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(void);
143 extern unsigned long this_cpu_load(void);
146 extern void calc_global_load(void);
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_stopped_or_traced(task) \
218 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
219 #define task_contributes_to_load(task) \
220 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
221 (task->flags & PF_FREEZING) == 0)
223 #define __set_task_state(tsk, state_value) \
224 do { (tsk)->state = (state_value); } while (0)
225 #define set_task_state(tsk, state_value) \
226 set_mb((tsk)->state, (state_value))
229 * set_current_state() includes a barrier so that the write of current->state
230 * is correctly serialised wrt the caller's subsequent test of whether to
233 * set_current_state(TASK_UNINTERRUPTIBLE);
234 * if (do_i_need_to_sleep())
237 * If the caller does not need such serialisation then use __set_current_state()
239 #define __set_current_state(state_value) \
240 do { current->state = (state_value); } while (0)
241 #define set_current_state(state_value) \
242 set_mb(current->state, (state_value))
244 /* Task command name length */
245 #define TASK_COMM_LEN 16
247 #include <linux/spinlock.h>
250 * This serializes "schedule()" and also protects
251 * the run-queue from deletions/modifications (but
252 * _adding_ to the beginning of the run-queue has
255 extern rwlock_t tasklist_lock;
256 extern spinlock_t mmlist_lock;
260 #ifdef CONFIG_PROVE_RCU
261 extern int lockdep_tasklist_lock_is_held(void);
262 #endif /* #ifdef CONFIG_PROVE_RCU */
264 extern void sched_init(void);
265 extern void sched_init_smp(void);
266 extern asmlinkage void schedule_tail(struct task_struct *prev);
267 extern void init_idle(struct task_struct *idle, int cpu);
268 extern void init_idle_bootup_task(struct task_struct *idle);
270 extern int runqueue_is_locked(int cpu);
271 extern void task_rq_unlock_wait(struct task_struct *p);
273 extern cpumask_var_t nohz_cpu_mask;
274 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
275 extern int select_nohz_load_balancer(int cpu);
276 extern int get_nohz_load_balancer(void);
277 extern int nohz_ratelimit(int cpu);
279 static inline int select_nohz_load_balancer(int cpu)
284 static inline int nohz_ratelimit(int cpu)
291 * Only dump TASK_* tasks. (0 for all tasks)
293 extern void show_state_filter(unsigned long state_filter);
295 static inline void show_state(void)
297 show_state_filter(0);
300 extern void show_regs(struct pt_regs *);
303 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
304 * task), SP is the stack pointer of the first frame that should be shown in the back
305 * trace (or NULL if the entire call-chain of the task should be shown).
307 extern void show_stack(struct task_struct *task, unsigned long *sp);
309 void io_schedule(void);
310 long io_schedule_timeout(long timeout);
312 extern void cpu_init (void);
313 extern void trap_init(void);
314 extern void update_process_times(int user);
315 extern void scheduler_tick(void);
317 extern void sched_show_task(struct task_struct *p);
319 #ifdef CONFIG_DETECT_SOFTLOCKUP
320 extern void softlockup_tick(void);
321 extern void touch_softlockup_watchdog(void);
322 extern void touch_softlockup_watchdog_sync(void);
323 extern void touch_all_softlockup_watchdogs(void);
324 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
326 size_t *lenp, loff_t *ppos);
327 extern unsigned int softlockup_panic;
328 extern int softlockup_thresh;
330 static inline void softlockup_tick(void)
333 static inline void touch_softlockup_watchdog(void)
336 static inline void touch_softlockup_watchdog_sync(void)
339 static inline void touch_all_softlockup_watchdogs(void)
344 #ifdef CONFIG_DETECT_HUNG_TASK
345 extern unsigned int sysctl_hung_task_panic;
346 extern unsigned long sysctl_hung_task_check_count;
347 extern unsigned long sysctl_hung_task_timeout_secs;
348 extern unsigned long sysctl_hung_task_warnings;
349 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
351 size_t *lenp, loff_t *ppos);
354 /* Attach to any functions which should be ignored in wchan output. */
355 #define __sched __attribute__((__section__(".sched.text")))
357 /* Linker adds these: start and end of __sched functions */
358 extern char __sched_text_start[], __sched_text_end[];
360 /* Is this address in the __sched functions? */
361 extern int in_sched_functions(unsigned long addr);
363 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
364 extern signed long schedule_timeout(signed long timeout);
365 extern signed long schedule_timeout_interruptible(signed long timeout);
366 extern signed long schedule_timeout_killable(signed long timeout);
367 extern signed long schedule_timeout_uninterruptible(signed long timeout);
368 asmlinkage void schedule(void);
369 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
372 struct user_namespace;
375 * Default maximum number of active map areas, this limits the number of vmas
376 * per mm struct. Users can overwrite this number by sysctl but there is a
379 * When a program's coredump is generated as ELF format, a section is created
380 * per a vma. In ELF, the number of sections is represented in unsigned short.
381 * This means the number of sections should be smaller than 65535 at coredump.
382 * Because the kernel adds some informative sections to a image of program at
383 * generating coredump, we need some margin. The number of extra sections is
384 * 1-3 now and depends on arch. We use "5" as safe margin, here.
386 #define MAPCOUNT_ELF_CORE_MARGIN (5)
387 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
389 extern int sysctl_max_map_count;
391 #include <linux/aio.h>
394 extern void arch_pick_mmap_layout(struct mm_struct *mm);
396 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
397 unsigned long, unsigned long);
399 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
400 unsigned long len, unsigned long pgoff,
401 unsigned long flags);
402 extern void arch_unmap_area(struct mm_struct *, unsigned long);
403 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
405 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
409 extern void set_dumpable(struct mm_struct *mm, int value);
410 extern int get_dumpable(struct mm_struct *mm);
414 #define MMF_DUMPABLE 0 /* core dump is permitted */
415 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
417 #define MMF_DUMPABLE_BITS 2
418 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
420 /* coredump filter bits */
421 #define MMF_DUMP_ANON_PRIVATE 2
422 #define MMF_DUMP_ANON_SHARED 3
423 #define MMF_DUMP_MAPPED_PRIVATE 4
424 #define MMF_DUMP_MAPPED_SHARED 5
425 #define MMF_DUMP_ELF_HEADERS 6
426 #define MMF_DUMP_HUGETLB_PRIVATE 7
427 #define MMF_DUMP_HUGETLB_SHARED 8
429 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
430 #define MMF_DUMP_FILTER_BITS 7
431 #define MMF_DUMP_FILTER_MASK \
432 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
433 #define MMF_DUMP_FILTER_DEFAULT \
434 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
435 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
437 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
438 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
440 # define MMF_DUMP_MASK_DEFAULT_ELF 0
442 /* leave room for more dump flags */
443 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
445 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
447 struct sighand_struct {
449 struct k_sigaction action[_NSIG];
451 wait_queue_head_t signalfd_wqh;
454 struct pacct_struct {
457 unsigned long ac_mem;
458 cputime_t ac_utime, ac_stime;
459 unsigned long ac_minflt, ac_majflt;
470 * struct task_cputime - collected CPU time counts
471 * @utime: time spent in user mode, in &cputime_t units
472 * @stime: time spent in kernel mode, in &cputime_t units
473 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
475 * This structure groups together three kinds of CPU time that are
476 * tracked for threads and thread groups. Most things considering
477 * CPU time want to group these counts together and treat all three
478 * of them in parallel.
