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 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 #define CLONE_NEWNET 0x40000000 /* New network namespace */
30 #define CLONE_IO 0x80000000 /* Clone io context */
35 #define SCHED_NORMAL 0
39 /* SCHED_ISO: reserved but not implemented yet */
41 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
42 #define SCHED_RESET_ON_FORK 0x40000000
50 #include <asm/param.h> /* for HZ */
52 #include <linux/capability.h>
53 #include <linux/threads.h>
54 #include <linux/kernel.h>
55 #include <linux/types.h>
56 #include <linux/timex.h>
57 #include <linux/jiffies.h>
58 #include <linux/rbtree.h>
59 #include <linux/thread_info.h>
60 #include <linux/cpumask.h>
61 #include <linux/errno.h>
62 #include <linux/nodemask.h>
63 #include <linux/mm_types.h>
65 #include <asm/system.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>
93 #include <asm/processor.h>
96 struct futex_pi_state;
97 struct robust_list_head;
100 struct perf_event_context;
103 * List of flags we want to share for kernel threads,
104 * if only because they are not used by them anyway.
106 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
109 * These are the constant used to fake the fixed-point load-average
110 * counting. Some notes:
111 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
112 * a load-average precision of 10 bits integer + 11 bits fractional
113 * - if you want to count load-averages more often, you need more
114 * precision, or rounding will get you. With 2-second counting freq,
115 * the EXP_n values would be 1981, 2034 and 2043 if still using only
118 extern unsigned long avenrun[]; /* Load averages */
119 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
121 #define FSHIFT 11 /* nr of bits of precision */
122 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
123 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
124 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
125 #define EXP_5 2014 /* 1/exp(5sec/5min) */
126 #define EXP_15 2037 /* 1/exp(5sec/15min) */
128 #define CALC_LOAD(load,exp,n) \
130 load += n*(FIXED_1-exp); \
133 extern unsigned long total_forks;
134 extern int nr_threads;
135 DECLARE_PER_CPU(unsigned long, process_counts);
136 extern int nr_processes(void);
137 extern unsigned long nr_running(void);
138 extern unsigned long nr_uninterruptible(void);
139 extern unsigned long nr_iowait(void);
140 extern unsigned long nr_iowait_cpu(int cpu);
141 extern unsigned long this_cpu_load(void);
144 extern void calc_global_load(unsigned long ticks);
146 extern unsigned long get_parent_ip(unsigned long addr);
151 #ifdef CONFIG_SCHED_DEBUG
152 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
153 extern void proc_sched_set_task(struct task_struct *p);
155 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
158 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
161 static inline void proc_sched_set_task(struct task_struct *p)
165 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
171 * Task state bitmask. NOTE! These bits are also
172 * encoded in fs/proc/array.c: get_task_state().
174 * We have two separate sets of flags: task->state
175 * is about runnability, while task->exit_state are
176 * about the task exiting. Confusing, but this way
177 * modifying one set can't modify the other one by
180 #define TASK_RUNNING 0
181 #define TASK_INTERRUPTIBLE 1
182 #define TASK_UNINTERRUPTIBLE 2
183 #define __TASK_STOPPED 4
184 #define __TASK_TRACED 8
185 /* in tsk->exit_state */
186 #define EXIT_ZOMBIE 16
188 /* in tsk->state again */
190 #define TASK_WAKEKILL 128
191 #define TASK_WAKING 256
192 #define TASK_STATE_MAX 512
194 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
196 extern char ___assert_task_state[1 - 2*!!(
197 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
199 /* Convenience macros for the sake of set_task_state */
200 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
201 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
202 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
204 /* Convenience macros for the sake of wake_up */
205 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
206 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
208 /* get_task_state() */
209 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
210 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
213 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
214 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
215 #define task_is_dead(task) ((task)->exit_state != 0)
216 #define task_is_stopped_or_traced(task) \
217 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
218 #define task_contributes_to_load(task) \
219 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
220 (task->flags & PF_FREEZING) == 0)
222 #define __set_task_state(tsk, state_value) \
223 do { (tsk)->state = (state_value); } while (0)
224 #define set_task_state(tsk, state_value) \
225 set_mb((tsk)->state, (state_value))
228 * set_current_state() includes a barrier so that the write of current->state
229 * is correctly serialised wrt the caller's subsequent test of whether to
232 * set_current_state(TASK_UNINTERRUPTIBLE);
233 * if (do_i_need_to_sleep())
236 * If the caller does not need such serialisation then use __set_current_state()
238 #define __set_current_state(state_value) \
239 do { current->state = (state_value); } while (0)
240 #define set_current_state(state_value) \
241 set_mb(current->state, (state_value))
243 /* Task command name length */
244 #define TASK_COMM_LEN 16
246 #include <linux/spinlock.h>
249 * This serializes "schedule()" and also protects
250 * the run-queue from deletions/modifications (but
251 * _adding_ to the beginning of the run-queue has
254 extern rwlock_t tasklist_lock;
255 extern spinlock_t mmlist_lock;
259 #ifdef CONFIG_PROVE_RCU
260 extern int lockdep_tasklist_lock_is_held(void);
261 #endif /* #ifdef CONFIG_PROVE_RCU */
263 extern void sched_init(void);
264 extern void sched_init_smp(void);
265 extern asmlinkage void schedule_tail(struct task_struct *prev);
266 extern void init_idle(struct task_struct *idle, int cpu);
267 extern void init_idle_bootup_task(struct task_struct *idle);
269 extern int runqueue_is_locked(int cpu);
271 extern cpumask_var_t nohz_cpu_mask;
272 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
273 extern void select_nohz_load_balancer(int stop_tick);
274 extern int get_nohz_timer_target(void);
276 static inline void select_nohz_load_balancer(int stop_tick) { }
280 * Only dump TASK_* tasks. (0 for all tasks)
282 extern void show_state_filter(unsigned long state_filter);
284 static inline void show_state(void)
286 show_state_filter(0);
289 extern void show_regs(struct pt_regs *);
292 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
293 * task), SP is the stack pointer of the first frame that should be shown in the back
294 * trace (or NULL if the entire call-chain of the task should be shown).
296 extern void show_stack(struct task_struct *task, unsigned long *sp);
298 void io_schedule(void);
299 long io_schedule_timeout(long timeout);
301 extern void cpu_init (void);
302 extern void trap_init(void);
303 extern void update_process_times(int user);
304 extern void scheduler_tick(void);
306 extern void sched_show_task(struct task_struct *p);
308 #ifdef CONFIG_LOCKUP_DETECTOR
309 extern void touch_softlockup_watchdog(void);
310 extern void touch_softlockup_watchdog_sync(void);
311 extern void touch_all_softlockup_watchdogs(void);
312 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
314 size_t *lenp, loff_t *ppos);
315 extern unsigned int softlockup_panic;
316 extern int softlockup_thresh;
317 void lockup_detector_init(void);
319 static inline void touch_softlockup_watchdog(void)
322 static inline void touch_softlockup_watchdog_sync(void)
325 static inline void touch_all_softlockup_watchdogs(void)
328 static inline void lockup_detector_init(void)
333 #ifdef CONFIG_DETECT_HUNG_TASK
334 extern unsigned int sysctl_hung_task_panic;
335 extern unsigned long sysctl_hung_task_check_count;
336 extern unsigned long sysctl_hung_task_timeout_secs;
337 extern unsigned long sysctl_hung_task_warnings;
338 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
340 size_t *lenp, loff_t *ppos);
342 /* Avoid need for ifdefs elsewhere in the code */
343 enum { sysctl_hung_task_timeout_secs = 0 };
346 /* Attach to any functions which should be ignored in wchan output. */
347 #define __sched __attribute__((__section__(".sched.text")))
349 /* Linker adds these: start and end of __sched functions */
350 extern char __sched_text_start[], __sched_text_end[];
352 /* Is this address in the __sched functions? */
353 extern int in_sched_functions(unsigned long addr);
355 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
356 extern signed long schedule_timeout(signed long timeout);
357 extern signed long schedule_timeout_interruptible(signed long timeout);
358 extern signed long schedule_timeout_killable(signed long timeout);
359 extern signed long schedule_timeout_uninterruptible(signed long timeout);
360 asmlinkage void schedule(void);
361 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
364 struct user_namespace;
367 * Default maximum number of active map areas, this limits the number of vmas
368 * per mm struct. Users can overwrite this number by sysctl but there is a
371 * When a program's coredump is generated as ELF format, a section is created
372 * per a vma. In ELF, the number of sections is represented in unsigned short.
