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);
148 extern u64 cpu_nr_migrations(int cpu);
150 extern unsigned long get_parent_ip(unsigned long addr);
155 #ifdef CONFIG_SCHED_DEBUG
156 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
157 extern void proc_sched_set_task(struct task_struct *p);
159 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
162 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
165 static inline void proc_sched_set_task(struct task_struct *p)
169 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
174 extern unsigned long long time_sync_thresh;
177 * Task state bitmask. NOTE! These bits are also
178 * encoded in fs/proc/array.c: get_task_state().
180 * We have two separate sets of flags: task->state
181 * is about runnability, while task->exit_state are
182 * about the task exiting. Confusing, but this way
183 * modifying one set can't modify the other one by
186 #define TASK_RUNNING 0
187 #define TASK_INTERRUPTIBLE 1
188 #define TASK_UNINTERRUPTIBLE 2
189 #define __TASK_STOPPED 4
190 #define __TASK_TRACED 8
191 /* in tsk->exit_state */
192 #define EXIT_ZOMBIE 16
194 /* in tsk->state again */
196 #define TASK_WAKEKILL 128
197 #define TASK_WAKING 256
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_stopped_or_traced(task) \
216 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
217 #define task_contributes_to_load(task) \
218 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
219 (task->flags & PF_FREEZING) == 0)
221 #define __set_task_state(tsk, state_value) \
222 do { (tsk)->state = (state_value); } while (0)
223 #define set_task_state(tsk, state_value) \
224 set_mb((tsk)->state, (state_value))
227 * set_current_state() includes a barrier so that the write of current->state
228 * is correctly serialised wrt the caller's subsequent test of whether to
231 * set_current_state(TASK_UNINTERRUPTIBLE);
232 * if (do_i_need_to_sleep())
235 * If the caller does not need such serialisation then use __set_current_state()
237 #define __set_current_state(state_value) \
238 do { current->state = (state_value); } while (0)
239 #define set_current_state(state_value) \
240 set_mb(current->state, (state_value))
242 /* Task command name length */
243 #define TASK_COMM_LEN 16
245 #include <linux/spinlock.h>
248 * This serializes "schedule()" and also protects
249 * the run-queue from deletions/modifications (but
250 * _adding_ to the beginning of the run-queue has
253 extern rwlock_t tasklist_lock;
254 extern spinlock_t mmlist_lock;
258 extern void sched_init(void);
259 extern void sched_init_smp(void);
260 extern asmlinkage void schedule_tail(struct task_struct *prev);
261 extern void init_idle(struct task_struct *idle, int cpu);
262 extern void init_idle_bootup_task(struct task_struct *idle);
264 extern int runqueue_is_locked(int cpu);
265 extern void task_rq_unlock_wait(struct task_struct *p);
267 extern cpumask_var_t nohz_cpu_mask;
268 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
269 extern int select_nohz_load_balancer(int cpu);
270 extern int get_nohz_load_balancer(void);
272 static inline int select_nohz_load_balancer(int cpu)
279 * Only dump TASK_* tasks. (0 for all tasks)
281 extern void show_state_filter(unsigned long state_filter);
283 static inline void show_state(void)
285 show_state_filter(0);
288 extern void show_regs(struct pt_regs *);
291 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
292 * task), SP is the stack pointer of the first frame that should be shown in the back
293 * trace (or NULL if the entire call-chain of the task should be shown).
295 extern void show_stack(struct task_struct *task, unsigned long *sp);
297 void io_schedule(void);
298 long io_schedule_timeout(long timeout);
300 extern void cpu_init (void);
301 extern void trap_init(void);
302 extern void update_process_times(int user);
303 extern void scheduler_tick(void);
305 extern void sched_show_task(struct task_struct *p);
307 #ifdef CONFIG_DETECT_SOFTLOCKUP
308 extern void softlockup_tick(void);
309 extern void touch_softlockup_watchdog(void);
310 extern void touch_all_softlockup_watchdogs(void);
311 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
312 struct file *filp, void __user *buffer,
313 size_t *lenp, loff_t *ppos);
314 extern unsigned int softlockup_panic;
315 extern int softlockup_thresh;
317 static inline void softlockup_tick(void)
320 static inline void touch_softlockup_watchdog(void)
323 static inline void touch_all_softlockup_watchdogs(void)
328 #ifdef CONFIG_DETECT_HUNG_TASK
329 extern unsigned int sysctl_hung_task_panic;
330 extern unsigned long sysctl_hung_task_check_count;
331 extern unsigned long sysctl_hung_task_timeout_secs;
332 extern unsigned long sysctl_hung_task_warnings;
333 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
334 struct file *filp, void __user *buffer,
335 size_t *lenp, loff_t *ppos);
338 /* Attach to any functions which should be ignored in wchan output. */
339 #define __sched __attribute__((__section__(".sched.text")))
341 /* Linker adds these: start and end of __sched functions */
342 extern char __sched_text_start[], __sched_text_end[];
344 /* Is this address in the __sched functions? */
345 extern int in_sched_functions(unsigned long addr);
347 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
348 extern signed long schedule_timeout(signed long timeout);
349 extern signed long schedule_timeout_interruptible(signed long timeout);
350 extern signed long schedule_timeout_killable(signed long timeout);
351 extern signed long schedule_timeout_uninterruptible(signed long timeout);
352 asmlinkage void __schedule(void);
353 asmlinkage void schedule(void);
354 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
357 struct user_namespace;
360 * Default maximum number of active map areas, this limits the number of vmas
361 * per mm struct. Users can overwrite this number by sysctl but there is a
364 * When a program's coredump is generated as ELF format, a section is created
365 * per a vma. In ELF, the number of sections is represented in unsigned short.
366 * This means the number of sections should be smaller than 65535 at coredump.
367 * Because the kernel adds some informative sections to a image of program at
368 * generating coredump, we need some margin. The number of extra sections is
369 * 1-3 now and depends on arch. We use "5" as safe margin, here.
371 #define MAPCOUNT_ELF_CORE_MARGIN (5)
372 #define DEFAULT_MAX_MAP_COUNT (USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
374 extern int sysctl_max_map_count;
376 #include <linux/aio.h>
379 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
380 unsigned long, unsigned long);
382 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
383 unsigned long len, unsigned long pgoff,
384 unsigned long flags);
385 extern void arch_unmap_area(struct mm_struct *, unsigned long);
386 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
388 #if USE_SPLIT_PTLOCKS
390 * The mm counters are not protected by its page_table_lock,
391 * so must be incremented atomically.
393 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
394 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
395 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
396 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
397 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
399 #else /* !USE_SPLIT_PTLOCKS */
401 * The mm counters are protected by its page_table_lock,
402 * so can be incremented directly.