480 struct task_cputime {
483 unsigned long long sum_exec_runtime;
485 /* Alternate field names when used to cache expirations. */
486 #define prof_exp stime
487 #define virt_exp utime
488 #define sched_exp sum_exec_runtime
490 #define INIT_CPUTIME \
491 (struct task_cputime) { \
492 .utime = cputime_zero, \
493 .stime = cputime_zero, \
494 .sum_exec_runtime = 0, \
498 * Disable preemption until the scheduler is running.
499 * Reset by start_kernel()->sched_init()->init_idle().
501 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
502 * before the scheduler is active -- see should_resched().
504 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
507 * struct thread_group_cputimer - thread group interval timer counts
508 * @cputime: thread group interval timers.
509 * @running: non-zero when there are timers running and
510 * @cputime receives updates.
511 * @lock: lock for fields in this struct.
513 * This structure contains the version of task_cputime, above, that is
514 * used for thread group CPU timer calculations.
516 struct thread_group_cputimer {
517 struct task_cputime cputime;
523 * NOTE! "signal_struct" does not have it's own
524 * locking, because a shared signal_struct always
525 * implies a shared sighand_struct, so locking
526 * sighand_struct is always a proper superset of
527 * the locking of signal_struct.
529 struct signal_struct {
533 wait_queue_head_t wait_chldexit; /* for wait4() */
535 /* current thread group signal load-balancing target: */
536 struct task_struct *curr_target;
538 /* shared signal handling: */
539 struct sigpending shared_pending;
541 /* thread group exit support */
544 * - notify group_exit_task when ->count is equal to notify_count
545 * - everyone except group_exit_task is stopped during signal delivery
546 * of fatal signals, group_exit_task processes the signal.
549 struct task_struct *group_exit_task;
551 /* thread group stop support, overloads group_exit_code too */
552 int group_stop_count;
553 unsigned int flags; /* see SIGNAL_* flags below */
555 /* POSIX.1b Interval Timers */
556 struct list_head posix_timers;
558 /* ITIMER_REAL timer for the process */
559 struct hrtimer real_timer;
560 struct pid *leader_pid;
561 ktime_t it_real_incr;
564 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
565 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
566 * values are defined to 0 and 1 respectively
568 struct cpu_itimer it[2];
571 * Thread group totals for process CPU timers.
572 * See thread_group_cputimer(), et al, for details.
574 struct thread_group_cputimer cputimer;
576 /* Earliest-expiration cache. */
577 struct task_cputime cputime_expires;
579 struct list_head cpu_timers[3];
581 struct pid *tty_old_pgrp;
583 /* boolean value for session group leader */
586 struct tty_struct *tty; /* NULL if no tty */
589 * Cumulative resource counters for dead threads in the group,
590 * and for reaped dead child processes forked by this group.
591 * Live threads maintain their own counters and add to these
592 * in __exit_signal, except for the group leader.
594 cputime_t utime, stime, cutime, cstime;
597 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
598 cputime_t prev_utime, prev_stime;
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 unsigned long maxrss, cmaxrss;
604 struct task_io_accounting ioac;
607 * Cumulative ns of schedule CPU time fo dead threads in the
608 * group, not including a zombie group leader, (This only differs
609 * from jiffies_to_ns(utime + stime) if sched_clock uses something
610 * other than jiffies.)
612 unsigned long long sum_sched_runtime;
615 * We don't bother to synchronize most readers of this at all,
616 * because there is no reader checking a limit that actually needs
617 * to get both rlim_cur and rlim_max atomically, and either one
618 * alone is a single word that can safely be read normally.
619 * getrlimit/setrlimit use task_lock(current->group_leader) to
620 * protect this instead of the siglock, because they really
621 * have no need to disable irqs.
623 struct rlimit rlim[RLIM_NLIMITS];
625 #ifdef CONFIG_BSD_PROCESS_ACCT
626 struct pacct_struct pacct; /* per-process accounting information */
628 #ifdef CONFIG_TASKSTATS
629 struct taskstats *stats;
633 struct tty_audit_buf *tty_audit_buf;
636 int oom_adj; /* OOM kill score adjustment (bit shift) */
639 /* Context switch must be unlocked if interrupts are to be enabled */
640 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
641 # define __ARCH_WANT_UNLOCKED_CTXSW
645 * Bits in flags field of signal_struct.
647 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
648 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
649 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
650 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
652 * Pending notifications to parent.
654 #define SIGNAL_CLD_STOPPED 0x00000010
655 #define SIGNAL_CLD_CONTINUED 0x00000020
656 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
658 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
660 /* If true, all threads except ->group_exit_task have pending SIGKILL */
661 static inline int signal_group_exit(const struct signal_struct *sig)
663 return (sig->flags & SIGNAL_GROUP_EXIT) ||
664 (sig->group_exit_task != NULL);
668 * Some day this will be a full-fledged user tracking system..
671 atomic_t __count; /* reference count */
672 atomic_t processes; /* How many processes does this user have? */
673 atomic_t files; /* How many open files does this user have? */
674 atomic_t sigpending; /* How many pending signals does this user have? */
675 #ifdef CONFIG_INOTIFY_USER
676 atomic_t inotify_watches; /* How many inotify watches does this user have? */
677 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
680 atomic_t epoll_watches; /* The number of file descriptors currently watched */
682 #ifdef CONFIG_POSIX_MQUEUE
683 /* protected by mq_lock */
684 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
686 unsigned long locked_shm; /* How many pages of mlocked shm ? */
689 struct key *uid_keyring; /* UID specific keyring */
690 struct key *session_keyring; /* UID's default session keyring */
693 /* Hash table maintenance information */
694 struct hlist_node uidhash_node;
696 struct user_namespace *user_ns;
698 #ifdef CONFIG_PERF_EVENTS
699 atomic_long_t locked_vm;
703 extern int uids_sysfs_init(void);
705 extern struct user_struct *find_user(uid_t);
707 extern struct user_struct root_user;
708 #define INIT_USER (&root_user)
711 struct backing_dev_info;
712 struct reclaim_state;
714 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
716 /* cumulative counters */
717 unsigned long pcount; /* # of times run on this cpu */
718 unsigned long long run_delay; /* time spent waiting on a runqueue */
721 unsigned long long last_arrival,/* when we last ran on a cpu */
722 last_queued; /* when we were last queued to run */
723 #ifdef CONFIG_SCHEDSTATS
725 unsigned int bkl_count;
728 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
730 #ifdef CONFIG_TASK_DELAY_ACCT
731 struct task_delay_info {
733 unsigned int flags; /* Private per-task flags */
735 /* For each stat XXX, add following, aligned appropriately
737 * struct timespec XXX_start, XXX_end;
741 * Atomicity of updates to XXX_delay, XXX_count protected by
742 * single lock above (split into XXX_lock if contention is an issue).
746 * XXX_count is incremented on every XXX operation, the delay
747 * associated with the operation is added to XXX_delay.
748 * XXX_delay contains the accumulated delay time in nanoseconds.
