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;
102 struct perf_event_context;
105 * List of flags we want to share for kernel threads,
106 * if only because they are not used by them anyway.
108 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
111 * These are the constant used to fake the fixed-point load-average
112 * counting. Some notes:
113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
114 * a load-average precision of 10 bits integer + 11 bits fractional
115 * - if you want to count load-averages more often, you need more
116 * precision, or rounding will get you. With 2-second counting freq,
117 * the EXP_n values would be 1981, 2034 and 2043 if still using only
120 extern unsigned long avenrun[]; /* Load averages */
121 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
123 #define FSHIFT 11 /* nr of bits of precision */
124 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
125 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
126 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
127 #define EXP_5 2014 /* 1/exp(5sec/5min) */
128 #define EXP_15 2037 /* 1/exp(5sec/15min) */
130 #define CALC_LOAD(load,exp,n) \
132 load += n*(FIXED_1-exp); \
135 extern unsigned long total_forks;
136 extern int nr_threads;
137 DECLARE_PER_CPU(unsigned long, process_counts);
138 extern int nr_processes(void);
139 extern unsigned long nr_running(void);
140 extern unsigned long nr_uninterruptible(void);
141 extern unsigned long nr_iowait(void);
142 extern unsigned long nr_iowait_cpu(int cpu);
143 extern unsigned long this_cpu_load(void);
146 extern void calc_global_load(unsigned long ticks);
148 extern unsigned long get_parent_ip(unsigned long addr);
153 #ifdef CONFIG_SCHED_DEBUG
154 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
155 extern void proc_sched_set_task(struct task_struct *p);
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
160 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
163 static inline void proc_sched_set_task(struct task_struct *p)
167 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
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
193 #define TASK_WAKING 256
194 #define TASK_STATE_MAX 512
196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
198 extern char ___assert_task_state[1 - 2*!!(
199 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
201 /* Convenience macros for the sake of set_task_state */
202 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
203 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
204 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
206 /* Convenience macros for the sake of wake_up */
207 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
208 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
210 /* get_task_state() */
211 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
215 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
216 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
217 #define task_is_dead(task) ((task)->exit_state != 0)
218 #define task_is_stopped_or_traced(task) \
219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
220 #define task_contributes_to_load(task) \
221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
222 (task->flags & PF_FREEZING) == 0)
224 #define __set_task_state(tsk, state_value) \
225 do { (tsk)->state = (state_value); } while (0)
226 #define set_task_state(tsk, state_value) \
227 set_mb((tsk)->state, (state_value))
230 * set_current_state() includes a barrier so that the write of current->state
231 * is correctly serialised wrt the caller's subsequent test of whether to
234 * set_current_state(TASK_UNINTERRUPTIBLE);
235 * if (do_i_need_to_sleep())
238 * If the caller does not need such serialisation then use __set_current_state()
240 #define __set_current_state(state_value) \
241 do { current->state = (state_value); } while (0)
242 #define set_current_state(state_value) \
243 set_mb(current->state, (state_value))
245 /* Task command name length */
246 #define TASK_COMM_LEN 16
248 #include <linux/spinlock.h>
251 * This serializes "schedule()" and also protects
252 * the run-queue from deletions/modifications (but
253 * _adding_ to the beginning of the run-queue has
256 extern rwlock_t tasklist_lock;
257 extern spinlock_t mmlist_lock;
261 #ifdef CONFIG_PROVE_RCU
262 extern int lockdep_tasklist_lock_is_held(void);
263 #endif /* #ifdef CONFIG_PROVE_RCU */
265 extern void sched_init(void);
266 extern void sched_init_smp(void);
267 extern asmlinkage void schedule_tail(struct task_struct *prev);
268 extern void init_idle(struct task_struct *idle, int cpu);
269 extern void init_idle_bootup_task(struct task_struct *idle);
271 extern int runqueue_is_locked(int cpu);
273 extern cpumask_var_t nohz_cpu_mask;
274 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
275 extern void select_nohz_load_balancer(int stop_tick);
276 extern int get_nohz_timer_target(void);
278 static inline void select_nohz_load_balancer(int stop_tick) { }
282 * Only dump TASK_* tasks. (0 for all tasks)
284 extern void show_state_filter(unsigned long state_filter);
286 static inline void show_state(void)
288 show_state_filter(0);
291 extern void show_regs(struct pt_regs *);
294 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
295 * task), SP is the stack pointer of the first frame that should be shown in the back
296 * trace (or NULL if the entire call-chain of the task should be shown).
298 extern void show_stack(struct task_struct *task, unsigned long *sp);
300 void io_schedule(void);
301 long io_schedule_timeout(long timeout);
303 extern void cpu_init (void);
304 extern void trap_init(void);
305 extern void update_process_times(int user);
306 extern void scheduler_tick(void);
308 extern void sched_show_task(struct task_struct *p);
310 #ifdef CONFIG_LOCKUP_DETECTOR
311 extern void touch_softlockup_watchdog(void);
312 extern void touch_softlockup_watchdog_sync(void);
313 extern void touch_all_softlockup_watchdogs(void);
314 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
316 size_t *lenp, loff_t *ppos);
317 extern unsigned int softlockup_panic;
318 extern int softlockup_thresh;
320 static inline void touch_softlockup_watchdog(void)
323 static inline void touch_softlockup_watchdog_sync(void)
326 static inline void touch_all_softlockup_watchdogs(void)
331 #ifdef CONFIG_DETECT_HUNG_TASK
332 extern unsigned int sysctl_hung_task_panic;
333 extern unsigned long sysctl_hung_task_check_count;
334 extern unsigned long sysctl_hung_task_timeout_secs;
335 extern unsigned long sysctl_hung_task_warnings;
336 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
338 size_t *lenp, loff_t *ppos);
340 /* Avoid need for ifdefs elsewhere in the code */
341 enum { sysctl_hung_task_timeout_secs = 0 };
344 /* Attach to any functions which should be ignored in wchan output. */
345 #define __sched __attribute__((__section__(".sched.text")))
347 /* Linker adds these: start and end of __sched functions */
348 extern char __sched_text_start[], __sched_text_end[];
350 /* Is this address in the __sched functions? */
351 extern int in_sched_functions(unsigned long addr);
353 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
354 extern signed long schedule_timeout(signed long timeout);
355 extern signed long schedule_timeout_interruptible(signed long timeout);
356 extern signed long schedule_timeout_killable(signed long timeout);
357 extern signed long schedule_timeout_uninterruptible(signed long timeout);
358 asmlinkage void schedule(void);
359 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
362 struct user_namespace;
365 * Default maximum number of active map areas, this limits the number of vmas
366 * per mm struct. Users can overwrite this number by sysctl but there is a
369 * When a program's coredump is generated as ELF format, a section is created
370 * per a vma. In ELF, the number of sections is represented in unsigned short.
371 * This means the number of sections should be smaller than 65535 at coredump.
372 * Because the kernel adds some informative sections to a image of program at
373 * generating coredump, we need some margin. The number of extra sections is
374 * 1-3 now and depends on arch. We use "5" as safe margin, here.
