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/path.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/kobject.h>
92 #include <linux/latencytop.h>
93 #include <linux/cred.h>
95 #include <asm/processor.h>
98 struct futex_pi_state;
99 struct robust_list_head;
103 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(void);
144 extern unsigned long this_cpu_load(void);
147 extern void calc_global_load(void);
149 extern unsigned long get_parent_ip(unsigned long addr);
154 #ifdef CONFIG_SCHED_DEBUG
155 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
156 extern void proc_sched_set_task(struct task_struct *p);
158 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
161 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
164 static inline void proc_sched_set_task(struct task_struct *p)
168 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
174 * Task state bitmask. NOTE! These bits are also
175 * encoded in fs/proc/array.c: get_task_state().
177 * We have two separate sets of flags: task->state
178 * is about runnability, while task->exit_state are
179 * about the task exiting. Confusing, but this way
180 * modifying one set can't modify the other one by
183 #define TASK_RUNNING 0
184 #define TASK_INTERRUPTIBLE 1
185 #define TASK_UNINTERRUPTIBLE 2
186 #define __TASK_STOPPED 4
187 #define __TASK_TRACED 8
188 /* in tsk->exit_state */
189 #define EXIT_ZOMBIE 16
191 /* in tsk->state again */
193 #define TASK_WAKEKILL 128
194 #define TASK_WAKING 256
195 #define TASK_STATE_MAX 512
197 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
199 extern char ___assert_task_state[1 - 2*!!(
200 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
202 /* Convenience macros for the sake of set_task_state */
203 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
204 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
205 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
207 /* Convenience macros for the sake of wake_up */
208 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
209 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
211 /* get_task_state() */
212 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
213 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
216 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
217 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
218 #define task_is_stopped_or_traced(task) \
219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
220 #define task_contributes_to_load(task) \
221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
222 (task->flags & PF_FREEZING) == 0)
224 #define __set_task_state(tsk, state_value) \
225 do { (tsk)->state = (state_value); } while (0)
226 #define set_task_state(tsk, state_value) \
227 set_mb((tsk)->state, (state_value))
230 * set_current_state() includes a barrier so that the write of current->state
231 * is correctly serialised wrt the caller's subsequent test of whether to
234 * set_current_state(TASK_UNINTERRUPTIBLE);
235 * if (do_i_need_to_sleep())
238 * If the caller does not need such serialisation then use __set_current_state()
240 #define __set_current_state(state_value) \
241 do { current->state = (state_value); } while (0)
242 #define set_current_state(state_value) \
243 set_mb(current->state, (state_value))
245 /* Task command name length */
246 #define TASK_COMM_LEN 16
248 #include <linux/spinlock.h>
251 * This serializes "schedule()" and also protects
252 * the run-queue from deletions/modifications (but
253 * _adding_ to the beginning of the run-queue has
256 extern rwlock_t tasklist_lock;
257 extern spinlock_t mmlist_lock;
261 extern void sched_init(void);
262 extern void sched_init_smp(void);
263 extern asmlinkage void schedule_tail(struct task_struct *prev);
264 extern void init_idle(struct task_struct *idle, int cpu);
265 extern void init_idle_bootup_task(struct task_struct *idle);
267 extern int runqueue_is_locked(int cpu);
268 extern void task_rq_unlock_wait(struct task_struct *p);
270 extern cpumask_var_t nohz_cpu_mask;
271 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
272 extern int select_nohz_load_balancer(int cpu);
273 extern int get_nohz_load_balancer(void);
275 static inline int select_nohz_load_balancer(int cpu)
282 * Only dump TASK_* tasks. (0 for all tasks)
284 extern void show_state_filter(unsigned long state_filter);
286 static inline void show_state(void)
288 show_state_filter(0);
291 extern void show_regs(struct pt_regs *);
294 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
295 * task), SP is the stack pointer of the first frame that should be shown in the back
296 * trace (or NULL if the entire call-chain of the task should be shown).
298 extern void show_stack(struct task_struct *task, unsigned long *sp);
300 void io_schedule(void);
301 long io_schedule_timeout(long timeout);
303 extern void cpu_init (void);
304 extern void trap_init(void);
305 extern void update_process_times(int user);
306 extern void scheduler_tick(void);
308 extern void sched_show_task(struct task_struct *p);
310 #ifdef CONFIG_DETECT_SOFTLOCKUP
311 extern void softlockup_tick(void);
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_dosoftlockup_thresh(struct ctl_table *table, int write,
317 size_t *lenp, loff_t *ppos);
318 extern unsigned int softlockup_panic;
319 extern int softlockup_thresh;
321 static inline void softlockup_tick(void)
324 static inline void touch_softlockup_watchdog(void)
327 static inline void touch_softlockup_watchdog_sync(void)
330 static inline void touch_all_softlockup_watchdogs(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);
345 /* Attach to any functions which should be ignored in wchan output. */
346 #define __sched __attribute__((__section__(".sched.text")))
348 /* Linker adds these: start and end of __sched functions */
349 extern char __sched_text_start[], __sched_text_end[];
351 /* Is this address in the __sched functions? */
352 extern int in_sched_functions(unsigned long addr);
354 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
355 extern signed long schedule_timeout(signed long timeout);
356 extern signed long schedule_timeout_interruptible(signed long timeout);
357 extern signed long schedule_timeout_killable(signed long timeout);
358 extern signed long schedule_timeout_uninterruptible(signed long timeout);
359 asmlinkage void schedule(void);
360 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
363 struct user_namespace;
366 * Default maximum number of active map areas, this limits the number of vmas
367 * per mm struct. Users can overwrite this number by sysctl but there is a
370 * When a program's coredump is generated as ELF format, a section is created
371 * per a vma. In ELF, the number of sections is represented in unsigned short.
372 * This means the number of sections should be smaller than 65535 at coredump.
373 * Because the kernel adds some informative sections to a image of program at
374 * generating coredump, we need some margin. The number of extra sections is
375 * 1-3 now and depends on arch. We use "5" as safe margin, here.
377 #define MAPCOUNT_ELF_CORE_MARGIN (5)
378 #define DEFAULT_MAX_MAP_COUNT (USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
380 extern int sysctl_max_map_count;
382 #include <linux/aio.h>
385 extern void arch_pick_mmap_layout(struct mm_struct *mm);
387 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
388 unsigned long, unsigned long);
390 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
391 unsigned long len, unsigned long pgoff,
392 unsigned long flags);
393 extern void arch_unmap_area(struct mm_struct *, unsigned long);
394 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
396 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399 #if USE_SPLIT_PTLOCKS
401 * The mm counters are not protected by its page_table_lock,
402 * so must be incremented atomically.
404 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
405 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
406 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
407 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
408 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
410 #else /* !USE_SPLIT_PTLOCKS */
412 * The mm counters are protected by its page_table_lock,
413 * so can be incremented directly.
