CPUs, use (which you should not) requires disabling IRQs.
1925 * Please use one of the three interfaces below.
1927 extern unsigned long long notrace sched_clock(void);
1929 * See the comment in kernel/sched_clock.c
1931 extern u64 cpu_clock(int cpu);
1932 extern u64 local_clock(void);
1933 extern u64 sched_clock_cpu(int cpu);
1936 extern void sched_clock_init(void);
1938 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1939 static inline void sched_clock_tick(void)
1943 static inline void sched_clock_idle_sleep_event(void)
1947 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1952 * Architectures can set this to 1 if they have specified
1953 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1954 * but then during bootup it turns out that sched_clock()
1955 * is reliable after all:
1957 extern int sched_clock_stable;
1959 extern void sched_clock_tick(void);
1960 extern void sched_clock_idle_sleep_event(void);
1961 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1964 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1966 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1967 * The reason for this explicit opt-in is not to have perf penalty with
1968 * slow sched_clocks.
1970 extern void enable_sched_clock_irqtime(void);
1971 extern void disable_sched_clock_irqtime(void);
1973 static inline void enable_sched_clock_irqtime(void) {}
1974 static inline void disable_sched_clock_irqtime(void) {}
1977 extern unsigned long long
1978 task_sched_runtime(struct task_struct *task);
1980 /* sched_exec is called by processes performing an exec */
1982 extern void sched_exec(void);
1984 #define sched_exec() {}
1987 extern void sched_clock_idle_sleep_event(void);
1988 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1990 #ifdef CONFIG_HOTPLUG_CPU
1991 extern void idle_task_exit(void);
1993 static inline void idle_task_exit(void) {}
1996 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1997 extern void wake_up_idle_cpu(int cpu);
1999 static inline void wake_up_idle_cpu(int cpu) { }
2002 extern unsigned int sysctl_sched_latency;
2003 extern unsigned int sysctl_sched_min_granularity;
2004 extern unsigned int sysctl_sched_wakeup_granularity;
2005 extern unsigned int sysctl_sched_child_runs_first;
2007 enum sched_tunable_scaling {
2008 SCHED_TUNABLESCALING_NONE,
2009 SCHED_TUNABLESCALING_LOG,
2010 SCHED_TUNABLESCALING_LINEAR,
2011 SCHED_TUNABLESCALING_END,
2013 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2015 #ifdef CONFIG_SCHED_DEBUG
2016 extern unsigned int sysctl_sched_migration_cost;
2017 extern unsigned int sysctl_sched_nr_migrate;
2018 extern unsigned int sysctl_sched_time_avg;
2019 extern unsigned int sysctl_timer_migration;
2020 extern unsigned int sysctl_sched_shares_window;
2022 int sched_proc_update_handler(struct ctl_table *table, int write,
2023 void __user *buffer, size_t *length,
2026 #ifdef CONFIG_SCHED_DEBUG
2027 static inline unsigned int get_sysctl_timer_migration(void)
2029 return sysctl_timer_migration;
2032 static inline unsigned int get_sysctl_timer_migration(void)
2037 extern unsigned int sysctl_sched_rt_period;
2038 extern int sysctl_sched_rt_runtime;
2040 int sched_rt_handler(struct ctl_table *table, int write,
2041 void __user *buffer, size_t *lenp,
2044 #ifdef CONFIG_SCHED_AUTOGROUP
2045 extern unsigned int sysctl_sched_autogroup_enabled;
2047 extern void sched_autogroup_create_attach(struct task_struct *p);
2048 extern void sched_autogroup_detach(struct task_struct *p);
2049 extern void sched_autogroup_fork(struct signal_struct *sig);
2050 extern void sched_autogroup_exit(struct signal_struct *sig);
2051 #ifdef CONFIG_PROC_FS
2052 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2053 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
2056 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2057 static inline void sched_autogroup_detach(struct task_struct *p) { }
2058 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2059 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2062 #ifdef CONFIG_CFS_BANDWIDTH
2063 extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2066 #ifdef CONFIG_RT_MUTEXES
2067 extern int rt_mutex_getprio(struct task_struct *p);
2068 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2069 extern void rt_mutex_adjust_pi(struct task_struct *p);
2071 static inline int rt_mutex_getprio(struct task_struct *p)
2073 return p->normal_prio;
2075 # define rt_mutex_adjust_pi(p) do { } while (0)
2078 extern bool yield_to(struct task_struct *p, bool preempt);
2079 extern void set_user_nice(struct task_struct *p, long nice);
2080 extern int task_prio(const struct task_struct *p);
2081 extern int task_nice(const struct task_struct *p);
2082 extern int can_nice(const struct task_struct *p, const int nice);
2083 extern int task_curr(const struct task_struct *p);
2084 extern int idle_cpu(int cpu);
2085 extern int sched_setscheduler(struct task_struct *, int,
2086 const struct sched_param *);
2087 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2088 const struct sched_param *);
2089 extern struct task_struct *idle_task(int cpu);
2090 extern struct task_struct *curr_task(int cpu);
2091 extern void set_curr_task(int cpu, struct task_struct *p);
2096 * The default (Linux) execution domain.