376 #define MAPCOUNT_ELF_CORE_MARGIN (5)
377 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
379 extern int sysctl_max_map_count;
381 #include <linux/aio.h>
384 extern void arch_pick_mmap_layout(struct mm_struct *mm);
386 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
387 unsigned long, unsigned long);
389 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
390 unsigned long len, unsigned long pgoff,
391 unsigned long flags);
392 extern void arch_unmap_area(struct mm_struct *, unsigned long);
393 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
395 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399 extern void set_dumpable(struct mm_struct *mm, int value);
400 extern int get_dumpable(struct mm_struct *mm);
404 #define MMF_DUMPABLE 0 /* core dump is permitted */
405 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
407 #define MMF_DUMPABLE_BITS 2
408 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
410 /* coredump filter bits */
411 #define MMF_DUMP_ANON_PRIVATE 2
412 #define MMF_DUMP_ANON_SHARED 3
413 #define MMF_DUMP_MAPPED_PRIVATE 4
414 #define MMF_DUMP_MAPPED_SHARED 5
415 #define MMF_DUMP_ELF_HEADERS 6
416 #define MMF_DUMP_HUGETLB_PRIVATE 7
417 #define MMF_DUMP_HUGETLB_SHARED 8
419 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
420 #define MMF_DUMP_FILTER_BITS 7
421 #define MMF_DUMP_FILTER_MASK \
422 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
423 #define MMF_DUMP_FILTER_DEFAULT \
424 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
425 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
427 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
428 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
430 # define MMF_DUMP_MASK_DEFAULT_ELF 0
432 /* leave room for more dump flags */
433 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
435 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
437 struct sighand_struct {
439 struct k_sigaction action[_NSIG];
441 wait_queue_head_t signalfd_wqh;
444 struct pacct_struct {
447 unsigned long ac_mem;
448 cputime_t ac_utime, ac_stime;
449 unsigned long ac_minflt, ac_majflt;
460 * struct task_cputime - collected CPU time counts
461 * @utime: time spent in user mode, in &cputime_t units
462 * @stime: time spent in kernel mode, in &cputime_t units
463 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
465 * This structure groups together three kinds of CPU time that are
466 * tracked for threads and thread groups. Most things considering
467 * CPU time want to group these counts together and treat all three
468 * of them in parallel.
470 struct task_cputime {
473 unsigned long long sum_exec_runtime;
475 /* Alternate field names when used to cache expirations. */
476 #define prof_exp stime
477 #define virt_exp utime
478 #define sched_exp sum_exec_runtime
480 #define INIT_CPUTIME \
481 (struct task_cputime) { \
482 .utime = cputime_zero, \
483 .stime = cputime_zero, \
484 .sum_exec_runtime = 0, \
488 * Disable preemption until the scheduler is running.
489 * Reset by start_kernel()->sched_init()->init_idle().
491 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
492 * before the scheduler is active -- see should_resched().
494 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
497 * struct thread_group_cputimer - thread group interval timer counts
498 * @cputime: thread group interval timers.
499 * @running: non-zero when there are timers running and
500 * @cputime receives updates.
501 * @lock: lock for fields in this struct.
503 * This structure contains the version of task_cputime, above, that is
504 * used for thread group CPU timer calculations.
506 struct thread_group_cputimer {
507 struct task_cputime cputime;
515 * NOTE! "signal_struct" does not have it's own
516 * locking, because a shared signal_struct always
517 * implies a shared sighand_struct, so locking
518 * sighand_struct is always a proper superset of
519 * the locking of signal_struct.
521 struct signal_struct {
526 wait_queue_head_t wait_chldexit; /* for wait4() */
528 /* current thread group signal load-balancing target: */
529 struct task_struct *curr_target;
531 /* shared signal handling: */
532 struct sigpending shared_pending;
534 /* thread group exit support */
537 * - notify group_exit_task when ->count is equal to notify_count
538 * - everyone except group_exit_task is stopped during signal delivery
539 * of fatal signals, group_exit_task processes the signal.
542 struct task_struct *group_exit_task;
544 /* thread group stop support, overloads group_exit_code too */
545 int group_stop_count;
546 unsigned int flags; /* see SIGNAL_* flags below */
548 /* POSIX.1b Interval Timers */
549 struct list_head posix_timers;
551 /* ITIMER_REAL timer for the process */
552 struct hrtimer real_timer;
553 struct pid *leader_pid;
554 ktime_t it_real_incr;
557 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
558 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
559 * values are defined to 0 and 1 respectively
561 struct cpu_itimer it[2];
564 * Thread group totals for process CPU timers.
565 * See thread_group_cputimer(), et al, for details.
567 struct thread_group_cputimer cputimer;
569 /* Earliest-expiration cache. */
570 struct task_cputime cputime_expires;
572 struct list_head cpu_timers[3];
574 struct pid *tty_old_pgrp;
576 /* boolean value for session group leader */
579 struct tty_struct *tty; /* NULL if no tty */
581 #ifdef CONFIG_SCHED_AUTOGROUP
582 struct autogroup *autogroup;
585 * Cumulative resource counters for dead threads in the group,
586 * and for reaped dead child processes forked by this group.
587 * Live threads maintain their own counters and add to these
588 * in __exit_signal, except for the group leader.
590 cputime_t utime, stime, cutime, cstime;
593 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
594 cputime_t prev_utime, prev_stime;
596 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
597 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
598 unsigned long inblock, oublock, cinblock, coublock;
599 unsigned long maxrss, cmaxrss;
600 struct task_io_accounting ioac;
603 * Cumulative ns of schedule CPU time fo dead threads in the
604 * group, not including a zombie group leader, (This only differs
605 * from jiffies_to_ns(utime + stime) if sched_clock uses something
606 * other than jiffies.)
608 unsigned long long sum_sched_runtime;
611 * We don't bother to synchronize most readers of this at all,
612 * because there is no reader checking a limit that actually needs
613 * to get both rlim_cur and rlim_max atomically, and either one
614 * alone is a single word that can safely be read normally.
615 * getrlimit/setrlimit use task_lock(current->group_leader) to
616 * protect this instead of the siglock, because they really
617 * have no need to disable irqs.
619 struct rlimit rlim[RLIM_NLIMITS];
621 #ifdef CONFIG_BSD_PROCESS_ACCT
622 struct pacct_struct pacct; /* per-process accounting information */
624 #ifdef CONFIG_TASKSTATS
625 struct taskstats *stats;
629 struct tty_audit_buf *tty_audit_buf;
632 int oom_adj; /* OOM kill score adjustment (bit shift) */
633 int oom_score_adj; /* OOM kill score adjustment */
635 struct mutex cred_guard_mutex; /* guard against foreign influences on
636 * credential calculations
637 * (notably. ptrace) */
640 /* Context switch must be unlocked if interrupts are to be enabled */
641 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
642 # define __ARCH_WANT_UNLOCKED_CTXSW
646 * Bits in flags field of signal_struct.
648 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
649 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
650 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
651 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
653 * Pending notifications to parent.
655 #define SIGNAL_CLD_STOPPED 0x00000010
656 #define SIGNAL_CLD_CONTINUED 0x00000020
657 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
659 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
661 /* If true, all threads except ->group_exit_task have pending SIGKILL */
662 static inline int signal_group_exit(const struct signal_struct *sig)
664 return (sig->flags & SIGNAL_GROUP_EXIT) ||
665 (sig->group_exit_task != NULL);
669 * Some day this will be a full-fledged user tracking system..