373 * This means the number of sections should be smaller than 65535 at coredump.
374 * Because the kernel adds some informative sections to a image of program at
375 * generating coredump, we need some margin. The number of extra sections is
376 * 1-3 now and depends on arch. We use "5" as safe margin, here.
378 #define MAPCOUNT_ELF_CORE_MARGIN (5)
379 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
381 extern int sysctl_max_map_count;
383 #include <linux/aio.h>
386 extern void arch_pick_mmap_layout(struct mm_struct *mm);
388 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
389 unsigned long, unsigned long);
391 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
392 unsigned long len, unsigned long pgoff,
393 unsigned long flags);
394 extern void arch_unmap_area(struct mm_struct *, unsigned long);
395 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
397 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
401 extern void set_dumpable(struct mm_struct *mm, int value);
402 extern int get_dumpable(struct mm_struct *mm);
406 #define MMF_DUMPABLE 0 /* core dump is permitted */
407 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
409 #define MMF_DUMPABLE_BITS 2
410 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
412 /* coredump filter bits */
413 #define MMF_DUMP_ANON_PRIVATE 2
414 #define MMF_DUMP_ANON_SHARED 3
415 #define MMF_DUMP_MAPPED_PRIVATE 4
416 #define MMF_DUMP_MAPPED_SHARED 5
417 #define MMF_DUMP_ELF_HEADERS 6
418 #define MMF_DUMP_HUGETLB_PRIVATE 7
419 #define MMF_DUMP_HUGETLB_SHARED 8
421 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
422 #define MMF_DUMP_FILTER_BITS 7
423 #define MMF_DUMP_FILTER_MASK \
424 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
425 #define MMF_DUMP_FILTER_DEFAULT \
426 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
427 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
429 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
430 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
432 # define MMF_DUMP_MASK_DEFAULT_ELF 0
434 /* leave room for more dump flags */
435 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
437 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
439 struct sighand_struct {
441 struct k_sigaction action[_NSIG];
443 wait_queue_head_t signalfd_wqh;
446 struct pacct_struct {
449 unsigned long ac_mem;
450 cputime_t ac_utime, ac_stime;
451 unsigned long ac_minflt, ac_majflt;
462 * struct task_cputime - collected CPU time counts
463 * @utime: time spent in user mode, in &cputime_t units
464 * @stime: time spent in kernel mode, in &cputime_t units
465 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
467 * This structure groups together three kinds of CPU time that are
468 * tracked for threads and thread groups. Most things considering
469 * CPU time want to group these counts together and treat all three
470 * of them in parallel.
472 struct task_cputime {
475 unsigned long long sum_exec_runtime;
477 /* Alternate field names when used to cache expirations. */
478 #define prof_exp stime
479 #define virt_exp utime
480 #define sched_exp sum_exec_runtime
482 #define INIT_CPUTIME \
483 (struct task_cputime) { \
484 .utime = cputime_zero, \
485 .stime = cputime_zero, \
486 .sum_exec_runtime = 0, \
490 * Disable preemption until the scheduler is running.
491 * Reset by start_kernel()->sched_init()->init_idle().
493 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
494 * before the scheduler is active -- see should_resched().
496 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
499 * struct thread_group_cputimer - thread group interval timer counts
500 * @cputime: thread group interval timers.
501 * @running: non-zero when there are timers running and
502 * @cputime receives updates.
503 * @lock: lock for fields in this struct.
505 * This structure contains the version of task_cputime, above, that is
506 * used for thread group CPU timer calculations.
508 struct thread_group_cputimer {
509 struct task_cputime cputime;
517 * NOTE! "signal_struct" does not have it's own
518 * locking, because a shared signal_struct always
519 * implies a shared sighand_struct, so locking
520 * sighand_struct is always a proper superset of
521 * the locking of signal_struct.
523 struct signal_struct {
528 wait_queue_head_t wait_chldexit; /* for wait4() */
530 /* current thread group signal load-balancing target: */
531 struct task_struct *curr_target;
533 /* shared signal handling: */
534 struct sigpending shared_pending;
536 /* thread group exit support */
539 * - notify group_exit_task when ->count is equal to notify_count
540 * - everyone except group_exit_task is stopped during signal delivery
541 * of fatal signals, group_exit_task processes the signal.
544 struct task_struct *group_exit_task;
546 /* thread group stop support, overloads group_exit_code too */
547 int group_stop_count;
548 unsigned int flags; /* see SIGNAL_* flags below */
550 /* POSIX.1b Interval Timers */
551 struct list_head posix_timers;
553 /* ITIMER_REAL timer for the process */
554 struct hrtimer real_timer;
555 struct pid *leader_pid;
556 ktime_t it_real_incr;
559 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
560 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
561 * values are defined to 0 and 1 respectively
563 struct cpu_itimer it[2];
566 * Thread group totals for process CPU timers.
567 * See thread_group_cputimer(), et al, for details.
569 struct thread_group_cputimer cputimer;
571 /* Earliest-expiration cache. */
572 struct task_cputime cputime_expires;
574 struct list_head cpu_timers[3];
576 struct pid *tty_old_pgrp;
578 /* boolean value for session group leader */
581 struct tty_struct *tty; /* NULL if no tty */
583 #ifdef CONFIG_SCHED_AUTOGROUP
584 struct autogroup *autogroup;
587 * Cumulative resource counters for dead threads in the group,
588 * and for reaped dead child processes forked by this group.
589 * Live threads maintain their own counters and add to these
590 * in __exit_signal, except for the group leader.
592 cputime_t utime, stime, cutime, cstime;
595 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
596 cputime_t prev_utime, prev_stime;
598 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
599 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
600 unsigned long inblock, oublock, cinblock, coublock;
601 unsigned long maxrss, cmaxrss;
602 struct task_io_accounting ioac;
605 * Cumulative ns of schedule CPU time fo dead threads in the
606 * group, not including a zombie group leader, (This only differs
607 * from jiffies_to_ns(utime + stime) if sched_clock uses something
608 * other than jiffies.)
610 unsigned long long sum_sched_runtime;
613 * We don't bother to synchronize most readers of this at all,
614 * because there is no reader checking a limit that actually needs
615 * to get both rlim_cur and rlim_max atomically, and either one
616 * alone is a single word that can safely be read normally.
617 * getrlimit/setrlimit use task_lock(current->group_leader) to
618 * protect this instead of the siglock, because they really
619 * have no need to disable irqs.
621 struct rlimit rlim[RLIM_NLIMITS];
623 #ifdef CONFIG_BSD_PROCESS_ACCT
624 struct pacct_struct pacct; /* per-process accounting information */
626 #ifdef CONFIG_TASKSTATS
627 struct taskstats *stats;
631 struct tty_audit_buf *tty_audit_buf;
634 int oom_adj; /* OOM kill score adjustment (bit shift) */
635 int oom_score_adj; /* OOM kill score adjustment */
637 struct mutex cred_guard_mutex; /* guard against foreign influences on
638 * credential calculations
639 * (notably. ptrace) */
642 /* Context switch must be unlocked if interrupts are to be enabled */
643 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
644 # define __ARCH_WANT_UNLOCKED_CTXSW
648 * Bits in flags field of signal_struct.
650 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
651 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
652 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
653 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
655 * Pending notifications to parent.