404 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
405 #define get_mm_counter(mm, member) ((mm)->_##member)
406 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
407 #define inc_mm_counter(mm, member) (mm)->_##member++
408 #define dec_mm_counter(mm, member) (mm)->_##member--
410 #endif /* !USE_SPLIT_PTLOCKS */
412 #define get_mm_rss(mm) \
413 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
414 #define update_hiwater_rss(mm) do { \
415 unsigned long _rss = get_mm_rss(mm); \
416 if ((mm)->hiwater_rss < _rss) \
417 (mm)->hiwater_rss = _rss; \
419 #define update_hiwater_vm(mm) do { \
420 if ((mm)->hiwater_vm < (mm)->total_vm) \
421 (mm)->hiwater_vm = (mm)->total_vm; \
424 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
426 return max(mm->hiwater_rss, get_mm_rss(mm));
429 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
430 struct mm_struct *mm)
432 unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
434 if (*maxrss < hiwater_rss)
435 *maxrss = hiwater_rss;
438 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
440 return max(mm->hiwater_vm, mm->total_vm);
443 extern void set_dumpable(struct mm_struct *mm, int value);
444 extern int get_dumpable(struct mm_struct *mm);
448 #define MMF_DUMPABLE 0 /* core dump is permitted */
449 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
451 #define MMF_DUMPABLE_BITS 2
452 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
454 /* coredump filter bits */
455 #define MMF_DUMP_ANON_PRIVATE 2
456 #define MMF_DUMP_ANON_SHARED 3
457 #define MMF_DUMP_MAPPED_PRIVATE 4
458 #define MMF_DUMP_MAPPED_SHARED 5
459 #define MMF_DUMP_ELF_HEADERS 6
460 #define MMF_DUMP_HUGETLB_PRIVATE 7
461 #define MMF_DUMP_HUGETLB_SHARED 8
463 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
464 #define MMF_DUMP_FILTER_BITS 7
465 #define MMF_DUMP_FILTER_MASK \
466 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
467 #define MMF_DUMP_FILTER_DEFAULT \
468 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
469 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
471 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
472 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
474 # define MMF_DUMP_MASK_DEFAULT_ELF 0
476 /* leave room for more dump flags */
477 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
479 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
481 struct sighand_struct {
483 struct k_sigaction action[_NSIG];
485 wait_queue_head_t signalfd_wqh;
488 struct pacct_struct {
491 unsigned long ac_mem;
492 cputime_t ac_utime, ac_stime;
493 unsigned long ac_minflt, ac_majflt;
497 * struct task_cputime - collected CPU time counts
498 * @utime: time spent in user mode, in &cputime_t units
499 * @stime: time spent in kernel mode, in &cputime_t units
500 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
502 * This structure groups together three kinds of CPU time that are
503 * tracked for threads and thread groups. Most things considering
504 * CPU time want to group these counts together and treat all three
505 * of them in parallel.
507 struct task_cputime {
510 unsigned long long sum_exec_runtime;
512 /* Alternate field names when used to cache expirations. */
513 #define prof_exp stime
514 #define virt_exp utime
515 #define sched_exp sum_exec_runtime
517 #define INIT_CPUTIME \
518 (struct task_cputime) { \
519 .utime = cputime_zero, \
520 .stime = cputime_zero, \
521 .sum_exec_runtime = 0, \
525 * Disable preemption until the scheduler is running.
526 * Reset by start_kernel()->sched_init()->init_idle().
528 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
529 * before the scheduler is active -- see should_resched().
531 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
534 * struct thread_group_cputimer - thread group interval timer counts
535 * @cputime: thread group interval timers.
536 * @running: non-zero when there are timers running and
537 * @cputime receives updates.
538 * @lock: lock for fields in this struct.
540 * This structure contains the version of task_cputime, above, that is
541 * used for thread group CPU timer calculations.
543 struct thread_group_cputimer {
544 struct task_cputime cputime;
550 * NOTE! "signal_struct" does not have it's own
551 * locking, because a shared signal_struct always
552 * implies a shared sighand_struct, so locking
553 * sighand_struct is always a proper superset of
554 * the locking of signal_struct.
556 struct signal_struct {
560 wait_queue_head_t wait_chldexit; /* for wait4() */
562 /* current thread group signal load-balancing target: */
563 struct task_struct *curr_target;
565 /* shared signal handling: */
566 struct sigpending shared_pending;
568 /* thread group exit support */
571 * - notify group_exit_task when ->count is equal to notify_count
572 * - everyone except group_exit_task is stopped during signal delivery
573 * of fatal signals, group_exit_task processes the signal.
576 struct task_struct *group_exit_task;
578 /* thread group stop support, overloads group_exit_code too */
579 int group_stop_count;
580 unsigned int flags; /* see SIGNAL_* flags below */
582 /* POSIX.1b Interval Timers */
583 struct list_head posix_timers;
585 /* ITIMER_REAL timer for the process */
586 struct hrtimer real_timer;
587 struct pid *leader_pid;
588 ktime_t it_real_incr;
590 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
591 cputime_t it_prof_expires, it_virt_expires;
592 cputime_t it_prof_incr, it_virt_incr;
595 * Thread group totals for process CPU timers.
596 * See thread_group_cputimer(), et al, for details.
598 struct thread_group_cputimer cputimer;
600 /* Earliest-expiration cache. */
601 struct task_cputime cputime_expires;
603 struct list_head cpu_timers[3];
605 struct pid *tty_old_pgrp;
607 /* boolean value for session group leader */
610 struct tty_struct *tty; /* NULL if no tty */
613 * Cumulative resource counters for dead threads in the group,
614 * and for reaped dead child processes forked by this group.
615 * Live threads maintain their own counters and add to these
616 * in __exit_signal, except for the group leader.
618 cputime_t utime, stime, cutime, cstime;
621 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
622 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
623 unsigned long inblock, oublock, cinblock, coublock;
624 unsigned long maxrss, cmaxrss;
625 struct task_io_accounting ioac;
628 * Cumulative ns of schedule CPU time fo dead threads in the
629 * group, not including a zombie group leader, (This only differs
630 * from jiffies_to_ns(utime + stime) if sched_clock uses something
631 * other than jiffies.)
633 unsigned long long sum_sched_runtime;
636 * We don't bother to synchronize most readers of this at all,
637 * because there is no reader checking a limit that actually needs
638 * to get both rlim_cur and rlim_max atomically, and either one
639 * alone is a single word that can safely be read normally.
640 * getrlimit/setrlimit use task_lock(current->group_leader) to
641 * protect this instead of the siglock, because they really
642 * have no need to disable irqs.
644 struct rlimit rlim[RLIM_NLIMITS];
646 #ifdef CONFIG_BSD_PROCESS_ACCT
647 struct pacct_struct pacct; /* per-process accounting information */
649 #ifdef CONFIG_TASKSTATS
650 struct taskstats *stats;
654 struct tty_audit_buf *tty_audit_buf;
657 int oom_adj; /* OOM kill score adjustment (bit shift) */
660 /* Context switch must be unlocked if interrupts are to be enabled */
661 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
662 # define __ARCH_WANT_UNLOCKED_CTXSW
666 * Bits in flags field of signal_struct.
668 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
669 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
670 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
671 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
673 * Pending notifications to parent.
675 #define SIGNAL_CLD_STOPPED 0x00000010
676 #define SIGNAL_CLD_CONTINUED 0x00000020
677 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
679 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
681 /* If true, all threads except ->group_exit_task have pending SIGKILL */
682 static inline int signal_group_exit(const struct signal_struct *sig)
684 return (sig->flags & SIGNAL_GROUP_EXIT) ||
685 (sig->group_exit_task != NULL);
689 * Some day this will be a full-fledged user tracking system..