672 atomic_t __count; /* reference count */
673 atomic_t processes; /* How many processes does this user have? */
674 atomic_t files; /* How many open files does this user have? */
675 atomic_t sigpending; /* How many pending signals does this user have? */
676 #ifdef CONFIG_INOTIFY_USER
677 atomic_t inotify_watches; /* How many inotify watches does this user have? */
678 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
680 #ifdef CONFIG_FANOTIFY
681 atomic_t fanotify_listeners;
684 atomic_t epoll_watches; /* The number of file descriptors currently watched */
686 #ifdef CONFIG_POSIX_MQUEUE
687 /* protected by mq_lock */
688 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
690 unsigned long locked_shm; /* How many pages of mlocked shm ? */
693 struct key *uid_keyring; /* UID specific keyring */
694 struct key *session_keyring; /* UID's default session keyring */
697 /* Hash table maintenance information */
698 struct hlist_node uidhash_node;
700 struct user_namespace *user_ns;
702 #ifdef CONFIG_PERF_EVENTS
703 atomic_long_t locked_vm;
707 extern int uids_sysfs_init(void);
709 extern struct user_struct *find_user(uid_t);
711 extern struct user_struct root_user;
712 #define INIT_USER (&root_user)
715 struct backing_dev_info;
716 struct reclaim_state;
718 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
720 /* cumulative counters */
721 unsigned long pcount; /* # of times run on this cpu */
722 unsigned long long run_delay; /* time spent waiting on a runqueue */
725 unsigned long long last_arrival,/* when we last ran on a cpu */
726 last_queued; /* when we were last queued to run */
727 #ifdef CONFIG_SCHEDSTATS
729 unsigned int bkl_count;
732 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
734 #ifdef CONFIG_TASK_DELAY_ACCT
735 struct task_delay_info {
737 unsigned int flags; /* Private per-task flags */
739 /* For each stat XXX, add following, aligned appropriately
741 * struct timespec XXX_start, XXX_end;
745 * Atomicity of updates to XXX_delay, XXX_count protected by
746 * single lock above (split into XXX_lock if contention is an issue).
750 * XXX_count is incremented on every XXX operation, the delay
751 * associated with the operation is added to XXX_delay.
752 * XXX_delay contains the accumulated delay time in nanoseconds.
754 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
755 u64 blkio_delay; /* wait for sync block io completion */
756 u64 swapin_delay; /* wait for swapin block io completion */
757 u32 blkio_count; /* total count of the number of sync block */
758 /* io operations performed */
759 u32 swapin_count; /* total count of the number of swapin block */
760 /* io operations performed */
762 struct timespec freepages_start, freepages_end;
763 u64 freepages_delay; /* wait for memory reclaim */
764 u32 freepages_count; /* total count of memory reclaim */
766 #endif /* CONFIG_TASK_DELAY_ACCT */
768 static inline int sched_info_on(void)
770 #ifdef CONFIG_SCHEDSTATS
772 #elif defined(CONFIG_TASK_DELAY_ACCT)
773 extern int delayacct_on;
788 * sched-domains (multiprocessor balancing) declarations:
792 * Increase resolution of nice-level calculations:
794 #define SCHED_LOAD_SHIFT 10
795 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
797 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
800 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
801 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
802 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
803 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
804 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
805 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
806 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
807 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
808 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
809 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
810 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
811 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
812 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
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;
832 if (!sched_mc_power_savings)
833 return SD_PREFER_SIBLING;
838 static inline int sd_balance_for_package_power(void)
840 if (sched_mc_power_savings | sched_smt_power_savings)
841 return SD_POWERSAVINGS_BALANCE;
843 return SD_PREFER_SIBLING;
846 extern int __weak arch_sd_sibiling_asym_packing(void);
849 * Optimise SD flags for power savings:
850 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
851 * Keep default SD flags if sched_{smt,mc}_power_saving=0
854 static inline int sd_power_saving_flags(void)
856 if (sched_mc_power_savings | sched_smt_power_savings)
857 return SD_BALANCE_NEWIDLE;
863 struct sched_group *next; /* Must be a circular list */
866 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
869 unsigned int cpu_power, cpu_power_orig;
870 unsigned int group_weight;
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 {
901 struct sched_domain_attr {
902 int relax_domain_level;
905 #define SD_ATTR_INIT (struct sched_domain_attr) { \
906 .relax_domain_level = -1, \
909 struct sched_domain {
910 /* These fields must be setup */
911 struct sched_domain *parent; /* top domain must be null terminated */
912 struct sched_domain *child; /* bottom domain must be null terminated */
913 struct sched_group *groups; /* the balancing groups of the domain */
914 unsigned long min_interval; /* Minimum balance interval ms */
915 unsigned long max_interval; /* Maximum balance interval ms */
916 unsigned int busy_factor; /* less balancing by factor if busy */
917 unsigned int imbalance_pct; /* No balance until over watermark */
918 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
919 unsigned int busy_idx;
920 unsigned int idle_idx;
921 unsigned int newidle_idx;
922 unsigned int wake_idx;
923 unsigned int forkexec_idx;
924 unsigned int smt_gain;
925 int flags; /* See SD_* */
926 enum sched_domain_level level;
928 /* Runtime fields. */
929 unsigned long last_balance; /* init to jiffies. units in jiffies */
930 unsigned int balance_interval; /* initialise to 1. units in ms. */
931 unsigned int nr_balance_failed; /* initialise to 0 */
935 #ifdef CONFIG_SCHEDSTATS
936 /* load_balance() stats */
937 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
938 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
939 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
940 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
946 /* Active load balancing */
947 unsigned int alb_count;
948 unsigned int alb_failed;
949 unsigned int alb_pushed;
951 /* SD_BALANCE_EXEC stats */
952 unsigned int sbe_count;
953 unsigned int sbe_balanced;
954 unsigned int sbe_pushed;
956 /* SD_BALANCE_FORK stats */
957 unsigned int sbf_count;
958 unsigned int sbf_balanced;
959 unsigned int sbf_pushed;
961 /* try_to_wake_up() stats */
962 unsigned int ttwu_wake_remote;
963 unsigned int ttwu_move_affine;
964 unsigned int ttwu_move_balance;
966 #ifdef CONFIG_SCHED_DEBUG
970 unsigned int span_weight;
972 * Span of all CPUs in this domain.