657 #define SIGNAL_CLD_STOPPED 0x00000010
658 #define SIGNAL_CLD_CONTINUED 0x00000020
659 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
661 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
663 /* If true, all threads except ->group_exit_task have pending SIGKILL */
664 static inline int signal_group_exit(const struct signal_struct *sig)
666 return (sig->flags & SIGNAL_GROUP_EXIT) ||
667 (sig->group_exit_task != NULL);
671 * Some day this will be a full-fledged user tracking system..
674 atomic_t __count; /* reference count */
675 atomic_t processes; /* How many processes does this user have? */
676 atomic_t files; /* How many open files does this user have? */
677 atomic_t sigpending; /* How many pending signals does this user have? */
678 #ifdef CONFIG_INOTIFY_USER
679 atomic_t inotify_watches; /* How many inotify watches does this user have? */
680 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
682 #ifdef CONFIG_FANOTIFY
683 atomic_t fanotify_listeners;
686 atomic_t epoll_watches; /* The number of file descriptors currently watched */
688 #ifdef CONFIG_POSIX_MQUEUE
689 /* protected by mq_lock */
690 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
692 unsigned long locked_shm; /* How many pages of mlocked shm ? */
695 struct key *uid_keyring; /* UID specific keyring */
696 struct key *session_keyring; /* UID's default session keyring */
699 /* Hash table maintenance information */
700 struct hlist_node uidhash_node;
702 struct user_namespace *user_ns;
704 #ifdef CONFIG_PERF_EVENTS
705 atomic_long_t locked_vm;
709 extern int uids_sysfs_init(void);
711 extern struct user_struct *find_user(uid_t);
713 extern struct user_struct root_user;
714 #define INIT_USER (&root_user)
717 struct backing_dev_info;
718 struct reclaim_state;
720 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
722 /* cumulative counters */
723 unsigned long pcount; /* # of times run on this cpu */
724 unsigned long long run_delay; /* time spent waiting on a runqueue */
727 unsigned long long last_arrival,/* when we last ran on a cpu */
728 last_queued; /* when we were last queued to run */
729 #ifdef CONFIG_SCHEDSTATS
731 unsigned int bkl_count;
734 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
736 #ifdef CONFIG_TASK_DELAY_ACCT
737 struct task_delay_info {
739 unsigned int flags; /* Private per-task flags */
741 /* For each stat XXX, add following, aligned appropriately
743 * struct timespec XXX_start, XXX_end;
747 * Atomicity of updates to XXX_delay, XXX_count protected by
748 * single lock above (split into XXX_lock if contention is an issue).
752 * XXX_count is incremented on every XXX operation, the delay
753 * associated with the operation is added to XXX_delay.
754 * XXX_delay contains the accumulated delay time in nanoseconds.
756 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
757 u64 blkio_delay; /* wait for sync block io completion */
758 u64 swapin_delay; /* wait for swapin block io completion */
759 u32 blkio_count; /* total count of the number of sync block */
760 /* io operations performed */
761 u32 swapin_count; /* total count of the number of swapin block */
762 /* io operations performed */
764 struct timespec freepages_start, freepages_end;
765 u64 freepages_delay; /* wait for memory reclaim */
766 u32 freepages_count; /* total count of memory reclaim */
768 #endif /* CONFIG_TASK_DELAY_ACCT */
770 static inline int sched_info_on(void)
772 #ifdef CONFIG_SCHEDSTATS
774 #elif defined(CONFIG_TASK_DELAY_ACCT)
775 extern int delayacct_on;
790 * sched-domains (multiprocessor balancing) declarations:
794 * Increase resolution of nice-level calculations:
796 #define SCHED_LOAD_SHIFT 10
797 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
799 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
802 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
803 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
804 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
805 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
806 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
807 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
808 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
809 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
810 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
811 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
812 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
813 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
814 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
816 enum powersavings_balance_level {
817 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
818 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
819 * first for long running threads
821 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
822 * cpu package for power savings
824 MAX_POWERSAVINGS_BALANCE_LEVELS
827 extern int sched_mc_power_savings, sched_smt_power_savings;
829 static inline int sd_balance_for_mc_power(void)
831 if (sched_smt_power_savings)
832 return SD_POWERSAVINGS_BALANCE;
834 if (!sched_mc_power_savings)
835 return SD_PREFER_SIBLING;
840 static inline int sd_balance_for_package_power(void)
842 if (sched_mc_power_savings | sched_smt_power_savings)
843 return SD_POWERSAVINGS_BALANCE;
845 return SD_PREFER_SIBLING;
848 extern int __weak arch_sd_sibiling_asym_packing(void);
851 * Optimise SD flags for power savings:
852 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
853 * Keep default SD flags if sched_{smt,mc}_power_saving=0
856 static inline int sd_power_saving_flags(void)
858 if (sched_mc_power_savings | sched_smt_power_savings)
859 return SD_BALANCE_NEWIDLE;
865 struct sched_group *next; /* Must be a circular list */
868 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
871 unsigned int cpu_power, cpu_power_orig;
872 unsigned int group_weight;
875 * The CPUs this group covers.
877 * NOTE: this field is variable length. (Allocated dynamically
878 * by attaching extra space to the end of the structure,
879 * depending on how many CPUs the kernel has booted up with)
881 * It is also be embedded into static data structures at build
882 * time. (See 'struct static_sched_group' in kernel/sched.c)
884 unsigned long cpumask[0];
887 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
889 return to_cpumask(sg->cpumask);
892 enum sched_domain_level {
903 struct sched_domain_attr {
904 int relax_domain_level;
907 #define SD_ATTR_INIT (struct sched_domain_attr) { \
908 .relax_domain_level = -1, \
911 struct sched_domain {
912 /* These fields must be setup */
913 struct sched_domain *parent; /* top domain must be null terminated */
914 struct sched_domain *child; /* bottom domain must be null terminated */
915 struct sched_group *groups; /* the balancing groups of the domain */
916 unsigned long min_interval; /* Minimum balance interval ms */
917 unsigned long max_interval; /* Maximum balance interval ms */
918 unsigned int busy_factor; /* less balancing by factor if busy */
919 unsigned int imbalance_pct; /* No balance until over watermark */
920 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
921 unsigned int busy_idx;
922 unsigned int idle_idx;
923 unsigned int newidle_idx;
924 unsigned int wake_idx;
925 unsigned int forkexec_idx;
926 unsigned int smt_gain;
927 int flags; /* See SD_* */
928 enum sched_domain_level level;
930 /* Runtime fields. */
931 unsigned long last_balance; /* init to jiffies. units in jiffies */
932 unsigned int balance_interval; /* initialise to 1. units in ms. */
933 unsigned int nr_balance_failed; /* initialise to 0 */
937 #ifdef CONFIG_SCHEDSTATS
938 /* load_balance() stats */
939 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
940 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
948 /* Active load balancing */
949 unsigned int alb_count;
950 unsigned int alb_failed;
951 unsigned int alb_pushed;
953 /* SD_BALANCE_EXEC stats */
954 unsigned int sbe_count;
955 unsigned int sbe_balanced;
956 unsigned int sbe_pushed;
958 /* SD_BALANCE_FORK stats */
959 unsigned int sbf_count;
960 unsigned int sbf_balanced;
961 unsigned int sbf_pushed;
963 /* try_to_wake_up() stats */
964 unsigned int ttwu_wake_remote;
965 unsigned int ttwu_move_affine;
966 unsigned int ttwu_move_balance;
968 #ifdef CONFIG_SCHED_DEBUG
972 unsigned int span_weight;
974 * Span of all CPUs in this domain.