750 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
751 u64 blkio_delay; /* wait for sync block io completion */
752 u64 swapin_delay; /* wait for swapin block io completion */
753 u32 blkio_count; /* total count of the number of sync block */
754 /* io operations performed */
755 u32 swapin_count; /* total count of the number of swapin block */
756 /* io operations performed */
758 struct timespec freepages_start, freepages_end;
759 u64 freepages_delay; /* wait for memory reclaim */
760 u32 freepages_count; /* total count of memory reclaim */
762 #endif /* CONFIG_TASK_DELAY_ACCT */
764 static inline int sched_info_on(void)
766 #ifdef CONFIG_SCHEDSTATS
768 #elif defined(CONFIG_TASK_DELAY_ACCT)
769 extern int delayacct_on;
784 * sched-domains (multiprocessor balancing) declarations:
788 * Increase resolution of nice-level calculations:
790 #define SCHED_LOAD_SHIFT 10
791 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
793 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
796 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
797 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
798 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
799 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
800 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
801 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
802 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
803 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
804 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
805 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
806 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
808 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
810 enum powersavings_balance_level {
811 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
812 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
813 * first for long running threads
815 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
816 * cpu package for power savings
818 MAX_POWERSAVINGS_BALANCE_LEVELS
821 extern int sched_mc_power_savings, sched_smt_power_savings;
823 static inline int sd_balance_for_mc_power(void)
825 if (sched_smt_power_savings)
826 return SD_POWERSAVINGS_BALANCE;
828 if (!sched_mc_power_savings)
829 return SD_PREFER_SIBLING;
834 static inline int sd_balance_for_package_power(void)
836 if (sched_mc_power_savings | sched_smt_power_savings)
837 return SD_POWERSAVINGS_BALANCE;
839 return SD_PREFER_SIBLING;
843 * Optimise SD flags for power savings:
844 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
845 * Keep default SD flags if sched_{smt,mc}_power_saving=0
848 static inline int sd_power_saving_flags(void)
850 if (sched_mc_power_savings | sched_smt_power_savings)
851 return SD_BALANCE_NEWIDLE;
857 struct sched_group *next; /* Must be a circular list */
860 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
863 unsigned int cpu_power;
866 * The CPUs this group covers.
868 * NOTE: this field is variable length. (Allocated dynamically
869 * by attaching extra space to the end of the structure,
870 * depending on how many CPUs the kernel has booted up with)
872 * It is also be embedded into static data structures at build
873 * time. (See 'struct static_sched_group' in kernel/sched.c)
875 unsigned long cpumask[0];
878 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
880 return to_cpumask(sg->cpumask);
883 enum sched_domain_level {
893 struct sched_domain_attr {
894 int relax_domain_level;
897 #define SD_ATTR_INIT (struct sched_domain_attr) { \
898 .relax_domain_level = -1, \
901 struct sched_domain {
902 /* These fields must be setup */
903 struct sched_domain *parent; /* top domain must be null terminated */
904 struct sched_domain *child; /* bottom domain must be null terminated */
905 struct sched_group *groups; /* the balancing groups of the domain */
906 unsigned long min_interval; /* Minimum balance interval ms */
907 unsigned long max_interval; /* Maximum balance interval ms */
908 unsigned int busy_factor; /* less balancing by factor if busy */
909 unsigned int imbalance_pct; /* No balance until over watermark */
910 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
911 unsigned int busy_idx;
912 unsigned int idle_idx;
913 unsigned int newidle_idx;
914 unsigned int wake_idx;
915 unsigned int forkexec_idx;
916 unsigned int smt_gain;
917 int flags; /* See SD_* */
918 enum sched_domain_level level;
920 /* Runtime fields. */
921 unsigned long last_balance; /* init to jiffies. units in jiffies */
922 unsigned int balance_interval; /* initialise to 1. units in ms. */
923 unsigned int nr_balance_failed; /* initialise to 0 */
927 #ifdef CONFIG_SCHEDSTATS
928 /* load_balance() stats */
929 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
931 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
932 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
933 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
934 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
935 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
936 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
938 /* Active load balancing */
939 unsigned int alb_count;
940 unsigned int alb_failed;
941 unsigned int alb_pushed;
943 /* SD_BALANCE_EXEC stats */
944 unsigned int sbe_count;
945 unsigned int sbe_balanced;
946 unsigned int sbe_pushed;
948 /* SD_BALANCE_FORK stats */
949 unsigned int sbf_count;
950 unsigned int sbf_balanced;
951 unsigned int sbf_pushed;
953 /* try_to_wake_up() stats */
954 unsigned int ttwu_wake_remote;
955 unsigned int ttwu_move_affine;
956 unsigned int ttwu_move_balance;
958 #ifdef CONFIG_SCHED_DEBUG
962 unsigned int span_weight;
964 * Span of all CPUs in this domain.
966 * NOTE: this field is variable length. (Allocated dynamically
967 * by attaching extra space to the end of the structure,
968 * depending on how many CPUs the kernel has booted up with)
970 * It is also be embedded into static data structures at build
971 * time. (See 'struct static_sched_domain' in kernel/sched.c)
973 unsigned long span[0];
976 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
978 return to_cpumask(sd->span);
981 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
982 struct sched_domain_attr *dattr_new);
984 /* Allocate an array of sched domains, for partition_sched_domains(). */
985 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
986 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
988 /* Test a flag in parent sched domain */
989 static inline int test_sd_parent(struct sched_domain *sd, int flag)
991 if (sd->parent && (sd->parent->flags & flag))
997 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
998 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1000 #else /* CONFIG_SMP */
1002 struct sched_domain_attr;
1005 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1006 struct sched_domain_attr *dattr_new)
1009 #endif /* !CONFIG_SMP */
1012 struct io_context; /* See blkdev.h */
1015 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1016 extern void prefetch_stack(struct task_struct *t);
1018 static inline void prefetch_stack(struct task_struct *t) { }
1021 struct audit_context; /* See audit.c */
1023 struct pipe_inode_info;
1024 struct uts_namespace;
1027 struct sched_domain;
1032 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1033 #define WF_FORK 0x02 /* child wakeup after fork */
1035 #define ENQUEUE_WAKEUP 1
1036 #define ENQUEUE_WAKING 2
1037 #define ENQUEUE_HEAD 4
1039 #define DEQUEUE_SLEEP 1
1041 struct sched_class {
1042 const struct sched_class *next;
1044 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1045 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1046 void (*yield_task) (struct rq *rq);
1048 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1050 struct task_struct * (*pick_next_task) (struct rq *rq);