974 * NOTE: this field is variable length. (Allocated dynamically
975 * by attaching extra space to the end of the structure,
976 * depending on how many CPUs the kernel has booted up with)
978 * It is also be embedded into static data structures at build
979 * time. (See 'struct static_sched_domain' in kernel/sched.c)
981 unsigned long span[0];
984 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
986 return to_cpumask(sd->span);
989 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
990 struct sched_domain_attr *dattr_new);
992 /* Allocate an array of sched domains, for partition_sched_domains(). */
993 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
994 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
996 /* Test a flag in parent sched domain */
997 static inline int test_sd_parent(struct sched_domain *sd, int flag)
999 if (sd->parent && (sd->parent->flags & flag))
1005 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1006 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1008 #else /* CONFIG_SMP */
1010 struct sched_domain_attr;
1013 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1014 struct sched_domain_attr *dattr_new)
1017 #endif /* !CONFIG_SMP */
1020 struct io_context; /* See blkdev.h */
1023 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1024 extern void prefetch_stack(struct task_struct *t);
1026 static inline void prefetch_stack(struct task_struct *t) { }
1029 struct audit_context; /* See audit.c */
1031 struct pipe_inode_info;
1032 struct uts_namespace;
1035 struct sched_domain;
1040 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1041 #define WF_FORK 0x02 /* child wakeup after fork */
1043 #define ENQUEUE_WAKEUP 1
1044 #define ENQUEUE_WAKING 2
1045 #define ENQUEUE_HEAD 4
1047 #define DEQUEUE_SLEEP 1
1049 struct sched_class {
1050 const struct sched_class *next;
1052 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1053 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1054 void (*yield_task) (struct rq *rq);
1056 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1058 struct task_struct * (*pick_next_task) (struct rq *rq);
1059 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1062 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1063 int sd_flag, int flags);
1065 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1066 void (*post_schedule) (struct rq *this_rq);
1067 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1068 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1070 void (*set_cpus_allowed)(struct task_struct *p,
1071 const struct cpumask *newmask);
1073 void (*rq_online)(struct rq *rq);
1074 void (*rq_offline)(struct rq *rq);
1077 void (*set_curr_task) (struct rq *rq);
1078 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1079 void (*task_fork) (struct task_struct *p);
1081 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1083 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1085 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1086 int oldprio, int running);
1088 unsigned int (*get_rr_interval) (struct rq *rq,
1089 struct task_struct *task);
1091 #ifdef CONFIG_FAIR_GROUP_SCHED
1092 void (*task_move_group) (struct task_struct *p, int on_rq);
1096 struct load_weight {
1097 unsigned long weight, inv_weight;
1100 #ifdef CONFIG_SCHEDSTATS
1101 struct sched_statistics {
1111 s64 sum_sleep_runtime;
1118 u64 nr_migrations_cold;
1119 u64 nr_failed_migrations_affine;
1120 u64 nr_failed_migrations_running;
1121 u64 nr_failed_migrations_hot;
1122 u64 nr_forced_migrations;
1125 u64 nr_wakeups_sync;
1126 u64 nr_wakeups_migrate;
1127 u64 nr_wakeups_local;
1128 u64 nr_wakeups_remote;
1129 u64 nr_wakeups_affine;
1130 u64 nr_wakeups_affine_attempts;
1131 u64 nr_wakeups_passive;
1132 u64 nr_wakeups_idle;
1136 struct sched_entity {
1137 struct load_weight load; /* for load-balancing */
1138 struct rb_node run_node;
1139 struct list_head group_node;
1143 u64 sum_exec_runtime;
1145 u64 prev_sum_exec_runtime;
1149 #ifdef CONFIG_SCHEDSTATS
1150 struct sched_statistics statistics;
1153 #ifdef CONFIG_FAIR_GROUP_SCHED
1154 struct sched_entity *parent;
1155 /* rq on which this entity is (to be) queued: */
1156 struct cfs_rq *cfs_rq;
1157 /* rq "owned" by this entity/group: */
1158 struct cfs_rq *my_q;
1162 struct sched_rt_entity {
1163 struct list_head run_list;
1164 unsigned long timeout;
1165 unsigned int time_slice;
1166 int nr_cpus_allowed;
1168 struct sched_rt_entity *back;
1169 #ifdef CONFIG_RT_GROUP_SCHED
1170 struct sched_rt_entity *parent;
1171 /* rq on which this entity is (to be) queued: */
1172 struct rt_rq *rt_rq;
1173 /* rq "owned" by this entity/group: */
1180 enum perf_event_task_context {
1181 perf_invalid_context = -1,
1182 perf_hw_context = 0,
1184 perf_nr_task_contexts,
1187 struct task_struct {
1188 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1191 unsigned int flags; /* per process flags, defined below */
1192 unsigned int ptrace;
1194 int lock_depth; /* BKL lock depth */
1197 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1202 int prio, static_prio, normal_prio;
1203 unsigned int rt_priority;
1204 const struct sched_class *sched_class;
1205 struct sched_entity se;
1206 struct sched_rt_entity rt;
1208 #ifdef CONFIG_PREEMPT_NOTIFIERS
1209 /* list of struct preempt_notifier: */
1210 struct hlist_head preempt_notifiers;
1214 * fpu_counter contains the number of consecutive context switches
1215 * that the FPU is used. If this is over a threshold, the lazy fpu
1216 * saving becomes unlazy to save the trap. This is an unsigned char
1217 * so that after 256 times the counter wraps and the behavior turns
1218 * lazy again; this to deal with bursty apps that only use FPU for
1221 unsigned char fpu_counter;
1222 #ifdef CONFIG_BLK_DEV_IO_TRACE
1223 unsigned int btrace_seq;
1226 unsigned int policy;
1227 cpumask_t cpus_allowed;
1229 #ifdef CONFIG_PREEMPT_RCU
1230 int rcu_read_lock_nesting;
1231 char rcu_read_unlock_special;
1232 struct list_head rcu_node_entry;
1233 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1234 #ifdef CONFIG_TREE_PREEMPT_RCU
1235 struct rcu_node *rcu_blocked_node;
1236 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1238 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1239 struct sched_info sched_info;
1242 struct list_head tasks;
1244 struct plist_node pushable_tasks;
1247 struct mm_struct *mm, *active_mm;
1248 #if defined(SPLIT_RSS_COUNTING)
1249 struct task_rss_stat rss_stat;
1253 int exit_code, exit_signal;
1254 int pdeath_signal; /* The signal sent when the parent dies */
1256 unsigned int personality;
1257 unsigned did_exec:1;
1258 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1260 unsigned in_iowait:1;
1263 /* Revert to default priority/policy when forking */
1264 unsigned sched_reset_on_fork:1;
1269 #ifdef CONFIG_CC_STACKPROTECTOR
1270 /* Canary value for the -fstack-protector gcc feature */
1271 unsigned long stack_canary;
1275 * pointers to (original) parent process, youngest child, younger sibling,
1276 * older sibling, respectively. (p->father can be replaced with
1277 * p->real_parent->pid)
1279 struct task_struct *real_parent; /* real parent process */
1280 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1282 * children/sibling forms the list of my natural children
1284 struct list_head children; /* list of my children */
1285 struct list_head sibling; /* linkage in my parent's children list */
1286 struct task_struct *group_leader; /* threadgroup leader */
1289 * ptraced is the list of tasks this task is using ptrace on.
1290 * This includes both natural children and PTRACE_ATTACH targets.