415 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
416 #define get_mm_counter(mm, member) ((mm)->_##member)
417 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
418 #define inc_mm_counter(mm, member) (mm)->_##member++
419 #define dec_mm_counter(mm, member) (mm)->_##member--
421 #endif /* !USE_SPLIT_PTLOCKS */
423 #define get_mm_rss(mm) \
424 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
425 #define update_hiwater_rss(mm) do { \
426 unsigned long _rss = get_mm_rss(mm); \
427 if ((mm)->hiwater_rss < _rss) \
428 (mm)->hiwater_rss = _rss; \
430 #define update_hiwater_vm(mm) do { \
431 if ((mm)->hiwater_vm < (mm)->total_vm) \
432 (mm)->hiwater_vm = (mm)->total_vm; \
435 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
437 return max(mm->hiwater_rss, get_mm_rss(mm));
440 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
441 struct mm_struct *mm)
443 unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
445 if (*maxrss < hiwater_rss)
446 *maxrss = hiwater_rss;
449 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
451 return max(mm->hiwater_vm, mm->total_vm);
454 extern void set_dumpable(struct mm_struct *mm, int value);
455 extern int get_dumpable(struct mm_struct *mm);
459 #define MMF_DUMPABLE 0 /* core dump is permitted */
460 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
462 #define MMF_DUMPABLE_BITS 2
463 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
465 /* coredump filter bits */
466 #define MMF_DUMP_ANON_PRIVATE 2
467 #define MMF_DUMP_ANON_SHARED 3
468 #define MMF_DUMP_MAPPED_PRIVATE 4
469 #define MMF_DUMP_MAPPED_SHARED 5
470 #define MMF_DUMP_ELF_HEADERS 6
471 #define MMF_DUMP_HUGETLB_PRIVATE 7
472 #define MMF_DUMP_HUGETLB_SHARED 8
474 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
475 #define MMF_DUMP_FILTER_BITS 7
476 #define MMF_DUMP_FILTER_MASK \
477 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
478 #define MMF_DUMP_FILTER_DEFAULT \
479 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
480 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
482 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
483 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
485 # define MMF_DUMP_MASK_DEFAULT_ELF 0
487 /* leave room for more dump flags */
488 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
490 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
492 struct sighand_struct {
494 struct k_sigaction action[_NSIG];
496 wait_queue_head_t signalfd_wqh;
499 struct pacct_struct {
502 unsigned long ac_mem;
503 cputime_t ac_utime, ac_stime;
504 unsigned long ac_minflt, ac_majflt;
515 * struct task_cputime - collected CPU time counts
516 * @utime: time spent in user mode, in &cputime_t units
517 * @stime: time spent in kernel mode, in &cputime_t units
518 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
520 * This structure groups together three kinds of CPU time that are
521 * tracked for threads and thread groups. Most things considering
522 * CPU time want to group these counts together and treat all three
523 * of them in parallel.
525 struct task_cputime {
528 unsigned long long sum_exec_runtime;
530 /* Alternate field names when used to cache expirations. */
531 #define prof_exp stime
532 #define virt_exp utime
533 #define sched_exp sum_exec_runtime
535 #define INIT_CPUTIME \
536 (struct task_cputime) { \
537 .utime = cputime_zero, \
538 .stime = cputime_zero, \
539 .sum_exec_runtime = 0, \
543 * Disable preemption until the scheduler is running.
544 * Reset by start_kernel()->sched_init()->init_idle().
546 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
547 * before the scheduler is active -- see should_resched().
549 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
552 * struct thread_group_cputimer - thread group interval timer counts
553 * @cputime: thread group interval timers.
554 * @running: non-zero when there are timers running and
555 * @cputime receives updates.
556 * @lock: lock for fields in this struct.
558 * This structure contains the version of task_cputime, above, that is
559 * used for thread group CPU timer calculations.
561 struct thread_group_cputimer {
562 struct task_cputime cputime;
568 * NOTE! "signal_struct" does not have it's own
569 * locking, because a shared signal_struct always
570 * implies a shared sighand_struct, so locking
571 * sighand_struct is always a proper superset of
572 * the locking of signal_struct.
574 struct signal_struct {
578 wait_queue_head_t wait_chldexit; /* for wait4() */
580 /* current thread group signal load-balancing target: */
581 struct task_struct *curr_target;
583 /* shared signal handling: */
584 struct sigpending shared_pending;
586 /* thread group exit support */
589 * - notify group_exit_task when ->count is equal to notify_count
590 * - everyone except group_exit_task is stopped during signal delivery
591 * of fatal signals, group_exit_task processes the signal.
594 struct task_struct *group_exit_task;
596 /* thread group stop support, overloads group_exit_code too */
597 int group_stop_count;
598 unsigned int flags; /* see SIGNAL_* flags below */
600 /* POSIX.1b Interval Timers */
601 struct list_head posix_timers;
603 /* ITIMER_REAL timer for the process */
604 struct hrtimer real_timer;
605 struct pid *leader_pid;
606 ktime_t it_real_incr;
609 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
610 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
611 * values are defined to 0 and 1 respectively
613 struct cpu_itimer it[2];
616 * Thread group totals for process CPU timers.
617 * See thread_group_cputimer(), et al, for details.
619 struct thread_group_cputimer cputimer;
621 /* Earliest-expiration cache. */
622 struct task_cputime cputime_expires;
624 struct list_head cpu_timers[3];
626 struct pid *tty_old_pgrp;
628 /* boolean value for session group leader */
631 struct tty_struct *tty; /* NULL if no tty */
634 * Cumulative resource counters for dead threads in the group,
635 * and for reaped dead child processes forked by this group.
636 * Live threads maintain their own counters and add to these
637 * in __exit_signal, except for the group leader.
639 cputime_t utime, stime, cutime, cstime;
642 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
643 cputime_t prev_utime, prev_stime;
645 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
646 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
647 unsigned long inblock, oublock, cinblock, coublock;
648 unsigned long maxrss, cmaxrss;
649 struct task_io_accounting ioac;
652 * Cumulative ns of schedule CPU time fo dead threads in the
653 * group, not including a zombie group leader, (This only differs
654 * from jiffies_to_ns(utime + stime) if sched_clock uses something
655 * other than jiffies.)
657 unsigned long long sum_sched_runtime;
660 * We don't bother to synchronize most readers of this at all,
661 * because there is no reader checking a limit that actually needs
662 * to get both rlim_cur and rlim_max atomically, and either one
663 * alone is a single word that can safely be read normally.
664 * getrlimit/setrlimit use task_lock(current->group_leader) to
665 * protect this instead of the siglock, because they really
666 * have no need to disable irqs.
668 struct rlimit rlim[RLIM_NLIMITS];
670 #ifdef CONFIG_BSD_PROCESS_ACCT
671 struct pacct_struct pacct; /* per-process accounting information */
673 #ifdef CONFIG_TASKSTATS
674 struct taskstats *stats;
678 struct tty_audit_buf *tty_audit_buf;
681 int oom_adj; /* OOM kill score adjustment (bit shift) */
684 /* Context switch must be unlocked if interrupts are to be enabled */
685 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
686 # define __ARCH_WANT_UNLOCKED_CTXSW
690 * Bits in flags field of signal_struct.
692 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
693 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
694 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
695 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
697 * Pending notifications to parent.
699 #define SIGNAL_CLD_STOPPED 0x00000010
700 #define SIGNAL_CLD_CONTINUED 0x00000020
701 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
703 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
705 /* If true, all threads except ->group_exit_task have pending SIGKILL */
706 static inline int signal_group_exit(const struct signal_struct *sig)
708 return (sig->flags & SIGNAL_GROUP_EXIT) ||
709 (sig->group_exit_task != NULL);
713 * Some day this will be a full-fledged user tracking system..