692 atomic_t __count; /* reference count */
693 atomic_t processes; /* How many processes does this user have? */
694 atomic_t files; /* How many open files does this user have? */
695 atomic_t sigpending; /* How many pending signals does this user have? */
696 #ifdef CONFIG_INOTIFY_USER
697 atomic_t inotify_watches; /* How many inotify watches does this user have? */
698 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
701 atomic_t epoll_watches; /* The number of file descriptors currently watched */
703 #ifdef CONFIG_POSIX_MQUEUE
704 /* protected by mq_lock */
705 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
707 unsigned long locked_shm; /* How many pages of mlocked shm ? */
710 struct key *uid_keyring; /* UID specific keyring */
711 struct key *session_keyring; /* UID's default session keyring */
714 /* Hash table maintenance information */
715 struct hlist_node uidhash_node;
717 struct user_namespace *user_ns;
719 #ifdef CONFIG_USER_SCHED
720 struct task_group *tg;
723 struct delayed_work work;
727 #ifdef CONFIG_PERF_EVENTS
728 atomic_long_t locked_vm;
732 extern int uids_sysfs_init(void);
734 extern struct user_struct *find_user(uid_t);
736 extern struct user_struct root_user;
737 #define INIT_USER (&root_user)
740 struct backing_dev_info;
741 struct reclaim_state;
743 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
745 /* cumulative counters */
746 unsigned long pcount; /* # of times run on this cpu */
747 unsigned long long run_delay; /* time spent waiting on a runqueue */
750 unsigned long long last_arrival,/* when we last ran on a cpu */
751 last_queued; /* when we were last queued to run */
752 #ifdef CONFIG_SCHEDSTATS
754 unsigned int bkl_count;
757 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
759 #ifdef CONFIG_TASK_DELAY_ACCT
760 struct task_delay_info {
762 unsigned int flags; /* Private per-task flags */
764 /* For each stat XXX, add following, aligned appropriately
766 * struct timespec XXX_start, XXX_end;
770 * Atomicity of updates to XXX_delay, XXX_count protected by
771 * single lock above (split into XXX_lock if contention is an issue).
775 * XXX_count is incremented on every XXX operation, the delay
776 * associated with the operation is added to XXX_delay.
777 * XXX_delay contains the accumulated delay time in nanoseconds.
779 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
780 u64 blkio_delay; /* wait for sync block io completion */
781 u64 swapin_delay; /* wait for swapin block io completion */
782 u32 blkio_count; /* total count of the number of sync block */
783 /* io operations performed */
784 u32 swapin_count; /* total count of the number of swapin block */
785 /* io operations performed */
787 struct timespec freepages_start, freepages_end;
788 u64 freepages_delay; /* wait for memory reclaim */
789 u32 freepages_count; /* total count of memory reclaim */
791 #endif /* CONFIG_TASK_DELAY_ACCT */
793 static inline int sched_info_on(void)
795 #ifdef CONFIG_SCHEDSTATS
797 #elif defined(CONFIG_TASK_DELAY_ACCT)
798 extern int delayacct_on;
813 * sched-domains (multiprocessor balancing) declarations:
817 * Increase resolution of nice-level calculations:
819 #define SCHED_LOAD_SHIFT 10
820 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
822 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
825 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
826 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
827 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
828 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
829 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
830 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
831 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
832 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
833 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
834 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
835 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
837 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
839 enum powersavings_balance_level {
840 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
841 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
842 * first for long running threads
844 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
845 * cpu package for power savings
847 MAX_POWERSAVINGS_BALANCE_LEVELS
850 extern int sched_mc_power_savings, sched_smt_power_savings;
852 static inline int sd_balance_for_mc_power(void)
854 if (sched_smt_power_savings)
855 return SD_POWERSAVINGS_BALANCE;
857 return SD_PREFER_SIBLING;
860 static inline int sd_balance_for_package_power(void)
862 if (sched_mc_power_savings | sched_smt_power_savings)
863 return SD_POWERSAVINGS_BALANCE;
865 return SD_PREFER_SIBLING;
869 * Optimise SD flags for power savings:
870 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
871 * Keep default SD flags if sched_{smt,mc}_power_saving=0
874 static inline int sd_power_saving_flags(void)
876 if (sched_mc_power_savings | sched_smt_power_savings)
877 return SD_BALANCE_NEWIDLE;
883 struct sched_group *next; /* Must be a circular list */
886 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
889 unsigned int cpu_power;
892 * The CPUs this group covers.
894 * NOTE: this field is variable length. (Allocated dynamically
895 * by attaching extra space to the end of the structure,
896 * depending on how many CPUs the kernel has booted up with)
898 * It is also be embedded into static data structures at build
899 * time. (See 'struct static_sched_group' in kernel/sched.c)
901 unsigned long cpumask[0];
904 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
906 return to_cpumask(sg->cpumask);
909 enum sched_domain_level {
919 struct sched_domain_attr {
920 int relax_domain_level;
923 #define SD_ATTR_INIT (struct sched_domain_attr) { \
924 .relax_domain_level = -1, \
927 struct sched_domain {
928 /* These fields must be setup */
929 struct sched_domain *parent; /* top domain must be null terminated */
930 struct sched_domain *child; /* bottom domain must be null terminated */
931 struct sched_group *groups; /* the balancing groups of the domain */
932 unsigned long min_interval; /* Minimum balance interval ms */
933 unsigned long max_interval; /* Maximum balance interval ms */
934 unsigned int busy_factor; /* less balancing by factor if busy */
935 unsigned int imbalance_pct; /* No balance until over watermark */
936 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
937 unsigned int busy_idx;
938 unsigned int idle_idx;
939 unsigned int newidle_idx;
940 unsigned int wake_idx;
941 unsigned int forkexec_idx;
942 unsigned int smt_gain;
943 int flags; /* See SD_* */
944 enum sched_domain_level level;
946 /* Runtime fields. */
947 unsigned long last_balance; /* init to jiffies. units in jiffies */
948 unsigned int balance_interval; /* initialise to 1. units in ms. */
949 unsigned int nr_balance_failed; /* initialise to 0 */
953 #ifdef CONFIG_SCHEDSTATS
954 /* load_balance() stats */
955 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
956 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
957 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
958 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
959 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
960 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
961 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
962 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
964 /* Active load balancing */
965 unsigned int alb_count;
966 unsigned int alb_failed;
967 unsigned int alb_pushed;
969 /* SD_BALANCE_EXEC stats */
970 unsigned int sbe_count;
971 unsigned int sbe_balanced;
972 unsigned int sbe_pushed;
974 /* SD_BALANCE_FORK stats */
975 unsigned int sbf_count;
976 unsigned int sbf_balanced;
977 unsigned int sbf_pushed;
979 /* try_to_wake_up() stats */
980 unsigned int ttwu_wake_remote;
981 unsigned int ttwu_move_affine;
982 unsigned int ttwu_move_balance;
984 #ifdef CONFIG_SCHED_DEBUG
989 * Span of all CPUs in this domain.
