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_counter_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 void calc_global_load(void);
144 extern u64 cpu_nr_migrations(int cpu);
146 extern unsigned long get_parent_ip(unsigned long addr);
151 #ifdef CONFIG_SCHED_DEBUG
152 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
153 extern void proc_sched_set_task(struct task_struct *p);
155 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
158 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
161 static inline void proc_sched_set_task(struct task_struct *p)
165 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
170 extern unsigned long long time_sync_thresh;
173 * Task state bitmask. NOTE! These bits are also
174 * encoded in fs/proc/array.c: get_task_state().
176 * We have two separate sets of flags: task->state
177 * is about runnability, while task->exit_state are
178 * about the task exiting. Confusing, but this way
179 * modifying one set can't modify the other one by
182 #define TASK_RUNNING 0
183 #define TASK_INTERRUPTIBLE 1
184 #define TASK_UNINTERRUPTIBLE 2
185 #define __TASK_STOPPED 4
186 #define __TASK_TRACED 8
187 /* in tsk->exit_state */
188 #define EXIT_ZOMBIE 16
190 /* in tsk->state again */
192 #define TASK_WAKEKILL 128
194 /* Convenience macros for the sake of set_task_state */
195 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
196 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
197 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
199 /* Convenience macros for the sake of wake_up */
200 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
201 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
203 /* get_task_state() */
204 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
205 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
208 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
209 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
210 #define task_is_stopped_or_traced(task) \
211 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
212 #define task_contributes_to_load(task) \
213 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
214 (task->flags & PF_FREEZING) == 0)
216 #define __set_task_state(tsk, state_value) \
217 do { (tsk)->state = (state_value); } while (0)
218 #define set_task_state(tsk, state_value) \
219 set_mb((tsk)->state, (state_value))
222 * set_current_state() includes a barrier so that the write of current->state
223 * is correctly serialised wrt the caller's subsequent test of whether to
226 * set_current_state(TASK_UNINTERRUPTIBLE);
227 * if (do_i_need_to_sleep())
230 * If the caller does not need such serialisation then use __set_current_state()
232 #define __set_current_state(state_value) \
233 do { current->state = (state_value); } while (0)
234 #define set_current_state(state_value) \
235 set_mb(current->state, (state_value))
237 /* Task command name length */
238 #define TASK_COMM_LEN 16
240 #include <linux/spinlock.h>
243 * This serializes "schedule()" and also protects
244 * the run-queue from deletions/modifications (but
245 * _adding_ to the beginning of the run-queue has
248 extern rwlock_t tasklist_lock;
249 extern spinlock_t mmlist_lock;
253 extern void sched_init(void);
254 extern void sched_init_smp(void);
255 extern asmlinkage void schedule_tail(struct task_struct *prev);
256 extern void init_idle(struct task_struct *idle, int cpu);
257 extern void init_idle_bootup_task(struct task_struct *idle);
259 extern int runqueue_is_locked(void);
260 extern void task_rq_unlock_wait(struct task_struct *p);
262 extern cpumask_var_t nohz_cpu_mask;
263 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
264 extern int select_nohz_load_balancer(int cpu);
265 extern int get_nohz_load_balancer(void);
267 static inline int select_nohz_load_balancer(int cpu)
274 * Only dump TASK_* tasks. (0 for all tasks)
276 extern void show_state_filter(unsigned long state_filter);
278 static inline void show_state(void)
280 show_state_filter(0);
283 extern void show_regs(struct pt_regs *);
286 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
287 * task), SP is the stack pointer of the first frame that should be shown in the back
288 * trace (or NULL if the entire call-chain of the task should be shown).
290 extern void show_stack(struct task_struct *task, unsigned long *sp);
292 void io_schedule(void);
293 long io_schedule_timeout(long timeout);
295 extern void cpu_init (void);
296 extern void trap_init(void);
297 extern void update_process_times(int user);
298 extern void scheduler_tick(void);
300 extern void sched_show_task(struct task_struct *p);
302 #ifdef CONFIG_DETECT_SOFTLOCKUP
303 extern void softlockup_tick(void);
304 extern void touch_softlockup_watchdog(void);
305 extern void touch_all_softlockup_watchdogs(void);
306 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
307 struct file *filp, void __user *buffer,
308 size_t *lenp, loff_t *ppos);
309 extern unsigned int softlockup_panic;
310 extern int softlockup_thresh;
312 static inline void softlockup_tick(void)
315 static inline void touch_softlockup_watchdog(void)
318 static inline void touch_all_softlockup_watchdogs(void)
323 #ifdef CONFIG_DETECT_HUNG_TASK
324 extern unsigned int sysctl_hung_task_panic;
325 extern unsigned long sysctl_hung_task_check_count;
326 extern unsigned long sysctl_hung_task_timeout_secs;
327 extern unsigned long sysctl_hung_task_warnings;
328 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
329 struct file *filp, void __user *buffer,
330 size_t *lenp, loff_t *ppos);
333 /* Attach to any functions which should be ignored in wchan output. */
334 #define __sched __attribute__((__section__(".sched.text")))
336 /* Linker adds these: start and end of __sched functions */
337 extern char __sched_text_start[], __sched_text_end[];
339 /* Is this address in the __sched functions? */
340 extern int in_sched_functions(unsigned long addr);
342 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
343 extern signed long schedule_timeout(signed long timeout);
344 extern signed long schedule_timeout_interruptible(signed long timeout);
345 extern signed long schedule_timeout_killable(signed long timeout);
346 extern signed long schedule_timeout_uninterruptible(signed long timeout);
347 asmlinkage void __schedule(void);
348 asmlinkage void schedule(void);
349 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
352 struct user_namespace;
355 * Default maximum number of active map areas, this limits the number of vmas
356 * per mm struct. Users can overwrite this number by sysctl but there is a
359 * When a program's coredump is generated as ELF format, a section is created
360 * per a vma. In ELF, the number of sections is represented in unsigned short.
361 * This means the number of sections should be smaller than 65535 at coredump.
362 * Because the kernel adds some informative sections to a image of program at
363 * generating coredump, we need some margin. The number of extra sections is
364 * 1-3 now and depends on arch. We use "5" as safe margin, here.
366 #define MAPCOUNT_ELF_CORE_MARGIN (5)
367 #define DEFAULT_MAX_MAP_COUNT (USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
369 extern int sysctl_max_map_count;
371 #include <linux/aio.h>
374 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
375 unsigned long, unsigned long);
377 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
378 unsigned long len, unsigned long pgoff,
379 unsigned long flags);
380 extern void arch_unmap_area(struct mm_struct *, unsigned long);
381 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
383 #if USE_SPLIT_PTLOCKS
385 * The mm counters are not protected by its page_table_lock,
386 * so must be incremented atomically.
388 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
389 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
390 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
391 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
392 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
394 #else /* !USE_SPLIT_PTLOCKS */
396 * The mm counters are protected by its page_table_lock,
397 * so can be incremented directly.