716 atomic_t __count; /* reference count */
717 atomic_t processes; /* How many processes does this user have? */
718 atomic_t files; /* How many open files does this user have? */
719 atomic_t sigpending; /* How many pending signals does this user have? */
720 #ifdef CONFIG_INOTIFY_USER
721 atomic_t inotify_watches; /* How many inotify watches does this user have? */
722 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
725 atomic_t epoll_watches; /* The number of file descriptors currently watched */
727 #ifdef CONFIG_POSIX_MQUEUE
728 /* protected by mq_lock */
729 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
731 unsigned long locked_shm; /* How many pages of mlocked shm ? */
734 struct key *uid_keyring; /* UID specific keyring */
735 struct key *session_keyring; /* UID's default session keyring */
738 /* Hash table maintenance information */
739 struct hlist_node uidhash_node;
741 struct user_namespace *user_ns;
743 #ifdef CONFIG_PERF_EVENTS
744 atomic_long_t locked_vm;
748 extern int uids_sysfs_init(void);
750 extern struct user_struct *find_user(uid_t);
752 extern struct user_struct root_user;
753 #define INIT_USER (&root_user)
756 struct backing_dev_info;
757 struct reclaim_state;
759 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
761 /* cumulative counters */
762 unsigned long pcount; /* # of times run on this cpu */
763 unsigned long long run_delay; /* time spent waiting on a runqueue */
766 unsigned long long last_arrival,/* when we last ran on a cpu */
767 last_queued; /* when we were last queued to run */
768 #ifdef CONFIG_SCHEDSTATS
770 unsigned int bkl_count;
773 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
775 #ifdef CONFIG_TASK_DELAY_ACCT
776 struct task_delay_info {
778 unsigned int flags; /* Private per-task flags */
780 /* For each stat XXX, add following, aligned appropriately
782 * struct timespec XXX_start, XXX_end;
786 * Atomicity of updates to XXX_delay, XXX_count protected by
787 * single lock above (split into XXX_lock if contention is an issue).
791 * XXX_count is incremented on every XXX operation, the delay
792 * associated with the operation is added to XXX_delay.
793 * XXX_delay contains the accumulated delay time in nanoseconds.
795 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
796 u64 blkio_delay; /* wait for sync block io completion */
797 u64 swapin_delay; /* wait for swapin block io completion */
798 u32 blkio_count; /* total count of the number of sync block */
799 /* io operations performed */
800 u32 swapin_count; /* total count of the number of swapin block */
801 /* io operations performed */
803 struct timespec freepages_start, freepages_end;
804 u64 freepages_delay; /* wait for memory reclaim */
805 u32 freepages_count; /* total count of memory reclaim */
807 #endif /* CONFIG_TASK_DELAY_ACCT */
809 static inline int sched_info_on(void)
811 #ifdef CONFIG_SCHEDSTATS
813 #elif defined(CONFIG_TASK_DELAY_ACCT)
814 extern int delayacct_on;
829 * sched-domains (multiprocessor balancing) declarations:
833 * Increase resolution of nice-level calculations:
835 #define SCHED_LOAD_SHIFT 10
836 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
838 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
841 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
842 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
843 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
844 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
845 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
846 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
847 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
848 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
849 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
850 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
851 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
853 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
855 enum powersavings_balance_level {
856 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
857 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
858 * first for long running threads
860 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
861 * cpu package for power savings
863 MAX_POWERSAVINGS_BALANCE_LEVELS
866 extern int sched_mc_power_savings, sched_smt_power_savings;
868 static inline int sd_balance_for_mc_power(void)
870 if (sched_smt_power_savings)
871 return SD_POWERSAVINGS_BALANCE;
873 return SD_PREFER_SIBLING;
876 static inline int sd_balance_for_package_power(void)
878 if (sched_mc_power_savings | sched_smt_power_savings)
879 return SD_POWERSAVINGS_BALANCE;
881 return SD_PREFER_SIBLING;
885 * Optimise SD flags for power savings:
886 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
887 * Keep default SD flags if sched_{smt,mc}_power_saving=0
890 static inline int sd_power_saving_flags(void)
892 if (sched_mc_power_savings | sched_smt_power_savings)
893 return SD_BALANCE_NEWIDLE;
899 struct sched_group *next; /* Must be a circular list */
902 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
905 unsigned int cpu_power;
908 * The CPUs this group covers.
910 * NOTE: this field is variable length. (Allocated dynamically
911 * by attaching extra space to the end of the structure,
912 * depending on how many CPUs the kernel has booted up with)
914 * It is also be embedded into static data structures at build
915 * time. (See 'struct static_sched_group' in kernel/sched.c)
917 unsigned long cpumask[0];
920 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
922 return to_cpumask(sg->cpumask);
925 enum sched_domain_level {
935 struct sched_domain_attr {
936 int relax_domain_level;
939 #define SD_ATTR_INIT (struct sched_domain_attr) { \
940 .relax_domain_level = -1, \
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_* */
960 enum sched_domain_level level;
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
1005 * Span of all CPUs in this domain.