976 * NOTE: this field is variable length. (Allocated dynamically
977 * by attaching extra space to the end of the structure,
978 * depending on how many CPUs the kernel has booted up with)
980 * It is also be embedded into static data structures at build
981 * time. (See 'struct static_sched_domain' in kernel/sched.c)
983 unsigned long span[0];
986 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
988 return to_cpumask(sd->span);
991 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
992 struct sched_domain_attr *dattr_new);
994 /* Allocate an array of sched domains, for partition_sched_domains(). */
995 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
996 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
998 /* Test a flag in parent sched domain */
999 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1001 if (sd->parent && (sd->parent->flags & flag))
1007 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1008 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1010 #else /* CONFIG_SMP */
1012 struct sched_domain_attr;
1015 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1016 struct sched_domain_attr *dattr_new)
1019 #endif /* !CONFIG_SMP */
1022 struct io_context; /* See blkdev.h */
1025 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1026 extern void prefetch_stack(struct task_struct *t);
1028 static inline void prefetch_stack(struct task_struct *t) { }
1031 struct audit_context; /* See audit.c */
1033 struct pipe_inode_info;
1034 struct uts_namespace;
1037 struct sched_domain;
1042 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1043 #define WF_FORK 0x02 /* child wakeup after fork */
1045 #define ENQUEUE_WAKEUP 1
1046 #define ENQUEUE_WAKING 2
1047 #define ENQUEUE_HEAD 4
1049 #define DEQUEUE_SLEEP 1
1051 struct sched_class {
1052 const struct sched_class *next;
1054 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1055 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1056 void (*yield_task) (struct rq *rq);
1058 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1060 struct task_struct * (*pick_next_task) (struct rq *rq);
1061 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1064 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1065 int sd_flag, int flags);
1067 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1068 void (*post_schedule) (struct rq *this_rq);
1069 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1070 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1072 void (*set_cpus_allowed)(struct task_struct *p,
1073 const struct cpumask *newmask);
1075 void (*rq_online)(struct rq *rq);
1076 void (*rq_offline)(struct rq *rq);
1079 void (*set_curr_task) (struct rq *rq);
1080 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1081 void (*task_fork) (struct task_struct *p);
1083 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1085 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1087 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1088 int oldprio, int running);
1090 unsigned int (*get_rr_interval) (struct rq *rq,
1091 struct task_struct *task);
1093 #ifdef CONFIG_FAIR_GROUP_SCHED
1094 void (*task_move_group) (struct task_struct *p, int on_rq);
1098 struct load_weight {
1099 unsigned long weight, inv_weight;
1102 #ifdef CONFIG_SCHEDSTATS
1103 struct sched_statistics {
1113 s64 sum_sleep_runtime;
1120 u64 nr_migrations_cold;
1121 u64 nr_failed_migrations_affine;
1122 u64 nr_failed_migrations_running;
1123 u64 nr_failed_migrations_hot;
1124 u64 nr_forced_migrations;
1127 u64 nr_wakeups_sync;
1128 u64 nr_wakeups_migrate;
1129 u64 nr_wakeups_local;
1130 u64 nr_wakeups_remote;
1131 u64 nr_wakeups_affine;
1132 u64 nr_wakeups_affine_attempts;
1133 u64 nr_wakeups_passive;
1134 u64 nr_wakeups_idle;
1138 struct sched_entity {
1139 struct load_weight load; /* for load-balancing */
1140 struct rb_node run_node;
1141 struct list_head group_node;
1145 u64 sum_exec_runtime;
1147 u64 prev_sum_exec_runtime;
1151 #ifdef CONFIG_SCHEDSTATS
1152 struct sched_statistics statistics;
1155 #ifdef CONFIG_FAIR_GROUP_SCHED
1156 struct sched_entity *parent;
1157 /* rq on which this entity is (to be) queued: */
1158 struct cfs_rq *cfs_rq;
1159 /* rq "owned" by this entity/group: */
1160 struct cfs_rq *my_q;
1164 struct sched_rt_entity {
1165 struct list_head run_list;
1166 unsigned long timeout;
1167 unsigned int time_slice;
1168 int nr_cpus_allowed;
1170 struct sched_rt_entity *back;
1171 #ifdef CONFIG_RT_GROUP_SCHED
1172 struct sched_rt_entity *parent;
1173 /* rq on which this entity is (to be) queued: */
1174 struct rt_rq *rt_rq;
1175 /* rq "owned" by this entity/group: */
1182 enum perf_event_task_context {
1183 perf_invalid_context = -1,
1184 perf_hw_context = 0,
1186 perf_nr_task_contexts,
1189 struct task_struct {
1190 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1193 unsigned int flags; /* per process flags, defined below */
1194 unsigned int ptrace;
1196 int lock_depth; /* BKL lock depth */
1199 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1204 int prio, static_prio, normal_prio;
1205 unsigned int rt_priority;
1206 const struct sched_class *sched_class;
1207 struct sched_entity se;
1208 struct sched_rt_entity rt;
1210 #ifdef CONFIG_PREEMPT_NOTIFIERS
1211 /* list of struct preempt_notifier: */
1212 struct hlist_head preempt_notifiers;
1216 * fpu_counter contains the number of consecutive context switches
1217 * that the FPU is used. If this is over a threshold, the lazy fpu
1218 * saving becomes unlazy to save the trap. This is an unsigned char
1219 * so that after 256 times the counter wraps and the behavior turns
1220 * lazy again; this to deal with bursty apps that only use FPU for
1223 unsigned char fpu_counter;
1224 #ifdef CONFIG_BLK_DEV_IO_TRACE
1225 unsigned int btrace_seq;
1228 unsigned int policy;
1229 cpumask_t cpus_allowed;
1231 #ifdef CONFIG_PREEMPT_RCU
1232 int rcu_read_lock_nesting;
1233 char rcu_read_unlock_special;
1234 struct list_head rcu_node_entry;
1235 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1236 #ifdef CONFIG_TREE_PREEMPT_RCU
1237 struct rcu_node *rcu_blocked_node;
1238 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1239 #ifdef CONFIG_RCU_BOOST
1240 struct rt_mutex *rcu_boost_mutex;
1241 #endif /* #ifdef CONFIG_RCU_BOOST */
1243 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1244 struct sched_info sched_info;
1247 struct list_head tasks;
1249 struct plist_node pushable_tasks;
1252 struct mm_struct *mm, *active_mm;
1253 #if defined(SPLIT_RSS_COUNTING)
1254 struct task_rss_stat rss_stat;
1258 int exit_code, exit_signal;
1259 int pdeath_signal; /* The signal sent when the parent dies */
1261 unsigned int personality;
1262 unsigned did_exec:1;
1263 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1265 unsigned in_iowait:1;
1268 /* Revert to default priority/policy when forking */
1269 unsigned sched_reset_on_fork:1;
1274 #ifdef CONFIG_CC_STACKPROTECTOR
1275 /* Canary value for the -fstack-protector gcc feature */
1276 unsigned long stack_canary;
1280 * pointers to (original) parent process, youngest child, younger sibling,
1281 * older sibling, respectively. (p->father can be replaced with
1282 * p->real_parent->pid)
1284 struct task_struct *real_parent; /* real parent process */
1285 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1287 * children/sibling forms the list of my natural children
1289 struct list_head children; /* list of my children */
1290 struct list_head sibling; /* linkage in my parent's children list */
1291 struct task_struct *group_leader; /* threadgroup leader */
1294 * ptraced is the list of tasks this task is using ptrace on.
1295 * This includes both natural children and PTRACE_ATTACH targets.