399 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
400 #define get_mm_counter(mm, member) ((mm)->_##member)
401 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
402 #define inc_mm_counter(mm, member) (mm)->_##member++
403 #define dec_mm_counter(mm, member) (mm)->_##member--
405 #endif /* !USE_SPLIT_PTLOCKS */
407 #define get_mm_rss(mm) \
408 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
409 #define update_hiwater_rss(mm) do { \
410 unsigned long _rss = get_mm_rss(mm); \
411 if ((mm)->hiwater_rss < _rss) \
412 (mm)->hiwater_rss = _rss; \
414 #define update_hiwater_vm(mm) do { \
415 if ((mm)->hiwater_vm < (mm)->total_vm) \
416 (mm)->hiwater_vm = (mm)->total_vm; \
419 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
421 return max(mm->hiwater_rss, get_mm_rss(mm));
424 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
426 return max(mm->hiwater_vm, mm->total_vm);
429 extern void set_dumpable(struct mm_struct *mm, int value);
430 extern int get_dumpable(struct mm_struct *mm);
434 #define MMF_DUMPABLE 0 /* core dump is permitted */
435 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
436 #define MMF_DUMPABLE_BITS 2
438 /* coredump filter bits */
439 #define MMF_DUMP_ANON_PRIVATE 2
440 #define MMF_DUMP_ANON_SHARED 3
441 #define MMF_DUMP_MAPPED_PRIVATE 4
442 #define MMF_DUMP_MAPPED_SHARED 5
443 #define MMF_DUMP_ELF_HEADERS 6
444 #define MMF_DUMP_HUGETLB_PRIVATE 7
445 #define MMF_DUMP_HUGETLB_SHARED 8
446 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
447 #define MMF_DUMP_FILTER_BITS 7
448 #define MMF_DUMP_FILTER_MASK \
449 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
450 #define MMF_DUMP_FILTER_DEFAULT \
451 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
452 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
454 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
455 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
457 # define MMF_DUMP_MASK_DEFAULT_ELF 0
460 struct sighand_struct {
462 struct k_sigaction action[_NSIG];
464 wait_queue_head_t signalfd_wqh;
467 struct pacct_struct {
470 unsigned long ac_mem;
471 cputime_t ac_utime, ac_stime;
472 unsigned long ac_minflt, ac_majflt;
476 * struct task_cputime - collected CPU time counts
477 * @utime: time spent in user mode, in &cputime_t units
478 * @stime: time spent in kernel mode, in &cputime_t units
479 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
481 * This structure groups together three kinds of CPU time that are
482 * tracked for threads and thread groups. Most things considering
483 * CPU time want to group these counts together and treat all three
484 * of them in parallel.
486 struct task_cputime {
489 unsigned long long sum_exec_runtime;
491 /* Alternate field names when used to cache expirations. */
492 #define prof_exp stime
493 #define virt_exp utime
494 #define sched_exp sum_exec_runtime
496 #define INIT_CPUTIME \
497 (struct task_cputime) { \
498 .utime = cputime_zero, \
499 .stime = cputime_zero, \
500 .sum_exec_runtime = 0, \
504 * Disable preemption until the scheduler is running.
505 * Reset by start_kernel()->sched_init()->init_idle().
507 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
508 * before the scheduler is active -- see should_resched().
510 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
513 * struct thread_group_cputimer - thread group interval timer counts
514 * @cputime: thread group interval timers.
515 * @running: non-zero when there are timers running and
516 * @cputime receives updates.
517 * @lock: lock for fields in this struct.
519 * This structure contains the version of task_cputime, above, that is
520 * used for thread group CPU timer calculations.
522 struct thread_group_cputimer {
523 struct task_cputime cputime;
529 * NOTE! "signal_struct" does not have it's own
530 * locking, because a shared signal_struct always
531 * implies a shared sighand_struct, so locking
532 * sighand_struct is always a proper superset of
533 * the locking of signal_struct.
535 struct signal_struct {
539 wait_queue_head_t wait_chldexit; /* for wait4() */
541 /* current thread group signal load-balancing target: */
542 struct task_struct *curr_target;
544 /* shared signal handling: */
545 struct sigpending shared_pending;
547 /* thread group exit support */
550 * - notify group_exit_task when ->count is equal to notify_count
551 * - everyone except group_exit_task is stopped during signal delivery
552 * of fatal signals, group_exit_task processes the signal.
555 struct task_struct *group_exit_task;
557 /* thread group stop support, overloads group_exit_code too */
558 int group_stop_count;
559 unsigned int flags; /* see SIGNAL_* flags below */
561 /* POSIX.1b Interval Timers */
562 struct list_head posix_timers;
564 /* ITIMER_REAL timer for the process */
565 struct hrtimer real_timer;
566 struct pid *leader_pid;
567 ktime_t it_real_incr;
569 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
570 cputime_t it_prof_expires, it_virt_expires;
571 cputime_t it_prof_incr, it_virt_incr;
574 * Thread group totals for process CPU timers.
575 * See thread_group_cputimer(), et al, for details.
577 struct thread_group_cputimer cputimer;
579 /* Earliest-expiration cache. */
580 struct task_cputime cputime_expires;
582 struct list_head cpu_timers[3];
584 struct pid *tty_old_pgrp;
586 /* boolean value for session group leader */
589 struct tty_struct *tty; /* NULL if no tty */
592 * Cumulative resource counters for dead threads in the group,
593 * and for reaped dead child processes forked by this group.
594 * Live threads maintain their own counters and add to these
595 * in __exit_signal, except for the group leader.
597 cputime_t utime, stime, cutime, cstime;
600 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
601 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
602 unsigned long inblock, oublock, cinblock, coublock;
603 struct task_io_accounting ioac;
606 * Cumulative ns of schedule CPU time fo dead threads in the
607 * group, not including a zombie group leader, (This only differs
608 * from jiffies_to_ns(utime + stime) if sched_clock uses something
609 * other than jiffies.)
611 unsigned long long sum_sched_runtime;
614 * We don't bother to synchronize most readers of this at all,
615 * because there is no reader checking a limit that actually needs
616 * to get both rlim_cur and rlim_max atomically, and either one
617 * alone is a single word that can safely be read normally.
618 * getrlimit/setrlimit use task_lock(current->group_leader) to
619 * protect this instead of the siglock, because they really
620 * have no need to disable irqs.
622 struct rlimit rlim[RLIM_NLIMITS];
624 #ifdef CONFIG_BSD_PROCESS_ACCT
625 struct pacct_struct pacct; /* per-process accounting information */
627 #ifdef CONFIG_TASKSTATS
628 struct taskstats *stats;
632 struct tty_audit_buf *tty_audit_buf;
636 /* Context switch must be unlocked if interrupts are to be enabled */
637 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
638 # define __ARCH_WANT_UNLOCKED_CTXSW
642 * Bits in flags field of signal_struct.
644 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
645 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
646 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
647 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
649 * Pending notifications to parent.
651 #define SIGNAL_CLD_STOPPED 0x00000010
652 #define SIGNAL_CLD_CONTINUED 0x00000020
653 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
655 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
657 /* If true, all threads except ->group_exit_task have pending SIGKILL */
658 static inline int signal_group_exit(const struct signal_struct *sig)
660 return (sig->flags & SIGNAL_GROUP_EXIT) ||
661 (sig->group_exit_task != NULL);
665 * Some day this will be a full-fledged user tracking system..
