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(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_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;
513 * NOTE! "signal_struct" does not have it's own
514 * locking, because a shared signal_struct always
515 * implies a shared sighand_struct, so locking
516 * sighand_struct is always a proper superset of
517 * the locking of signal_struct.
519 struct signal_struct {
524 wait_queue_head_t wait_chldexit; /* for wait4() */
526 /* current thread group signal load-balancing target: */
527 struct task_struct *curr_target;
529 /* shared signal handling: */
530 struct sigpending shared_pending;
532 /* thread group exit support */
535 * - notify group_exit_task when ->count is equal to notify_count
536 * - everyone except group_exit_task is stopped during signal delivery
537 * of fatal signals, group_exit_task processes the signal.
540 struct task_struct *group_exit_task;
542 /* thread group stop support, overloads group_exit_code too */
543 int group_stop_count;
544 unsigned int flags; /* see SIGNAL_* flags below */
546 /* POSIX.1b Interval Timers */
547 struct list_head posix_timers;
549 /* ITIMER_REAL timer for the process */
550 struct hrtimer real_timer;
551 struct pid *leader_pid;
552 ktime_t it_real_incr;
555 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
556 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
557 * values are defined to 0 and 1 respectively
559 struct cpu_itimer it[2];
562 * Thread group totals for process CPU timers.
563 * See thread_group_cputimer(), et al, for details.
565 struct thread_group_cputimer cputimer;
567 /* Earliest-expiration cache. */
568 struct task_cputime cputime_expires;
570 struct list_head cpu_timers[3];
572 struct pid *tty_old_pgrp;
574 /* boolean value for session group leader */
577 struct tty_struct *tty; /* NULL if no tty */
580 * Cumulative resource counters for dead threads in the group,
581 * and for reaped dead child processes forked by this group.
582 * Live threads maintain their own counters and add to these
583 * in __exit_signal, except for the group leader.
585 cputime_t utime, stime, cutime, cstime;
588 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
589 cputime_t prev_utime, prev_stime;
591 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
592 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
593 unsigned long inblock, oublock, cinblock, coublock;
594 unsigned long maxrss, cmaxrss;
595 struct task_io_accounting ioac;
598 * Cumulative ns of schedule CPU time fo dead threads in the
599 * group, not including a zombie group leader, (This only differs
600 * from jiffies_to_ns(utime + stime) if sched_clock uses something
601 * other than jiffies.)
603 unsigned long long sum_sched_runtime;
606 * We don't bother to synchronize most readers of this at all,
607 * because there is no reader checking a limit that actually needs
608 * to get both rlim_cur and rlim_max atomically, and either one
609 * alone is a single word that can safely be read normally.
610 * getrlimit/setrlimit use task_lock(current->group_leader) to
611 * protect this instead of the siglock, because they really
612 * have no need to disable irqs.
614 struct rlimit rlim[RLIM_NLIMITS];
616 #ifdef CONFIG_BSD_PROCESS_ACCT
617 struct pacct_struct pacct; /* per-process accounting information */
619 #ifdef CONFIG_TASKSTATS
620 struct taskstats *stats;
624 struct tty_audit_buf *tty_audit_buf;
627 int oom_adj; /* OOM kill score adjustment (bit shift) */
628 int oom_score_adj; /* OOM kill score adjustment */
631 /* Context switch must be unlocked if interrupts are to be enabled */
632 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
633 # define __ARCH_WANT_UNLOCKED_CTXSW
637 * Bits in flags field of signal_struct.
639 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
640 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
641 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
642 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
644 * Pending notifications to parent.
646 #define SIGNAL_CLD_STOPPED 0x00000010
647 #define SIGNAL_CLD_CONTINUED 0x00000020
648 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
650 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
652 /* If true, all threads except ->group_exit_task have pending SIGKILL */
653 static inline int signal_group_exit(const struct signal_struct *sig)
655 return (sig->flags & SIGNAL_GROUP_EXIT) ||
656 (sig->group_exit_task != NULL);
660 * Some day this will be a full-fledged user tracking system..