991 * NOTE: this field is variable length. (Allocated dynamically
992 * by attaching extra space to the end of the structure,
993 * depending on how many CPUs the kernel has booted up with)
995 * It is also be embedded into static data structures at build
996 * time. (See 'struct static_sched_domain' in kernel/sched.c)
998 unsigned long span[0];
1001 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1003 return to_cpumask(sd->span);
1006 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
1007 struct sched_domain_attr *dattr_new);
1009 /* Test a flag in parent sched domain */
1010 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1012 if (sd->parent && (sd->parent->flags & flag))
1018 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1019 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1021 #else /* CONFIG_SMP */
1023 struct sched_domain_attr;
1026 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
1027 struct sched_domain_attr *dattr_new)
1030 #endif /* !CONFIG_SMP */
1033 struct io_context; /* See blkdev.h */
1036 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1037 extern void prefetch_stack(struct task_struct *t);
1039 static inline void prefetch_stack(struct task_struct *t) { }
1042 struct audit_context; /* See audit.c */
1044 struct pipe_inode_info;
1045 struct uts_namespace;
1048 struct sched_domain;
1053 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1054 #define WF_FORK 0x02 /* child wakeup after fork */
1056 struct sched_class {
1057 const struct sched_class *next;
1059 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
1060 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1061 void (*yield_task) (struct rq *rq);
1063 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1065 struct task_struct * (*pick_next_task) (struct rq *rq);
1066 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1069 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1071 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1072 struct rq *busiest, unsigned long max_load_move,
1073 struct sched_domain *sd, enum cpu_idle_type idle,
1074 int *all_pinned, int *this_best_prio);
1076 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1077 struct rq *busiest, struct sched_domain *sd,
1078 enum cpu_idle_type idle);
1079 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1080 void (*post_schedule) (struct rq *this_rq);
1081 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1083 void (*set_cpus_allowed)(struct task_struct *p,
1084 const struct cpumask *newmask);
1086 void (*rq_online)(struct rq *rq);
1087 void (*rq_offline)(struct rq *rq);
1090 void (*set_curr_task) (struct rq *rq);
1091 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1092 void (*task_new) (struct rq *rq, struct task_struct *p);
1094 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1096 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1098 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1099 int oldprio, int running);
1101 unsigned int (*get_rr_interval) (struct task_struct *task);
1103 #ifdef CONFIG_FAIR_GROUP_SCHED
1104 void (*moved_group) (struct task_struct *p);
1108 struct load_weight {
1109 unsigned long weight, inv_weight;
1113 * CFS stats for a schedulable entity (task, task-group etc)
1115 * Current field usage histogram:
1122 struct sched_entity {
1123 struct load_weight load; /* for load-balancing */
1124 struct rb_node run_node;
1125 struct list_head group_node;
1129 u64 sum_exec_runtime;
1131 u64 prev_sum_exec_runtime;
1143 #ifdef CONFIG_SCHEDSTATS
1153 s64 sum_sleep_runtime;
1160 u64 nr_migrations_cold;
1161 u64 nr_failed_migrations_affine;
1162 u64 nr_failed_migrations_running;
1163 u64 nr_failed_migrations_hot;
1164 u64 nr_forced_migrations;
1165 u64 nr_forced2_migrations;
1168 u64 nr_wakeups_sync;
1169 u64 nr_wakeups_migrate;
1170 u64 nr_wakeups_local;
1171 u64 nr_wakeups_remote;
1172 u64 nr_wakeups_affine;
1173 u64 nr_wakeups_affine_attempts;
1174 u64 nr_wakeups_passive;
1175 u64 nr_wakeups_idle;
1178 #ifdef CONFIG_FAIR_GROUP_SCHED
1179 struct sched_entity *parent;
1180 /* rq on which this entity is (to be) queued: */
1181 struct cfs_rq *cfs_rq;
1182 /* rq "owned" by this entity/group: */
1183 struct cfs_rq *my_q;
1187 struct sched_rt_entity {
1188 struct list_head run_list;
1189 unsigned long timeout;
1190 unsigned int time_slice;
1191 int nr_cpus_allowed;
1193 struct sched_rt_entity *back;
1194 #ifdef CONFIG_RT_GROUP_SCHED
1195 struct sched_rt_entity *parent;
1196 /* rq on which this entity is (to be) queued: */
1197 struct rt_rq *rt_rq;
1198 /* rq "owned" by this entity/group: */
1205 struct task_struct {
1206 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1209 unsigned int flags; /* per process flags, defined below */
1210 unsigned int ptrace;
1212 int lock_depth; /* BKL lock depth */
1215 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1220 int prio, static_prio, normal_prio;
1221 unsigned int rt_priority;
1222 const struct sched_class *sched_class;
1223 struct sched_entity se;
1224 struct sched_rt_entity rt;
1226 #ifdef CONFIG_PREEMPT_NOTIFIERS
1227 /* list of struct preempt_notifier: */
1228 struct hlist_head preempt_notifiers;
1232 * fpu_counter contains the number of consecutive context switches
1233 * that the FPU is used. If this is over a threshold, the lazy fpu
1234 * saving becomes unlazy to save the trap. This is an unsigned char
1235 * so that after 256 times the counter wraps and the behavior turns
1236 * lazy again; this to deal with bursty apps that only use FPU for
1239 unsigned char fpu_counter;
1240 #ifdef CONFIG_BLK_DEV_IO_TRACE
1241 unsigned int btrace_seq;
1244 unsigned int policy;
1245 cpumask_t cpus_allowed;
1247 #ifdef CONFIG_TREE_PREEMPT_RCU
1248 int rcu_read_lock_nesting;
1249 char rcu_read_unlock_special;
1250 struct rcu_node *rcu_blocked_node;
1251 struct list_head rcu_node_entry;
1252 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1254 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1255 struct sched_info sched_info;
1258 struct list_head tasks;
1259 struct plist_node pushable_tasks;
1261 struct mm_struct *mm, *active_mm;
1264 struct linux_binfmt *binfmt;
1266 int exit_code, exit_signal;
1267 int pdeath_signal; /* The signal sent when the parent dies */
1269 unsigned int personality;
1270 unsigned did_exec:1;
1271 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1273 unsigned in_iowait:1;
1276 /* Revert to default priority/policy when forking */
1277 unsigned sched_reset_on_fork:1;
1282 #ifdef CONFIG_CC_STACKPROTECTOR
1283 /* Canary value for the -fstack-protector gcc feature */
1284 unsigned long stack_canary;
1288 * pointers to (original) parent process, youngest child, younger sibling,
1289 * older sibling, respectively. (p->father can be replaced with
1290 * p->real_parent->pid)
1292 struct task_struct *real_parent; /* real parent process */
1293 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1295 * children/sibling forms the list of my natural children
1297 struct list_head children; /* list of my children */
1298 struct list_head sibling; /* linkage in my parent's children list */
1299 struct task_struct *group_leader; /* threadgroup leader */
1302 * ptraced is the list of tasks this task is using ptrace on.
1303 * This includes both natural children and PTRACE_ATTACH targets.
1304 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1306 struct list_head ptraced;
1307 struct list_head ptrace_entry;
1310 * This is the tracer handle for the ptrace BTS extension.
1311 * This field actually belongs to the ptracer task.