668 atomic_t __count; /* reference count */
669 atomic_t processes; /* How many processes does this user have? */
670 atomic_t files; /* How many open files does this user have? */
671 atomic_t sigpending; /* How many pending signals does this user have? */
672 #ifdef CONFIG_INOTIFY_USER
673 atomic_t inotify_watches; /* How many inotify watches does this user have? */
674 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
677 atomic_t epoll_watches; /* The number of file descriptors currently watched */
679 #ifdef CONFIG_POSIX_MQUEUE
680 /* protected by mq_lock */
681 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
683 unsigned long locked_shm; /* How many pages of mlocked shm ? */
686 struct key *uid_keyring; /* UID specific keyring */
687 struct key *session_keyring; /* UID's default session keyring */
690 /* Hash table maintenance information */
691 struct hlist_node uidhash_node;
693 struct user_namespace *user_ns;
695 #ifdef CONFIG_USER_SCHED
696 struct task_group *tg;
699 struct delayed_work work;
703 #ifdef CONFIG_PERF_COUNTERS
704 atomic_long_t locked_vm;
708 extern int uids_sysfs_init(void);
710 extern struct user_struct *find_user(uid_t);
712 extern struct user_struct root_user;
713 #define INIT_USER (&root_user)
716 struct backing_dev_info;
717 struct reclaim_state;
719 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
721 /* cumulative counters */
722 unsigned long pcount; /* # of times run on this cpu */
723 unsigned long long run_delay; /* time spent waiting on a runqueue */
726 unsigned long long last_arrival,/* when we last ran on a cpu */
727 last_queued; /* when we were last queued to run */
728 #ifdef CONFIG_SCHEDSTATS
730 unsigned int bkl_count;
733 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
735 #ifdef CONFIG_TASK_DELAY_ACCT
736 struct task_delay_info {
738 unsigned int flags; /* Private per-task flags */
740 /* For each stat XXX, add following, aligned appropriately
742 * struct timespec XXX_start, XXX_end;
746 * Atomicity of updates to XXX_delay, XXX_count protected by
747 * single lock above (split into XXX_lock if contention is an issue).
751 * XXX_count is incremented on every XXX operation, the delay
752 * associated with the operation is added to XXX_delay.
753 * XXX_delay contains the accumulated delay time in nanoseconds.
755 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
756 u64 blkio_delay; /* wait for sync block io completion */
757 u64 swapin_delay; /* wait for swapin block io completion */
758 u32 blkio_count; /* total count of the number of sync block */
759 /* io operations performed */
760 u32 swapin_count; /* total count of the number of swapin block */
761 /* io operations performed */
763 struct timespec freepages_start, freepages_end;
764 u64 freepages_delay; /* wait for memory reclaim */
765 u32 freepages_count; /* total count of memory reclaim */
767 #endif /* CONFIG_TASK_DELAY_ACCT */
769 static inline int sched_info_on(void)
771 #ifdef CONFIG_SCHEDSTATS
773 #elif defined(CONFIG_TASK_DELAY_ACCT)
774 extern int delayacct_on;
789 * sched-domains (multiprocessor balancing) declarations:
793 * Increase resolution of nice-level calculations:
795 #define SCHED_LOAD_SHIFT 10
796 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
798 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
801 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
802 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
803 #define SD_BALANCE_EXEC 4 /* Balance on exec */
804 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
805 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
806 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
807 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
808 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
809 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
810 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
811 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
812 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
814 enum powersavings_balance_level {
815 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
816 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
817 * first for long running threads
819 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
820 * cpu package for power savings
822 MAX_POWERSAVINGS_BALANCE_LEVELS
825 extern int sched_mc_power_savings, sched_smt_power_savings;
827 static inline int sd_balance_for_mc_power(void)
829 if (sched_smt_power_savings)
830 return SD_POWERSAVINGS_BALANCE;
835 static inline int sd_balance_for_package_power(void)
837 if (sched_mc_power_savings | sched_smt_power_savings)
838 return SD_POWERSAVINGS_BALANCE;
844 * Optimise SD flags for power savings:
845 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
846 * Keep default SD flags if sched_{smt,mc}_power_saving=0
849 static inline int sd_power_saving_flags(void)
851 if (sched_mc_power_savings | sched_smt_power_savings)
852 return SD_BALANCE_NEWIDLE;
858 struct sched_group *next; /* Must be a circular list */
861 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
862 * single CPU. This is read only (except for setup, hotplug CPU).
863 * Note : Never change cpu_power without recompute its reciprocal
865 unsigned int __cpu_power;
867 * reciprocal value of cpu_power to avoid expensive divides
868 * (see include/linux/reciprocal_div.h)
870 u32 reciprocal_cpu_power;
873 * The CPUs this group covers.
875 * NOTE: this field is variable length. (Allocated dynamically
876 * by attaching extra space to the end of the structure,
877 * depending on how many CPUs the kernel has booted up with)
879 * It is also be embedded into static data structures at build
880 * time. (See 'struct static_sched_group' in kernel/sched.c)
882 unsigned long cpumask[0];
885 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
887 return to_cpumask(sg->cpumask);
890 enum sched_domain_level {
900 struct sched_domain_attr {
901 int relax_domain_level;
904 #define SD_ATTR_INIT (struct sched_domain_attr) { \
905 .relax_domain_level = -1, \
908 struct sched_domain {
909 /* These fields must be setup */
910 struct sched_domain *parent; /* top domain must be null terminated */
911 struct sched_domain *child; /* bottom domain must be null terminated */
912 struct sched_group *groups; /* the balancing groups of the domain */
913 unsigned long min_interval; /* Minimum balance interval ms */
914 unsigned long max_interval; /* Maximum balance interval ms */
915 unsigned int busy_factor; /* less balancing by factor if busy */
916 unsigned int imbalance_pct; /* No balance until over watermark */
917 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
918 unsigned int busy_idx;
919 unsigned int idle_idx;
920 unsigned int newidle_idx;
921 unsigned int wake_idx;
922 unsigned int forkexec_idx;
923 int flags; /* See SD_* */
924 enum sched_domain_level level;
926 /* Runtime fields. */
927 unsigned long last_balance; /* init to jiffies. units in jiffies */
928 unsigned int balance_interval; /* initialise to 1. units in ms. */
929 unsigned int nr_balance_failed; /* initialise to 0 */
933 #ifdef CONFIG_SCHEDSTATS
934 /* load_balance() stats */
935 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
936 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
937 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
938 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
939 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
940 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
944 /* Active load balancing */
945 unsigned int alb_count;
946 unsigned int alb_failed;
947 unsigned int alb_pushed;
949 /* SD_BALANCE_EXEC stats */
950 unsigned int sbe_count;
951 unsigned int sbe_balanced;
952 unsigned int sbe_pushed;
954 /* SD_BALANCE_FORK stats */
955 unsigned int sbf_count;
956 unsigned int sbf_balanced;
957 unsigned int sbf_pushed;
959 /* try_to_wake_up() stats */
960 unsigned int ttwu_wake_remote;
961 unsigned int ttwu_move_affine;
962 unsigned int ttwu_move_balance;
964 #ifdef CONFIG_SCHED_DEBUG
969 * Span of all CPUs in this domain.