663 atomic_t __count; /* reference count */
664 atomic_t processes; /* How many processes does this user have? */
665 atomic_t files; /* How many open files does this user have? */
666 atomic_t sigpending; /* How many pending signals does this user have? */
667 #ifdef CONFIG_INOTIFY_USER
668 atomic_t inotify_watches; /* How many inotify watches does this user have? */
669 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
672 atomic_t epoll_watches; /* The number of file descriptors currently watched */
674 #ifdef CONFIG_POSIX_MQUEUE
675 /* protected by mq_lock */
676 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
678 unsigned long locked_shm; /* How many pages of mlocked shm ? */
681 struct key *uid_keyring; /* UID specific keyring */
682 struct key *session_keyring; /* UID's default session keyring */
685 /* Hash table maintenance information */
686 struct hlist_node uidhash_node;
688 struct user_namespace *user_ns;
690 #ifdef CONFIG_PERF_EVENTS
691 atomic_long_t locked_vm;
695 extern int uids_sysfs_init(void);
697 extern struct user_struct *find_user(uid_t);
699 extern struct user_struct root_user;
700 #define INIT_USER (&root_user)
703 struct backing_dev_info;
704 struct reclaim_state;
706 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
708 /* cumulative counters */
709 unsigned long pcount; /* # of times run on this cpu */
710 unsigned long long run_delay; /* time spent waiting on a runqueue */
713 unsigned long long last_arrival,/* when we last ran on a cpu */
714 last_queued; /* when we were last queued to run */
715 #ifdef CONFIG_SCHEDSTATS
717 unsigned int bkl_count;
720 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
722 #ifdef CONFIG_TASK_DELAY_ACCT
723 struct task_delay_info {
725 unsigned int flags; /* Private per-task flags */
727 /* For each stat XXX, add following, aligned appropriately
729 * struct timespec XXX_start, XXX_end;
733 * Atomicity of updates to XXX_delay, XXX_count protected by
734 * single lock above (split into XXX_lock if contention is an issue).
738 * XXX_count is incremented on every XXX operation, the delay
739 * associated with the operation is added to XXX_delay.
740 * XXX_delay contains the accumulated delay time in nanoseconds.
742 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
743 u64 blkio_delay; /* wait for sync block io completion */
744 u64 swapin_delay; /* wait for swapin block io completion */
745 u32 blkio_count; /* total count of the number of sync block */
746 /* io operations performed */
747 u32 swapin_count; /* total count of the number of swapin block */
748 /* io operations performed */
750 struct timespec freepages_start, freepages_end;
751 u64 freepages_delay; /* wait for memory reclaim */
752 u32 freepages_count; /* total count of memory reclaim */
754 #endif /* CONFIG_TASK_DELAY_ACCT */
756 static inline int sched_info_on(void)
758 #ifdef CONFIG_SCHEDSTATS
760 #elif defined(CONFIG_TASK_DELAY_ACCT)
761 extern int delayacct_on;
776 * sched-domains (multiprocessor balancing) declarations:
780 * Increase resolution of nice-level calculations:
782 #define SCHED_LOAD_SHIFT 10
783 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
785 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
788 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
789 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
790 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
791 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
792 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
793 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
794 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
795 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
796 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
797 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
798 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
799 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
800 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
802 enum powersavings_balance_level {
803 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
804 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
805 * first for long running threads
807 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
808 * cpu package for power savings
810 MAX_POWERSAVINGS_BALANCE_LEVELS
813 extern int sched_mc_power_savings, sched_smt_power_savings;
815 static inline int sd_balance_for_mc_power(void)
817 if (sched_smt_power_savings)
818 return SD_POWERSAVINGS_BALANCE;
820 if (!sched_mc_power_savings)
821 return SD_PREFER_SIBLING;
826 static inline int sd_balance_for_package_power(void)
828 if (sched_mc_power_savings | sched_smt_power_savings)
829 return SD_POWERSAVINGS_BALANCE;
831 return SD_PREFER_SIBLING;
834 extern int __weak arch_sd_sibiling_asym_packing(void);
837 * Optimise SD flags for power savings:
838 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
839 * Keep default SD flags if sched_{smt,mc}_power_saving=0
842 static inline int sd_power_saving_flags(void)
844 if (sched_mc_power_savings | sched_smt_power_savings)
845 return SD_BALANCE_NEWIDLE;
851 struct sched_group *next; /* Must be a circular list */
854 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
857 unsigned int cpu_power, cpu_power_orig;
860 * The CPUs this group covers.