1007 * NOTE: this field is variable length. (Allocated dynamically
1008 * by attaching extra space to the end of the structure,
1009 * depending on how many CPUs the kernel has booted up with)
1011 * It is also be embedded into static data structures at build
1012 * time. (See 'struct static_sched_domain' in kernel/sched.c)
1014 unsigned long span[0];
1017 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1019 return to_cpumask(sd->span);
1022 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1023 struct sched_domain_attr *dattr_new);
1025 /* Allocate an array of sched domains, for partition_sched_domains(). */
1026 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1027 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1029 /* Test a flag in parent sched domain */
1030 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1032 if (sd->parent && (sd->parent->flags & flag))
1038 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1039 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1041 #else /* CONFIG_SMP */
1043 struct sched_domain_attr;
1046 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1047 struct sched_domain_attr *dattr_new)
1050 #endif /* !CONFIG_SMP */
1053 struct io_context; /* See blkdev.h */
1056 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1057 extern void prefetch_stack(struct task_struct *t);
1059 static inline void prefetch_stack(struct task_struct *t) { }
1062 struct audit_context; /* See audit.c */
1064 struct pipe_inode_info;
1065 struct uts_namespace;
1068 struct sched_domain;
1073 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1074 #define WF_FORK 0x02 /* child wakeup after fork */
1076 struct sched_class {
1077 const struct sched_class *next;
1079 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup,
1081 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1082 void (*yield_task) (struct rq *rq);
1084 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1086 struct task_struct * (*pick_next_task) (struct rq *rq);
1087 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1090 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1092 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1093 void (*post_schedule) (struct rq *this_rq);
1094 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1095 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1097 void (*set_cpus_allowed)(struct task_struct *p,
1098 const struct cpumask *newmask);
1100 void (*rq_online)(struct rq *rq);
1101 void (*rq_offline)(struct rq *rq);
1104 void (*set_curr_task) (struct rq *rq);
1105 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1106 void (*task_fork) (struct task_struct *p);
1108 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1110 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1112 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1113 int oldprio, int running);
1115 unsigned int (*get_rr_interval) (struct rq *rq,
1116 struct task_struct *task);
1118 #ifdef CONFIG_FAIR_GROUP_SCHED
1119 void (*moved_group) (struct task_struct *p, int on_rq);
1123 struct load_weight {
1124 unsigned long weight, inv_weight;
1128 * CFS stats for a schedulable entity (task, task-group etc)
1130 * Current field usage histogram:
1137 struct sched_entity {
1138 struct load_weight load; /* for load-balancing */
1139 struct rb_node run_node;
1140 struct list_head group_node;
1144 u64 sum_exec_runtime;
1146 u64 prev_sum_exec_runtime;
1156 #ifdef CONFIG_SCHEDSTATS
1166 s64 sum_sleep_runtime;
1173 u64 nr_migrations_cold;
1174 u64 nr_failed_migrations_affine;
1175 u64 nr_failed_migrations_running;
1176 u64 nr_failed_migrations_hot;
1177 u64 nr_forced_migrations;
1180 u64 nr_wakeups_sync;
1181 u64 nr_wakeups_migrate;
1182 u64 nr_wakeups_local;
1183 u64 nr_wakeups_remote;
1184 u64 nr_wakeups_affine;
1185 u64 nr_wakeups_affine_attempts;
1186 u64 nr_wakeups_passive;
1187 u64 nr_wakeups_idle;
1190 #ifdef CONFIG_FAIR_GROUP_SCHED
1191 struct sched_entity *parent;
1192 /* rq on which this entity is (to be) queued: */
1193 struct cfs_rq *cfs_rq;
1194 /* rq "owned" by this entity/group: */
1195 struct cfs_rq *my_q;
1199 struct sched_rt_entity {
1200 struct list_head run_list;
1201 unsigned long timeout;
1202 unsigned int time_slice;
1203 int nr_cpus_allowed;
1205 struct sched_rt_entity *back;
1206 #ifdef CONFIG_RT_GROUP_SCHED
1207 struct sched_rt_entity *parent;
1208 /* rq on which this entity is (to be) queued: */
1209 struct rt_rq *rt_rq;
1210 /* rq "owned" by this entity/group: */
1217 struct task_struct {
1218 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1221 unsigned int flags; /* per process flags, defined below */
1222 unsigned int ptrace;
1224 int lock_depth; /* BKL lock depth */
1227 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1232 int prio, static_prio, normal_prio;
1233 unsigned int rt_priority;
1234 const struct sched_class *sched_class;
1235 struct sched_entity se;
1236 struct sched_rt_entity rt;
1238 #ifdef CONFIG_PREEMPT_NOTIFIERS
1239 /* list of struct preempt_notifier: */
1240 struct hlist_head preempt_notifiers;
1244 * fpu_counter contains the number of consecutive context switches
1245 * that the FPU is used. If this is over a threshold, the lazy fpu
1246 * saving becomes unlazy to save the trap. This is an unsigned char
1247 * so that after 256 times the counter wraps and the behavior turns
1248 * lazy again; this to deal with bursty apps that only use FPU for
1251 unsigned char fpu_counter;
1252 #ifdef CONFIG_BLK_DEV_IO_TRACE
1253 unsigned int btrace_seq;
1256 unsigned int policy;
1257 cpumask_t cpus_allowed;
1259 #ifdef CONFIG_TREE_PREEMPT_RCU
1260 int rcu_read_lock_nesting;
1261 char rcu_read_unlock_special;
1262 struct rcu_node *rcu_blocked_node;
1263 struct list_head rcu_node_entry;
1264 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1266 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1267 struct sched_info sched_info;
1270 struct list_head tasks;
1271 struct plist_node pushable_tasks;
1273 struct mm_struct *mm, *active_mm;
1277 int exit_code, exit_signal;
1278 int pdeath_signal; /* The signal sent when the parent dies */
1280 unsigned int personality;
1281 unsigned did_exec:1;
1282 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1284 unsigned in_iowait:1;
1287 /* Revert to default priority/policy when forking */
1288 unsigned sched_reset_on_fork:1;
1293 #ifdef CONFIG_CC_STACKPROTECTOR
1294 /* Canary value for the -fstack-protector gcc feature */
1295 unsigned long stack_canary;
1299 * pointers to (original) parent process, youngest child, younger sibling,
1300 * older sibling, respectively. (p->father can be replaced with
1301 * p->real_parent->pid)
1303 struct task_struct *real_parent; /* real parent process */
1304 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1306 * children/sibling forms the list of my natural children
1308 struct list_head children; /* list of my children */
1309 struct list_head sibling; /* linkage in my parent's children list */
1310 struct task_struct *group_leader; /* threadgroup leader */
1313 * ptraced is the list of tasks this task is using ptrace on.
1314 * This includes both natural children and PTRACE_ATTACH targets.
1315 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1317 struct list_head ptraced;
1318 struct list_head ptrace_entry;
1321 * This is the tracer handle for the ptrace BTS extension.
1322 * This field actually belongs to the ptracer task.
1324 struct bts_context *bts;
1326 /* PID/PID hash table linkage. */
1327 struct pid_link pids[PIDTYPE_MAX];
1328 struct list_head thread_group;
1330 struct completion *vfork_done; /* for vfork() */
1331 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1332 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1334 cputime_t utime, stime, utimescaled, stimescaled;
1336 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1337 cputime_t prev_utime, prev_stime;
1339 unsigned long nvcsw, nivcsw; /* context switch counts */
1340 struct timespec start_time; /* monotonic time */
1341 struct timespec real_start_time; /* boot based time */
1342 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1343 unsigned long min_flt, maj_flt;
1345 struct task_cputime cputime_expires;
1346 struct list_head cpu_timers[3];
1348 /* process credentials */
1349 const struct cred *real_cred; /* objective and real subjective task
1350 * credentials (COW) */
1351 const struct cred *cred; /* effective (overridable) subjective task
1352 * credentials (COW) */
1353 struct mutex cred_guard_mutex; /* guard against foreign influences on
1354 * credential calculations
1355 * (notably. ptrace) */
1356 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1358 char comm[TASK_COMM_LEN]; /* executable name excluding path
1359 - access with [gs]et_task_comm (which lock
1360 it with task_lock())