971 * NOTE: this field is variable length. (Allocated dynamically
972 * by attaching extra space to the end of the structure,
973 * depending on how many CPUs the kernel has booted up with)
975 * It is also be embedded into static data structures at build
976 * time. (See 'struct static_sched_domain' in kernel/sched.c)
978 unsigned long span[0];
981 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
983 return to_cpumask(sd->span);
986 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
987 struct sched_domain_attr *dattr_new);
989 /* Test a flag in parent sched domain */
990 static inline int test_sd_parent(struct sched_domain *sd, int flag)
992 if (sd->parent && (sd->parent->flags & flag))
998 #else /* CONFIG_SMP */
1000 struct sched_domain_attr;
1003 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
1004 struct sched_domain_attr *dattr_new)
1007 #endif /* !CONFIG_SMP */
1009 struct io_context; /* See blkdev.h */
1012 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1013 extern void prefetch_stack(struct task_struct *t);
1015 static inline void prefetch_stack(struct task_struct *t) { }
1018 struct audit_context; /* See audit.c */
1020 struct pipe_inode_info;
1021 struct uts_namespace;
1024 struct sched_domain;
1026 struct sched_class {
1027 const struct sched_class *next;
1029 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
1030 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1031 void (*yield_task) (struct rq *rq);
1033 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
1035 struct task_struct * (*pick_next_task) (struct rq *rq);
1036 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1039 int (*select_task_rq)(struct task_struct *p, int sync);
1041 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1042 struct rq *busiest, unsigned long max_load_move,
1043 struct sched_domain *sd, enum cpu_idle_type idle,
1044 int *all_pinned, int *this_best_prio);
1046 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1047 struct rq *busiest, struct sched_domain *sd,
1048 enum cpu_idle_type idle);
1049 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1050 void (*post_schedule) (struct rq *this_rq);
1051 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1053 void (*set_cpus_allowed)(struct task_struct *p,
1054 const struct cpumask *newmask);
1056 void (*rq_online)(struct rq *rq);
1057 void (*rq_offline)(struct rq *rq);
1060 void (*set_curr_task) (struct rq *rq);
1061 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1062 void (*task_new) (struct rq *rq, struct task_struct *p);
1064 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1066 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1068 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1069 int oldprio, int running);
1071 #ifdef CONFIG_FAIR_GROUP_SCHED
1072 void (*moved_group) (struct task_struct *p);
1076 struct load_weight {
1077 unsigned long weight, inv_weight;
1081 * CFS stats for a schedulable entity (task, task-group etc)
1083 * Current field usage histogram:
1090 struct sched_entity {
1091 struct load_weight load; /* for load-balancing */
1092 struct rb_node run_node;
1093 struct list_head group_node;
1097 u64 sum_exec_runtime;
1099 u64 prev_sum_exec_runtime;
1109 #ifdef CONFIG_SCHEDSTATS
1119 s64 sum_sleep_runtime;
1126 u64 nr_migrations_cold;
1127 u64 nr_failed_migrations_affine;
1128 u64 nr_failed_migrations_running;
1129 u64 nr_failed_migrations_hot;
1130 u64 nr_forced_migrations;
1131 u64 nr_forced2_migrations;
1134 u64 nr_wakeups_sync;
1135 u64 nr_wakeups_migrate;
1136 u64 nr_wakeups_local;
1137 u64 nr_wakeups_remote;
1138 u64 nr_wakeups_affine;
1139 u64 nr_wakeups_affine_attempts;
1140 u64 nr_wakeups_passive;
1141 u64 nr_wakeups_idle;
1144 #ifdef CONFIG_FAIR_GROUP_SCHED
1145 struct sched_entity *parent;
1146 /* rq on which this entity is (to be) queued: */
1147 struct cfs_rq *cfs_rq;
1148 /* rq "owned" by this entity/group: */
1149 struct cfs_rq *my_q;
1153 struct sched_rt_entity {
1154 struct list_head run_list;
1155 unsigned long timeout;
1156 unsigned int time_slice;
1157 int nr_cpus_allowed;
1159 struct sched_rt_entity *back;
1160 #ifdef CONFIG_RT_GROUP_SCHED
1161 struct sched_rt_entity *parent;
1162 /* rq on which this entity is (to be) queued: */
1163 struct rt_rq *rt_rq;
1164 /* rq "owned" by this entity/group: */
1169 struct task_struct {
1170 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1173 unsigned int flags; /* per process flags, defined below */
1174 unsigned int ptrace;
1176 int lock_depth; /* BKL lock depth */
1179 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1184 int prio, static_prio, normal_prio;
1185 unsigned int rt_priority;
1186 const struct sched_class *sched_class;
1187 struct sched_entity se;
1188 struct sched_rt_entity rt;
1190 #ifdef CONFIG_PREEMPT_NOTIFIERS
1191 /* list of struct preempt_notifier: */
1192 struct hlist_head preempt_notifiers;
1196 * fpu_counter contains the number of consecutive context switches
1197 * that the FPU is used. If this is over a threshold, the lazy fpu
1198 * saving becomes unlazy to save the trap. This is an unsigned char
1199 * so that after 256 times the counter wraps and the behavior turns
1200 * lazy again; this to deal with bursty apps that only use FPU for
1203 unsigned char fpu_counter;
1204 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1205 #ifdef CONFIG_BLK_DEV_IO_TRACE
1206 unsigned int btrace_seq;
1209 unsigned int policy;
1210 cpumask_t cpus_allowed;
1212 #ifdef CONFIG_PREEMPT_RCU
1213 int rcu_read_lock_nesting;
1214 int rcu_flipctr_idx;
1215 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1217 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1218 struct sched_info sched_info;
1221 struct list_head tasks;
1222 struct plist_node pushable_tasks;
1224 struct mm_struct *mm, *active_mm;
1227 struct linux_binfmt *binfmt;
1229 int exit_code, exit_signal;
1230 int pdeath_signal; /* The signal sent when the parent dies */
1232 unsigned int personality;
1233 unsigned did_exec:1;
1234 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1236 unsigned in_iowait:1;
1239 /* Revert to default priority/policy when forking */
1240 unsigned sched_reset_on_fork:1;
1245 #ifdef CONFIG_CC_STACKPROTECTOR
1246 /* Canary value for the -fstack-protector gcc feature */
1247 unsigned long stack_canary;
1251 * pointers to (original) parent process, youngest child, younger sibling,
1252 * older sibling, respectively. (p->father can be replaced with
1253 * p->real_parent->pid)
1255 struct task_struct *real_parent; /* real parent process */
1256 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1258 * children/sibling forms the list of my natural children
1260 struct list_head children; /* list of my children */
1261 struct list_head sibling; /* linkage in my parent's children list */
1262 struct task_struct *group_leader; /* threadgroup leader */
1265 * ptraced is the list of tasks this task is using ptrace on.
1266 * This includes both natural children and PTRACE_ATTACH targets.
1267 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1269 struct list_head ptraced;
1270 struct list_head ptrace_entry;
1273 * This is the tracer handle for the ptrace BTS extension.
1274 * This field actually belongs to the ptracer task.