1296 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1298 struct list_head ptraced;
1299 struct list_head ptrace_entry;
1301 /* PID/PID hash table linkage. */
1302 struct pid_link pids[PIDTYPE_MAX];
1303 struct list_head thread_group;
1305 struct completion *vfork_done; /* for vfork() */
1306 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1307 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1309 cputime_t utime, stime, utimescaled, stimescaled;
1311 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1312 cputime_t prev_utime, prev_stime;
1314 unsigned long nvcsw, nivcsw; /* context switch counts */
1315 struct timespec start_time; /* monotonic time */
1316 struct timespec real_start_time; /* boot based time */
1317 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1318 unsigned long min_flt, maj_flt;
1320 struct task_cputime cputime_expires;
1321 struct list_head cpu_timers[3];
1323 /* process credentials */
1324 const struct cred __rcu *real_cred; /* objective and real subjective task
1325 * credentials (COW) */
1326 const struct cred __rcu *cred; /* effective (overridable) subjective task
1327 * credentials (COW) */
1328 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1330 char comm[TASK_COMM_LEN]; /* executable name excluding path
1331 - access with [gs]et_task_comm (which lock
1332 it with task_lock())
1333 - initialized normally by setup_new_exec */
1334 /* file system info */
1335 int link_count, total_link_count;
1336 #ifdef CONFIG_SYSVIPC
1338 struct sysv_sem sysvsem;
1340 #ifdef CONFIG_DETECT_HUNG_TASK
1341 /* hung task detection */
1342 unsigned long last_switch_count;
1344 /* CPU-specific state of this task */
1345 struct thread_struct thread;
1346 /* filesystem information */
1347 struct fs_struct *fs;
1348 /* open file information */
1349 struct files_struct *files;
1351 struct nsproxy *nsproxy;
1352 /* signal handlers */
1353 struct signal_struct *signal;
1354 struct sighand_struct *sighand;
1356 sigset_t blocked, real_blocked;
1357 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1358 struct sigpending pending;
1360 unsigned long sas_ss_sp;
1362 int (*notifier)(void *priv);
1363 void *notifier_data;
1364 sigset_t *notifier_mask;
1365 struct audit_context *audit_context;
1366 #ifdef CONFIG_AUDITSYSCALL
1368 unsigned int sessionid;
1372 /* Thread group tracking */
1375 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1377 spinlock_t alloc_lock;
1379 #ifdef CONFIG_GENERIC_HARDIRQS
1380 /* IRQ handler threads */
1381 struct irqaction *irqaction;
1384 /* Protection of the PI data structures: */
1385 raw_spinlock_t pi_lock;
1387 #ifdef CONFIG_RT_MUTEXES
1388 /* PI waiters blocked on a rt_mutex held by this task */
1389 struct plist_head pi_waiters;
1390 /* Deadlock detection and priority inheritance handling */
1391 struct rt_mutex_waiter *pi_blocked_on;
1394 #ifdef CONFIG_DEBUG_MUTEXES
1395 /* mutex deadlock detection */
1396 struct mutex_waiter *blocked_on;
1398 #ifdef CONFIG_TRACE_IRQFLAGS
1399 unsigned int irq_events;
1400 unsigned long hardirq_enable_ip;
1401 unsigned long hardirq_disable_ip;
1402 unsigned int hardirq_enable_event;
1403 unsigned int hardirq_disable_event;
1404 int hardirqs_enabled;
1405 int hardirq_context;
1406 unsigned long softirq_disable_ip;
1407 unsigned long softirq_enable_ip;
1408 unsigned int softirq_disable_event;
1409 unsigned int softirq_enable_event;
1410 int softirqs_enabled;
1411 int softirq_context;
1413 #ifdef CONFIG_LOCKDEP
1414 # define MAX_LOCK_DEPTH 48UL
1417 unsigned int lockdep_recursion;
1418 struct held_lock held_locks[MAX_LOCK_DEPTH];
1419 gfp_t lockdep_reclaim_gfp;
1422 /* journalling filesystem info */
1425 /* stacked block device info */
1426 struct bio_list *bio_list;
1429 struct reclaim_state *reclaim_state;
1431 struct backing_dev_info *backing_dev_info;
1433 struct io_context *io_context;
1435 unsigned long ptrace_message;
1436 siginfo_t *last_siginfo; /* For ptrace use. */
1437 struct task_io_accounting ioac;
1438 #if defined(CONFIG_TASK_XACCT)
1439 u64 acct_rss_mem1; /* accumulated rss usage */
1440 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1441 cputime_t acct_timexpd; /* stime + utime since last update */
1443 #ifdef CONFIG_CPUSETS
1444 nodemask_t mems_allowed; /* Protected by alloc_lock */
1445 int mems_allowed_change_disable;
1446 int cpuset_mem_spread_rotor;
1447 int cpuset_slab_spread_rotor;
1449 #ifdef CONFIG_CGROUPS
1450 /* Control Group info protected by css_set_lock */
1451 struct css_set __rcu *cgroups;
1452 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1453 struct list_head cg_list;
1456 struct robust_list_head __user *robust_list;
1457 #ifdef CONFIG_COMPAT
1458 struct compat_robust_list_head __user *compat_robust_list;
1460 struct list_head pi_state_list;
1461 struct futex_pi_state *pi_state_cache;
1463 #ifdef CONFIG_PERF_EVENTS
1464 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1465 struct mutex perf_event_mutex;
1466 struct list_head perf_event_list;
1469 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1472 atomic_t fs_excl; /* holding fs exclusive resources */
1473 struct rcu_head rcu;
1476 * cache last used pipe for splice
1478 struct pipe_inode_info *splice_pipe;
1479 #ifdef CONFIG_TASK_DELAY_ACCT
1480 struct task_delay_info *delays;
1482 #ifdef CONFIG_FAULT_INJECTION
1485 struct prop_local_single dirties;
1486 #ifdef CONFIG_LATENCYTOP
1487 int latency_record_count;
1488 struct latency_record latency_record[LT_SAVECOUNT];
1491 * time slack values; these are used to round up poll() and
1492 * select() etc timeout values. These are in nanoseconds.
1494 unsigned long timer_slack_ns;
1495 unsigned long default_timer_slack_ns;
1497 struct list_head *scm_work_list;
1498 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1499 /* Index of current stored address in ret_stack */
1501 /* Stack of return addresses for return function tracing */
1502 struct ftrace_ret_stack *ret_stack;
1503 /* time stamp for last schedule */
1504 unsigned long long ftrace_timestamp;
1506 * Number of functions that haven't been traced
1507 * because of depth overrun.
1509 atomic_t trace_overrun;
1510 /* Pause for the tracing */
1511 atomic_t tracing_graph_pause;
1513 #ifdef CONFIG_TRACING
1514 /* state flags for use by tracers */
1515 unsigned long trace;
1516 /* bitmask of trace recursion */
1517 unsigned long trace_recursion;
1518 #endif /* CONFIG_TRACING */
1519 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1520 struct memcg_batch_info {
1521 int do_batch; /* incremented when batch uncharge started */
1522 struct mem_cgroup *memcg; /* target memcg of uncharge */
1523 unsigned long bytes; /* uncharged usage */
1524 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1529 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1530 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1533 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1534 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1535 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1536 * values are inverted: lower p->prio value means higher priority.
1538 * The MAX_USER_RT_PRIO value allows the actual maximum
1539 * RT priority to be separate from the value exported to
1540 * user-space. This allows kernel threads to set their
1541 * priority to a value higher than any user task. Note:
1542 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1545 #define MAX_USER_RT_PRIO 100
1546 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1548 #define MAX_PRIO (MAX_RT_PRIO + 40)
1549 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1551 static inline int rt_prio(int prio)
1553 if (unlikely(prio < MAX_RT_PRIO))
1558 static inline int rt_task(struct task_struct *p)
1560 return rt_prio(p->prio);
1563 static inline struct pid *task_pid(struct task_struct *task)
1565 return task->pids[PIDTYPE_PID].pid;
1568 static inline struct pid *task_tgid(struct task_struct *task)
1570 return task->group_leader->pids[PIDTYPE_PID].pid;
1574 * Without tasklist or rcu lock it is not safe to dereference
1575 * the result of task_pgrp/task_session even if task == current,
1576 * we can race with another thread doing sys_setsid/sys_setpgid.