1051 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1054 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1055 int sd_flag, int flags);
1057 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1058 void (*post_schedule) (struct rq *this_rq);
1059 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1060 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1062 void (*set_cpus_allowed)(struct task_struct *p,
1063 const struct cpumask *newmask);
1065 void (*rq_online)(struct rq *rq);
1066 void (*rq_offline)(struct rq *rq);
1069 void (*set_curr_task) (struct rq *rq);
1070 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1071 void (*task_fork) (struct task_struct *p);
1073 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1075 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1077 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1078 int oldprio, int running);
1080 unsigned int (*get_rr_interval) (struct rq *rq,
1081 struct task_struct *task);
1083 #ifdef CONFIG_FAIR_GROUP_SCHED
1084 void (*moved_group) (struct task_struct *p, int on_rq);
1088 struct load_weight {
1089 unsigned long weight, inv_weight;
1092 #ifdef CONFIG_SCHEDSTATS
1093 struct sched_statistics {
1103 s64 sum_sleep_runtime;
1110 u64 nr_migrations_cold;
1111 u64 nr_failed_migrations_affine;
1112 u64 nr_failed_migrations_running;
1113 u64 nr_failed_migrations_hot;
1114 u64 nr_forced_migrations;
1117 u64 nr_wakeups_sync;
1118 u64 nr_wakeups_migrate;
1119 u64 nr_wakeups_local;
1120 u64 nr_wakeups_remote;
1121 u64 nr_wakeups_affine;
1122 u64 nr_wakeups_affine_attempts;
1123 u64 nr_wakeups_passive;
1124 u64 nr_wakeups_idle;
1128 struct sched_entity {
1129 struct load_weight load; /* for load-balancing */
1130 struct rb_node run_node;
1131 struct list_head group_node;
1135 u64 sum_exec_runtime;
1137 u64 prev_sum_exec_runtime;
1141 #ifdef CONFIG_SCHEDSTATS
1142 struct sched_statistics statistics;
1145 #ifdef CONFIG_FAIR_GROUP_SCHED
1146 struct sched_entity *parent;
1147 /* rq on which this entity is (to be) queued: */
1148 struct cfs_rq *cfs_rq;
1149 /* rq "owned" by this entity/group: */
1150 struct cfs_rq *my_q;
1154 struct sched_rt_entity {
1155 struct list_head run_list;
1156 unsigned long timeout;
1157 unsigned int time_slice;
1158 int nr_cpus_allowed;
1160 struct sched_rt_entity *back;
1161 #ifdef CONFIG_RT_GROUP_SCHED
1162 struct sched_rt_entity *parent;
1163 /* rq on which this entity is (to be) queued: */
1164 struct rt_rq *rt_rq;
1165 /* rq "owned" by this entity/group: */
1172 struct task_struct {
1173 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1176 unsigned int flags; /* per process flags, defined below */
1177 unsigned int ptrace;
1179 int lock_depth; /* BKL lock depth */
1182 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1187 int prio, static_prio, normal_prio;
1188 unsigned int rt_priority;
1189 const struct sched_class *sched_class;
1190 struct sched_entity se;
1191 struct sched_rt_entity rt;
1193 #ifdef CONFIG_PREEMPT_NOTIFIERS
1194 /* list of struct preempt_notifier: */
1195 struct hlist_head preempt_notifiers;
1199 * fpu_counter contains the number of consecutive context switches
1200 * that the FPU is used. If this is over a threshold, the lazy fpu
1201 * saving becomes unlazy to save the trap. This is an unsigned char
1202 * so that after 256 times the counter wraps and the behavior turns
1203 * lazy again; this to deal with bursty apps that only use FPU for
1206 unsigned char fpu_counter;
1207 #ifdef CONFIG_BLK_DEV_IO_TRACE
1208 unsigned int btrace_seq;
1211 unsigned int policy;
1212 cpumask_t cpus_allowed;
1214 #ifdef CONFIG_TREE_PREEMPT_RCU
1215 int rcu_read_lock_nesting;
1216 char rcu_read_unlock_special;
1217 struct rcu_node *rcu_blocked_node;
1218 struct list_head rcu_node_entry;
1219 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1221 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1222 struct sched_info sched_info;
1225 struct list_head tasks;
1226 struct plist_node pushable_tasks;
1228 struct mm_struct *mm, *active_mm;
1229 #if defined(SPLIT_RSS_COUNTING)
1230 struct task_rss_stat rss_stat;
1234 int exit_code, exit_signal;
1235 int pdeath_signal; /* The signal sent when the parent dies */
1237 unsigned int personality;
1238 unsigned did_exec:1;
1239 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1241 unsigned in_iowait:1;
1244 /* Revert to default priority/policy when forking */
1245 unsigned sched_reset_on_fork:1;
1250 #ifdef CONFIG_CC_STACKPROTECTOR
1251 /* Canary value for the -fstack-protector gcc feature */
1252 unsigned long stack_canary;
1256 * pointers to (original) parent process, youngest child, younger sibling,
1257 * older sibling, respectively. (p->father can be replaced with
1258 * p->real_parent->pid)
1260 struct task_struct *real_parent; /* real parent process */
1261 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1263 * children/sibling forms the list of my natural children
1265 struct list_head children; /* list of my children */
1266 struct list_head sibling; /* linkage in my parent's children list */
1267 struct task_struct *group_leader; /* threadgroup leader */
1270 * ptraced is the list of tasks this task is using ptrace on.
1271 * This includes both natural children and PTRACE_ATTACH targets.
1272 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1274 struct list_head ptraced;
1275 struct list_head ptrace_entry;
1277 /* PID/PID hash table linkage. */
1278 struct pid_link pids[PIDTYPE_MAX];
1279 struct list_head thread_group;
1281 struct completion *vfork_done; /* for vfork() */
1282 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1283 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1285 cputime_t utime, stime, utimescaled, stimescaled;
1287 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1288 cputime_t prev_utime, prev_stime;
1290 unsigned long nvcsw, nivcsw; /* context switch counts */
1291 struct timespec start_time; /* monotonic time */
1292 struct timespec real_start_time; /* boot based time */
1293 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1294 unsigned long min_flt, maj_flt;
1296 struct task_cputime cputime_expires;
1297 struct list_head cpu_timers[3];
1299 /* process credentials */
1300 const struct cred *real_cred; /* objective and real subjective task
1301 * credentials (COW) */
1302 const struct cred *cred; /* effective (overridable) subjective task
1303 * credentials (COW) */
1304 struct mutex cred_guard_mutex; /* guard against foreign influences on
1305 * credential calculations
1306 * (notably. ptrace) */
1307 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1309 char comm[TASK_COMM_LEN]; /* executable name excluding path
1310 - access with [gs]et_task_comm (which lock
1311 it with task_lock())
1312 - initialized normally by setup_new_exec */
1313 /* file system info */
1314 int link_count, total_link_count;
1315 #ifdef CONFIG_SYSVIPC
1317 struct sysv_sem sysvsem;
1319 #ifdef CONFIG_DETECT_HUNG_TASK
1320 /* hung task detection */
1321 unsigned long last_switch_count;
1323 /* CPU-specific state of this task */
1324 struct thread_struct thread;
1325 /* filesystem information */
1326 struct fs_struct *fs;
1327 /* open file information */
1328 struct files_struct *files;
1330 struct nsproxy *nsproxy;
1331 /* signal handlers */
1332 struct signal_struct *signal;
1333 struct sighand_struct *sighand;
1335 sigset_t blocked, real_blocked;
1336 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1337 struct sigpending pending;
1339 unsigned long sas_ss_sp;
1341 int (*notifier)(void *priv);
1342 void *notifier_data;
1343 sigset_t *notifier_mask;
1344 struct audit_context *audit_context;
1345 #ifdef CONFIG_AUDITSYSCALL
1347 unsigned int sessionid;
1351 /* Thread group tracking */