1276 struct bts_context *bts;
1278 /* PID/PID hash table linkage. */
1279 struct pid_link pids[PIDTYPE_MAX];
1280 struct list_head thread_group;
1282 struct completion *vfork_done; /* for vfork() */
1283 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1284 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1286 cputime_t utime, stime, utimescaled, stimescaled;
1288 cputime_t prev_utime, prev_stime;
1289 unsigned long nvcsw, nivcsw; /* context switch counts */
1290 struct timespec start_time; /* monotonic time */
1291 struct timespec real_start_time; /* boot based time */
1292 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1293 unsigned long min_flt, maj_flt;
1295 struct task_cputime cputime_expires;
1296 struct list_head cpu_timers[3];
1298 /* process credentials */
1299 const struct cred *real_cred; /* objective and real subjective task
1300 * credentials (COW) */
1301 const struct cred *cred; /* effective (overridable) subjective task
1302 * credentials (COW) */
1303 struct mutex cred_guard_mutex; /* guard against foreign influences on
1304 * credential calculations
1305 * (notably. ptrace) */
1307 char comm[TASK_COMM_LEN]; /* executable name excluding path
1308 - access with [gs]et_task_comm (which lock
1309 it with task_lock())
1310 - initialized normally by flush_old_exec */
1311 /* file system info */
1312 int link_count, total_link_count;
1313 #ifdef CONFIG_SYSVIPC
1315 struct sysv_sem sysvsem;
1317 #ifdef CONFIG_DETECT_HUNG_TASK
1318 /* hung task detection */
1319 unsigned long last_switch_count;
1321 /* CPU-specific state of this task */
1322 struct thread_struct thread;
1323 /* filesystem information */
1324 struct fs_struct *fs;
1325 /* open file information */
1326 struct files_struct *files;
1328 struct nsproxy *nsproxy;
1329 /* signal handlers */
1330 struct signal_struct *signal;
1331 struct sighand_struct *sighand;
1333 sigset_t blocked, real_blocked;
1334 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1335 struct sigpending pending;
1337 unsigned long sas_ss_sp;
1339 int (*notifier)(void *priv);
1340 void *notifier_data;
1341 sigset_t *notifier_mask;
1342 struct audit_context *audit_context;
1343 #ifdef CONFIG_AUDITSYSCALL
1345 unsigned int sessionid;
1349 /* Thread group tracking */
1352 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1354 spinlock_t alloc_lock;
1356 #ifdef CONFIG_GENERIC_HARDIRQS
1357 /* IRQ handler threads */
1358 struct irqaction *irqaction;
1361 /* Protection of the PI data structures: */
1364 #ifdef CONFIG_RT_MUTEXES
1365 /* PI waiters blocked on a rt_mutex held by this task */
1366 struct plist_head pi_waiters;
1367 /* Deadlock detection and priority inheritance handling */
1368 struct rt_mutex_waiter *pi_blocked_on;
1371 #ifdef CONFIG_DEBUG_MUTEXES
1372 /* mutex deadlock detection */
1373 struct mutex_waiter *blocked_on;
1375 #ifdef CONFIG_TRACE_IRQFLAGS
1376 unsigned int irq_events;
1377 int hardirqs_enabled;
1378 unsigned long hardirq_enable_ip;
1379 unsigned int hardirq_enable_event;
1380 unsigned long hardirq_disable_ip;
1381 unsigned int hardirq_disable_event;
1382 int softirqs_enabled;
1383 unsigned long softirq_disable_ip;
1384 unsigned int softirq_disable_event;
1385 unsigned long softirq_enable_ip;
1386 unsigned int softirq_enable_event;
1387 int hardirq_context;
1388 int softirq_context;
1390 #ifdef CONFIG_LOCKDEP
1391 # define MAX_LOCK_DEPTH 48UL
1394 unsigned int lockdep_recursion;
1395 struct held_lock held_locks[MAX_LOCK_DEPTH];
1396 gfp_t lockdep_reclaim_gfp;
1399 /* journalling filesystem info */
1402 /* stacked block device info */
1403 struct bio *bio_list, **bio_tail;
1406 struct reclaim_state *reclaim_state;
1408 struct backing_dev_info *backing_dev_info;
1410 struct io_context *io_context;
1412 unsigned long ptrace_message;
1413 siginfo_t *last_siginfo; /* For ptrace use. */
1414 struct task_io_accounting ioac;
1415 #if defined(CONFIG_TASK_XACCT)
1416 u64 acct_rss_mem1; /* accumulated rss usage */
1417 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1418 cputime_t acct_timexpd; /* stime + utime since last update */
1420 #ifdef CONFIG_CPUSETS
1421 nodemask_t mems_allowed; /* Protected by alloc_lock */
1422 int cpuset_mem_spread_rotor;
1424 #ifdef CONFIG_CGROUPS
1425 /* Control Group info protected by css_set_lock */
1426 struct css_set *cgroups;
1427 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1428 struct list_head cg_list;
1431 struct robust_list_head __user *robust_list;
1432 #ifdef CONFIG_COMPAT
1433 struct compat_robust_list_head __user *compat_robust_list;
1435 struct list_head pi_state_list;
1436 struct futex_pi_state *pi_state_cache;
1438 #ifdef CONFIG_PERF_COUNTERS
1439 struct perf_counter_context *perf_counter_ctxp;
1440 struct mutex perf_counter_mutex;
1441 struct list_head perf_counter_list;
1444 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1447 atomic_t fs_excl; /* holding fs exclusive resources */
1448 struct rcu_head rcu;
1451 * cache last used pipe for splice
1453 struct pipe_inode_info *splice_pipe;
1454 #ifdef CONFIG_TASK_DELAY_ACCT
1455 struct task_delay_info *delays;
1457 #ifdef CONFIG_FAULT_INJECTION
1460 struct prop_local_single dirties;
1461 #ifdef CONFIG_LATENCYTOP
1462 int latency_record_count;
1463 struct latency_record latency_record[LT_SAVECOUNT];
1466 * time slack values; these are used to round up poll() and
1467 * select() etc timeout values. These are in nanoseconds.
1469 unsigned long timer_slack_ns;
1470 unsigned long default_timer_slack_ns;
1472 struct list_head *scm_work_list;
1473 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1474 /* Index of current stored adress in ret_stack */
1476 /* Stack of return addresses for return function tracing */
1477 struct ftrace_ret_stack *ret_stack;
1478 /* time stamp for last schedule */
1479 unsigned long long ftrace_timestamp;
1481 * Number of functions that haven't been traced
1482 * because of depth overrun.
1484 atomic_t trace_overrun;
1485 /* Pause for the tracing */
1486 atomic_t tracing_graph_pause;
1488 #ifdef CONFIG_TRACING
1489 /* state flags for use by tracers */
1490 unsigned long trace;
1491 /* bitmask of trace recursion */
1492 unsigned long trace_recursion;
1493 #endif /* CONFIG_TRACING */
1496 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1497 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1500 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1501 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1502 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1503 * values are inverted: lower p->prio value means higher priority.
1505 * The MAX_USER_RT_PRIO value allows the actual maximum
1506 * RT priority to be separate from the value exported to
1507 * user-space. This allows kernel threads to set their
1508 * priority to a value higher than any user task. Note:
1509 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1512 #define MAX_USER_RT_PRIO 100
1513 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1515 #define MAX_PRIO (MAX_RT_PRIO + 40)
1516 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1518 static inline int rt_prio(int prio)
1520 if (unlikely(prio < MAX_RT_PRIO))
1525 static inline int rt_task(struct task_struct *p)
1527 return rt_prio(p->prio);
1530 static inline struct pid *task_pid(struct task_struct *task)
1532 return task->pids[PIDTYPE_PID].pid;
1535 static inline struct pid *task_tgid(struct task_struct *task)
1537 return task->group_leader->pids[PIDTYPE_PID].pid;
1541 * Without tasklist or rcu lock it is not safe to dereference
1542 * the result of task_pgrp/task_session even if task == current,
1543 * we can race with another thread doing sys_setsid/sys_setpgid.