1578 static inline struct pid *task_pgrp(struct task_struct *task)
1580 return task->group_leader->pids[PIDTYPE_PGID].pid;
1583 static inline struct pid *task_session(struct task_struct *task)
1585 return task->group_leader->pids[PIDTYPE_SID].pid;
1588 struct pid_namespace;
1591 * the helpers to get the task's different pids as they are seen
1592 * from various namespaces
1594 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1595 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1597 * task_xid_nr_ns() : id seen from the ns specified;
1599 * set_task_vxid() : assigns a virtual id to a task;
1601 * see also pid_nr() etc in include/linux/pid.h
1603 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1604 struct pid_namespace *ns);
1606 static inline pid_t task_pid_nr(struct task_struct *tsk)
1611 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1612 struct pid_namespace *ns)
1614 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1617 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1619 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1623 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1628 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1630 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1632 return pid_vnr(task_tgid(tsk));
1636 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1637 struct pid_namespace *ns)
1639 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1642 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1644 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1648 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1649 struct pid_namespace *ns)
1651 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1654 static inline pid_t task_session_vnr(struct task_struct *tsk)
1656 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1659 /* obsolete, do not use */
1660 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1662 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1666 * pid_alive - check that a task structure is not stale
1667 * @p: Task structure to be checked.
1669 * Test if a process is not yet dead (at most zombie state)
1670 * If pid_alive fails, then pointers within the task structure
1671 * can be stale and must not be dereferenced.
1673 static inline int pid_alive(struct task_struct *p)
1675 return p->pids[PIDTYPE_PID].pid != NULL;
1679 * is_global_init - check if a task structure is init
1680 * @tsk: Task structure to be checked.
1682 * Check if a task structure is the first user space task the kernel created.
1684 static inline int is_global_init(struct task_struct *tsk)
1686 return tsk->pid == 1;
1690 * is_container_init:
1691 * check whether in the task is init in its own pid namespace.
1693 extern int is_container_init(struct task_struct *tsk);
1695 extern struct pid *cad_pid;
1697 extern void free_task(struct task_struct *tsk);
1698 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1700 extern void __put_task_struct(struct task_struct *t);
1702 static inline void put_task_struct(struct task_struct *t)
1704 if (atomic_dec_and_test(&t->usage))
1705 __put_task_struct(t);
1708 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1709 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1714 #define PF_KSOFTIRQD 0x00000001 /* I am ksoftirqd */
1715 #define PF_STARTING 0x00000002 /* being created */
1716 #define PF_EXITING 0x00000004 /* getting shut down */
1717 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1718 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1719 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1720 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1721 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1722 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1723 #define PF_DUMPCORE 0x00000200 /* dumped core */
1724 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1725 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1726 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1727 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1728 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1729 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1730 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1731 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1732 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1733 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1734 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1735 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1736 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1737 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1738 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1739 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1740 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1741 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1742 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1743 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1744 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1747 * Only the _current_ task can read/write to tsk->flags, but other
1748 * tasks can access tsk->flags in readonly mode for example
1749 * with tsk_used_math (like during threaded core dumping).
1750 * There is however an exception to this rule during ptrace
1751 * or during fork: the ptracer task is allowed to write to the
1752 * child->flags of its traced child (same goes for fork, the parent
1753 * can write to the child->flags), because we're guaranteed the
1754 * child is not running and in turn not changing child->flags
1755 * at the same time the parent does it.
1757 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1758 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1759 #define clear_used_math() clear_stopped_child_used_math(current)
1760 #define set_used_math() set_stopped_child_used_math(current)
1761 #define conditional_stopped_child_used_math(condition, child) \
1762 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1763 #define conditional_used_math(condition) \
1764 conditional_stopped_child_used_math(condition, current)
1765 #define copy_to_stopped_child_used_math(child) \
1766 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1767 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1768 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1769 #define used_math() tsk_used_math(current)
1771 #ifdef CONFIG_PREEMPT_RCU
1773 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1774 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1775 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1777 static inline void rcu_copy_process(struct task_struct *p)
1779 p->rcu_read_lock_nesting = 0;
1780 p->rcu_read_unlock_special = 0;
1781 #ifdef CONFIG_TREE_PREEMPT_RCU
1782 p->rcu_blocked_node = NULL;
1783 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1784 #ifdef CONFIG_RCU_BOOST
1785 p->rcu_boost_mutex = NULL;
1786 #endif /* #ifdef CONFIG_RCU_BOOST */
1787 INIT_LIST_HEAD(&p->rcu_node_entry);
1792 static inline void rcu_copy_process(struct task_struct *p)
1799 extern int set_cpus_allowed_ptr(struct task_struct *p,
1800 const struct cpumask *new_mask);
1802 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1803 const struct cpumask *new_mask)
1805 if (!cpumask_test_cpu(0, new_mask))
1811 #ifndef CONFIG_CPUMASK_OFFSTACK
1812 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1814 return set_cpus_allowed_ptr(p, &new_mask);
1819 * Do not use outside of architecture code which knows its limitations.
1821 * sched_clock() has no promise of monotonicity or bounded drift between
1822 * CPUs, use (which you should not) requires disabling IRQs.
1824 * Please use one of the three interfaces below.
1826 extern unsigned long long notrace sched_clock(void);
1828 * See the comment in kernel/sched_clock.c
1830 extern u64 cpu_clock(int cpu);
1831 extern u64 local_clock(void);
1832 extern u64 sched_clock_cpu(int cpu);
1835 extern void sched_clock_init(void);
1837 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1838 static inline void sched_clock_tick(void)
1842 static inline void sched_clock_idle_sleep_event(void)
1846 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1851 * Architectures can set this to 1 if they have specified
1852 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1853 * but then during bootup it turns out that sched_clock()
1854 * is reliable after all:
1856 extern int sched_clock_stable;
1858 extern void sched_clock_tick(void);
1859 extern void sched_clock_idle_sleep_event(void);
1860 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1863 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1865 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1866 * The reason for this explicit opt-in is not to have perf penalty with
1867 * slow sched_clocks.
1869 extern void enable_sched_clock_irqtime(void);
1870 extern void disable_sched_clock_irqtime(void);
1872 static inline void enable_sched_clock_irqtime(void) {}
1873 static inline void disable_sched_clock_irqtime(void) {}
1876 extern unsigned long long
1877 task_sched_runtime(struct task_struct *task);
1878 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1880 /* sched_exec is called by processes performing an exec */
1882 extern void sched_exec(void);
1884 #define sched_exec() {}
1887 extern void sched_clock_idle_sleep_event(void);
1888 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1890 #ifdef CONFIG_HOTPLUG_CPU
1891 extern void idle_task_exit(void);
1893 static inline void idle_task_exit(void) {}
1896 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1897 extern void wake_up_idle_cpu(int cpu);
1899 static inline void wake_up_idle_cpu(int cpu) { }
1902 extern unsigned int sysctl_sched_latency;
1903 extern unsigned int sysctl_sched_min_granularity;
1904 extern unsigned int sysctl_sched_wakeup_granularity;
1905 extern unsigned int sysctl_sched_child_runs_first;
1907 enum sched_tunable_scaling {
1908 SCHED_TUNABLESCALING_NONE,
1909 SCHED_TUNABLESCALING_LOG,
1910 SCHED_TUNABLESCALING_LINEAR,
1911 SCHED_TUNABLESCALING_END,
1913 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1915 #ifdef CONFIG_SCHED_DEBUG
1916 extern unsigned int sysctl_sched_migration_cost;
1917 extern unsigned int sysctl_sched_nr_migrate;
1918 extern unsigned int sysctl_sched_time_avg;
1919 extern unsigned int sysctl_timer_migration;
1920 extern unsigned int sysctl_sched_shares_window;
1922 int sched_proc_update_handler(struct ctl_table *table, int write,
1923 void __user *buffer, size_t *length,
1926 #ifdef CONFIG_SCHED_DEBUG
1927 static inline unsigned int get_sysctl_timer_migration(void)
1929 return sysctl_timer_migration;
1932 static inline unsigned int get_sysctl_timer_migration(void)
1937 extern unsigned int sysctl_sched_rt_period;
1938 extern int sysctl_sched_rt_runtime;
1940 int sched_rt_handler(struct ctl_table *table, int write,
1941 void __user *buffer, size_t *lenp,
1944 extern unsigned int sysctl_sched_compat_yield;
1946 #ifdef CONFIG_SCHED_AUTOGROUP
1947 extern unsigned int sysctl_sched_autogroup_enabled;
1949 extern void sched_autogroup_create_attach(struct task_struct *p);
1950 extern void sched_autogroup_detach(struct task_struct *p);
1951 extern void sched_autogroup_fork(struct signal_struct *sig);
1952 extern void sched_autogroup_exit(struct signal_struct *sig);
1953 #ifdef CONFIG_PROC_FS
1954 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1955 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
1958 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1959 static inline void sched_autogroup_detach(struct task_struct *p) { }
1960 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1961 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1964 #ifdef CONFIG_RT_MUTEXES
1965 extern int rt_mutex_getprio(struct task_struct *p);
1966 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1967 extern void rt_mutex_adjust_pi(struct task_struct *p);
1969 static inline int rt_mutex_getprio(struct task_struct *p)
1971 return p->normal_prio;
1973 # define rt_mutex_adjust_pi(p) do { } while (0)
1976 extern void set_user_nice(struct task_struct *p, long nice);
1977 extern int task_prio(const struct task_struct *p);
1978 extern int task_nice(const struct task_struct *p);
1979 extern int can_nice(const struct task_struct *p, const int nice);
1980 extern int task_curr(const struct task_struct *p);
1981 extern int idle_cpu(int cpu);
1982 extern int sched_setscheduler(struct task_struct *, int,
1983 const struct sched_param *);
1984 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1985 const struct sched_param *);
1986 extern struct task_struct *idle_task(int cpu);
1987 extern struct task_struct *curr_task(int cpu);
1988 extern void set_curr_task(int cpu, struct task_struct *p);
1993 * The default (Linux) execution domain.