1361 - initialized normally by setup_new_exec */
1362 /* file system info */
1363 int link_count, total_link_count;
1364 #ifdef CONFIG_SYSVIPC
1366 struct sysv_sem sysvsem;
1368 #ifdef CONFIG_DETECT_HUNG_TASK
1369 /* hung task detection */
1370 unsigned long last_switch_count;
1372 /* CPU-specific state of this task */
1373 struct thread_struct thread;
1374 /* filesystem information */
1375 struct fs_struct *fs;
1376 /* open file information */
1377 struct files_struct *files;
1379 struct nsproxy *nsproxy;
1380 /* signal handlers */
1381 struct signal_struct *signal;
1382 struct sighand_struct *sighand;
1384 sigset_t blocked, real_blocked;
1385 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1386 struct sigpending pending;
1388 unsigned long sas_ss_sp;
1390 int (*notifier)(void *priv);
1391 void *notifier_data;
1392 sigset_t *notifier_mask;
1393 struct audit_context *audit_context;
1394 #ifdef CONFIG_AUDITSYSCALL
1396 unsigned int sessionid;
1400 /* Thread group tracking */
1403 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1405 spinlock_t alloc_lock;
1407 #ifdef CONFIG_GENERIC_HARDIRQS
1408 /* IRQ handler threads */
1409 struct irqaction *irqaction;
1412 /* Protection of the PI data structures: */
1413 raw_spinlock_t pi_lock;
1415 #ifdef CONFIG_RT_MUTEXES
1416 /* PI waiters blocked on a rt_mutex held by this task */
1417 struct plist_head pi_waiters;
1418 /* Deadlock detection and priority inheritance handling */
1419 struct rt_mutex_waiter *pi_blocked_on;
1422 #ifdef CONFIG_DEBUG_MUTEXES
1423 /* mutex deadlock detection */
1424 struct mutex_waiter *blocked_on;
1426 #ifdef CONFIG_TRACE_IRQFLAGS
1427 unsigned int irq_events;
1428 unsigned long hardirq_enable_ip;
1429 unsigned long hardirq_disable_ip;
1430 unsigned int hardirq_enable_event;
1431 unsigned int hardirq_disable_event;
1432 int hardirqs_enabled;
1433 int hardirq_context;
1434 unsigned long softirq_disable_ip;
1435 unsigned long softirq_enable_ip;
1436 unsigned int softirq_disable_event;
1437 unsigned int softirq_enable_event;
1438 int softirqs_enabled;
1439 int softirq_context;
1441 #ifdef CONFIG_LOCKDEP
1442 # define MAX_LOCK_DEPTH 48UL
1445 unsigned int lockdep_recursion;
1446 struct held_lock held_locks[MAX_LOCK_DEPTH];
1447 gfp_t lockdep_reclaim_gfp;
1450 /* journalling filesystem info */
1453 /* stacked block device info */
1454 struct bio *bio_list, **bio_tail;
1457 struct reclaim_state *reclaim_state;
1459 struct backing_dev_info *backing_dev_info;
1461 struct io_context *io_context;
1463 unsigned long ptrace_message;
1464 siginfo_t *last_siginfo; /* For ptrace use. */
1465 struct task_io_accounting ioac;
1466 #if defined(CONFIG_TASK_XACCT)
1467 u64 acct_rss_mem1; /* accumulated rss usage */
1468 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1469 cputime_t acct_timexpd; /* stime + utime since last update */
1471 #ifdef CONFIG_CPUSETS
1472 nodemask_t mems_allowed; /* Protected by alloc_lock */
1473 int cpuset_mem_spread_rotor;
1475 #ifdef CONFIG_CGROUPS
1476 /* Control Group info protected by css_set_lock */
1477 struct css_set *cgroups;
1478 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1479 struct list_head cg_list;
1482 struct robust_list_head __user *robust_list;
1483 #ifdef CONFIG_COMPAT
1484 struct compat_robust_list_head __user *compat_robust_list;
1486 struct list_head pi_state_list;
1487 struct futex_pi_state *pi_state_cache;
1489 #ifdef CONFIG_PERF_EVENTS
1490 struct perf_event_context *perf_event_ctxp;
1491 struct mutex perf_event_mutex;
1492 struct list_head perf_event_list;
1495 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1498 atomic_t fs_excl; /* holding fs exclusive resources */
1499 struct rcu_head rcu;
1502 * cache last used pipe for splice
1504 struct pipe_inode_info *splice_pipe;
1505 #ifdef CONFIG_TASK_DELAY_ACCT
1506 struct task_delay_info *delays;
1508 #ifdef CONFIG_FAULT_INJECTION
1511 struct prop_local_single dirties;
1512 #ifdef CONFIG_LATENCYTOP
1513 int latency_record_count;
1514 struct latency_record latency_record[LT_SAVECOUNT];
1517 * time slack values; these are used to round up poll() and
1518 * select() etc timeout values. These are in nanoseconds.
1520 unsigned long timer_slack_ns;
1521 unsigned long default_timer_slack_ns;
1523 struct list_head *scm_work_list;
1524 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1525 /* Index of current stored adress in ret_stack */
1527 /* Stack of return addresses for return function tracing */
1528 struct ftrace_ret_stack *ret_stack;
1529 /* time stamp for last schedule */
1530 unsigned long long ftrace_timestamp;
1532 * Number of functions that haven't been traced
1533 * because of depth overrun.
1535 atomic_t trace_overrun;
1536 /* Pause for the tracing */
1537 atomic_t tracing_graph_pause;
1539 #ifdef CONFIG_TRACING
1540 /* state flags for use by tracers */
1541 unsigned long trace;
1542 /* bitmask of trace recursion */
1543 unsigned long trace_recursion;
1544 #endif /* CONFIG_TRACING */
1545 unsigned long stack_start;
1546 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1547 struct memcg_batch_info {
1548 int do_batch; /* incremented when batch uncharge started */
1549 struct mem_cgroup *memcg; /* target memcg of uncharge */
1550 unsigned long bytes; /* uncharged usage */
1551 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1556 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1557 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1560 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1561 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1562 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1563 * values are inverted: lower p->prio value means higher priority.
1565 * The MAX_USER_RT_PRIO value allows the actual maximum
1566 * RT priority to be separate from the value exported to
1567 * user-space. This allows kernel threads to set their
1568 * priority to a value higher than any user task. Note:
1569 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1572 #define MAX_USER_RT_PRIO 100
1573 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1575 #define MAX_PRIO (MAX_RT_PRIO + 40)
1576 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1578 static inline int rt_prio(int prio)
1580 if (unlikely(prio < MAX_RT_PRIO))
1585 static inline int rt_task(struct task_struct *p)
1587 return rt_prio(p->prio);
1590 static inline struct pid *task_pid(struct task_struct *task)
1592 return task->pids[PIDTYPE_PID].pid;
1595 static inline struct pid *task_tgid(struct task_struct *task)
1597 return task->group_leader->pids[PIDTYPE_PID].pid;
1601 * Without tasklist or rcu lock it is not safe to dereference
1602 * the result of task_pgrp/task_session even if task == current,
1603 * we can race with another thread doing sys_setsid/sys_setpgid.
1605 static inline struct pid *task_pgrp(struct task_struct *task)
1607 return task->group_leader->pids[PIDTYPE_PGID].pid;
1610 static inline struct pid *task_session(struct task_struct *task)
1612 return task->group_leader->pids[PIDTYPE_SID].pid;
1615 struct pid_namespace;
1618 * the helpers to get the task's different pids as they are seen
1619 * from various namespaces
1621 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1622 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1624 * task_xid_nr_ns() : id seen from the ns specified;
1626 * set_task_vxid() : assigns a virtual id to a task;
1628 * see also pid_nr() etc in include/linux/pid.h
1630 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1631 struct pid_namespace *ns);
1633 static inline pid_t task_pid_nr(struct task_struct *tsk)
1638 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1639 struct pid_namespace *ns)
1641 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1644 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1646 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1650 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1655 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1657 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1659 return pid_vnr(task_tgid(tsk));
1663 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1664 struct pid_namespace *ns)
1666 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1669 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1671 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1675 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1676 struct pid_namespace *ns)
1678 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1681 static inline pid_t task_session_vnr(struct task_struct *tsk)
1683 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1686 /* obsolete, do not use */
1687 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1689 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1693 * pid_alive - check that a task structure is not stale
1694 * @p: Task structure to be checked.