1354 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1356 spinlock_t alloc_lock;
1358 #ifdef CONFIG_GENERIC_HARDIRQS
1359 /* IRQ handler threads */
1360 struct irqaction *irqaction;
1363 /* Protection of the PI data structures: */
1364 raw_spinlock_t pi_lock;
1366 #ifdef CONFIG_RT_MUTEXES
1367 /* PI waiters blocked on a rt_mutex held by this task */
1368 struct plist_head pi_waiters;
1369 /* Deadlock detection and priority inheritance handling */
1370 struct rt_mutex_waiter *pi_blocked_on;
1373 #ifdef CONFIG_DEBUG_MUTEXES
1374 /* mutex deadlock detection */
1375 struct mutex_waiter *blocked_on;
1377 #ifdef CONFIG_TRACE_IRQFLAGS
1378 unsigned int irq_events;
1379 unsigned long hardirq_enable_ip;
1380 unsigned long hardirq_disable_ip;
1381 unsigned int hardirq_enable_event;
1382 unsigned int hardirq_disable_event;
1383 int hardirqs_enabled;
1384 int hardirq_context;
1385 unsigned long softirq_disable_ip;
1386 unsigned long softirq_enable_ip;
1387 unsigned int softirq_disable_event;
1388 unsigned int softirq_enable_event;
1389 int softirqs_enabled;
1390 int softirq_context;
1392 #ifdef CONFIG_LOCKDEP
1393 # define MAX_LOCK_DEPTH 48UL
1396 unsigned int lockdep_recursion;
1397 struct held_lock held_locks[MAX_LOCK_DEPTH];
1398 gfp_t lockdep_reclaim_gfp;
1401 /* journalling filesystem info */
1404 /* stacked block device info */
1405 struct bio_list *bio_list;
1408 struct reclaim_state *reclaim_state;
1410 struct backing_dev_info *backing_dev_info;
1412 struct io_context *io_context;
1414 unsigned long ptrace_message;
1415 siginfo_t *last_siginfo; /* For ptrace use. */
1416 struct task_io_accounting ioac;
1417 #if defined(CONFIG_TASK_XACCT)
1418 u64 acct_rss_mem1; /* accumulated rss usage */
1419 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1420 cputime_t acct_timexpd; /* stime + utime since last update */
1422 #ifdef CONFIG_CPUSETS
1423 nodemask_t mems_allowed; /* Protected by alloc_lock */
1424 int mems_allowed_change_disable;
1425 int cpuset_mem_spread_rotor;
1427 #ifdef CONFIG_CGROUPS
1428 /* Control Group info protected by css_set_lock */
1429 struct css_set *cgroups;
1430 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1431 struct list_head cg_list;
1434 struct robust_list_head __user *robust_list;
1435 #ifdef CONFIG_COMPAT
1436 struct compat_robust_list_head __user *compat_robust_list;
1438 struct list_head pi_state_list;
1439 struct futex_pi_state *pi_state_cache;
1441 #ifdef CONFIG_PERF_EVENTS
1442 struct perf_event_context *perf_event_ctxp;
1443 struct mutex perf_event_mutex;
1444 struct list_head perf_event_list;
1447 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1450 atomic_t fs_excl; /* holding fs exclusive resources */
1451 struct rcu_head rcu;
1454 * cache last used pipe for splice
1456 struct pipe_inode_info *splice_pipe;
1457 #ifdef CONFIG_TASK_DELAY_ACCT
1458 struct task_delay_info *delays;
1460 #ifdef CONFIG_FAULT_INJECTION
1463 struct prop_local_single dirties;
1464 #ifdef CONFIG_LATENCYTOP
1465 int latency_record_count;
1466 struct latency_record latency_record[LT_SAVECOUNT];
1469 * time slack values; these are used to round up poll() and
1470 * select() etc timeout values. These are in nanoseconds.
1472 unsigned long timer_slack_ns;
1473 unsigned long default_timer_slack_ns;
1475 struct list_head *scm_work_list;
1476 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1477 /* Index of current stored address in ret_stack */
1479 /* Stack of return addresses for return function tracing */
1480 struct ftrace_ret_stack *ret_stack;
1481 /* time stamp for last schedule */
1482 unsigned long long ftrace_timestamp;
1484 * Number of functions that haven't been traced
1485 * because of depth overrun.
1487 atomic_t trace_overrun;
1488 /* Pause for the tracing */
1489 atomic_t tracing_graph_pause;
1491 #ifdef CONFIG_TRACING
1492 /* state flags for use by tracers */
1493 unsigned long trace;
1494 /* bitmask of trace recursion */
1495 unsigned long trace_recursion;
1496 #endif /* CONFIG_TRACING */
1497 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1498 struct memcg_batch_info {
1499 int do_batch; /* incremented when batch uncharge started */
1500 struct mem_cgroup *memcg; /* target memcg of uncharge */
1501 unsigned long bytes; /* uncharged usage */
1502 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1507 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1508 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1511 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1512 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1513 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1514 * values are inverted: lower p->prio value means higher priority.
1516 * The MAX_USER_RT_PRIO value allows the actual maximum
1517 * RT priority to be separate from the value exported to
1518 * user-space. This allows kernel threads to set their
1519 * priority to a value higher than any user task. Note:
1520 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1523 #define MAX_USER_RT_PRIO 100
1524 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1526 #define MAX_PRIO (MAX_RT_PRIO + 40)
1527 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1529 static inline int rt_prio(int prio)
1531 if (unlikely(prio < MAX_RT_PRIO))
1536 static inline int rt_task(struct task_struct *p)
1538 return rt_prio(p->prio);
1541 static inline struct pid *task_pid(struct task_struct *task)
1543 return task->pids[PIDTYPE_PID].pid;
1546 static inline struct pid *task_tgid(struct task_struct *task)
1548 return task->group_leader->pids[PIDTYPE_PID].pid;
1552 * Without tasklist or rcu lock it is not safe to dereference
1553 * the result of task_pgrp/task_session even if task == current,
1554 * we can race with another thread doing sys_setsid/sys_setpgid.
1556 static inline struct pid *task_pgrp(struct task_struct *task)
1558 return task->group_leader->pids[PIDTYPE_PGID].pid;
1561 static inline struct pid *task_session(struct task_struct *task)
1563 return task->group_leader->pids[PIDTYPE_SID].pid;
1566 struct pid_namespace;
1569 * the helpers to get the task's different pids as they are seen
1570 * from various namespaces
1572 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1573 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1575 * task_xid_nr_ns() : id seen from the ns specified;
1577 * set_task_vxid() : assigns a virtual id to a task;
1579 * see also pid_nr() etc in include/linux/pid.h
1581 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1582 struct pid_namespace *ns);
1584 static inline pid_t task_pid_nr(struct task_struct *tsk)
1589 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1590 struct pid_namespace *ns)
1592 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1595 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1597 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1601 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1606 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1608 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1610 return pid_vnr(task_tgid(tsk));
1614 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1615 struct pid_namespace *ns)
1617 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1620 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1622 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1626 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1627 struct pid_namespace *ns)
1629 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1632 static inline pid_t task_session_vnr(struct task_struct *tsk)
1634 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1637 /* obsolete, do not use */
1638 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1640 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1644 * pid_alive - check that a task structure is not stale
1645 * @p: Task structure to be checked.