1313 struct bts_context *bts;
1315 /* PID/PID hash table linkage. */
1316 struct pid_link pids[PIDTYPE_MAX];
1317 struct list_head thread_group;
1319 struct completion *vfork_done; /* for vfork() */
1320 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1321 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1323 cputime_t utime, stime, utimescaled, stimescaled;
1325 cputime_t prev_utime, prev_stime;
1326 unsigned long nvcsw, nivcsw; /* context switch counts */
1327 struct timespec start_time; /* monotonic time */
1328 struct timespec real_start_time; /* boot based time */
1329 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1330 unsigned long min_flt, maj_flt;
1332 struct task_cputime cputime_expires;
1333 struct list_head cpu_timers[3];
1335 /* process credentials */
1336 const struct cred *real_cred; /* objective and real subjective task
1337 * credentials (COW) */
1338 const struct cred *cred; /* effective (overridable) subjective task
1339 * credentials (COW) */
1340 struct mutex cred_guard_mutex; /* guard against foreign influences on
1341 * credential calculations
1342 * (notably. ptrace) */
1343 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1345 char comm[TASK_COMM_LEN]; /* executable name excluding path
1346 - access with [gs]et_task_comm (which lock
1347 it with task_lock())
1348 - initialized normally by flush_old_exec */
1349 /* file system info */
1350 int link_count, total_link_count;
1351 #ifdef CONFIG_SYSVIPC
1353 struct sysv_sem sysvsem;
1355 #ifdef CONFIG_DETECT_HUNG_TASK
1356 /* hung task detection */
1357 unsigned long last_switch_count;
1359 /* CPU-specific state of this task */
1360 struct thread_struct thread;
1361 /* filesystem information */
1362 struct fs_struct *fs;
1363 /* open file information */
1364 struct files_struct *files;
1366 struct nsproxy *nsproxy;
1367 /* signal handlers */
1368 struct signal_struct *signal;
1369 struct sighand_struct *sighand;
1371 sigset_t blocked, real_blocked;
1372 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1373 struct sigpending pending;
1375 unsigned long sas_ss_sp;
1377 int (*notifier)(void *priv);
1378 void *notifier_data;
1379 sigset_t *notifier_mask;
1380 struct audit_context *audit_context;
1381 #ifdef CONFIG_AUDITSYSCALL
1383 unsigned int sessionid;
1387 /* Thread group tracking */
1390 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1392 spinlock_t alloc_lock;
1394 #ifdef CONFIG_GENERIC_HARDIRQS
1395 /* IRQ handler threads */
1396 struct irqaction *irqaction;
1399 /* Protection of the PI data structures: */
1402 #ifdef CONFIG_RT_MUTEXES
1403 /* PI waiters blocked on a rt_mutex held by this task */
1404 struct plist_head pi_waiters;
1405 /* Deadlock detection and priority inheritance handling */
1406 struct rt_mutex_waiter *pi_blocked_on;
1409 #ifdef CONFIG_DEBUG_MUTEXES
1410 /* mutex deadlock detection */
1411 struct mutex_waiter *blocked_on;
1413 #ifdef CONFIG_TRACE_IRQFLAGS
1414 unsigned int irq_events;
1415 int hardirqs_enabled;
1416 unsigned long hardirq_enable_ip;
1417 unsigned int hardirq_enable_event;
1418 unsigned long hardirq_disable_ip;
1419 unsigned int hardirq_disable_event;
1420 int softirqs_enabled;
1421 unsigned long softirq_disable_ip;
1422 unsigned int softirq_disable_event;
1423 unsigned long softirq_enable_ip;
1424 unsigned int softirq_enable_event;
1425 int hardirq_context;
1426 int softirq_context;
1428 #ifdef CONFIG_LOCKDEP
1429 # define MAX_LOCK_DEPTH 48UL
1432 unsigned int lockdep_recursion;
1433 struct held_lock held_locks[MAX_LOCK_DEPTH];
1434 gfp_t lockdep_reclaim_gfp;
1437 /* journalling filesystem info */
1440 /* stacked block device info */
1441 struct bio *bio_list, **bio_tail;
1444 struct reclaim_state *reclaim_state;
1446 struct backing_dev_info *backing_dev_info;
1448 struct io_context *io_context;
1450 unsigned long ptrace_message;
1451 siginfo_t *last_siginfo; /* For ptrace use. */
1452 struct task_io_accounting ioac;
1453 #if defined(CONFIG_TASK_XACCT)
1454 u64 acct_rss_mem1; /* accumulated rss usage */
1455 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1456 cputime_t acct_timexpd; /* stime + utime since last update */
1458 #ifdef CONFIG_CPUSETS
1459 nodemask_t mems_allowed; /* Protected by alloc_lock */
1460 int cpuset_mem_spread_rotor;
1462 #ifdef CONFIG_CGROUPS
1463 /* Control Group info protected by css_set_lock */
1464 struct css_set *cgroups;
1465 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1466 struct list_head cg_list;
1469 struct robust_list_head __user *robust_list;
1470 #ifdef CONFIG_COMPAT
1471 struct compat_robust_list_head __user *compat_robust_list;
1473 struct list_head pi_state_list;
1474 struct futex_pi_state *pi_state_cache;
1476 #ifdef CONFIG_PERF_EVENTS
1477 struct perf_event_context *perf_event_ctxp;
1478 struct mutex perf_event_mutex;
1479 struct list_head perf_event_list;
1482 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1485 atomic_t fs_excl; /* holding fs exclusive resources */
1486 struct rcu_head rcu;
1489 * cache last used pipe for splice
1491 struct pipe_inode_info *splice_pipe;
1492 #ifdef CONFIG_TASK_DELAY_ACCT
1493 struct task_delay_info *delays;
1495 #ifdef CONFIG_FAULT_INJECTION
1498 struct prop_local_single dirties;
1499 #ifdef CONFIG_LATENCYTOP
1500 int latency_record_count;
1501 struct latency_record latency_record[LT_SAVECOUNT];
1504 * time slack values; these are used to round up poll() and
1505 * select() etc timeout values. These are in nanoseconds.
1507 unsigned long timer_slack_ns;
1508 unsigned long default_timer_slack_ns;
1510 struct list_head *scm_work_list;
1511 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1512 /* Index of current stored adress in ret_stack */
1514 /* Stack of return addresses for return function tracing */
1515 struct ftrace_ret_stack *ret_stack;
1516 /* time stamp for last schedule */
1517 unsigned long long ftrace_timestamp;
1519 * Number of functions that haven't been traced
1520 * because of depth overrun.
1522 atomic_t trace_overrun;
1523 /* Pause for the tracing */
1524 atomic_t tracing_graph_pause;
1526 #ifdef CONFIG_TRACING
1527 /* state flags for use by tracers */
1528 unsigned long trace;
1529 /* bitmask of trace recursion */
1530 unsigned long trace_recursion;
1531 #endif /* CONFIG_TRACING */
1532 unsigned long stack_start;
1535 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1536 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1539 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1540 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1541 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1542 * values are inverted: lower p->prio value means higher priority.
1544 * The MAX_USER_RT_PRIO value allows the actual maximum
1545 * RT priority to be separate from the value exported to
1546 * user-space. This allows kernel threads to set their
1547 * priority to a value higher than any user task. Note:
1548 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1551 #define MAX_USER_RT_PRIO 100
1552 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1554 #define MAX_PRIO (MAX_RT_PRIO + 40)
1555 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1557 static inline int rt_prio(int prio)
1559 if (unlikely(prio < MAX_RT_PRIO))
1564 static inline int rt_task(struct task_struct *p)
1566 return rt_prio(p->prio);
1569 static inline struct pid *task_pid(struct task_struct *task)
1571 return task->pids[PIDTYPE_PID].pid;
1574 static inline struct pid *task_tgid(struct task_struct *task)
1576 return task->group_leader->pids[PIDTYPE_PID].pid;
1580 * Without tasklist or rcu lock it is not safe to dereference
1581 * the result of task_pgrp/task_session even if task == current,
1582 * we can race with another thread doing sys_setsid/sys_setpgid.