1545 static inline struct pid *task_pgrp(struct task_struct *task)
1547 return task->group_leader->pids[PIDTYPE_PGID].pid;
1550 static inline struct pid *task_session(struct task_struct *task)
1552 return task->group_leader->pids[PIDTYPE_SID].pid;
1555 struct pid_namespace;
1558 * the helpers to get the task's different pids as they are seen
1559 * from various namespaces
1561 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1562 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1564 * task_xid_nr_ns() : id seen from the ns specified;
1566 * set_task_vxid() : assigns a virtual id to a task;
1568 * see also pid_nr() etc in include/linux/pid.h
1570 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1571 struct pid_namespace *ns);
1573 static inline pid_t task_pid_nr(struct task_struct *tsk)
1578 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1579 struct pid_namespace *ns)
1581 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1584 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1586 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1590 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1595 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1597 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1599 return pid_vnr(task_tgid(tsk));
1603 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1604 struct pid_namespace *ns)
1606 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1609 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1611 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1615 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1616 struct pid_namespace *ns)
1618 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1621 static inline pid_t task_session_vnr(struct task_struct *tsk)
1623 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1626 /* obsolete, do not use */
1627 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1629 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1633 * pid_alive - check that a task structure is not stale
1634 * @p: Task structure to be checked.
1636 * Test if a process is not yet dead (at most zombie state)
1637 * If pid_alive fails, then pointers within the task structure
1638 * can be stale and must not be dereferenced.
1640 static inline int pid_alive(struct task_struct *p)
1642 return p->pids[PIDTYPE_PID].pid != NULL;
1646 * is_global_init - check if a task structure is init
1647 * @tsk: Task structure to be checked.
1649 * Check if a task structure is the first user space task the kernel created.
1651 static inline int is_global_init(struct task_struct *tsk)
1653 return tsk->pid == 1;
1657 * is_container_init:
1658 * check whether in the task is init in its own pid namespace.
1660 extern int is_container_init(struct task_struct *tsk);
1662 extern struct pid *cad_pid;
1664 extern void free_task(struct task_struct *tsk);
1665 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1667 extern void __put_task_struct(struct task_struct *t);
1669 static inline void put_task_struct(struct task_struct *t)
1671 if (atomic_dec_and_test(&t->usage))
1672 __put_task_struct(t);
1675 extern cputime_t task_utime(struct task_struct *p);
1676 extern cputime_t task_stime(struct task_struct *p);
1677 extern cputime_t task_gtime(struct task_struct *p);
1682 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1683 /* Not implemented yet, only for 486*/
1684 #define PF_STARTING 0x00000002 /* being created */
1685 #define PF_EXITING 0x00000004 /* getting shut down */
1686 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1687 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1688 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1689 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1690 #define PF_DUMPCORE 0x00000200 /* dumped core */
1691 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1692 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1693 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1694 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1695 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1696 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1697 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1698 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1699 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1700 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1701 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1702 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1703 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1704 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1705 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1706 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1707 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1708 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1709 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1710 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1711 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1714 * Only the _current_ task can read/write to tsk->flags, but other
1715 * tasks can access tsk->flags in readonly mode for example
1716 * with tsk_used_math (like during threaded core dumping).
1717 * There is however an exception to this rule during ptrace
1718 * or during fork: the ptracer task is allowed to write to the
1719 * child->flags of its traced child (same goes for fork, the parent
1720 * can write to the child->flags), because we're guaranteed the
1721 * child is not running and in turn not changing child->flags
1722 * at the same time the parent does it.
1724 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1725 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1726 #define clear_used_math() clear_stopped_child_used_math(current)
1727 #define set_used_math() set_stopped_child_used_math(current)
1728 #define conditional_stopped_child_used_math(condition, child) \
1729 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1730 #define conditional_used_math(condition) \
1731 conditional_stopped_child_used_math(condition, current)
1732 #define copy_to_stopped_child_used_math(child) \
1733 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1734 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1735 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1736 #define used_math() tsk_used_math(current)
1739 extern int set_cpus_allowed_ptr(struct task_struct *p,
1740 const struct cpumask *new_mask);
1742 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1743 const struct cpumask *new_mask)
1745 if (!cpumask_test_cpu(0, new_mask))
1750 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1752 return set_cpus_allowed_ptr(p, &new_mask);
1756 * Architectures can set this to 1 if they have specified
1757 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1758 * but then during bootup it turns out that sched_clock()
1759 * is reliable after all:
1761 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1762 extern int sched_clock_stable;
1765 extern unsigned long long sched_clock(void);
1767 extern void sched_clock_init(void);
1768 extern u64 sched_clock_cpu(int cpu);
1770 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1771 static inline void sched_clock_tick(void)
1775 static inline void sched_clock_idle_sleep_event(void)
1779 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1783 extern void sched_clock_tick(void);
1784 extern void sched_clock_idle_sleep_event(void);
1785 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1789 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1790 * clock constructed from sched_clock():
1792 extern unsigned long long cpu_clock(int cpu);
1794 extern unsigned long long
1795 task_sched_runtime(struct task_struct *task);
1796 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1798 /* sched_exec is called by processes performing an exec */
1800 extern void sched_exec(void);
1802 #define sched_exec() {}
1805 extern void sched_clock_idle_sleep_event(void);
1806 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1808 #ifdef CONFIG_HOTPLUG_CPU
1809 extern void idle_task_exit(void);
1811 static inline void idle_task_exit(void) {}
1814 extern void sched_idle_next(void);
1816 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1817 extern void wake_up_idle_cpu(int cpu);
1819 static inline void wake_up_idle_cpu(int cpu) { }
1822 extern unsigned int sysctl_sched_latency;
1823 extern unsigned int sysctl_sched_min_granularity;
1824 extern unsigned int sysctl_sched_wakeup_granularity;
1825 extern unsigned int sysctl_sched_shares_ratelimit;
1826 extern unsigned int sysctl_sched_shares_thresh;
1827 #ifdef CONFIG_SCHED_DEBUG
1828 extern unsigned int sysctl_sched_child_runs_first;
1829 extern unsigned int sysctl_sched_features;
1830 extern unsigned int sysctl_sched_migration_cost;
1831 extern unsigned int sysctl_sched_nr_migrate;
1832 extern unsigned int sysctl_timer_migration;
1834 int sched_nr_latency_handler(struct ctl_table *table, int write,
1835 struct file *file, void __user *buffer, size_t *length,
1838 #ifdef CONFIG_SCHED_DEBUG
1839 static inline unsigned int get_sysctl_timer_migration(void)
1841 return sysctl_timer_migration;
1844 static inline unsigned int get_sysctl_timer_migration(void)
1849 extern unsigned int sysctl_sched_rt_period;
1850 extern int sysctl_sched_rt_runtime;
1852 int sched_rt_handler(struct ctl_table *table, int write,
1853 struct file *filp, void __user *buffer, size_t *lenp,
1856 extern unsigned int sysctl_sched_compat_yield;
1858 #ifdef CONFIG_RT_MUTEXES
1859 extern int rt_mutex_getprio(struct task_struct *p);
1860 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1861 extern void rt_mutex_adjust_pi(struct task_struct *p);
1863 static inline int rt_mutex_getprio(struct task_struct *p)
1865 return p->normal_prio;
1867 # define rt_mutex_adjust_pi(p) do { } while (0)
1870 extern void set_user_nice(struct task_struct *p, long nice);
1871 extern int task_prio(const struct task_struct *p);
1872 extern int task_nice(const struct task_struct *p);
1873 extern int can_nice(const struct task_struct *p, const int nice);
1874 extern int task_curr(const struct task_struct *p);
1875 extern int idle_cpu(int cpu);
1876 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1877 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1878 struct sched_param *);
1879 extern struct task_struct *idle_task(int cpu);
1880 extern struct task_struct *curr_task(int cpu);
1881 extern void set_curr_task(int cpu, struct task_struct *p);
1886 * The default (Linux) execution domain.