1647 * Test if a process is not yet dead (at most zombie state)
1648 * If pid_alive fails, then pointers within the task structure
1649 * can be stale and must not be dereferenced.
1651 static inline int pid_alive(struct task_struct *p)
1653 return p->pids[PIDTYPE_PID].pid != NULL;
1657 * is_global_init - check if a task structure is init
1658 * @tsk: Task structure to be checked.
1660 * Check if a task structure is the first user space task the kernel created.
1662 static inline int is_global_init(struct task_struct *tsk)
1664 return tsk->pid == 1;
1668 * is_container_init:
1669 * check whether in the task is init in its own pid namespace.
1671 extern int is_container_init(struct task_struct *tsk);
1673 extern struct pid *cad_pid;
1675 extern void free_task(struct task_struct *tsk);
1676 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1678 extern void __put_task_struct(struct task_struct *t);
1680 static inline void put_task_struct(struct task_struct *t)
1682 if (atomic_dec_and_test(&t->usage))
1683 __put_task_struct(t);
1686 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1687 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1692 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1693 /* Not implemented yet, only for 486*/
1694 #define PF_STARTING 0x00000002 /* being created */
1695 #define PF_EXITING 0x00000004 /* getting shut down */
1696 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1697 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1698 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1699 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1700 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1701 #define PF_DUMPCORE 0x00000200 /* dumped core */
1702 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1703 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1704 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1705 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1706 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1707 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1708 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1709 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1710 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1711 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1712 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1713 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1714 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1715 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1716 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1717 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1718 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1719 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1720 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1721 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1722 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1723 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1726 * Only the _current_ task can read/write to tsk->flags, but other
1727 * tasks can access tsk->flags in readonly mode for example
1728 * with tsk_used_math (like during threaded core dumping).
1729 * There is however an exception to this rule during ptrace
1730 * or during fork: the ptracer task is allowed to write to the
1731 * child->flags of its traced child (same goes for fork, the parent
1732 * can write to the child->flags), because we're guaranteed the
1733 * child is not running and in turn not changing child->flags
1734 * at the same time the parent does it.
1736 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1737 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1738 #define clear_used_math() clear_stopped_child_used_math(current)
1739 #define set_used_math() set_stopped_child_used_math(current)
1740 #define conditional_stopped_child_used_math(condition, child) \
1741 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1742 #define conditional_used_math(condition) \
1743 conditional_stopped_child_used_math(condition, current)
1744 #define copy_to_stopped_child_used_math(child) \
1745 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1746 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1747 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1748 #define used_math() tsk_used_math(current)
1750 #ifdef CONFIG_TREE_PREEMPT_RCU
1752 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1753 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1755 static inline void rcu_copy_process(struct task_struct *p)
1757 p->rcu_read_lock_nesting = 0;
1758 p->rcu_read_unlock_special = 0;
1759 p->rcu_blocked_node = NULL;
1760 INIT_LIST_HEAD(&p->rcu_node_entry);
1765 static inline void rcu_copy_process(struct task_struct *p)
1772 extern int set_cpus_allowed_ptr(struct task_struct *p,
1773 const struct cpumask *new_mask);
1775 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1776 const struct cpumask *new_mask)
1778 if (!cpumask_test_cpu(0, new_mask))
1784 #ifndef CONFIG_CPUMASK_OFFSTACK
1785 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1787 return set_cpus_allowed_ptr(p, &new_mask);
1792 * Architectures can set this to 1 if they have specified
1793 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1794 * but then during bootup it turns out that sched_clock()
1795 * is reliable after all:
1797 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1798 extern int sched_clock_stable;
1801 /* ftrace calls sched_clock() directly */
1802 extern unsigned long long notrace sched_clock(void);
1804 extern void sched_clock_init(void);
1805 extern u64 sched_clock_cpu(int cpu);
1807 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1808 static inline void sched_clock_tick(void)
1812 static inline void sched_clock_idle_sleep_event(void)
1816 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1820 extern void sched_clock_tick(void);
1821 extern void sched_clock_idle_sleep_event(void);
1822 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1826 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1827 * clock constructed from sched_clock():
1829 extern unsigned long long cpu_clock(int cpu);
1831 extern unsigned long long
1832 task_sched_runtime(struct task_struct *task);
1833 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1835 /* sched_exec is called by processes performing an exec */
1837 extern void sched_exec(void);
1839 #define sched_exec() {}
1842 extern void sched_clock_idle_sleep_event(void);
1843 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1845 #ifdef CONFIG_HOTPLUG_CPU
1846 extern void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p);
1847 extern void idle_task_exit(void);
1849 static inline void idle_task_exit(void) {}
1852 extern void sched_idle_next(void);
1854 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1855 extern void wake_up_idle_cpu(int cpu);
1857 static inline void wake_up_idle_cpu(int cpu) { }
1860 extern unsigned int sysctl_sched_latency;
1861 extern unsigned int sysctl_sched_min_granularity;
1862 extern unsigned int sysctl_sched_wakeup_granularity;
1863 extern unsigned int sysctl_sched_shares_ratelimit;
1864 extern unsigned int sysctl_sched_shares_thresh;
1865 extern unsigned int sysctl_sched_child_runs_first;
1867 enum sched_tunable_scaling {
1868 SCHED_TUNABLESCALING_NONE,
1869 SCHED_TUNABLESCALING_LOG,
1870 SCHED_TUNABLESCALING_LINEAR,
1871 SCHED_TUNABLESCALING_END,
1873 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1875 #ifdef CONFIG_SCHED_DEBUG
1876 extern unsigned int sysctl_sched_migration_cost;
1877 extern unsigned int sysctl_sched_nr_migrate;
1878 extern unsigned int sysctl_sched_time_avg;
1879 extern unsigned int sysctl_timer_migration;
1881 int sched_proc_update_handler(struct ctl_table *table, int write,
1882 void __user *buffer, size_t *length,
1885 #ifdef CONFIG_SCHED_DEBUG
1886 static inline unsigned int get_sysctl_timer_migration(void)
1888 return sysctl_timer_migration;
1891 static inline unsigned int get_sysctl_timer_migration(void)
1896 extern unsigned int sysctl_sched_rt_period;
1897 extern int sysctl_sched_rt_runtime;
1899 int sched_rt_handler(struct ctl_table *table, int write,
1900 void __user *buffer, size_t *lenp,
1903 extern unsigned int sysctl_sched_compat_yield;
1905 #ifdef CONFIG_RT_MUTEXES
1906 extern int rt_mutex_getprio(struct task_struct *p);
1907 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1908 extern void rt_mutex_adjust_pi(struct task_struct *p);
1910 static inline int rt_mutex_getprio(struct task_struct *p)
1912 return p->normal_prio;
1914 # define rt_mutex_adjust_pi(p) do { } while (0)
1917 extern void set_user_nice(struct task_struct *p, long nice);
1918 extern int task_prio(const struct task_struct *p);
1919 extern int task_nice(const struct task_struct *p);
1920 extern int can_nice(const struct task_struct *p, const int nice);
1921 extern int task_curr(const struct task_struct *p);
1922 extern int idle_cpu(int cpu);
1923 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1924 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1925 struct sched_param *);
1926 extern struct task_struct *idle_task(int cpu);
1927 extern struct task_struct *curr_task(int cpu);
1928 extern void set_curr_task(int cpu, struct task_struct *p);
1933 * The default (Linux) execution domain.