1995 extern struct exec_domain default_exec_domain;
1997 union thread_union {
1998 struct thread_info thread_info;
1999 unsigned long stack[THREAD_SIZE/sizeof(long)];
2002 #ifndef __HAVE_ARCH_KSTACK_END
2003 static inline int kstack_end(void *addr)
2005 /* Reliable end of stack detection:
2006 * Some APM bios versions misalign the stack
2008 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2012 extern union thread_union init_thread_union;
2013 extern struct task_struct init_task;
2015 extern struct mm_struct init_mm;
2017 extern struct pid_namespace init_pid_ns;
2020 * find a task by one of its numerical ids
2022 * find_task_by_pid_ns():
2023 * finds a task by its pid in the specified namespace
2024 * find_task_by_vpid():
2025 * finds a task by its virtual pid
2027 * see also find_vpid() etc in include/linux/pid.h
2030 extern struct task_struct *find_task_by_vpid(pid_t nr);
2031 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2032 struct pid_namespace *ns);
2034 extern void __set_special_pids(struct pid *pid);
2036 /* per-UID process charging. */
2037 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2038 static inline struct user_struct *get_uid(struct user_struct *u)
2040 atomic_inc(&u->__count);
2043 extern void free_uid(struct user_struct *);
2044 extern void release_uids(struct user_namespace *ns);
2046 #include <asm/current.h>
2048 extern void do_timer(unsigned long ticks);
2050 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2051 extern int wake_up_process(struct task_struct *tsk);
2052 extern void wake_up_new_task(struct task_struct *tsk,
2053 unsigned long clone_flags);
2055 extern void kick_process(struct task_struct *tsk);
2057 static inline void kick_process(struct task_struct *tsk) { }
2059 extern void sched_fork(struct task_struct *p, int clone_flags);
2060 extern void sched_dead(struct task_struct *p);
2062 extern void proc_caches_init(void);
2063 extern void flush_signals(struct task_struct *);
2064 extern void __flush_signals(struct task_struct *);
2065 extern void ignore_signals(struct task_struct *);
2066 extern void flush_signal_handlers(struct task_struct *, int force_default);
2067 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2069 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2071 unsigned long flags;
2074 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2075 ret = dequeue_signal(tsk, mask, info);
2076 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2081 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2083 extern void unblock_all_signals(void);
2084 extern void release_task(struct task_struct * p);
2085 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2086 extern int force_sigsegv(int, struct task_struct *);
2087 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2088 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2089 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2090 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2091 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2092 extern int kill_pid(struct pid *pid, int sig, int priv);
2093 extern int kill_proc_info(int, struct siginfo *, pid_t);
2094 extern int do_notify_parent(struct task_struct *, int);
2095 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2096 extern void force_sig(int, struct task_struct *);
2097 extern int send_sig(int, struct task_struct *, int);
2098 extern int zap_other_threads(struct task_struct *p);
2099 extern struct sigqueue *sigqueue_alloc(void);
2100 extern void sigqueue_free(struct sigqueue *);
2101 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2102 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2103 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2105 static inline int kill_cad_pid(int sig, int priv)
2107 return kill_pid(cad_pid, sig, priv);
2110 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2111 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2112 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2113 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2116 * True if we are on the alternate signal stack.
2118 static inline int on_sig_stack(unsigned long sp)
2120 #ifdef CONFIG_STACK_GROWSUP
2121 return sp >= current->sas_ss_sp &&
2122 sp - current->sas_ss_sp < current->sas_ss_size;
2124 return sp > current->sas_ss_sp &&
2125 sp - current->sas_ss_sp <= current->sas_ss_size;
2129 static inline int sas_ss_flags(unsigned long sp)
2131 return (current->sas_ss_size == 0 ? SS_DISABLE
2132 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2136 * Routines for handling mm_structs
2138 extern struct mm_struct * mm_alloc(void);
2140 /* mmdrop drops the mm and the page tables */
2141 extern void __mmdrop(struct mm_struct *);
2142 static inline void mmdrop(struct mm_struct * mm)
2144 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2148 /* mmput gets rid of the mappings and all user-space */
2149 extern void mmput(struct mm_struct *);
2150 /* Grab a reference to a task's mm, if it is not already going away */
2151 extern struct mm_struct *get_task_mm(struct task_struct *task);
2152 /* Remove the current tasks stale references to the old mm_struct */
2153 extern void mm_release(struct task_struct *, struct mm_struct *);
2154 /* Allocate a new mm structure and copy contents from tsk->mm */
2155 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2157 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2158 struct task_struct *, struct pt_regs *);
2159 extern void flush_thread(void);
2160 extern void exit_thread(void);
2162 extern void exit_files(struct task_struct *);
2163 extern void __cleanup_sighand(struct sighand_struct *);
2165 extern void exit_itimers(struct signal_struct *);
2166 extern void flush_itimer_signals(void);
2168 extern NORET_TYPE void do_group_exit(int);
2170 extern void daemonize(const char *, ...);
2171 extern int allow_signal(int);
2172 extern int disallow_signal(int);
2174 extern int do_execve(const char *,
2175 const char __user * const __user *,
2176 const char __user * const __user *, struct pt_regs *);
2177 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2178 struct task_struct *fork_idle(int);
2180 extern void set_task_comm(struct task_struct *tsk, char *from);
2181 extern char *get_task_comm(char *to, struct task_struct *tsk);
2184 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2186 static inline unsigned long wait_task_inactive(struct task_struct *p,
2193 #define next_task(p) \
2194 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2196 #define for_each_process(p) \
2197 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2199 extern bool current_is_single_threaded(void);
2202 * Careful: do_each_thread/while_each_thread is a double loop so
2203 * 'break' will not work as expected - use goto instead.
2205 #define do_each_thread(g, t) \
2206 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2208 #define while_each_thread(g, t) \
2209 while ((t = next_thread(t)) != g)
2211 static inline int get_nr_threads(struct task_struct *tsk)
2213 return tsk->signal->nr_threads;
2216 /* de_thread depends on thread_group_leader not being a pid based check */
2217 #define thread_group_leader(p) (p == p->group_leader)
2219 /* Do to the insanities of de_thread it is possible for a process
2220 * to have the pid of the thread group leader without actually being
2221 * the thread group leader. For iteration through the pids in proc
2222 * all we care about is that we have a task with the appropriate
2223 * pid, we don't actually care if we have the right task.