1696 * Test if a process is not yet dead (at most zombie state)
1697 * If pid_alive fails, then pointers within the task structure
1698 * can be stale and must not be dereferenced.
1700 static inline int pid_alive(struct task_struct *p)
1702 return p->pids[PIDTYPE_PID].pid != NULL;
1706 * is_global_init - check if a task structure is init
1707 * @tsk: Task structure to be checked.
1709 * Check if a task structure is the first user space task the kernel created.
1711 static inline int is_global_init(struct task_struct *tsk)
1713 return tsk->pid == 1;
1717 * is_container_init:
1718 * check whether in the task is init in its own pid namespace.
1720 extern int is_container_init(struct task_struct *tsk);
1722 extern struct pid *cad_pid;
1724 extern void free_task(struct task_struct *tsk);
1725 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1727 extern void __put_task_struct(struct task_struct *t);
1729 static inline void put_task_struct(struct task_struct *t)
1731 if (atomic_dec_and_test(&t->usage))
1732 __put_task_struct(t);
1735 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1736 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1741 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1742 /* Not implemented yet, only for 486*/
1743 #define PF_STARTING 0x00000002 /* being created */
1744 #define PF_EXITING 0x00000004 /* getting shut down */
1745 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1746 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1747 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1748 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1749 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1750 #define PF_DUMPCORE 0x00000200 /* dumped core */
1751 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1752 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1753 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1754 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1755 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1756 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1757 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1758 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1759 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1760 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1761 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1762 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1763 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1764 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1765 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1766 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1767 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1768 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1769 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1770 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1771 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1772 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1775 * Only the _current_ task can read/write to tsk->flags, but other
1776 * tasks can access tsk->flags in readonly mode for example
1777 * with tsk_used_math (like during threaded core dumping).
1778 * There is however an exception to this rule during ptrace
1779 * or during fork: the ptracer task is allowed to write to the
1780 * child->flags of its traced child (same goes for fork, the parent
1781 * can write to the child->flags), because we're guaranteed the
1782 * child is not running and in turn not changing child->flags
1783 * at the same time the parent does it.
1785 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1786 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1787 #define clear_used_math() clear_stopped_child_used_math(current)
1788 #define set_used_math() set_stopped_child_used_math(current)
1789 #define conditional_stopped_child_used_math(condition, child) \
1790 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1791 #define conditional_used_math(condition) \
1792 conditional_stopped_child_used_math(condition, current)
1793 #define copy_to_stopped_child_used_math(child) \
1794 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1795 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1796 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1797 #define used_math() tsk_used_math(current)
1799 #ifdef CONFIG_TREE_PREEMPT_RCU
1801 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1802 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1804 static inline void rcu_copy_process(struct task_struct *p)
1806 p->rcu_read_lock_nesting = 0;
1807 p->rcu_read_unlock_special = 0;
1808 p->rcu_blocked_node = NULL;
1809 INIT_LIST_HEAD(&p->rcu_node_entry);
1814 static inline void rcu_copy_process(struct task_struct *p)
1821 extern int set_cpus_allowed_ptr(struct task_struct *p,
1822 const struct cpumask *new_mask);
1824 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1825 const struct cpumask *new_mask)
1827 if (!cpumask_test_cpu(0, new_mask))
1833 #ifndef CONFIG_CPUMASK_OFFSTACK
1834 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1836 return set_cpus_allowed_ptr(p, &new_mask);
1841 * Architectures can set this to 1 if they have specified
1842 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1843 * but then during bootup it turns out that sched_clock()
1844 * is reliable after all:
1846 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1847 extern int sched_clock_stable;
1850 /* ftrace calls sched_clock() directly */
1851 extern unsigned long long notrace sched_clock(void);
1853 extern void sched_clock_init(void);
1854 extern u64 sched_clock_cpu(int cpu);
1856 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1857 static inline void sched_clock_tick(void)
1861 static inline void sched_clock_idle_sleep_event(void)
1865 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1869 extern void sched_clock_tick(void);
1870 extern void sched_clock_idle_sleep_event(void);
1871 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1875 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1876 * clock constructed from sched_clock():
1878 extern unsigned long long cpu_clock(int cpu);
1880 extern unsigned long long
1881 task_sched_runtime(struct task_struct *task);
1882 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1884 /* sched_exec is called by processes performing an exec */
1886 extern void sched_exec(void);
1888 #define sched_exec() {}
1891 extern void sched_clock_idle_sleep_event(void);
1892 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1894 #ifdef CONFIG_HOTPLUG_CPU
1895 extern void idle_task_exit(void);
1897 static inline void idle_task_exit(void) {}
1900 extern void sched_idle_next(void);
1902 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1903 extern void wake_up_idle_cpu(int cpu);
1905 static inline void wake_up_idle_cpu(int cpu) { }
1908 extern unsigned int sysctl_sched_latency;
1909 extern unsigned int sysctl_sched_min_granularity;
1910 extern unsigned int sysctl_sched_wakeup_granularity;
1911 extern unsigned int sysctl_sched_shares_ratelimit;
1912 extern unsigned int sysctl_sched_shares_thresh;
1913 extern unsigned int sysctl_sched_child_runs_first;
1915 enum sched_tunable_scaling {
1916 SCHED_TUNABLESCALING_NONE,
1917 SCHED_TUNABLESCALING_LOG,
1918 SCHED_TUNABLESCALING_LINEAR,
1919 SCHED_TUNABLESCALING_END,
1921 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1923 #ifdef CONFIG_SCHED_DEBUG
1924 extern unsigned int sysctl_sched_migration_cost;
1925 extern unsigned int sysctl_sched_nr_migrate;
1926 extern unsigned int sysctl_sched_time_avg;
1927 extern unsigned int sysctl_timer_migration;
1929 int sched_proc_update_handler(struct ctl_table *table, int write,
1930 void __user *buffer, size_t *length,
1933 #ifdef CONFIG_SCHED_DEBUG
1934 static inline unsigned int get_sysctl_timer_migration(void)
1936 return sysctl_timer_migration;
1939 static inline unsigned int get_sysctl_timer_migration(void)
1944 extern unsigned int sysctl_sched_rt_period;
1945 extern int sysctl_sched_rt_runtime;
1947 int sched_rt_handler(struct ctl_table *table, int write,
1948 void __user *buffer, size_t *lenp,
1951 extern unsigned int sysctl_sched_compat_yield;
1953 #ifdef CONFIG_RT_MUTEXES
1954 extern int rt_mutex_getprio(struct task_struct *p);
1955 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1956 extern void rt_mutex_adjust_pi(struct task_struct *p);
1958 static inline int rt_mutex_getprio(struct task_struct *p)
1960 return p->normal_prio;
1962 # define rt_mutex_adjust_pi(p) do { } while (0)
1965 extern void set_user_nice(struct task_struct *p, long nice);
1966 extern int task_prio(const struct task_struct *p);
1967 extern int task_nice(const struct task_struct *p);
1968 extern int can_nice(const struct task_struct *p, const int nice);
1969 extern int task_curr(const struct task_struct *p);
1970 extern int idle_cpu(int cpu);
1971 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1972 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1973 struct sched_param *);
1974 extern struct task_struct *idle_task(int cpu);
1975 extern struct task_struct *curr_task(int cpu);
1976 extern void set_curr_task(int cpu, struct task_struct *p);
1981 * The default (Linux) execution domain.