1291 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1293 struct list_head ptraced;
1294 struct list_head ptrace_entry;
1296 /* PID/PID hash table linkage. */
1297 struct pid_link pids[PIDTYPE_MAX];
1298 struct list_head thread_group;
1300 struct completion *vfork_done; /* for vfork() */
1301 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1302 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1304 cputime_t utime, stime, utimescaled, stimescaled;
1306 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1307 cputime_t prev_utime, prev_stime;
1309 unsigned long nvcsw, nivcsw; /* context switch counts */
1310 struct timespec start_time; /* monotonic time */
1311 struct timespec real_start_time; /* boot based time */
1312 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1313 unsigned long min_flt, maj_flt;
1315 struct task_cputime cputime_expires;
1316 struct list_head cpu_timers[3];
1318 /* process credentials */
1319 const struct cred __rcu *real_cred; /* objective and real subjective task
1320 * credentials (COW) */
1321 const struct cred __rcu *cred; /* effective (overridable) subjective task
1322 * credentials (COW) */
1323 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1325 char comm[TASK_COMM_LEN]; /* executable name excluding path
1326 - access with [gs]et_task_comm (which lock
1327 it with task_lock())
1328 - initialized normally by setup_new_exec */
1329 /* file system info */
1330 int link_count, total_link_count;
1331 #ifdef CONFIG_SYSVIPC
1333 struct sysv_sem sysvsem;
1335 #ifdef CONFIG_DETECT_HUNG_TASK
1336 /* hung task detection */
1337 unsigned long last_switch_count;
1339 /* CPU-specific state of this task */
1340 struct thread_struct thread;
1341 /* filesystem information */
1342 struct fs_struct *fs;
1343 /* open file information */
1344 struct files_struct *files;
1346 struct nsproxy *nsproxy;
1347 /* signal handlers */
1348 struct signal_struct *signal;
1349 struct sighand_struct *sighand;
1351 sigset_t blocked, real_blocked;
1352 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1353 struct sigpending pending;
1355 unsigned long sas_ss_sp;
1357 int (*notifier)(void *priv);
1358 void *notifier_data;
1359 sigset_t *notifier_mask;
1360 struct audit_context *audit_context;
1361 #ifdef CONFIG_AUDITSYSCALL
1363 unsigned int sessionid;
1367 /* Thread group tracking */
1370 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1372 spinlock_t alloc_lock;
1374 #ifdef CONFIG_GENERIC_HARDIRQS
1375 /* IRQ handler threads */
1376 struct irqaction *irqaction;
1379 /* Protection of the PI data structures: */
1380 raw_spinlock_t pi_lock;
1382 #ifdef CONFIG_RT_MUTEXES
1383 /* PI waiters blocked on a rt_mutex held by this task */
1384 struct plist_head pi_waiters;
1385 /* Deadlock detection and priority inheritance handling */
1386 struct rt_mutex_waiter *pi_blocked_on;
1389 #ifdef CONFIG_DEBUG_MUTEXES
1390 /* mutex deadlock detection */
1391 struct mutex_waiter *blocked_on;
1393 #ifdef CONFIG_TRACE_IRQFLAGS
1394 unsigned int irq_events;
1395 unsigned long hardirq_enable_ip;
1396 unsigned long hardirq_disable_ip;
1397 unsigned int hardirq_enable_event;
1398 unsigned int hardirq_disable_event;
1399 int hardirqs_enabled;
1400 int hardirq_context;
1401 unsigned long softirq_disable_ip;
1402 unsigned long softirq_enable_ip;
1403 unsigned int softirq_disable_event;
1404 unsigned int softirq_enable_event;
1405 int softirqs_enabled;
1406 int softirq_context;
1408 #ifdef CONFIG_LOCKDEP
1409 # define MAX_LOCK_DEPTH 48UL
1412 unsigned int lockdep_recursion;
1413 struct held_lock held_locks[MAX_LOCK_DEPTH];
1414 gfp_t lockdep_reclaim_gfp;
1417 /* journalling filesystem info */
1420 /* stacked block device info */
1421 struct bio_list *bio_list;
1424 struct reclaim_state *reclaim_state;
1426 struct backing_dev_info *backing_dev_info;
1428 struct io_context *io_context;
1430 unsigned long ptrace_message;
1431 siginfo_t *last_siginfo; /* For ptrace use. */
1432 struct task_io_accounting ioac;
1433 #if defined(CONFIG_TASK_XACCT)
1434 u64 acct_rss_mem1; /* accumulated rss usage */
1435 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1436 cputime_t acct_timexpd; /* stime + utime since last update */
1438 #ifdef CONFIG_CPUSETS
1439 nodemask_t mems_allowed; /* Protected by alloc_lock */
1440 int mems_allowed_change_disable;
1441 int cpuset_mem_spread_rotor;
1442 int cpuset_slab_spread_rotor;
1444 #ifdef CONFIG_CGROUPS
1445 /* Control Group info protected by css_set_lock */
1446 struct css_set __rcu *cgroups;
1447 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1448 struct list_head cg_list;
1451 struct robust_list_head __user *robust_list;
1452 #ifdef CONFIG_COMPAT
1453 struct compat_robust_list_head __user *compat_robust_list;
1455 struct list_head pi_state_list;
1456 struct futex_pi_state *pi_state_cache;
1458 #ifdef CONFIG_PERF_EVENTS
1459 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1460 struct mutex perf_event_mutex;
1461 struct list_head perf_event_list;
1464 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1467 atomic_t fs_excl; /* holding fs exclusive resources */
1468 struct rcu_head rcu;
1471 * cache last used pipe for splice
1473 struct pipe_inode_info *splice_pipe;
1474 #ifdef CONFIG_TASK_DELAY_ACCT
1475 struct task_delay_info *delays;
1477 #ifdef CONFIG_FAULT_INJECTION
1480 struct prop_local_single dirties;
1481 #ifdef CONFIG_LATENCYTOP
1482 int latency_record_count;
1483 struct latency_record latency_record[LT_SAVECOUNT];
1486 * time slack values; these are used to round up poll() and
1487 * select() etc timeout values. These are in nanoseconds.
1489 unsigned long timer_slack_ns;
1490 unsigned long default_timer_slack_ns;
1492 struct list_head *scm_work_list;
1493 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1494 /* Index of current stored address in ret_stack */
1496 /* Stack of return addresses for return function tracing */
1497 struct ftrace_ret_stack *ret_stack;
1498 /* time stamp for last schedule */
1499 unsigned long long ftrace_timestamp;
1501 * Number of functions that haven't been traced
1502 * because of depth overrun.
1504 atomic_t trace_overrun;
1505 /* Pause for the tracing */
1506 atomic_t tracing_graph_pause;
1508 #ifdef CONFIG_TRACING
1509 /* state flags for use by tracers */
1510 unsigned long trace;
1511 /* bitmask of trace recursion */
1512 unsigned long trace_recursion;
1513 #endif /* CONFIG_TRACING */
1514 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1515 struct memcg_batch_info {
1516 int do_batch; /* incremented when batch uncharge started */
1517 struct mem_cgroup *memcg; /* target memcg of uncharge */
1518 unsigned long bytes; /* uncharged usage */
1519 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1524 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1525 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1528 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1529 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1530 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1531 * values are inverted: lower p->prio value means higher priority.
1533 * The MAX_USER_RT_PRIO value allows the actual maximum
1534 * RT priority to be separate from the value exported to
1535 * user-space. This allows kernel threads to set their
1536 * priority to a value higher than any user task. Note:
1537 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1540 #define MAX_USER_RT_PRIO 100
1541 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1543 #define MAX_PRIO (MAX_RT_PRIO + 40)
1544 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1546 static inline int rt_prio(int prio)
1548 if (unlikely(prio < MAX_RT_PRIO))
1553 static inline int rt_task(struct task_struct *p)
1555 return rt_prio(p->prio);
1558 static inline struct pid *task_pid(struct task_struct *task)
1560 return task->pids[PIDTYPE_PID].pid;
1563 static inline struct pid *task_tgid(struct task_struct *task)
1565 return task->group_leader->pids[PIDTYPE_PID].pid;
1569 * Without tasklist or rcu lock it is not safe to dereference
1570 * the result of task_pgrp/task_session even if task == current,
1571 * we can race with another thread doing sys_setsid/sys_setpgid.
1573 static inline struct pid *task_pgrp(struct task_struct *task)
1575 return task->group_leader->pids[PIDTYPE_PGID].pid;
1578 static inline struct pid *task_session(struct task_struct *task)
1580 return task->group_leader->pids[PIDTYPE_SID].pid;
1583 struct pid_namespace;
1586 * the helpers to get the task's different pids as they are seen
1587 * from various namespaces
1589 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1590 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1592 * task_xid_nr_ns() : id seen from the ns specified;
1594 * set_task_vxid() : assigns a virtual id to a task;
1596 * see also pid_nr() etc in include/linux/pid.h
1598 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1599 struct pid_namespace *ns);
1601 static inline pid_t task_pid_nr(struct task_struct *tsk)
1606 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1607 struct pid_namespace *ns)
1609 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1612 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1614 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1618 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1623 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1625 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1627 return pid_vnr(task_tgid(tsk));
1631 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1632 struct pid_namespace *ns)
1634 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1637 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1639 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1643 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1644 struct pid_namespace *ns)
1646 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1649 static inline pid_t task_session_vnr(struct task_struct *tsk)
1651 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1654 /* obsolete, do not use */
1655 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1657 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1661 * pid_alive - check that a task structure is not stale
1662 * @p: Task structure to be checked.
1664 * Test if a process is not yet dead (at most zombie state)
1665 * If pid_alive fails, then pointers within the task structure
1666 * can be stale and must not be dereferenced.