1888 extern struct exec_domain default_exec_domain;
1890 union thread_union {
1891 struct thread_info thread_info;
1892 unsigned long stack[THREAD_SIZE/sizeof(long)];
1895 #ifndef __HAVE_ARCH_KSTACK_END
1896 static inline int kstack_end(void *addr)
1898 /* Reliable end of stack detection:
1899 * Some APM bios versions misalign the stack
1901 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1905 extern union thread_union init_thread_union;
1906 extern struct task_struct init_task;
1908 extern struct mm_struct init_mm;
1910 extern struct pid_namespace init_pid_ns;
1913 * find a task by one of its numerical ids
1915 * find_task_by_pid_ns():
1916 * finds a task by its pid in the specified namespace
1917 * find_task_by_vpid():
1918 * finds a task by its virtual pid
1920 * see also find_vpid() etc in include/linux/pid.h
1923 extern struct task_struct *find_task_by_vpid(pid_t nr);
1924 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1925 struct pid_namespace *ns);
1927 extern void __set_special_pids(struct pid *pid);
1929 /* per-UID process charging. */
1930 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1931 static inline struct user_struct *get_uid(struct user_struct *u)
1933 atomic_inc(&u->__count);
1936 extern void free_uid(struct user_struct *);
1937 extern void release_uids(struct user_namespace *ns);
1939 #include <asm/current.h>
1941 extern void do_timer(unsigned long ticks);
1943 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1944 extern int wake_up_process(struct task_struct *tsk);
1945 extern void wake_up_new_task(struct task_struct *tsk,
1946 unsigned long clone_flags);
1948 extern void kick_process(struct task_struct *tsk);
1950 static inline void kick_process(struct task_struct *tsk) { }
1952 extern void sched_fork(struct task_struct *p, int clone_flags);
1953 extern void sched_dead(struct task_struct *p);
1955 extern void proc_caches_init(void);
1956 extern void flush_signals(struct task_struct *);
1957 extern void __flush_signals(struct task_struct *);
1958 extern void ignore_signals(struct task_struct *);
1959 extern void flush_signal_handlers(struct task_struct *, int force_default);
1960 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1962 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1964 unsigned long flags;
1967 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1968 ret = dequeue_signal(tsk, mask, info);
1969 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1974 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1976 extern void unblock_all_signals(void);
1977 extern void release_task(struct task_struct * p);
1978 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1979 extern int force_sigsegv(int, struct task_struct *);
1980 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1981 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1982 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1983 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1984 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1985 extern int kill_pid(struct pid *pid, int sig, int priv);
1986 extern int kill_proc_info(int, struct siginfo *, pid_t);
1987 extern int do_notify_parent(struct task_struct *, int);
1988 extern void force_sig(int, struct task_struct *);
1989 extern void force_sig_specific(int, struct task_struct *);
1990 extern int send_sig(int, struct task_struct *, int);
1991 extern void zap_other_threads(struct task_struct *p);
1992 extern struct sigqueue *sigqueue_alloc(void);
1993 extern void sigqueue_free(struct sigqueue *);
1994 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
1995 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1996 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1998 static inline int kill_cad_pid(int sig, int priv)
2000 return kill_pid(cad_pid, sig, priv);
2003 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2004 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2005 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2006 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2008 static inline int is_si_special(const struct siginfo *info)
2010 return info <= SEND_SIG_FORCED;
2013 /* True if we are on the alternate signal stack. */
2015 static inline int on_sig_stack(unsigned long sp)
2017 return (sp - current->sas_ss_sp < current->sas_ss_size);
2020 static inline int sas_ss_flags(unsigned long sp)
2022 return (current->sas_ss_size == 0 ? SS_DISABLE
2023 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2027 * Routines for handling mm_structs
2029 extern struct mm_struct * mm_alloc(void);
2031 /* mmdrop drops the mm and the page tables */
2032 extern void __mmdrop(struct mm_struct *);
2033 static inline void mmdrop(struct mm_struct * mm)
2035 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2039 /* mmput gets rid of the mappings and all user-space */
2040 extern void mmput(struct mm_struct *);
2041 /* Grab a reference to a task's mm, if it is not already going away */
2042 extern struct mm_struct *get_task_mm(struct task_struct *task);
2043 /* Remove the current tasks stale references to the old mm_struct */
2044 extern void mm_release(struct task_struct *, struct mm_struct *);
2045 /* Allocate a new mm structure and copy contents from tsk->mm */
2046 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2048 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2049 struct task_struct *, struct pt_regs *);
2050 extern void flush_thread(void);
2051 extern void exit_thread(void);
2053 extern void exit_files(struct task_struct *);
2054 extern void __cleanup_signal(struct signal_struct *);
2055 extern void __cleanup_sighand(struct sighand_struct *);
2057 extern void exit_itimers(struct signal_struct *);
2058 extern void flush_itimer_signals(void);
2060 extern NORET_TYPE void do_group_exit(int);
2062 extern void daemonize(const char *, ...);
2063 extern int allow_signal(int);
2064 extern int disallow_signal(int);
2066 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2067 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2068 struct task_struct *fork_idle(int);
2070 extern void set_task_comm(struct task_struct *tsk, char *from);
2071 extern char *get_task_comm(char *to, struct task_struct *tsk);
2074 extern void wait_task_context_switch(struct task_struct *p);
2075 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2077 static inline void wait_task_context_switch(struct task_struct *p) {}
2078 static inline unsigned long wait_task_inactive(struct task_struct *p,
2085 #define next_task(p) \
2086 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2088 #define for_each_process(p) \
2089 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2091 extern bool is_single_threaded(struct task_struct *);
2094 * Careful: do_each_thread/while_each_thread is a double loop so
2095 * 'break' will not work as expected - use goto instead.
2097 #define do_each_thread(g, t) \
2098 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2100 #define while_each_thread(g, t) \
2101 while ((t = next_thread(t)) != g)
2103 /* de_thread depends on thread_group_leader not being a pid based check */
2104 #define thread_group_leader(p) (p == p->group_leader)
2106 /* Do to the insanities of de_thread it is possible for a process
2107 * to have the pid of the thread group leader without actually being
2108 * the thread group leader. For iteration through the pids in proc
2109 * all we care about is that we have a task with the appropriate
2110 * pid, we don't actually care if we have the right task.
2112 static inline int has_group_leader_pid(struct task_struct *p)
2114 return p->pid == p->tgid;
2118 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2120 return p1->tgid == p2->tgid;
2123 static inline struct task_struct *next_thread(const struct task_struct *p)
2125 return list_entry_rcu(p->thread_group.next,
2126 struct task_struct, thread_group);
2129 static inline int thread_group_empty(struct task_struct *p)
2131 return list_empty(&p->thread_group);
2134 #define delay_group_leader(p) \
2135 (thread_group_leader(p) && !thread_group_empty(p))
2137 static inline int task_detached(struct task_struct *p)
2139 return p->exit_signal == -1;
2143 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2144 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2145 * pins the final release of task.io_context. Also protects ->cpuset and
2146 * ->cgroup.subsys[].