2225 static inline int has_group_leader_pid(struct task_struct *p)
2227 return p->pid == p->tgid;
2231 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2233 return p1->tgid == p2->tgid;
2236 static inline struct task_struct *next_thread(const struct task_struct *p)
2238 return list_entry_rcu(p->thread_group.next,
2239 struct task_struct, thread_group);
2242 static inline int thread_group_empty(struct task_struct *p)
2244 return list_empty(&p->thread_group);
2247 #define delay_group_leader(p) \
2248 (thread_group_leader(p) && !thread_group_empty(p))
2250 static inline int task_detached(struct task_struct *p)
2252 return p->exit_signal == -1;
2256 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2257 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2258 * pins the final release of task.io_context. Also protects ->cpuset and
2259 * ->cgroup.subsys[].
2261 * Nests both inside and outside of read_lock(&tasklist_lock).
2262 * It must not be nested with write_lock_irq(&tasklist_lock),
2263 * neither inside nor outside.
2265 static inline void task_lock(struct task_struct *p)
2267 spin_lock(&p->alloc_lock);
2270 static inline void task_unlock(struct task_struct *p)
2272 spin_unlock(&p->alloc_lock);
2275 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2276 unsigned long *flags);
2278 #define lock_task_sighand(tsk, flags) \
2279 ({ struct sighand_struct *__ss; \
2280 __cond_lock(&(tsk)->sighand->siglock, \
2281 (__ss = __lock_task_sighand(tsk, flags))); \
2285 static inline void unlock_task_sighand(struct task_struct *tsk,
2286 unsigned long *flags)
2288 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2291 #ifndef __HAVE_THREAD_FUNCTIONS
2293 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2294 #define task_stack_page(task) ((task)->stack)
2296 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2298 *task_thread_info(p) = *task_thread_info(org);
2299 task_thread_info(p)->task = p;
2302 static inline unsigned long *end_of_stack(struct task_struct *p)
2304 return (unsigned long *)(task_thread_info(p) + 1);
2309 static inline int object_is_on_stack(void *obj)
2311 void *stack = task_stack_page(current);
2313 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2316 extern void thread_info_cache_init(void);
2318 #ifdef CONFIG_DEBUG_STACK_USAGE
2319 static inline unsigned long stack_not_used(struct task_struct *p)
2321 unsigned long *n = end_of_stack(p);
2323 do { /* Skip over canary */
2327 return (unsigned long)n - (unsigned long)end_of_stack(p);
2331 /* set thread flags in other task's structures
2332 * - see asm/thread_info.h for TIF_xxxx flags available
2334 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2336 set_ti_thread_flag(task_thread_info(tsk), flag);
2339 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2341 clear_ti_thread_flag(task_thread_info(tsk), flag);
2344 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2346 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2349 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2351 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2354 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2356 return test_ti_thread_flag(task_thread_info(tsk), flag);
2359 static inline void set_tsk_need_resched(struct task_struct *tsk)
2361 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2364 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2366 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2369 static inline int test_tsk_need_resched(struct task_struct *tsk)
2371 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2374 static inline int restart_syscall(void)
2376 set_tsk_thread_flag(current, TIF_SIGPENDING);
2377 return -ERESTARTNOINTR;
2380 static inline int signal_pending(struct task_struct *p)
2382 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2385 static inline int __fatal_signal_pending(struct task_struct *p)
2387 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2390 static inline int fatal_signal_pending(struct task_struct *p)
2392 return signal_pending(p) && __fatal_signal_pending(p);
2395 static inline int signal_pending_state(long state, struct task_struct *p)
2397 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2399 if (!signal_pending(p))
2402 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2405 static inline int need_resched(void)
2407 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2411 * cond_resched() and cond_resched_lock(): latency reduction via
2412 * explicit rescheduling in places that are safe. The return
2413 * value indicates whether a reschedule was done in fact.
2414 * cond_resched_lock() will drop the spinlock before scheduling,
2415 * cond_resched_softirq() will enable bhs before scheduling.
2417 extern int _cond_resched(void);
2419 #define cond_resched() ({ \
2420 __might_sleep(__FILE__, __LINE__, 0); \
2424 extern int __cond_resched_lock(spinlock_t *lock);
2426 #ifdef CONFIG_PREEMPT
2427 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2429 #define PREEMPT_LOCK_OFFSET 0
2432 #define cond_resched_lock(lock) ({ \
2433 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2434 __cond_resched_lock(lock); \
2437 extern int __cond_resched_softirq(void);
2439 #define cond_resched_softirq() ({ \
2440 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2441 __cond_resched_softirq(); \
2445 * Does a critical section need to be broken due to another
2446 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2447 * but a general need for low latency)
2449 static inline int spin_needbreak(spinlock_t *lock)
2451 #ifdef CONFIG_PREEMPT
2452 return spin_is_contended(lock);
2459 * Thread group CPU time accounting.
2461 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2462 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2464 static inline void thread_group_cputime_init(struct signal_struct *sig)
2466 spin_lock_init(&sig->cputimer.lock);
2470 * Reevaluate whether the task has signals pending delivery.
2471 * Wake the task if so.
2472 * This is required every time the blocked sigset_t changes.
2473 * callers must hold sighand->siglock.
2475 extern void recalc_sigpending_and_wake(struct task_struct *t);
2476 extern void recalc_sigpending(void);
2478 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2481 * Wrappers for p->thread_info->cpu access. No-op on UP.
2485 static inline unsigned int task_cpu(const struct task_struct *p)
2487 return task_thread_info(p)->cpu;
2490 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2494 static inline unsigned int task_cpu(const struct task_struct *p)
2499 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2503 #endif /* CONFIG_SMP */
2505 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2506 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2508 extern void normalize_rt_tasks(void);
2510 #ifdef CONFIG_CGROUP_SCHED
2512 extern struct task_group root_task_group;
2514 extern struct task_group *sched_create_group(struct task_group *parent);
2515 extern void sched_destroy_group(struct task_group *tg);
2516 extern void sched_move_task(struct task_struct *tsk);
2517 #ifdef CONFIG_FAIR_GROUP_SCHED
2518 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2519 extern unsigned long sched_group_shares(struct task_group *tg);
2521 #ifdef CONFIG_RT_GROUP_SCHED
2522 extern int sched_group_set_rt_runtime(struct task_group *tg,
2523 long rt_runtime_us);
2524 extern long sched_group_rt_runtime(struct task_group *tg);
2525 extern int sched_group_set_rt_period(struct task_group *tg,
2527 extern long sched_group_rt_period(struct task_group *tg);
2528 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2532 extern int task_can_switch_user(struct user_struct *up,
2533 struct task_struct *tsk);
2535 #ifdef CONFIG_TASK_XACCT
2536 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2538 tsk->ioac.rchar += amt;
2541 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2543 tsk->ioac.wchar += amt;
2546 static inline void inc_syscr(struct task_struct *tsk)
2551 static inline void inc_syscw(struct task_struct *tsk)
2556 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2560 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2564 static inline void inc_syscr(struct task_struct *tsk)
2568 static inline void inc_syscw(struct task_struct *tsk)
2573 #ifndef TASK_SIZE_OF
2574 #define TASK_SIZE_OF(tsk) TASK_SIZE
2578 * Call the function if the target task is executing on a CPU right now:
2580 extern void task_oncpu_function_call(struct task_struct *p,
2581 void (*func) (void *info), void *info);
2584 #ifdef CONFIG_MM_OWNER
2585 extern void mm_update_next_owner(struct mm_struct *mm);
2586 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2588 static inline void mm_update_next_owner(struct mm_struct *mm)
2592 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2595 #endif /* CONFIG_MM_OWNER */
2597 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2600 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2603 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2606 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2609 static inline unsigned long rlimit(unsigned int limit)
2611 return task_rlimit(current, limit);
2614 static inline unsigned long rlimit_max(unsigned int limit)
2616 return task_rlimit_max(current, limit);
2619 #endif /* __KERNEL__ */