862 * NOTE: this field is variable length. (Allocated dynamically
863 * by attaching extra space to the end of the structure,
864 * depending on how many CPUs the kernel has booted up with)
866 * It is also be embedded into static data structures at build
867 * time. (See 'struct static_sched_group' in kernel/sched.c)
869 unsigned long cpumask[0];
872 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
874 return to_cpumask(sg->cpumask);
877 enum sched_domain_level {
887 struct sched_domain_attr {
888 int relax_domain_level;
891 #define SD_ATTR_INIT (struct sched_domain_attr) { \
892 .relax_domain_level = -1, \
895 struct sched_domain {
896 /* These fields must be setup */
897 struct sched_domain *parent; /* top domain must be null terminated */
898 struct sched_domain *child; /* bottom domain must be null terminated */
899 struct sched_group *groups; /* the balancing groups of the domain */
900 unsigned long min_interval; /* Minimum balance interval ms */
901 unsigned long max_interval; /* Maximum balance interval ms */
902 unsigned int busy_factor; /* less balancing by factor if busy */
903 unsigned int imbalance_pct; /* No balance until over watermark */
904 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
905 unsigned int busy_idx;
906 unsigned int idle_idx;
907 unsigned int newidle_idx;
908 unsigned int wake_idx;
909 unsigned int forkexec_idx;
910 unsigned int smt_gain;
911 int flags; /* See SD_* */
912 enum sched_domain_level level;
914 /* Runtime fields. */
915 unsigned long last_balance; /* init to jiffies. units in jiffies */
916 unsigned int balance_interval; /* initialise to 1. units in ms. */
917 unsigned int nr_balance_failed; /* initialise to 0 */
921 #ifdef CONFIG_SCHEDSTATS
922 /* load_balance() stats */
923 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
924 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
925 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
926 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
927 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
928 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
929 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
932 /* Active load balancing */
933 unsigned int alb_count;
934 unsigned int alb_failed;
935 unsigned int alb_pushed;
937 /* SD_BALANCE_EXEC stats */
938 unsigned int sbe_count;
939 unsigned int sbe_balanced;
940 unsigned int sbe_pushed;
942 /* SD_BALANCE_FORK stats */
943 unsigned int sbf_count;
944 unsigned int sbf_balanced;
945 unsigned int sbf_pushed;
947 /* try_to_wake_up() stats */
948 unsigned int ttwu_wake_remote;
949 unsigned int ttwu_move_affine;
950 unsigned int ttwu_move_balance;
952 #ifdef CONFIG_SCHED_DEBUG
956 unsigned int span_weight;
958 * Span of all CPUs in this domain.
960 * NOTE: this field is variable length. (Allocated dynamically
961 * by attaching extra space to the end of the structure,
962 * depending on how many CPUs the kernel has booted up with)
964 * It is also be embedded into static data structures at build
965 * time. (See 'struct static_sched_domain' in kernel/sched.c)
967 unsigned long span[0];
970 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
972 return to_cpumask(sd->span);
975 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
976 struct sched_domain_attr *dattr_new);
978 /* Allocate an array of sched domains, for partition_sched_domains(). */
979 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
980 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
982 /* Test a flag in parent sched domain */
983 static inline int test_sd_parent(struct sched_domain *sd, int flag)
985 if (sd->parent && (sd->parent->flags & flag))
991 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
992 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
994 #else /* CONFIG_SMP */
996 struct sched_domain_attr;
999 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1000 struct sched_domain_attr *dattr_new)
1003 #endif /* !CONFIG_SMP */
1006 struct io_context; /* See blkdev.h */
1009 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1010 extern void prefetch_stack(struct task_struct *t);
1012 static inline void prefetch_stack(struct task_struct *t) { }
1015 struct audit_context; /* See audit.c */
1017 struct pipe_inode_info;
1018 struct uts_namespace;
1021 struct sched_domain;
1026 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1027 #define WF_FORK 0x02 /* child wakeup after fork */
1029 #define ENQUEUE_WAKEUP 1
1030 #define ENQUEUE_WAKING 2
1031 #define ENQUEUE_HEAD 4
1033 #define DEQUEUE_SLEEP 1