1668 static inline int pid_alive(struct task_struct *p)
1670 return p->pids[PIDTYPE_PID].pid != NULL;
1674 * is_global_init - check if a task structure is init
1675 * @tsk: Task structure to be checked.
1677 * Check if a task structure is the first user space task the kernel created.
1679 static inline int is_global_init(struct task_struct *tsk)
1681 return tsk->pid == 1;
1685 * is_container_init:
1686 * check whether in the task is init in its own pid namespace.
1688 extern int is_container_init(struct task_struct *tsk);
1690 extern struct pid *cad_pid;
1692 extern void free_task(struct task_struct *tsk);
1693 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1695 extern void __put_task_struct(struct task_struct *t);
1697 static inline void put_task_struct(struct task_struct *t)
1699 if (atomic_dec_and_test(&t->usage))
1700 __put_task_struct(t);
1703 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1704 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1709 #define PF_KSOFTIRQD 0x00000001 /* I am ksoftirqd */
1710 #define PF_STARTING 0x00000002 /* being created */
1711 #define PF_EXITING 0x00000004 /* getting shut down */
1712 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1713 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1714 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1715 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1716 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1717 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1718 #define PF_DUMPCORE 0x00000200 /* dumped core */
1719 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1720 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1721 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1722 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1723 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1724 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1725 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1726 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1727 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1728 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1729 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1730 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1731 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1732 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1733 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1734 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1735 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1736 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1737 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1738 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1739 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1742 * Only the _current_ task can read/write to tsk->flags, but other
1743 * tasks can access tsk->flags in readonly mode for example
1744 * with tsk_used_math (like during threaded core dumping).
1745 * There is however an exception to this rule during ptrace
1746 * or during fork: the ptracer task is allowed to write to the
1747 * child->flags of its traced child (same goes for fork, the parent
1748 * can write to the child->flags), because we're guaranteed the
1749 * child is not running and in turn not changing child->flags
1750 * at the same time the parent does it.
1752 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1753 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1754 #define clear_used_math() clear_stopped_child_used_math(current)
1755 #define set_used_math() set_stopped_child_used_math(current)
1756 #define conditional_stopped_child_used_math(condition, child) \
1757 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1758 #define conditional_used_math(condition) \
1759 conditional_stopped_child_used_math(condition, current)
1760 #define copy_to_stopped_child_used_math(child) \
1761 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1762 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1763 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1764 #define used_math() tsk_used_math(current)
1766 #ifdef CONFIG_PREEMPT_RCU
1768 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1769 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1771 static inline void rcu_copy_process(struct task_struct *p)
1773 p->rcu_read_lock_nesting = 0;
1774 p->rcu_read_unlock_special = 0;
1775 #ifdef CONFIG_TREE_PREEMPT_RCU
1776 p->rcu_blocked_node = NULL;
1778 INIT_LIST_HEAD(&p->rcu_node_entry);
1783 static inline void rcu_copy_process(struct task_struct *p)
1790 extern int set_cpus_allowed_ptr(struct task_struct *p,
1791 const struct cpumask *new_mask);
1793 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1794 const struct cpumask *new_mask)
1796 if (!cpumask_test_cpu(0, new_mask))
1802 #ifndef CONFIG_CPUMASK_OFFSTACK
1803 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1805 return set_cpus_allowed_ptr(p, &new_mask);
1810 * Do not use outside of architecture code which knows its limitations.
1812 * sched_clock() has no promise of monotonicity or bounded drift between
1813 * CPUs, use (which you should not) requires disabling IRQs.
1815 * Please use one of the three interfaces below.
1817 extern unsigned long long notrace sched_clock(void);
1819 * See the comment in kernel/sched_clock.c
1821 extern u64 cpu_clock(int cpu);
1822 extern u64 local_clock(void);
1823 extern u64 sched_clock_cpu(int cpu);
1826 extern void sched_clock_init(void);
1828 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1829 static inline void sched_clock_tick(void)
1833 static inline void sched_clock_idle_sleep_event(void)
1837 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1842 * Architectures can set this to 1 if they have specified
1843 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1844 * but then during bootup it turns out that sched_clock()
1845 * is reliable after all:
1847 extern int sched_clock_stable;
1849 extern void sched_clock_tick(void);
1850 extern void sched_clock_idle_sleep_event(void);
1851 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1854 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1856 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1857 * The reason for this explicit opt-in is not to have perf penalty with
1858 * slow sched_clocks.
1860 extern void enable_sched_clock_irqtime(void);
1861 extern void disable_sched_clock_irqtime(void);
1863 static inline void enable_sched_clock_irqtime(void) {}
1864 static inline void disable_sched_clock_irqtime(void) {}
1867 extern unsigned long long
1868 task_sched_runtime(struct task_struct *task);
1869 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1871 /* sched_exec is called by processes performing an exec */
1873 extern void sched_exec(void);
1875 #define sched_exec() {}
1878 extern void sched_clock_idle_sleep_event(void);
1879 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1881 #ifdef CONFIG_HOTPLUG_CPU
1882 extern void idle_task_exit(void);
1884 static inline void idle_task_exit(void) {}
1887 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1888 extern void wake_up_idle_cpu(int cpu);
1890 static inline void wake_up_idle_cpu(int cpu) { }
1893 extern unsigned int sysctl_sched_latency;
1894 extern unsigned int sysctl_sched_min_granularity;
1895 extern unsigned int sysctl_sched_wakeup_granularity;
1896 extern unsigned int sysctl_sched_child_runs_first;
1898 enum sched_tunable_scaling {
1899 SCHED_TUNABLESCALING_NONE,
1900 SCHED_TUNABLESCALING_LOG,
1901 SCHED_TUNABLESCALING_LINEAR,
1902 SCHED_TUNABLESCALING_END,
1904 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1906 #ifdef CONFIG_SCHED_DEBUG
1907 extern unsigned int sysctl_sched_migration_cost;
1908 extern unsigned int sysctl_sched_nr_migrate;
1909 extern unsigned int sysctl_sched_time_avg;
1910 extern unsigned int sysctl_timer_migration;
1911 extern unsigned int sysctl_sched_shares_window;
1913 int sched_proc_update_handler(struct ctl_table *table, int write,
1914 void __user *buffer, size_t *length,
1917 #ifdef CONFIG_SCHED_DEBUG
1918 static inline unsigned int get_sysctl_timer_migration(void)
1920 return sysctl_timer_migration;
1923 static inline unsigned int get_sysctl_timer_migration(void)
1928 extern unsigned int sysctl_sched_rt_period;
1929 extern int sysctl_sched_rt_runtime;
1931 int sched_rt_handler(struct ctl_table *table, int write,
1932 void __user *buffer, size_t *lenp,
1935 extern unsigned int sysctl_sched_compat_yield;
1937 #ifdef CONFIG_SCHED_AUTOGROUP
1938 extern unsigned int sysctl_sched_autogroup_enabled;
1940 extern void sched_autogroup_create_attach(struct task_struct *p);
1941 extern void sched_autogroup_detach(struct task_struct *p);
1942 extern void sched_autogroup_fork(struct signal_struct *sig);
1943 extern void sched_autogroup_exit(struct signal_struct *sig);
1944 #ifdef CONFIG_PROC_FS
1945 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1946 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
1949 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1950 static inline void sched_autogroup_detach(struct task_struct *p) { }
1951 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1952 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1955 #ifdef CONFIG_RT_MUTEXES
1956 extern int rt_mutex_getprio(struct task_struct *p);
1957 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1958 extern void rt_mutex_adjust_pi(struct task_struct *p);
1960 static inline int rt_mutex_getprio(struct task_struct *p)
1962 return p->normal_prio;
1964 # define rt_mutex_adjust_pi(p) do { } while (0)
1967 extern void set_user_nice(struct task_struct *p, long nice);
1968 extern int task_prio(const struct task_struct *p);
1969 extern int task_nice(const struct task_struct *p);
1970 extern int can_nice(const struct task_struct *p, const int nice);
1971 extern int task_curr(const struct task_struct *p);
1972 extern int idle_cpu(int cpu);
1973 extern int sched_setscheduler(struct task_struct *, int,
1974 const struct sched_param *);
1975 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1976 const struct sched_param *);
1977 extern struct task_struct *idle_task(int cpu);
1978 extern struct task_struct *curr_task(int cpu);
1979 extern void set_curr_task(int cpu, struct task_struct *p);
1984 * The default (Linux) execution domain.