2098 extern struct exec_domain default_exec_domain;
2100 union thread_union {
2101 struct thread_info thread_info;
2102 unsigned long stack[THREAD_SIZE/sizeof(long)];
2105 #ifndef __HAVE_ARCH_KSTACK_END
2106 static inline int kstack_end(void *addr)
2108 /* Reliable end of stack detection:
2109 * Some APM bios versions misalign the stack
2111 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2115 extern union thread_union init_thread_union;
2116 extern struct task_struct init_task;
2118 extern struct mm_struct init_mm;
2120 extern struct pid_namespace init_pid_ns;
2123 * find a task by one of its numerical ids
2125 * find_task_by_pid_ns():
2126 * finds a task by its pid in the specified namespace
2127 * find_task_by_vpid():
2128 * finds a task by its virtual pid
2130 * see also find_vpid() etc in include/linux/pid.h
2133 extern struct task_struct *find_task_by_vpid(pid_t nr);
2134 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2135 struct pid_namespace *ns);
2137 extern void __set_special_pids(struct pid *pid);
2139 /* per-UID process charging. */
2140 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2141 static inline struct user_struct *get_uid(struct user_struct *u)
2143 atomic_inc(&u->__count);
2146 extern void free_uid(struct user_struct *);
2147 extern void release_uids(struct user_namespace *ns);
2149 #include <asm/current.h>
2151 extern void xtime_update(unsigned long ticks);
2153 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2154 extern int wake_up_process(struct task_struct *tsk);
2155 extern void wake_up_new_task(struct task_struct *tsk);
2157 extern void kick_process(struct task_struct *tsk);
2159 static inline void kick_process(struct task_struct *tsk) { }
2161 extern void sched_fork(struct task_struct *p);
2162 extern void sched_dead(struct task_struct *p);
2164 extern void proc_caches_init(void);
2165 extern void flush_signals(struct task_struct *);
2166 extern void __flush_signals(struct task_struct *);
2167 extern void ignore_signals(struct task_struct *);
2168 extern void flush_signal_handlers(struct task_struct *, int force_default);
2169 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2171 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2173 unsigned long flags;
2176 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2177 ret = dequeue_signal(tsk, mask, info);
2178 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2183 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2185 extern void unblock_all_signals(void);
2186 extern void release_task(struct task_struct * p);
2187 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2188 extern int force_sigsegv(int, struct task_struct *);
2189 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2190 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2191 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2192 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2193 const struct cred *, u32);
2194 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2195 extern int kill_pid(struct pid *pid, int sig, int priv);
2196 extern int kill_proc_info(int, struct siginfo *, pid_t);
2197 extern __must_check bool do_notify_parent(struct task_struct *, int);
2198 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2199 extern void force_sig(int, struct task_struct *);
2200 extern int send_sig(int, struct task_struct *, int);
2201 extern int zap_other_threads(struct task_struct *p);
2202 extern struct sigqueue *sigqueue_alloc(void);
2203 extern void sigqueue_free(struct sigqueue *);
2204 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2205 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2206 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2208 static inline int kill_cad_pid(int sig, int priv)
2210 return kill_pid(cad_pid, sig, priv);
2213 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2214 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2215 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2216 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2219 * True if we are on the alternate signal stack.