1584 static inline struct pid *task_pgrp(struct task_struct *task)
1586 return task->group_leader->pids[PIDTYPE_PGID].pid;
1589 static inline struct pid *task_session(struct task_struct *task)
1591 return task->group_leader->pids[PIDTYPE_SID].pid;
1594 struct pid_namespace;
1597 * the helpers to get the task's different pids as they are seen
1598 * from various namespaces
1600 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1601 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1603 * task_xid_nr_ns() : id seen from the ns specified;
1605 * set_task_vxid() : assigns a virtual id to a task;
1607 * see also pid_nr() etc in include/linux/pid.h
1609 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1610 struct pid_namespace *ns);
1612 static inline pid_t task_pid_nr(struct task_struct *tsk)
1617 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1618 struct pid_namespace *ns)
1620 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1623 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1625 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1629 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1634 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1636 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1638 return pid_vnr(task_tgid(tsk));
1642 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1643 struct pid_namespace *ns)
1645 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1648 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1650 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1654 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1655 struct pid_namespace *ns)
1657 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1660 static inline pid_t task_session_vnr(struct task_struct *tsk)
1662 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1665 /* obsolete, do not use */
1666 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1668 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1672 * pid_alive - check that a task structure is not stale
1673 * @p: Task structure to be checked.
1675 * Test if a process is not yet dead (at most zombie state)
1676 * If pid_alive fails, then pointers within the task structure
1677 * can be stale and must not be dereferenced.
1679 static inline int pid_alive(struct task_struct *p)
1681 return p->pids[PIDTYPE_PID].pid != NULL;
1685 * is_global_init - check if a task structure is init
1686 * @tsk: Task structure to be checked.
1688 * Check if a task structure is the first user space task the kernel created.
1690 static inline int is_global_init(struct task_struct *tsk)
1692 return tsk->pid == 1;
1696 * is_container_init:
1697 * check whether in the task is init in its own pid namespace.
1699 extern int is_container_init(struct task_struct *tsk);
1701 extern struct pid *cad_pid;
1703 extern void free_task(struct task_struct *tsk);
1704 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1706 extern void __put_task_struct(struct task_struct *t);
1708 static inline void put_task_struct(struct task_struct *t)
1710 if (atomic_dec_and_test(&t->usage))
1711 __put_task_struct(t);
1714 extern cputime_t task_utime(struct task_struct *p);
1715 extern cputime_t task_stime(struct task_struct *p);
1716 extern cputime_t task_gtime(struct task_struct *p);
1721 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1722 /* Not implemented yet, only for 486*/
1723 #define PF_STARTING 0x00000002 /* being created */
1724 #define PF_EXITING 0x00000004 /* getting shut down */
1725 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1726 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1727 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1728 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1729 #define PF_DUMPCORE 0x00000200 /* dumped core */
1730 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1731 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1732 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1733 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1734 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1735 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1736 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1737 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1738 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1739 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1740 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1741 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1742 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1743 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1744 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1745 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1746 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1747 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1748 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1749 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1750 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1753 * Only the _current_ task can read/write to tsk->flags, but other
1754 * tasks can access tsk->flags in readonly mode for example
1755 * with tsk_used_math (like during threaded core dumping).
1756 * There is however an exception to this rule during ptrace
1757 * or during fork: the ptracer task is allowed to write to the
1758 * child->flags of its traced child (same goes for fork, the parent
1759 * can write to the child->flags), because we're guaranteed the
1760 * child is not running and in turn not changing child->flags
1761 * at the same time the parent does it.
1763 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1764 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1765 #define clear_used_math() clear_stopped_child_used_math(current)
1766 #define set_used_math() set_stopped_child_used_math(current)
1767 #define conditional_stopped_child_used_math(condition, child) \
1768 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1769 #define conditional_used_math(condition) \
1770 conditional_stopped_child_used_math(condition, current)
1771 #define copy_to_stopped_child_used_math(child) \
1772 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1773 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1774 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1775 #define used_math() tsk_used_math(current)
1777 #ifdef CONFIG_TREE_PREEMPT_RCU
1779 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1780 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1782 static inline void rcu_copy_process(struct task_struct *p)
1784 p->rcu_read_lock_nesting = 0;
1785 p->rcu_read_unlock_special = 0;
1786 p->rcu_blocked_node = NULL;
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))
1810 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1812 return set_cpus_allowed_ptr(p, &new_mask);
1816 * Architectures can set this to 1 if they have specified
1817 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1818 * but then during bootup it turns out that sched_clock()
1819 * is reliable after all:
1821 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1822 extern int sched_clock_stable;
1825 extern unsigned long long sched_clock(void);
1827 extern void sched_clock_init(void);
1828 extern u64 sched_clock_cpu(int cpu);
1830 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1831 static inline void sched_clock_tick(void)
1835 static inline void sched_clock_idle_sleep_event(void)
1839 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1843 extern void sched_clock_tick(void);
1844 extern void sched_clock_idle_sleep_event(void);
1845 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1849 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1850 * clock constructed from sched_clock():
1852 extern unsigned long long cpu_clock(int cpu);
1854 extern unsigned long long
1855 task_sched_runtime(struct task_struct *task);
1856 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1858 /* sched_exec is called by processes performing an exec */
1860 extern void sched_exec(void);
1862 #define sched_exec() {}
1865 extern void sched_clock_idle_sleep_event(void);
1866 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1868 #ifdef CONFIG_HOTPLUG_CPU
1869 extern void idle_task_exit(void);
1871 static inline void idle_task_exit(void) {}
1874 extern void sched_idle_next(void);
1876 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1877 extern void wake_up_idle_cpu(int cpu);
1879 static inline void wake_up_idle_cpu(int cpu) { }
1882 extern unsigned int sysctl_sched_latency;
1883 extern unsigned int sysctl_sched_min_granularity;
1884 extern unsigned int sysctl_sched_wakeup_granularity;
1885 extern unsigned int sysctl_sched_shares_ratelimit;
1886 extern unsigned int sysctl_sched_shares_thresh;
1887 extern unsigned int sysctl_sched_child_runs_first;
1888 #ifdef CONFIG_SCHED_DEBUG
1889 extern unsigned int sysctl_sched_features;
1890 extern unsigned int sysctl_sched_migration_cost;
1891 extern unsigned int sysctl_sched_nr_migrate;
1892 extern unsigned int sysctl_sched_time_avg;
1893 extern unsigned int sysctl_timer_migration;
1895 int sched_nr_latency_handler(struct ctl_table *table, int write,
1896 struct file *file, void __user *buffer, size_t *length,
1899 #ifdef CONFIG_SCHED_DEBUG
1900 static inline unsigned int get_sysctl_timer_migration(void)
1902 return sysctl_timer_migration;
1905 static inline unsigned int get_sysctl_timer_migration(void)
1910 extern unsigned int sysctl_sched_rt_period;
1911 extern int sysctl_sched_rt_runtime;
1913 int sched_rt_handler(struct ctl_table *table, int write,
1914 struct file *filp, void __user *buffer, size_t *lenp,
1917 extern unsigned int sysctl_sched_compat_yield;
1919 #ifdef CONFIG_RT_MUTEXES
1920 extern int rt_mutex_getprio(struct task_struct *p);
1921 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1922 extern void rt_mutex_adjust_pi(struct task_struct *p);
1924 static inline int rt_mutex_getprio(struct task_struct *p)
1926 return p->normal_prio;
1928 # define rt_mutex_adjust_pi(p) do { } while (0)
1931 extern void set_user_nice(struct task_struct *p, long nice);
1932 extern int task_prio(const struct task_struct *p);
1933 extern int task_nice(const struct task_struct *p);
1934 extern int can_nice(const struct task_struct *p, const int nice);
1935 extern int task_curr(const struct task_struct *p);
1936 extern int idle_cpu(int cpu);
1937 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1938 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1939 struct sched_param *);
1940 extern struct task_struct *idle_task(int cpu);
1941 extern struct task_struct *curr_task(int cpu);
1942 extern void set_curr_task(int cpu, struct task_struct *p);
1947 * The default (Linux) execution domain.