1983 extern struct exec_domain default_exec_domain;
1985 union thread_union {
1986 struct thread_info thread_info;
1987 unsigned long stack[THREAD_SIZE/sizeof(long)];
1990 #ifndef __HAVE_ARCH_KSTACK_END
1991 static inline int kstack_end(void *addr)
1993 /* Reliable end of stack detection:
1994 * Some APM bios versions misalign the stack
1996 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2000 extern union thread_union init_thread_union;
2001 extern struct task_struct init_task;
2003 extern struct mm_struct init_mm;
2005 extern struct pid_namespace init_pid_ns;
2008 * find a task by one of its numerical ids
2010 * find_task_by_pid_ns():
2011 * finds a task by its pid in the specified namespace
2012 * find_task_by_vpid():
2013 * finds a task by its virtual pid
2015 * see also find_vpid() etc in include/linux/pid.h
2018 extern struct task_struct *find_task_by_vpid(pid_t nr);
2019 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2020 struct pid_namespace *ns);
2022 extern void __set_special_pids(struct pid *pid);
2024 /* per-UID process charging. */
2025 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2026 static inline struct user_struct *get_uid(struct user_struct *u)
2028 atomic_inc(&u->__count);
2031 extern void free_uid(struct user_struct *);
2032 extern void release_uids(struct user_namespace *ns);
2034 #include <asm/current.h>
2036 extern void do_timer(unsigned long ticks);
2038 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2039 extern int wake_up_process(struct task_struct *tsk);
2040 extern void wake_up_new_task(struct task_struct *tsk,
2041 unsigned long clone_flags);
2043 extern void kick_process(struct task_struct *tsk);
2045 static inline void kick_process(struct task_struct *tsk) { }
2047 extern void sched_fork(struct task_struct *p, int clone_flags);
2048 extern void sched_dead(struct task_struct *p);
2050 extern void proc_caches_init(void);
2051 extern void flush_signals(struct task_struct *);
2052 extern void __flush_signals(struct task_struct *);
2053 extern void ignore_signals(struct task_struct *);
2054 extern void flush_signal_handlers(struct task_struct *, int force_default);
2055 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2057 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2059 unsigned long flags;
2062 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2063 ret = dequeue_signal(tsk, mask, info);
2064 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2069 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2071 extern void unblock_all_signals(void);
2072 extern void release_task(struct task_struct * p);
2073 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2074 extern int force_sigsegv(int, struct task_struct *);
2075 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2076 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2077 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2078 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2079 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2080 extern int kill_pid(struct pid *pid, int sig, int priv);
2081 extern int kill_proc_info(int, struct siginfo *, pid_t);
2082 extern int do_notify_parent(struct task_struct *, int);
2083 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2084 extern void force_sig(int, struct task_struct *);
2085 extern int send_sig(int, struct task_struct *, int);
2086 extern void zap_other_threads(struct task_struct *p);
2087 extern struct sigqueue *sigqueue_alloc(void);
2088 extern void sigqueue_free(struct sigqueue *);
2089 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2090 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2091 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2093 static inline int kill_cad_pid(int sig, int priv)
2095 return kill_pid(cad_pid, sig, priv);
2098 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2099 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2100 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2101 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2104 * True if we are on the alternate signal stack.
2106 static inline int on_sig_stack(unsigned long sp)
2108 #ifdef CONFIG_STACK_GROWSUP
2109 return sp >= current->sas_ss_sp &&
2110 sp - current->sas_ss_sp < current->sas_ss_size;
2112 return sp > current->sas_ss_sp &&
2113 sp - current->sas_ss_sp <= current->sas_ss_size;
2117 static inline int sas_ss_flags(unsigned long sp)
2119 return (current->sas_ss_size == 0 ? SS_DISABLE
2120 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2124 * Routines for handling mm_structs
2126 extern struct mm_struct * mm_alloc(void);
2128 /* mmdrop drops the mm and the page tables */
2129 extern void __mmdrop(struct mm_struct *);
2130 static inline void mmdrop(struct mm_struct * mm)
2132 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2136 /* mmput gets rid of the mappings and all user-space */
2137 extern void mmput(struct mm_struct *);
2138 /* Grab a reference to a task's mm, if it is not already going away */
2139 extern struct mm_struct *get_task_mm(struct task_struct *task);
2140 /* Remove the current tasks stale references to the old mm_struct */
2141 extern void mm_release(struct task_struct *, struct mm_struct *);
2142 /* Allocate a new mm structure and copy contents from tsk->mm */
2143 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2145 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2146 struct task_struct *, struct pt_regs *);
2147 extern void flush_thread(void);
2148 extern void exit_thread(void);
2150 extern void exit_files(struct task_struct *);
2151 extern void __cleanup_signal(struct signal_struct *);
2152 extern void __cleanup_sighand(struct sighand_struct *);
2154 extern void exit_itimers(struct signal_struct *);
2155 extern void flush_itimer_signals(void);
2157 extern NORET_TYPE void do_group_exit(int);
2159 extern void daemonize(const char *, ...);
2160 extern int allow_signal(int);
2161 extern int disallow_signal(int);
2163 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2164 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2165 struct task_struct *fork_idle(int);
2167 extern void set_task_comm(struct task_struct *tsk, char *from);
2168 extern char *get_task_comm(char *to, struct task_struct *tsk);
2171 extern void wait_task_context_switch(struct task_struct *p);
2172 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2174 static inline void wait_task_context_switch(struct task_struct *p) {}
2175 static inline unsigned long wait_task_inactive(struct task_struct *p,
2182 #define next_task(p) \
2183 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2185 #define for_each_process(p) \
2186 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2188 extern bool current_is_single_threaded(void);
2191 * Careful: do_each_thread/while_each_thread is a double loop so
2192 * 'break' will not work as expected - use goto instead.
2194 #define do_each_thread(g, t) \
2195 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2197 #define while_each_thread(g, t) \
2198 while ((t = next_thread(t)) != g)
2200 /* de_thread depends on thread_group_leader not being a pid based check */
2201 #define thread_group_leader(p) (p == p->group_leader)
2203 /* Do to the insanities of de_thread it is possible for a process
2204 * to have the pid of the thread group leader without actually being
2205 * the thread group leader. For iteration through the pids in proc
2206 * all we care about is that we have a task with the appropriate
2207 * pid, we don't actually care if we have the right task.
2209 static inline int has_group_leader_pid(struct task_struct *p)
2211 return p->pid == p->tgid;
2215 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2217 return p1->tgid == p2->tgid;
2220 static inline struct task_struct *next_thread(const struct task_struct *p)
2222 return list_entry_rcu(p->thread_group.next,
2223 struct task_struct, thread_group);
2226 static inline int thread_group_empty(struct task_struct *p)
2228 return list_empty(&p->thread_group);
2231 #define delay_group_leader(p) \
2232 (thread_group_leader(p) && !thread_group_empty(p))
2234 static inline int task_detached(struct task_struct *p)
2236 return p->exit_signal == -1;
2240 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2241 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2242 * pins the final release of task.io_context. Also protects ->cpuset and
2243 * ->cgroup.subsys[].
2245 * Nests both inside and outside of read_lock(&tasklist_lock).