2221 static inline int on_sig_stack(unsigned long sp)
2223 #ifdef CONFIG_STACK_GROWSUP
2224 return sp >= current->sas_ss_sp &&
2225 sp - current->sas_ss_sp < current->sas_ss_size;
2227 return sp > current->sas_ss_sp &&
2228 sp - current->sas_ss_sp <= current->sas_ss_size;
2232 static inline int sas_ss_flags(unsigned long sp)
2234 return (current->sas_ss_size == 0 ? SS_DISABLE
2235 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2239 * Routines for handling mm_structs
2241 extern struct mm_struct * mm_alloc(void);
2243 /* mmdrop drops the mm and the page tables */
2244 extern void __mmdrop(struct mm_struct *);
2245 static inline void mmdrop(struct mm_struct * mm)
2247 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2251 /* mmput gets rid of the mappings and all user-space */
2252 extern void mmput(struct mm_struct *);
2253 /* Grab a reference to a task's mm, if it is not already going away */
2254 extern struct mm_struct *get_task_mm(struct task_struct *task);
2255 /* Remove the current tasks stale references to the old mm_struct */
2256 extern void mm_release(struct task_struct *, struct mm_struct *);
2257 /* Allocate a new mm structure and copy contents from tsk->mm */
2258 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2260 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2261 struct task_struct *, struct pt_regs *);
2262 extern void flush_thread(void);
2263 extern void exit_thread(void);
2265 extern void exit_files(struct task_struct *);
2266 extern void __cleanup_sighand(struct sighand_struct *);
2268 extern void exit_itimers(struct signal_struct *);
2269 extern void flush_itimer_signals(void);
2271 extern NORET_TYPE void do_group_exit(int);
2273 extern void daemonize(const char *, ...);
2274 extern int allow_signal(int);
2275 extern int disallow_signal(int);
2277 extern int do_execve(const char *,
2278 const char __user * const __user *,
2279 const char __user * const __user *, struct pt_regs *);
2280 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2281 struct task_struct *fork_idle(int);
2283 extern void set_task_comm(struct task_struct *tsk, char *from);
2284 extern char *get_task_comm(char *to, struct task_struct *tsk);
2287 void scheduler_ipi(void);
2288 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2290 static inline void scheduler_ipi(void) { }
2291 static inline unsigned long wait_task_inactive(struct task_struct *p,
2298 #define next_task(p) \
2299 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2301 #define for_each_process(p) \
2302 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2304 extern bool current_is_single_threaded(void);
2307 * Careful: do_each_thread/while_each_thread is a double loop so
2308 * 'break' will not work as expected - use goto instead.
2310 #define do_each_thread(g, t) \
2311 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2313 #define while_each_thread(g, t) \
2314 while ((t = next_thread(t)) != g)
2316 static inline int get_nr_threads(struct task_struct *tsk)
2318 return tsk->signal->nr_threads;
2321 static inline bool thread_group_leader(struct task_struct *p)
2323 return p->exit_signal >= 0;
2326 /* Do to the insanities of de_thread it is possible for a process
2327 * to have the pid of the thread group leader without actually being
2328 * the thread group leader. For iteration through the pids in proc
2329 * all we care about is that we have a task with the appropriate
2330 * pid, we don't actually care if we have the right task.
2332 static inline int has_group_leader_pid(struct task_struct *p)
2334 return p->pid == p->tgid;
2338 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2340 return p1->tgid == p2->tgid;
2343 static inline struct task_struct *next_thread(const struct task_struct *p)
2345 return list_entry_rcu(p->thread_group.next,
2346 struct task_struct, thread_group);
2349 static inline int thread_group_empty(struct task_struct *p)
2351 return list_empty(&p->thread_group);
2354 #define delay_group_leader(p) \
2355 (thread_group_leader(p) && !thread_group_empty(p))
2358 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2359 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2360 * pins the final release of task.io_context. Also protects ->cpuset and
2361 * ->cgroup.subsys[].
2363 * Nests both inside and outside of read_lock(&tasklist_lock).