1949 extern struct exec_domain default_exec_domain;
1951 union thread_union {
1952 struct thread_info thread_info;
1953 unsigned long stack[THREAD_SIZE/sizeof(long)];
1956 #ifndef __HAVE_ARCH_KSTACK_END
1957 static inline int kstack_end(void *addr)
1959 /* Reliable end of stack detection:
1960 * Some APM bios versions misalign the stack
1962 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1966 extern union thread_union init_thread_union;
1967 extern struct task_struct init_task;
1969 extern struct mm_struct init_mm;
1971 extern struct pid_namespace init_pid_ns;
1974 * find a task by one of its numerical ids
1976 * find_task_by_pid_ns():
1977 * finds a task by its pid in the specified namespace
1978 * find_task_by_vpid():
1979 * finds a task by its virtual pid
1981 * see also find_vpid() etc in include/linux/pid.h
1984 extern struct task_struct *find_task_by_vpid(pid_t nr);
1985 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1986 struct pid_namespace *ns);
1988 extern void __set_special_pids(struct pid *pid);
1990 /* per-UID process charging. */
1991 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1992 static inline struct user_struct *get_uid(struct user_struct *u)
1994 atomic_inc(&u->__count);
1997 extern void free_uid(struct user_struct *);
1998 extern void release_uids(struct user_namespace *ns);
2000 #include <asm/current.h>
2002 extern void do_timer(unsigned long ticks);
2004 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2005 extern int wake_up_process(struct task_struct *tsk);
2006 extern void wake_up_new_task(struct task_struct *tsk,
2007 unsigned long clone_flags);
2009 extern void kick_process(struct task_struct *tsk);
2011 static inline void kick_process(struct task_struct *tsk) { }
2013 extern void sched_fork(struct task_struct *p, int clone_flags);
2014 extern void sched_dead(struct task_struct *p);
2016 extern void proc_caches_init(void);
2017 extern void flush_signals(struct task_struct *);
2018 extern void __flush_signals(struct task_struct *);
2019 extern void ignore_signals(struct task_struct *);
2020 extern void flush_signal_handlers(struct task_struct *, int force_default);
2021 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2023 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2025 unsigned long flags;
2028 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2029 ret = dequeue_signal(tsk, mask, info);
2030 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2035 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2037 extern void unblock_all_signals(void);
2038 extern void release_task(struct task_struct * p);
2039 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2040 extern int force_sigsegv(int, struct task_struct *);
2041 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2042 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2043 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2044 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2045 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2046 extern int kill_pid(struct pid *pid, int sig, int priv);
2047 extern int kill_proc_info(int, struct siginfo *, pid_t);
2048 extern int do_notify_parent(struct task_struct *, int);
2049 extern void force_sig(int, struct task_struct *);
2050 extern void force_sig_specific(int, struct task_struct *);
2051 extern int send_sig(int, struct task_struct *, int);
2052 extern void zap_other_threads(struct task_struct *p);
2053 extern struct sigqueue *sigqueue_alloc(void);
2054 extern void sigqueue_free(struct sigqueue *);
2055 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2056 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2057 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2059 static inline int kill_cad_pid(int sig, int priv)
2061 return kill_pid(cad_pid, sig, priv);
2064 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2065 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2066 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2067 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2069 static inline int is_si_special(const struct siginfo *info)
2071 return info <= SEND_SIG_FORCED;
2074 /* True if we are on the alternate signal stack. */
2076 static inline int on_sig_stack(unsigned long sp)
2078 return (sp - current->sas_ss_sp < current->sas_ss_size);
2081 static inline int sas_ss_flags(unsigned long sp)
2083 return (current->sas_ss_size == 0 ? SS_DISABLE
2084 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2088 * Routines for handling mm_structs
2090 extern struct mm_struct * mm_alloc(void);
2092 /* mmdrop drops the mm and the page tables */
2093 extern void __mmdrop(struct mm_struct *);
2094 static inline void mmdrop(struct mm_struct * mm)
2096 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2100 /* mmput gets rid of the mappings and all user-space */
2101 extern void mmput(struct mm_struct *);
2102 /* Grab a reference to a task's mm, if it is not already going away */
2103 extern struct mm_struct *get_task_mm(struct task_struct *task);
2104 /* Remove the current tasks stale references to the old mm_struct */
2105 extern void mm_release(struct task_struct *, struct mm_struct *);
2106 /* Allocate a new mm structure and copy contents from tsk->mm */
2107 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2109 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2110 struct task_struct *, struct pt_regs *);
2111 extern void flush_thread(void);
2112 extern void exit_thread(void);
2114 extern void exit_files(struct task_struct *);
2115 extern void __cleanup_signal(struct signal_struct *);
2116 extern void __cleanup_sighand(struct sighand_struct *);
2118 extern void exit_itimers(struct signal_struct *);
2119 extern void flush_itimer_signals(void);
2121 extern NORET_TYPE void do_group_exit(int);
2123 extern void daemonize(const char *, ...);
2124 extern int allow_signal(int);
2125 extern int disallow_signal(int);
2127 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2128 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2129 struct task_struct *fork_idle(int);
2131 extern void set_task_comm(struct task_struct *tsk, char *from);
2132 extern char *get_task_comm(char *to, struct task_struct *tsk);
2135 extern void wait_task_context_switch(struct task_struct *p);
2136 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2138 static inline void wait_task_context_switch(struct task_struct *p) {}
2139 static inline unsigned long wait_task_inactive(struct task_struct *p,
2146 #define next_task(p) \
2147 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2149 #define for_each_process(p) \
2150 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2152 extern bool current_is_single_threaded(void);
2155 * Careful: do_each_thread/while_each_thread is a double loop so
2156 * 'break' will not work as expected - use goto instead.
2158 #define do_each_thread(g, t) \
2159 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2161 #define while_each_thread(g, t) \
2162 while ((t = next_thread(t)) != g)
2164 /* de_thread depends on thread_group_leader not being a pid based check */
2165 #define thread_group_leader(p) (p == p->group_leader)
2167 /* Do to the insanities of de_thread it is possible for a process
2168 * to have the pid of the thread group leader without actually being
2169 * the thread group leader. For iteration through the pids in proc
2170 * all we care about is that we have a task with the appropriate
2171 * pid, we don't actually care if we have the right task.
2173 static inline int has_group_leader_pid(struct task_struct *p)
2175 return p->pid == p->tgid;
2179 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2181 return p1->tgid == p2->tgid;
2184 static inline struct task_struct *next_thread(const struct task_struct *p)
2186 return list_entry_rcu(p->thread_group.next,
2187 struct task_struct, thread_group);
2190 static inline int thread_group_empty(struct task_struct *p)
2192 return list_empty(&p->thread_group);
2195 #define delay_group_leader(p) \
2196 (thread_group_leader(p) && !thread_group_empty(p))
2198 static inline int task_detached(struct task_struct *p)
2200 return p->exit_signal == -1;
2204 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2205 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2206 * pins the final release of task.io_context. Also protects ->cpuset and
2207 * ->cgroup.subsys[].