2148 * Nests both inside and outside of read_lock(&tasklist_lock).
2149 * It must not be nested with write_lock_irq(&tasklist_lock),
2150 * neither inside nor outside.
2152 static inline void task_lock(struct task_struct *p)
2154 spin_lock(&p->alloc_lock);
2157 static inline void task_unlock(struct task_struct *p)
2159 spin_unlock(&p->alloc_lock);
2162 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2163 unsigned long *flags);
2165 static inline void unlock_task_sighand(struct task_struct *tsk,
2166 unsigned long *flags)
2168 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2171 #ifndef __HAVE_THREAD_FUNCTIONS
2173 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2174 #define task_stack_page(task) ((task)->stack)
2176 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2178 *task_thread_info(p) = *task_thread_info(org);
2179 task_thread_info(p)->task = p;
2182 static inline unsigned long *end_of_stack(struct task_struct *p)
2184 return (unsigned long *)(task_thread_info(p) + 1);
2189 static inline int object_is_on_stack(void *obj)
2191 void *stack = task_stack_page(current);
2193 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2196 extern void thread_info_cache_init(void);
2198 #ifdef CONFIG_DEBUG_STACK_USAGE
2199 static inline unsigned long stack_not_used(struct task_struct *p)
2201 unsigned long *n = end_of_stack(p);
2203 do { /* Skip over canary */
2207 return (unsigned long)n - (unsigned long)end_of_stack(p);
2211 /* set thread flags in other task's structures
2212 * - see asm/thread_info.h for TIF_xxxx flags available
2214 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2216 set_ti_thread_flag(task_thread_info(tsk), flag);
2219 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2221 clear_ti_thread_flag(task_thread_info(tsk), flag);
2224 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2226 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2229 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2231 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2234 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2236 return test_ti_thread_flag(task_thread_info(tsk), flag);
2239 static inline void set_tsk_need_resched(struct task_struct *tsk)
2241 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2244 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2246 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2249 static inline int test_tsk_need_resched(struct task_struct *tsk)
2251 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2254 static inline int restart_syscall(void)
2256 set_tsk_thread_flag(current, TIF_SIGPENDING);
2257 return -ERESTARTNOINTR;
2260 static inline int signal_pending(struct task_struct *p)
2262 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2265 extern int __fatal_signal_pending(struct task_struct *p);
2267 static inline int fatal_signal_pending(struct task_struct *p)
2269 return signal_pending(p) && __fatal_signal_pending(p);
2272 static inline int signal_pending_state(long state, struct task_struct *p)
2274 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2276 if (!signal_pending(p))
2279 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2282 static inline int need_resched(void)
2284 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2288 * cond_resched() and cond_resched_lock(): latency reduction via
2289 * explicit rescheduling in places that are safe. The return
2290 * value indicates whether a reschedule was done in fact.
2291 * cond_resched_lock() will drop the spinlock before scheduling,
2292 * cond_resched_softirq() will enable bhs before scheduling.
2294 extern int _cond_resched(void);
2296 #define cond_resched() ({ \
2297 __might_sleep(__FILE__, __LINE__, 0); \
2301 extern int __cond_resched_lock(spinlock_t *lock);
2303 #ifdef CONFIG_PREEMPT
2304 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2306 #define PREEMPT_LOCK_OFFSET 0
2309 #define cond_resched_lock(lock) ({ \
2310 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2311 __cond_resched_lock(lock); \
2314 extern int __cond_resched_softirq(void);
2316 #define cond_resched_softirq() ({ \
2317 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2318 __cond_resched_softirq(); \
2322 * Does a critical section need to be broken due to another
2323 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2324 * but a general need for low latency)
2326 static inline int spin_needbreak(spinlock_t *lock)
2328 #ifdef CONFIG_PREEMPT
2329 return spin_is_contended(lock);
2336 * Thread group CPU time accounting.
2338 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2339 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2341 static inline void thread_group_cputime_init(struct signal_struct *sig)
2343 sig->cputimer.cputime = INIT_CPUTIME;
2344 spin_lock_init(&sig->cputimer.lock);
2345 sig->cputimer.running = 0;
2348 static inline void thread_group_cputime_free(struct signal_struct *sig)
2353 * Reevaluate whether the task has signals pending delivery.
2354 * Wake the task if so.
2355 * This is required every time the blocked sigset_t changes.
2356 * callers must hold sighand->siglock.
2358 extern void recalc_sigpending_and_wake(struct task_struct *t);
2359 extern void recalc_sigpending(void);
2361 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2364 * Wrappers for p->thread_info->cpu access. No-op on UP.
2368 static inline unsigned int task_cpu(const struct task_struct *p)
2370 return task_thread_info(p)->cpu;
2373 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2377 static inline unsigned int task_cpu(const struct task_struct *p)
2382 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2386 #endif /* CONFIG_SMP */
2388 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2390 #ifdef CONFIG_TRACING
2392 __trace_special(void *__tr, void *__data,
2393 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2396 __trace_special(void *__tr, void *__data,
2397 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2402 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2403 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2405 extern void normalize_rt_tasks(void);
2407 #ifdef CONFIG_GROUP_SCHED
2409 extern struct task_group init_task_group;
2410 #ifdef CONFIG_USER_SCHED
2411 extern struct task_group root_task_group;
2412 extern void set_tg_uid(struct user_struct *user);
2415 extern struct task_group *sched_create_group(struct task_group *parent);
2416 extern void sched_destroy_group(struct task_group *tg);
2417 extern void sched_move_task(struct task_struct *tsk);
2418 #ifdef CONFIG_FAIR_GROUP_SCHED
2419 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2420 extern unsigned long sched_group_shares(struct task_group *tg);
2422 #ifdef CONFIG_RT_GROUP_SCHED
2423 extern int sched_group_set_rt_runtime(struct task_group *tg,
2424 long rt_runtime_us);
2425 extern long sched_group_rt_runtime(struct task_group *tg);
2426 extern int sched_group_set_rt_period(struct task_group *tg,
2428 extern long sched_group_rt_period(struct task_group *tg);
2429 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2433 extern int task_can_switch_user(struct user_struct *up,
2434 struct task_struct *tsk);
2436 #ifdef CONFIG_TASK_XACCT
2437 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2439 tsk->ioac.rchar += amt;
2442 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2444 tsk->ioac.wchar += amt;
2447 static inline void inc_syscr(struct task_struct *tsk)
2452 static inline void inc_syscw(struct task_struct *tsk)
2457 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2461 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2465 static inline void inc_syscr(struct task_struct *tsk)
2469 static inline void inc_syscw(struct task_struct *tsk)
2474 #ifndef TASK_SIZE_OF
2475 #define TASK_SIZE_OF(tsk) TASK_SIZE
2479 * Call the function if the target task is executing on a CPU right now:
2481 extern void task_oncpu_function_call(struct task_struct *p,
2482 void (*func) (void *info), void *info);
2485 #ifdef CONFIG_MM_OWNER
2486 extern void mm_update_next_owner(struct mm_struct *mm);
2487 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2489 static inline void mm_update_next_owner(struct mm_struct *mm)
2493 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2496 #endif /* CONFIG_MM_OWNER */
2498 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2500 #endif /* __KERNEL__ */