1986 extern struct exec_domain default_exec_domain;
1988 union thread_union {
1989 struct thread_info thread_info;
1990 unsigned long stack[THREAD_SIZE/sizeof(long)];
1993 #ifndef __HAVE_ARCH_KSTACK_END
1994 static inline int kstack_end(void *addr)
1996 /* Reliable end of stack detection:
1997 * Some APM bios versions misalign the stack
1999 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2003 extern union thread_union init_thread_union;
2004 extern struct task_struct init_task;
2006 extern struct mm_struct init_mm;
2008 extern struct pid_namespace init_pid_ns;
2011 * find a task by one of its numerical ids
2013 * find_task_by_pid_ns():
2014 * finds a task by its pid in the specified namespace
2015 * find_task_by_vpid():
2016 * finds a task by its virtual pid
2018 * see also find_vpid() etc in include/linux/pid.h
2021 extern struct task_struct *find_task_by_vpid(pid_t nr);
2022 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2023 struct pid_namespace *ns);
2025 extern void __set_special_pids(struct pid *pid);
2027 /* per-UID process charging. */
2028 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2029 static inline struct user_struct *get_uid(struct user_struct *u)
2031 atomic_inc(&u->__count);
2034 extern void free_uid(struct user_struct *);
2035 extern void release_uids(struct user_namespace *ns);
2037 #include <asm/current.h>
2039 extern void do_timer(unsigned long ticks);
2041 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2042 extern int wake_up_process(struct task_struct *tsk);
2043 extern void wake_up_new_task(struct task_struct *tsk,
2044 unsigned long clone_flags);
2046 extern void kick_process(struct task_struct *tsk);
2048 static inline void kick_process(struct task_struct *tsk) { }
2050 extern void sched_fork(struct task_struct *p, int clone_flags);
2051 extern void sched_dead(struct task_struct *p);
2053 extern void proc_caches_init(void);
2054 extern void flush_signals(struct task_struct *);
2055 extern void __flush_signals(struct task_struct *);
2056 extern void ignore_signals(struct task_struct *);
2057 extern void flush_signal_handlers(struct task_struct *, int force_default);
2058 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2060 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2062 unsigned long flags;
2065 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2066 ret = dequeue_signal(tsk, mask, info);
2067 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2072 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2074 extern void unblock_all_signals(void);
2075 extern void release_task(struct task_struct * p);
2076 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2077 extern int force_sigsegv(int, struct task_struct *);
2078 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2079 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2080 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2081 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2082 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2083 extern int kill_pid(struct pid *pid, int sig, int priv);
2084 extern int kill_proc_info(int, struct siginfo *, pid_t);
2085 extern int do_notify_parent(struct task_struct *, int);
2086 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2087 extern void force_sig(int, struct task_struct *);
2088 extern int send_sig(int, struct task_struct *, int);
2089 extern int zap_other_threads(struct task_struct *p);
2090 extern struct sigqueue *sigqueue_alloc(void);
2091 extern void sigqueue_free(struct sigqueue *);
2092 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2093 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2094 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2096 static inline int kill_cad_pid(int sig, int priv)
2098 return kill_pid(cad_pid, sig, priv);
2101 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2102 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2103 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2104 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2107 * True if we are on the alternate signal stack.
2109 static inline int on_sig_stack(unsigned long sp)
2111 #ifdef CONFIG_STACK_GROWSUP
2112 return sp >= current->sas_ss_sp &&
2113 sp - current->sas_ss_sp < current->sas_ss_size;
2115 return sp > current->sas_ss_sp &&
2116 sp - current->sas_ss_sp <= current->sas_ss_size;
2120 static inline int sas_ss_flags(unsigned long sp)
2122 return (current->sas_ss_size == 0 ? SS_DISABLE
2123 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2127 * Routines for handling mm_structs
2129 extern struct mm_struct * mm_alloc(void);
2131 /* mmdrop drops the mm and the page tables */
2132 extern void __mmdrop(struct mm_struct *);
2133 static inline void mmdrop(struct mm_struct * mm)
2135 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2139 /* mmput gets rid of the mappings and all user-space */
2140 extern void mmput(struct mm_struct *);
2141 /* Grab a reference to a task's mm, if it is not already going away */
2142 extern struct mm_struct *get_task_mm(struct task_struct *task);
2143 /* Remove the current tasks stale references to the old mm_struct */
2144 extern void mm_release(struct task_struct *, struct mm_struct *);
2145 /* Allocate a new mm structure and copy contents from tsk->mm */
2146 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2148 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2149 struct task_struct *, struct pt_regs *);
2150 extern void flush_thread(void);
2151 extern void exit_thread(void);
2153 extern void exit_files(struct task_struct *);
2154 extern void __cleanup_sighand(struct sighand_struct *);
2156 extern void exit_itimers(struct signal_struct *);
2157 extern void flush_itimer_signals(void);
2159 extern NORET_TYPE void do_group_exit(int);
2161 extern void daemonize(const char *, ...);
2162 extern int allow_signal(int);
2163 extern int disallow_signal(int);
2165 extern int do_execve(const char *,
2166 const char __user * const __user *,
2167 const char __user * const __user *, struct pt_regs *);
2168 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2169 struct task_struct *fork_idle(int);
2171 extern void set_task_comm(struct task_struct *tsk, char *from);
2172 extern char *get_task_comm(char *to, struct task_struct *tsk);
2175 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2177 static inline unsigned long wait_task_inactive(struct task_struct *p,
2184 #define next_task(p) \
2185 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2187 #define for_each_process(p) \
2188 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2190 extern bool current_is_single_threaded(void);
2193 * Careful: do_each_thread/while_each_thread is a double loop so
2194 * 'break' will not work as expected - use goto instead.
2196 #define do_each_thread(g, t) \
2197 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2199 #define while_each_thread(g, t) \
2200 while ((t = next_thread(t)) != g)
2202 static inline int get_nr_threads(struct task_struct *tsk)
2204 return tsk->signal->nr_threads;
2207 /* de_thread depends on thread_group_leader not being a pid based check */
2208 #define thread_group_leader(p) (p == p->group_leader)
2210 /* Do to the insanities of de_thread it is possible for a process
2211 * to have the pid of the thread group leader without actually being
2212 * the thread group leader. For iteration through the pids in proc
2213 * all we care about is that we have a task with the appropriate
2214 * pid, we don't actually care if we have the right task.