2209 * Nests both inside and outside of read_lock(&tasklist_lock).
2210 * It must not be nested with write_lock_irq(&tasklist_lock),
2211 * neither inside nor outside.
2213 static inline void task_lock(struct task_struct *p)
2215 spin_lock(&p->alloc_lock);
2218 static inline void task_unlock(struct task_struct *p)
2220 spin_unlock(&p->alloc_lock);
2223 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2224 unsigned long *flags);
2226 static inline void unlock_task_sighand(struct task_struct *tsk,
2227 unsigned long *flags)
2229 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2232 #ifndef __HAVE_THREAD_FUNCTIONS
2234 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2235 #define task_stack_page(task) ((task)->stack)
2237 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2239 *task_thread_info(p) = *task_thread_info(org);
2240 task_thread_info(p)->task = p;
2243 static inline unsigned long *end_of_stack(struct task_struct *p)
2245 return (unsigned long *)(task_thread_info(p) + 1);
2250 static inline int object_is_on_stack(void *obj)
2252 void *stack = task_stack_page(current);
2254 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2257 extern void thread_info_cache_init(void);
2259 #ifdef CONFIG_DEBUG_STACK_USAGE
2260 static inline unsigned long stack_not_used(struct task_struct *p)
2262 unsigned long *n = end_of_stack(p);
2264 do { /* Skip over canary */
2268 return (unsigned long)n - (unsigned long)end_of_stack(p);
2272 /* set thread flags in other task's structures
2273 * - see asm/thread_info.h for TIF_xxxx flags available
2275 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2277 set_ti_thread_flag(task_thread_info(tsk), flag);
2280 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2282 clear_ti_thread_flag(task_thread_info(tsk), flag);
2285 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2287 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2290 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2292 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2295 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2297 return test_ti_thread_flag(task_thread_info(tsk), flag);
2300 static inline void set_tsk_need_resched(struct task_struct *tsk)
2302 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2305 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2307 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2310 static inline int test_tsk_need_resched(struct task_struct *tsk)
2312 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2315 static inline int restart_syscall(void)
2317 set_tsk_thread_flag(current, TIF_SIGPENDING);
2318 return -ERESTARTNOINTR;
2321 static inline int signal_pending(struct task_struct *p)
2323 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2326 extern int __fatal_signal_pending(struct task_struct *p);
2328 static inline int fatal_signal_pending(struct task_struct *p)
2330 return signal_pending(p) && __fatal_signal_pending(p);
2333 static inline int signal_pending_state(long state, struct task_struct *p)
2335 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2337 if (!signal_pending(p))
2340 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2343 static inline int need_resched(void)
2345 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2349 * cond_resched() and cond_resched_lock(): latency reduction via
2350 * explicit rescheduling in places that are safe. The return
2351 * value indicates whether a reschedule was done in fact.
2352 * cond_resched_lock() will drop the spinlock before scheduling,
2353 * cond_resched_softirq() will enable bhs before scheduling.
2355 extern int _cond_resched(void);
2357 #define cond_resched() ({ \
2358 __might_sleep(__FILE__, __LINE__, 0); \
2362 extern int __cond_resched_lock(spinlock_t *lock);
2364 #ifdef CONFIG_PREEMPT
2365 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2367 #define PREEMPT_LOCK_OFFSET 0
2370 #define cond_resched_lock(lock) ({ \
2371 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2372 __cond_resched_lock(lock); \
2375 extern int __cond_resched_softirq(void);
2377 #define cond_resched_softirq() ({ \
2378 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2379 __cond_resched_softirq(); \
2383 * Does a critical section need to be broken due to another
2384 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2385 * but a general need for low latency)
2387 static inline int spin_needbreak(spinlock_t *lock)
2389 #ifdef CONFIG_PREEMPT
2390 return spin_is_contended(lock);
2397 * Thread group CPU time accounting.
2399 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2400 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2402 static inline void thread_group_cputime_init(struct signal_struct *sig)
2404 sig->cputimer.cputime = INIT_CPUTIME;
2405 spin_lock_init(&sig->cputimer.lock);
2406 sig->cputimer.running = 0;
2409 static inline void thread_group_cputime_free(struct signal_struct *sig)
2414 * Reevaluate whether the task has signals pending delivery.
2415 * Wake the task if so.
2416 * This is required every time the blocked sigset_t changes.
2417 * callers must hold sighand->siglock.
2419 extern void recalc_sigpending_and_wake(struct task_struct *t);
2420 extern void recalc_sigpending(void);
2422 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2425 * Wrappers for p->thread_info->cpu access. No-op on UP.
2429 static inline unsigned int task_cpu(const struct task_struct *p)
2431 return task_thread_info(p)->cpu;
2434 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2438 static inline unsigned int task_cpu(const struct task_struct *p)
2443 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2447 #endif /* CONFIG_SMP */
2449 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2451 #ifdef CONFIG_TRACING
2453 __trace_special(void *__tr, void *__data,
2454 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2457 __trace_special(void *__tr, void *__data,
2458 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2463 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2464 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2466 extern void normalize_rt_tasks(void);
2468 #ifdef CONFIG_GROUP_SCHED
2470 extern struct task_group init_task_group;
2471 #ifdef CONFIG_USER_SCHED
2472 extern struct task_group root_task_group;
2473 extern void set_tg_uid(struct user_struct *user);
2476 extern struct task_group *sched_create_group(struct task_group *parent);
2477 extern void sched_destroy_group(struct task_group *tg);
2478 extern void sched_move_task(struct task_struct *tsk);
2479 #ifdef CONFIG_FAIR_GROUP_SCHED
2480 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2481 extern unsigned long sched_group_shares(struct task_group *tg);
2483 #ifdef CONFIG_RT_GROUP_SCHED
2484 extern int sched_group_set_rt_runtime(struct task_group *tg,
2485 long rt_runtime_us);
2486 extern long sched_group_rt_runtime(struct task_group *tg);
2487 extern int sched_group_set_rt_period(struct task_group *tg,
2489 extern long sched_group_rt_period(struct task_group *tg);
2490 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2494 extern int task_can_switch_user(struct user_struct *up,
2495 struct task_struct *tsk);
2497 #ifdef CONFIG_TASK_XACCT
2498 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2500 tsk->ioac.rchar += amt;
2503 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2505 tsk->ioac.wchar += amt;
2508 static inline void inc_syscr(struct task_struct *tsk)
2513 static inline void inc_syscw(struct task_struct *tsk)
2518 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2522 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2526 static inline void inc_syscr(struct task_struct *tsk)
2530 static inline void inc_syscw(struct task_struct *tsk)
2535 #ifndef TASK_SIZE_OF
2536 #define TASK_SIZE_OF(tsk) TASK_SIZE
2540 * Call the function if the target task is executing on a CPU right now:
2542 extern void task_oncpu_function_call(struct task_struct *p,
2543 void (*func) (void *info), void *info);
2546 #ifdef CONFIG_MM_OWNER
2547 extern void mm_update_next_owner(struct mm_struct *mm);
2548 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2550 static inline void mm_update_next_owner(struct mm_struct *mm)
2554 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2557 #endif /* CONFIG_MM_OWNER */
2559 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2561 #endif /* __KERNEL__ */