1935 extern struct exec_domain default_exec_domain;
1937 union thread_union {
1938 struct thread_info thread_info;
1939 unsigned long stack[THREAD_SIZE/sizeof(long)];
1942 #ifndef __HAVE_ARCH_KSTACK_END
1943 static inline int kstack_end(void *addr)
1945 /* Reliable end of stack detection:
1946 * Some APM bios versions misalign the stack
1948 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1952 extern union thread_union init_thread_union;
1953 extern struct task_struct init_task;
1955 extern struct mm_struct init_mm;
1957 extern struct pid_namespace init_pid_ns;
1960 * find a task by one of its numerical ids
1962 * find_task_by_pid_ns():
1963 * finds a task by its pid in the specified namespace
1964 * find_task_by_vpid():
1965 * finds a task by its virtual pid
1967 * see also find_vpid() etc in include/linux/pid.h
1970 extern struct task_struct *find_task_by_vpid(pid_t nr);
1971 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1972 struct pid_namespace *ns);
1974 extern void __set_special_pids(struct pid *pid);
1976 /* per-UID process charging. */
1977 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1978 static inline struct user_struct *get_uid(struct user_struct *u)
1980 atomic_inc(&u->__count);
1983 extern void free_uid(struct user_struct *);
1984 extern void release_uids(struct user_namespace *ns);
1986 #include <asm/current.h>
1988 extern void do_timer(unsigned long ticks);
1990 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1991 extern int wake_up_process(struct task_struct *tsk);
1992 extern void wake_up_new_task(struct task_struct *tsk,
1993 unsigned long clone_flags);
1995 extern void kick_process(struct task_struct *tsk);
1997 static inline void kick_process(struct task_struct *tsk) { }
1999 extern void sched_fork(struct task_struct *p, int clone_flags);
2000 extern void sched_dead(struct task_struct *p);
2002 extern void proc_caches_init(void);
2003 extern void flush_signals(struct task_struct *);
2004 extern void __flush_signals(struct task_struct *);
2005 extern void ignore_signals(struct task_struct *);
2006 extern void flush_signal_handlers(struct task_struct *, int force_default);
2007 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2009 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2011 unsigned long flags;
2014 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2015 ret = dequeue_signal(tsk, mask, info);
2016 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2021 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2023 extern void unblock_all_signals(void);
2024 extern void release_task(struct task_struct * p);
2025 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2026 extern int force_sigsegv(int, struct task_struct *);
2027 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2028 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2029 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2030 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2031 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2032 extern int kill_pid(struct pid *pid, int sig, int priv);
2033 extern int kill_proc_info(int, struct siginfo *, pid_t);
2034 extern int do_notify_parent(struct task_struct *, int);
2035 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2036 extern void force_sig(int, struct task_struct *);
2037 extern int send_sig(int, struct task_struct *, int);
2038 extern void zap_other_threads(struct task_struct *p);
2039 extern struct sigqueue *sigqueue_alloc(void);
2040 extern void sigqueue_free(struct sigqueue *);
2041 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2042 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2043 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2045 static inline int kill_cad_pid(int sig, int priv)
2047 return kill_pid(cad_pid, sig, priv);
2050 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2051 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2052 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2053 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2056 * True if we are on the alternate signal stack.
2058 static inline int on_sig_stack(unsigned long sp)
2060 #ifdef CONFIG_STACK_GROWSUP
2061 return sp >= current->sas_ss_sp &&
2062 sp - current->sas_ss_sp < current->sas_ss_size;
2064 return sp > current->sas_ss_sp &&
2065 sp - current->sas_ss_sp <= current->sas_ss_size;
2069 static inline int sas_ss_flags(unsigned long sp)
2071 return (current->sas_ss_size == 0 ? SS_DISABLE
2072 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2076 * Routines for handling mm_structs
2078 extern struct mm_struct * mm_alloc(void);
2080 /* mmdrop drops the mm and the page tables */
2081 extern void __mmdrop(struct mm_struct *);
2082 static inline void mmdrop(struct mm_struct * mm)
2084 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2088 /* mmput gets rid of the mappings and all user-space */
2089 extern void mmput(struct mm_struct *);
2090 /* Grab a reference to a task's mm, if it is not already going away */
2091 extern struct mm_struct *get_task_mm(struct task_struct *task);
2092 /* Remove the current tasks stale references to the old mm_struct */
2093 extern void mm_release(struct task_struct *, struct mm_struct *);
2094 /* Allocate a new mm structure and copy contents from tsk->mm */
2095 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2097 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2098 struct task_struct *, struct pt_regs *);
2099 extern void flush_thread(void);
2100 extern void exit_thread(void);
2102 extern void exit_files(struct task_struct *);
2103 extern void __cleanup_signal(struct signal_struct *);
2104 extern void __cleanup_sighand(struct sighand_struct *);
2106 extern void exit_itimers(struct signal_struct *);
2107 extern void flush_itimer_signals(void);
2109 extern NORET_TYPE void do_group_exit(int);
2111 extern void daemonize(const char *, ...);
2112 extern int allow_signal(int);
2113 extern int disallow_signal(int);
2115 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2116 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2117 struct task_struct *fork_idle(int);
2119 extern void set_task_comm(struct task_struct *tsk, char *from);
2120 extern char *get_task_comm(char *to, struct task_struct *tsk);
2123 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2125 static inline unsigned long wait_task_inactive(struct task_struct *p,
2132 #define next_task(p) \
2133 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2135 #define for_each_process(p) \
2136 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2138 extern bool current_is_single_threaded(void);
2141 * Careful: do_each_thread/while_each_thread is a double loop so
2142 * 'break' will not work as expected - use goto instead.
2144 #define do_each_thread(g, t) \
2145 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2147 #define while_each_thread(g, t) \
2148 while ((t = next_thread(t)) != g)
2150 /* de_thread depends on thread_group_leader not being a pid based check */
2151 #define thread_group_leader(p) (p == p->group_leader)
2153 /* Do to the insanities of de_thread it is possible for a process
2154 * to have the pid of the thread group leader without actually being
2155 * the thread group leader. For iteration through the pids in proc
2156 * all we care about is that we have a task with the appropriate
2157 * pid, we don't actually care if we have the right task.
2159 static inline int has_group_leader_pid(struct task_struct *p)
2161 return p->pid == p->tgid;
2165 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2167 return p1->tgid == p2->tgid;
2170 static inline struct task_struct *next_thread(const struct task_struct *p)
2172 return list_entry_rcu(p->thread_group.next,
2173 struct task_struct, thread_group);
2176 static inline int thread_group_empty(struct task_struct *p)
2178 return list_empty(&p->thread_group);
2181 #define delay_group_leader(p) \
2182 (thread_group_leader(p) && !thread_group_empty(p))
2184 static inline int task_detached(struct task_struct *p)
2186 return p->exit_signal == -1;
2190 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2191 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2192 * pins the final release of task.io_context. Also protects ->cpuset and
2193 * ->cgroup.subsys[].