2216 static inline int has_group_leader_pid(struct task_struct *p)
2218 return p->pid == p->tgid;
2222 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2224 return p1->tgid == p2->tgid;
2227 static inline struct task_struct *next_thread(const struct task_struct *p)
2229 return list_entry_rcu(p->thread_group.next,
2230 struct task_struct, thread_group);
2233 static inline int thread_group_empty(struct task_struct *p)
2235 return list_empty(&p->thread_group);
2238 #define delay_group_leader(p) \
2239 (thread_group_leader(p) && !thread_group_empty(p))
2241 static inline int task_detached(struct task_struct *p)
2243 return p->exit_signal == -1;
2247 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2248 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2249 * pins the final release of task.io_context. Also protects ->cpuset and
2250 * ->cgroup.subsys[].
2252 * Nests both inside and outside of read_lock(&tasklist_lock).
2253 * It must not be nested with write_lock_irq(&tasklist_lock),
2254 * neither inside nor outside.
2256 static inline void task_lock(struct task_struct *p)
2258 spin_lock(&p->alloc_lock);
2261 static inline void task_unlock(struct task_struct *p)
2263 spin_unlock(&p->alloc_lock);
2266 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2267 unsigned long *flags);
2269 #define lock_task_sighand(tsk, flags) \
2270 ({ struct sighand_struct *__ss; \
2271 __cond_lock(&(tsk)->sighand->siglock, \
2272 (__ss = __lock_task_sighand(tsk, flags))); \
2276 static inline void unlock_task_sighand(struct task_struct *tsk,
2277 unsigned long *flags)
2279 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2282 #ifndef __HAVE_THREAD_FUNCTIONS
2284 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2285 #define task_stack_page(task) ((task)->stack)
2287 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2289 *task_thread_info(p) = *task_thread_info(org);
2290 task_thread_info(p)->task = p;
2293 static inline unsigned long *end_of_stack(struct task_struct *p)
2295 return (unsigned long *)(task_thread_info(p) + 1);
2300 static inline int object_is_on_stack(void *obj)
2302 void *stack = task_stack_page(current);
2304 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2307 extern void thread_info_cache_init(void);
2309 #ifdef CONFIG_DEBUG_STACK_USAGE
2310 static inline unsigned long stack_not_used(struct task_struct *p)
2312 unsigned long *n = end_of_stack(p);
2314 do { /* Skip over canary */
2318 return (unsigned long)n - (unsigned long)end_of_stack(p);
2322 /* set thread flags in other task's structures
2323 * - see asm/thread_info.h for TIF_xxxx flags available
2325 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2327 set_ti_thread_flag(task_thread_info(tsk), flag);
2330 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2332 clear_ti_thread_flag(task_thread_info(tsk), flag);
2335 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2337 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2340 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2342 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2345 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2347 return test_ti_thread_flag(task_thread_info(tsk), flag);
2350 static inline void set_tsk_need_resched(struct task_struct *tsk)
2352 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2355 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2357 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2360 static inline int test_tsk_need_resched(struct task_struct *tsk)
2362 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2365 static inline int restart_syscall(void)
2367 set_tsk_thread_flag(current, TIF_SIGPENDING);
2368 return -ERESTARTNOINTR;
2371 static inline int signal_pending(struct task_struct *p)
2373 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2376 static inline int __fatal_signal_pending(struct task_struct *p)
2378 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2381 static inline int fatal_signal_pending(struct task_struct *p)
2383 return signal_pending(p) && __fatal_signal_pending(p);
2386 static inline int signal_pending_state(long state, struct task_struct *p)
2388 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2390 if (!signal_pending(p))
2393 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2396 static inline int need_resched(void)
2398 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2402 * cond_resched() and cond_resched_lock(): latency reduction via
2403 * explicit rescheduling in places that are safe. The return
2404 * value indicates whether a reschedule was done in fact.
2405 * cond_resched_lock() will drop the spinlock before scheduling,
2406 * cond_resched_softirq() will enable bhs before scheduling.
2408 extern int _cond_resched(void);
2410 #define cond_resched() ({ \
2411 __might_sleep(__FILE__, __LINE__, 0); \
2415 extern int __cond_resched_lock(spinlock_t *lock);
2417 #ifdef CONFIG_PREEMPT
2418 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2420 #define PREEMPT_LOCK_OFFSET 0
2423 #define cond_resched_lock(lock) ({ \
2424 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2425 __cond_resched_lock(lock); \
2428 extern int __cond_resched_softirq(void);
2430 #define cond_resched_softirq() ({ \
2431 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2432 __cond_resched_softirq(); \
2436 * Does a critical section need to be broken due to another
2437 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2438 * but a general need for low latency)
2440 static inline int spin_needbreak(spinlock_t *lock)
2442 #ifdef CONFIG_PREEMPT
2443 return spin_is_contended(lock);
2450 * Thread group CPU time accounting.
2452 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2453 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2455 static inline void thread_group_cputime_init(struct signal_struct *sig)
2457 spin_lock_init(&sig->cputimer.lock);
2461 * Reevaluate whether the task has signals pending delivery.
2462 * Wake the task if so.
2463 * This is required every time the blocked sigset_t changes.
2464 * callers must hold sighand->siglock.
2466 extern void recalc_sigpending_and_wake(struct task_struct *t);
2467 extern void recalc_sigpending(void);
2469 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2472 * Wrappers for p->thread_info->cpu access. No-op on UP.
2476 static inline unsigned int task_cpu(const struct task_struct *p)
2478 return task_thread_info(p)->cpu;
2481 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2485 static inline unsigned int task_cpu(const struct task_struct *p)
2490 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2494 #endif /* CONFIG_SMP */
2496 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2497 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2499 extern void normalize_rt_tasks(void);
2501 #ifdef CONFIG_CGROUP_SCHED
2503 extern struct task_group init_task_group;
2505 extern struct task_group *sched_create_group(struct task_group *parent);
2506 extern void sched_destroy_group(struct task_group *tg);
2507 extern void sched_move_task(struct task_struct *tsk);
2508 #ifdef CONFIG_FAIR_GROUP_SCHED
2509 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2510 extern unsigned long sched_group_shares(struct task_group *tg);
2512 #ifdef CONFIG_RT_GROUP_SCHED
2513 extern int sched_group_set_rt_runtime(struct task_group *tg,
2514 long rt_runtime_us);
2515 extern long sched_group_rt_runtime(struct task_group *tg);
2516 extern int sched_group_set_rt_period(struct task_group *tg,
2518 extern long sched_group_rt_period(struct task_group *tg);
2519 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2523 extern int task_can_switch_user(struct user_struct *up,
2524 struct task_struct *tsk);
2526 #ifdef CONFIG_TASK_XACCT
2527 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2529 tsk->ioac.rchar += amt;
2532 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2534 tsk->ioac.wchar += amt;
2537 static inline void inc_syscr(struct task_struct *tsk)
2542 static inline void inc_syscw(struct task_struct *tsk)
2547 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2551 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2555 static inline void inc_syscr(struct task_struct *tsk)
2559 static inline void inc_syscw(struct task_struct *tsk)
2564 #ifndef TASK_SIZE_OF
2565 #define TASK_SIZE_OF(tsk) TASK_SIZE
2569 * Call the function if the target task is executing on a CPU right now:
2571 extern void task_oncpu_function_call(struct task_struct *p,
2572 void (*func) (void *info), void *info);
2575 #ifdef CONFIG_MM_OWNER
2576 extern void mm_update_next_owner(struct mm_struct *mm);
2577 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2579 static inline void mm_update_next_owner(struct mm_struct *mm)
2583 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2586 #endif /* CONFIG_MM_OWNER */
2588 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2591 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2594 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2597 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2600 static inline unsigned long rlimit(unsigned int limit)
2602 return task_rlimit(current, limit);
2605 static inline unsigned long rlimit_max(unsigned int limit)
2607 return task_rlimit_max(current, limit);
2610 #endif /* __KERNEL__ */