2364 * It must not be nested with write_lock_irq(&tasklist_lock),
2365 * neither inside nor outside.
2367 static inline void task_lock(struct task_struct *p)
2369 spin_lock(&p->alloc_lock);
2372 static inline void task_unlock(struct task_struct *p)
2374 spin_unlock(&p->alloc_lock);
2377 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2378 unsigned long *flags);
2380 #define lock_task_sighand(tsk, flags) \
2381 ({ struct sighand_struct *__ss; \
2382 __cond_lock(&(tsk)->sighand->siglock, \
2383 (__ss = __lock_task_sighand(tsk, flags))); \
2387 static inline void unlock_task_sighand(struct task_struct *tsk,
2388 unsigned long *flags)
2390 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2393 /* See the declaration of threadgroup_fork_lock in signal_struct. */
2394 #ifdef CONFIG_CGROUPS
2395 static inline void threadgroup_fork_read_lock(struct task_struct *tsk)
2397 down_read(&tsk->signal->threadgroup_fork_lock);
2399 static inline void threadgroup_fork_read_unlock(struct task_struct *tsk)
2401 up_read(&tsk->signal->threadgroup_fork_lock);
2403 static inline void threadgroup_fork_write_lock(struct task_struct *tsk)
2405 down_write(&tsk->signal->threadgroup_fork_lock);
2407 static inline void threadgroup_fork_write_unlock(struct task_struct *tsk)
2409 up_write(&tsk->signal->threadgroup_fork_lock);
2412 static inline void threadgroup_fork_read_lock(struct task_struct *tsk) {}
2413 static inline void threadgroup_fork_read_unlock(struct task_struct *tsk) {}
2414 static inline void threadgroup_fork_write_lock(struct task_struct *tsk) {}
2415 static inline void threadgroup_fork_write_unlock(struct task_struct *tsk) {}
2418 #ifndef __HAVE_THREAD_FUNCTIONS
2420 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2421 #define task_stack_page(task) ((task)->stack)
2423 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2425 *task_thread_info(p) = *task_thread_info(org);
2426 task_thread_info(p)->task = p;
2429 static inline unsigned long *end_of_stack(struct task_struct *p)
2431 return (unsigned long *)(task_thread_info(p) + 1);
2436 static inline int object_is_on_stack(void *obj)
2438 void *stack = task_stack_page(current);
2440 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2443 extern void thread_info_cache_init(void);
2445 #ifdef CONFIG_DEBUG_STACK_USAGE
2446 static inline unsigned long stack_not_used(struct task_struct *p)
2448 unsigned long *n = end_of_stack(p);
2450 do { /* Skip over canary */
2454 return (unsigned long)n - (unsigned long)end_of_stack(p);
2458 /* set thread flags in other task's structures
2459 * - see asm/thread_info.h for TIF_xxxx flags available
2461 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2463 set_ti_thread_flag(task_thread_info(tsk), flag);
2466 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2468 clear_ti_thread_flag(task_thread_info(tsk), flag);
2471 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2473 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2476 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2478 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2481 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2483 return test_ti_thread_flag(task_thread_info(tsk), flag);
2486 static inline void set_tsk_need_resched(struct task_struct *tsk)
2488 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2491 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2493 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2496 static inline int test_tsk_need_resched(struct task_struct *tsk)
2498 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2501 static inline int restart_syscall(void)
2503 set_tsk_thread_flag(current, TIF_SIGPENDING);
2504 return -ERESTARTNOINTR;
2507 static inline int signal_pending(struct task_struct *p)
2509 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2512 static inline int __fatal_signal_pending(struct task_struct *p)
2514 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2517 static inline int fatal_signal_pending(struct task_struct *p)
2519 return signal_pending(p) && __fatal_signal_pending(p);
2522 static inline int signal_pending_state(long state, struct task_struct *p)
2524 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2526 if (!signal_pending(p))
2529 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2532 static inline int need_resched(void)
2534 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2538 * cond_resched() and cond_resched_lock(): latency reduction via
2539 * explicit rescheduling in places that are safe. The return
2540 * value indicates whether a reschedule was done in fact.
2541 * cond_resched_lock() will drop the spinlock before scheduling,
2542 * cond_resched_softirq() will enable bhs before scheduling.