2195 * Nests both inside and outside of read_lock(&tasklist_lock).
2196 * It must not be nested with write_lock_irq(&tasklist_lock),
2197 * neither inside nor outside.
2199 static inline void task_lock(struct task_struct *p)
2201 spin_lock(&p->alloc_lock);
2204 static inline void task_unlock(struct task_struct *p)
2206 spin_unlock(&p->alloc_lock);
2209 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2210 unsigned long *flags);
2212 static inline void unlock_task_sighand(struct task_struct *tsk,
2213 unsigned long *flags)
2215 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2218 #ifndef __HAVE_THREAD_FUNCTIONS
2220 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2221 #define task_stack_page(task) ((task)->stack)
2223 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2225 *task_thread_info(p) = *task_thread_info(org);
2226 task_thread_info(p)->task = p;
2229 static inline unsigned long *end_of_stack(struct task_struct *p)
2231 return (unsigned long *)(task_thread_info(p) + 1);
2236 static inline int object_is_on_stack(void *obj)
2238 void *stack = task_stack_page(current);
2240 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2243 extern void thread_info_cache_init(void);
2245 #ifdef CONFIG_DEBUG_STACK_USAGE
2246 static inline unsigned long stack_not_used(struct task_struct *p)
2248 unsigned long *n = end_of_stack(p);
2250 do { /* Skip over canary */
2254 return (unsigned long)n - (unsigned long)end_of_stack(p);
2258 /* set thread flags in other task's structures
2259 * - see asm/thread_info.h for TIF_xxxx flags available
2261 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2263 set_ti_thread_flag(task_thread_info(tsk), flag);
2266 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2268 clear_ti_thread_flag(task_thread_info(tsk), flag);
2271 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2273 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2276 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2278 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2281 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2283 return test_ti_thread_flag(task_thread_info(tsk), flag);
2286 static inline void set_tsk_need_resched(struct task_struct *tsk)
2288 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2291 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2293 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2296 static inline int test_tsk_need_resched(struct task_struct *tsk)
2298 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2301 static inline int restart_syscall(void)
2303 set_tsk_thread_flag(current, TIF_SIGPENDING);
2304 return -ERESTARTNOINTR;
2307 static inline int signal_pending(struct task_struct *p)
2309 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2312 static inline int __fatal_signal_pending(struct task_struct *p)
2314 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2317 static inline int fatal_signal_pending(struct task_struct *p)
2319 return signal_pending(p) && __fatal_signal_pending(p);
2322 static inline int signal_pending_state(long state, struct task_struct *p)
2324 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2326 if (!signal_pending(p))
2329 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2332 static inline int need_resched(void)
2334 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2338 * cond_resched() and cond_resched_lock(): latency reduction via
2339 * explicit rescheduling in places that are safe. The return
2340 * value indicates whether a reschedule was done in fact.
2341 * cond_resched_lock() will drop the spinlock before scheduling,
2342 * cond_resched_softirq() will enable bhs before scheduling.
2344 extern int _cond_resched(void);
2346 #define cond_resched() ({ \
2347 __might_sleep(__FILE__, __LINE__, 0); \
2351 extern int __cond_resched_lock(spinlock_t *lock);
2353 #ifdef CONFIG_PREEMPT
2354 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2356 #define PREEMPT_LOCK_OFFSET 0
2359 #define cond_resched_lock(lock) ({ \
2360 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2361 __cond_resched_lock(lock); \
2364 extern int __cond_resched_softirq(void);
2366 #define cond_resched_softirq() ({ \
2367 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2368 __cond_resched_softirq(); \
2372 * Does a critical section need to be broken due to another
2373 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2374 * but a general need for low latency)
2376 static inline int spin_needbreak(spinlock_t *lock)
2378 #ifdef CONFIG_PREEMPT
2379 return spin_is_contended(lock);
2386 * Thread group CPU time accounting.
2388 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2389 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2391 static inline void thread_group_cputime_init(struct signal_struct *sig)
2393 spin_lock_init(&sig->cputimer.lock);
2396 static inline void thread_group_cputime_free(struct signal_struct *sig)
2401 * Reevaluate whether the task has signals pending delivery.
2402 * Wake the task if so.
2403 * This is required every time the blocked sigset_t changes.
2404 * callers must hold sighand->siglock.
2406 extern void recalc_sigpending_and_wake(struct task_struct *t);
2407 extern void recalc_sigpending(void);
2409 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2412 * Wrappers for p->thread_info->cpu access. No-op on UP.
2416 static inline unsigned int task_cpu(const struct task_struct *p)
2418 return task_thread_info(p)->cpu;
2421 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2425 static inline unsigned int task_cpu(const struct task_struct *p)
2430 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2434 #endif /* CONFIG_SMP */
2436 #ifdef CONFIG_TRACING
2438 __trace_special(void *__tr, void *__data,
2439 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2442 __trace_special(void *__tr, void *__data,
2443 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2448 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2449 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2451 extern void normalize_rt_tasks(void);
2453 #ifdef CONFIG_CGROUP_SCHED
2455 extern struct task_group init_task_group;
2457 extern struct task_group *sched_create_group(struct task_group *parent);
2458 extern void sched_destroy_group(struct task_group *tg);
2459 extern void sched_move_task(struct task_struct *tsk);
2460 #ifdef CONFIG_FAIR_GROUP_SCHED
2461 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2462 extern unsigned long sched_group_shares(struct task_group *tg);
2464 #ifdef CONFIG_RT_GROUP_SCHED
2465 extern int sched_group_set_rt_runtime(struct task_group *tg,
2466 long rt_runtime_us);
2467 extern long sched_group_rt_runtime(struct task_group *tg);
2468 extern int sched_group_set_rt_period(struct task_group *tg,
2470 extern long sched_group_rt_period(struct task_group *tg);
2471 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2475 extern int task_can_switch_user(struct user_struct *up,
2476 struct task_struct *tsk);
2478 #ifdef CONFIG_TASK_XACCT
2479 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2481 tsk->ioac.rchar += amt;
2484 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2486 tsk->ioac.wchar += amt;
2489 static inline void inc_syscr(struct task_struct *tsk)
2494 static inline void inc_syscw(struct task_struct *tsk)
2499 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2503 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2507 static inline void inc_syscr(struct task_struct *tsk)
2511 static inline void inc_syscw(struct task_struct *tsk)
2516 #ifndef TASK_SIZE_OF
2517 #define TASK_SIZE_OF(tsk) TASK_SIZE
2521 * Call the function if the target task is executing on a CPU right now:
2523 extern void task_oncpu_function_call(struct task_struct *p,
2524 void (*func) (void *info), void *info);
2527 #ifdef CONFIG_MM_OWNER
2528 extern void mm_update_next_owner(struct mm_struct *mm);
2529 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2531 static inline void mm_update_next_owner(struct mm_struct *mm)
2535 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2538 #endif /* CONFIG_MM_OWNER */
2540 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2543 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2546 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2549 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2552 static inline unsigned long rlimit(unsigned int limit)
2554 return task_rlimit(current, limit);
2557 static inline unsigned long rlimit_max(unsigned int limit)
2559 return task_rlimit_max(current, limit);
2562 #endif /* __KERNEL__ */