2246 * It must not be nested with write_lock_irq(&tasklist_lock),
2247 * neither inside nor outside.
2249 static inline void task_lock(struct task_struct *p)
2251 spin_lock(&p->alloc_lock);
2254 static inline void task_unlock(struct task_struct *p)
2256 spin_unlock(&p->alloc_lock);
2259 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2260 unsigned long *flags);
2262 static inline void unlock_task_sighand(struct task_struct *tsk,
2263 unsigned long *flags)
2265 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2268 #ifndef __HAVE_THREAD_FUNCTIONS
2270 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2271 #define task_stack_page(task) ((task)->stack)
2273 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2275 *task_thread_info(p) = *task_thread_info(org);
2276 task_thread_info(p)->task = p;
2279 static inline unsigned long *end_of_stack(struct task_struct *p)
2281 return (unsigned long *)(task_thread_info(p) + 1);
2286 static inline int object_is_on_stack(void *obj)
2288 void *stack = task_stack_page(current);
2290 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2293 extern void thread_info_cache_init(void);
2295 #ifdef CONFIG_DEBUG_STACK_USAGE
2296 static inline unsigned long stack_not_used(struct task_struct *p)
2298 unsigned long *n = end_of_stack(p);
2300 do { /* Skip over canary */
2304 return (unsigned long)n - (unsigned long)end_of_stack(p);
2308 /* set thread flags in other task's structures
2309 * - see asm/thread_info.h for TIF_xxxx flags available
2311 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2313 set_ti_thread_flag(task_thread_info(tsk), flag);
2316 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2318 clear_ti_thread_flag(task_thread_info(tsk), flag);
2321 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2323 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2326 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2328 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2331 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2333 return test_ti_thread_flag(task_thread_info(tsk), flag);
2336 static inline void set_tsk_need_resched(struct task_struct *tsk)
2338 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2341 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2343 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2346 static inline int test_tsk_need_resched(struct task_struct *tsk)
2348 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2351 static inline int restart_syscall(void)
2353 set_tsk_thread_flag(current, TIF_SIGPENDING);
2354 return -ERESTARTNOINTR;
2357 static inline int signal_pending(struct task_struct *p)
2359 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2362 static inline int __fatal_signal_pending(struct task_struct *p)
2364 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2367 static inline int fatal_signal_pending(struct task_struct *p)
2369 return signal_pending(p) && __fatal_signal_pending(p);
2372 static inline int signal_pending_state(long state, struct task_struct *p)
2374 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2376 if (!signal_pending(p))
2379 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2382 static inline int need_resched(void)
2384 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2388 * cond_resched() and cond_resched_lock(): latency reduction via
2389 * explicit rescheduling in places that are safe. The return
2390 * value indicates whether a reschedule was done in fact.
2391 * cond_resched_lock() will drop the spinlock before scheduling,
2392 * cond_resched_softirq() will enable bhs before scheduling.
2394 extern int _cond_resched(void);
2396 #define cond_resched() ({ \
2397 __might_sleep(__FILE__, __LINE__, 0); \
2401 extern int __cond_resched_lock(spinlock_t *lock);
2403 #ifdef CONFIG_PREEMPT
2404 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2406 #define PREEMPT_LOCK_OFFSET 0
2409 #define cond_resched_lock(lock) ({ \
2410 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2411 __cond_resched_lock(lock); \
2414 extern int __cond_resched_softirq(void);
2416 #define cond_resched_softirq() ({ \
2417 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2418 __cond_resched_softirq(); \
2422 * Does a critical section need to be broken due to another
2423 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2424 * but a general need for low latency)
2426 static inline int spin_needbreak(spinlock_t *lock)
2428 #ifdef CONFIG_PREEMPT
2429 return spin_is_contended(lock);
2436 * Thread group CPU time accounting.
2438 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2439 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2441 static inline void thread_group_cputime_init(struct signal_struct *sig)
2443 sig->cputimer.cputime = INIT_CPUTIME;
2444 spin_lock_init(&sig->cputimer.lock);
2445 sig->cputimer.running = 0;
2448 static inline void thread_group_cputime_free(struct signal_struct *sig)
2453 * Reevaluate whether the task has signals pending delivery.
2454 * Wake the task if so.
2455 * This is required every time the blocked sigset_t changes.
2456 * callers must hold sighand->siglock.
2458 extern void recalc_sigpending_and_wake(struct task_struct *t);
2459 extern void recalc_sigpending(void);
2461 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2464 * Wrappers for p->thread_info->cpu access. No-op on UP.
2468 static inline unsigned int task_cpu(const struct task_struct *p)
2470 return task_thread_info(p)->cpu;
2473 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2477 static inline unsigned int task_cpu(const struct task_struct *p)
2482 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2486 #endif /* CONFIG_SMP */
2488 #ifdef CONFIG_TRACING
2490 __trace_special(void *__tr, void *__data,
2491 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2494 __trace_special(void *__tr, void *__data,
2495 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2500 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2501 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2503 extern void normalize_rt_tasks(void);
2505 #ifdef CONFIG_CGROUP_SCHED
2507 extern struct task_group init_task_group;
2509 extern struct task_group *sched_create_group(struct task_group *parent);
2510 extern void sched_destroy_group(struct task_group *tg);
2511 extern void sched_move_task(struct task_struct *tsk);
2512 #ifdef CONFIG_FAIR_GROUP_SCHED
2513 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2514 extern unsigned long sched_group_shares(struct task_group *tg);
2516 #ifdef CONFIG_RT_GROUP_SCHED
2517 extern int sched_group_set_rt_runtime(struct task_group *tg,
2518 long rt_runtime_us);
2519 extern long sched_group_rt_runtime(struct task_group *tg);
2520 extern int sched_group_set_rt_period(struct task_group *tg,
2522 extern long sched_group_rt_period(struct task_group *tg);
2523 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2527 extern int task_can_switch_user(struct user_struct *up,
2528 struct task_struct *tsk);
2530 #ifdef CONFIG_TASK_XACCT
2531 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2533 tsk->ioac.rchar += amt;
2536 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2538 tsk->ioac.wchar += amt;
2541 static inline void inc_syscr(struct task_struct *tsk)
2546 static inline void inc_syscw(struct task_struct *tsk)
2551 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2555 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2559 static inline void inc_syscr(struct task_struct *tsk)
2563 static inline void inc_syscw(struct task_struct *tsk)
2568 #ifndef TASK_SIZE_OF
2569 #define TASK_SIZE_OF(tsk) TASK_SIZE
2573 * Call the function if the target task is executing on a CPU right now:
2575 extern void task_oncpu_function_call(struct task_struct *p,
2576 void (*func) (void *info), void *info);
2579 #ifdef CONFIG_MM_OWNER
2580 extern void mm_update_next_owner(struct mm_struct *mm);
2581 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2583 static inline void mm_update_next_owner(struct mm_struct *mm)
2587 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2590 #endif /* CONFIG_MM_OWNER */
2592 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2595 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2598 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2601 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2604 static inline unsigned long rlimit(unsigned int limit)
2606 return task_rlimit(current, limit);
2609 static inline unsigned long rlimit_max(unsigned int limit)
2611 return task_rlimit_max(current, limit);
2614 #endif /* __KERNEL__ */