2544 extern int _cond_resched(void);
2546 #define cond_resched() ({ \
2547 __might_sleep(__FILE__, __LINE__, 0); \
2551 extern int __cond_resched_lock(spinlock_t *lock);
2553 #ifdef CONFIG_PREEMPT_COUNT
2554 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2556 #define PREEMPT_LOCK_OFFSET 0
2559 #define cond_resched_lock(lock) ({ \
2560 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2561 __cond_resched_lock(lock); \
2564 extern int __cond_resched_softirq(void);
2566 #define cond_resched_softirq() ({ \
2567 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2568 __cond_resched_softirq(); \
2572 * Does a critical section need to be broken due to another
2573 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2574 * but a general need for low latency)
2576 static inline int spin_needbreak(spinlock_t *lock)
2578 #ifdef CONFIG_PREEMPT
2579 return spin_is_contended(lock);
2586 * Thread group CPU time accounting.
2588 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2589 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2591 static inline void thread_group_cputime_init(struct signal_struct *sig)
2593 raw_spin_lock_init(&sig->cputimer.lock);
2597 * Reevaluate whether the task has signals pending delivery.
2598 * Wake the task if so.
2599 * This is required every time the blocked sigset_t changes.
2600 * callers must hold sighand->siglock.
2602 extern void recalc_sigpending_and_wake(struct task_struct *t);
2603 extern void recalc_sigpending(void);
2605 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2607 static inline void signal_wake_up(struct task_struct *t, bool resume)
2609 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2611 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2613 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2617 * Wrappers for p->thread_info->cpu access. No-op on UP.
2621 static inline unsigned int task_cpu(const struct task_struct *p)
2623 return task_thread_info(p)->cpu;
2626 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2630 static inline unsigned int task_cpu(const struct task_struct *p)
2635 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2639 #endif /* CONFIG_SMP */
2641 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2642 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2644 extern void normalize_rt_tasks(void);
2646 #ifdef CONFIG_CGROUP_SCHED
2648 extern struct task_group root_task_group;
2650 extern struct task_group *sched_create_group(struct task_group *parent);
2651 extern void sched_destroy_group(struct task_group *tg);
2652 extern void sched_move_task(struct task_struct *tsk);
2653 #ifdef CONFIG_FAIR_GROUP_SCHED
2654 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2655 extern unsigned long sched_group_shares(struct task_group *tg);
2657 #ifdef CONFIG_RT_GROUP_SCHED
2658 extern int sched_group_set_rt_runtime(struct task_group *tg,
2659 long rt_runtime_us);
2660 extern long sched_group_rt_runtime(struct task_group *tg);
2661 extern int sched_group_set_rt_period(struct task_group *tg,
2663 extern long sched_group_rt_period(struct task_group *tg);
2664 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2666 #endif /* CONFIG_CGROUP_SCHED */
2668 extern int task_can_switch_user(struct user_struct *up,
2669 struct task_struct *tsk);
2671 #ifdef CONFIG_TASK_XACCT
2672 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2674 tsk->ioac.rchar += amt;
2677 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2679 tsk->ioac.wchar += amt;
2682 static inline void inc_syscr(struct task_struct *tsk)
2687 static inline void inc_syscw(struct task_struct *tsk)
2692 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2696 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2700 static inline void inc_syscr(struct task_struct *tsk)
2704 static inline void inc_syscw(struct task_struct *tsk)
2709 #ifndef TASK_SIZE_OF
2710 #define TASK_SIZE_OF(tsk) TASK_SIZE
2713 #ifdef CONFIG_MM_OWNER
2714 extern void mm_update_next_owner(struct mm_struct *mm);
2715 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2717 static inline void mm_update_next_owner(struct mm_struct *mm)
2721 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2724 #endif /* CONFIG_MM_OWNER */
2726 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2729 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2732 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2735 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2738 static inline unsigned long rlimit(unsigned int limit)
2740 return task_rlimit(current, limit);
2743 static inline unsigned long rlimit_max(unsigned int limit)
2745 return task_rlimit_max(current, limit);
2748 #endif /* __KERNEL__ */