1035 struct sched_class {
1036 const struct sched_class *next;
1038 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1039 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1040 void (*yield_task) (struct rq *rq);
1042 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1044 struct task_struct * (*pick_next_task) (struct rq *rq);
1045 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1048 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1049 int sd_flag, int flags);
1051 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1052 void (*post_schedule) (struct rq *this_rq);
1053 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1054 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1056 void (*set_cpus_allowed)(struct task_struct *p,
1057 const struct cpumask *newmask);
1059 void (*rq_online)(struct rq *rq);
1060 void (*rq_offline)(struct rq *rq);
1063 void (*set_curr_task) (struct rq *rq);
1064 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1065 void (*task_fork) (struct task_struct *p);
1067 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1069 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1071 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1072 int oldprio, int running);
1074 unsigned int (*get_rr_interval) (struct rq *rq,
1075 struct task_struct *task);
1077 #ifdef CONFIG_FAIR_GROUP_SCHED
1078 void (*moved_group) (struct task_struct *p, int on_rq);
1082 struct load_weight {
1083 unsigned long weight, inv_weight;
1086 #ifdef CONFIG_SCHEDSTATS
1087 struct sched_statistics {
1097 s64 sum_sleep_runtime;
1104 u64 nr_migrations_cold;
1105 u64 nr_failed_migrations_affine;
1106 u64 nr_failed_migrations_running;
1107 u64 nr_failed_migrations_hot;
1108 u64 nr_forced_migrations;
1111 u64 nr_wakeups_sync;
1112 u64 nr_wakeups_migrate;
1113 u64 nr_wakeups_local;
1114 u64 nr_wakeups_remote;
1115 u64 nr_wakeups_affine;
1116 u64 nr_wakeups_affine_attempts;
1117 u64 nr_wakeups_passive;
1118 u64 nr_wakeups_idle;
1122 struct sched_entity {
1123 struct load_weight load; /* for load-balancing */
1124 struct rb_node run_node;
1125 struct list_head group_node;
1129 u64 sum_exec_runtime;
1131 u64 prev_sum_exec_runtime;
1135 #ifdef CONFIG_SCHEDSTATS
1136 struct sched_statistics statistics;
1139 #ifdef CONFIG_FAIR_GROUP_SCHED
1140 struct sched_entity *parent;
1141 /* rq on which this entity is (to be) queued: */
1142 struct cfs_rq *cfs_rq;
1143 /* rq "owned" by this entity/group: */
1144 struct cfs_rq *my_q;
1148 struct sched_rt_entity {
1149 struct list_head run_list;
1150 unsigned long timeout;
1151 unsigned int time_slice;
1152 int nr_cpus_allowed;
1154 struct sched_rt_entity *back;
1155 #ifdef CONFIG_RT_GROUP_SCHED
1156 struct sched_rt_entity *parent;
1157 /* rq on which this entity is (to be) queued: */
1158 struct rt_rq *rt_rq;
1159 /* rq "owned" by this entity/group: */
1166 struct task_struct {
1167 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1170 unsigned int flags; /* per process flags, defined below */
1171 unsigned int ptrace;
1173 int lock_depth; /* BKL lock depth */
1176 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1181 int prio, static_prio, normal_prio;
1182 unsigned int rt_priority;
1183 const struct sched_class *sched_class;
1184 struct sched_entity se;
1185 struct sched_rt_entity rt;
1187 #ifdef CONFIG_PREEMPT_NOTIFIERS
1188 /* list of struct preempt_notifier: */
1189 struct hlist_head preempt_notifiers;
1193 * fpu_counter contains the number of consecutive context switches
1194 * that the FPU is used. If this is over a threshold, the lazy fpu
1195 * saving becomes unlazy to save the trap. This is an unsigned char
1196 * so that after 256 times the counter wraps and the behavior turns
1197 * lazy again; this to deal with bursty apps that only use FPU for
1200 unsigned char fpu_counter;
1201 #ifdef CONFIG_BLK_DEV_IO_TRACE
1202 unsigned int btrace_seq;
1205 unsigned int policy;
1206 cpumask_t cpus_allowed;
1208 #ifdef CONFIG_TREE_PREEMPT_RCU
1209 int rcu_read_lock_nesting;
1210 char rcu_read_unlock_special;
1211 struct rcu_node *rcu_blocked_node;
1212 struct list_head rcu_node_entry;
1213 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1215 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1216 struct sched_info sched_info;
1219 struct list_head tasks;
1220 struct plist_node pushable_tasks;
1222 struct mm_struct *mm, *active_mm;
1223 #if defined(SPLIT_RSS_COUNTING)
1224 struct task_rss_stat rss_stat;
1228 int exit_code, exit_signal;
1229 int pdeath_signal; /* The signal sent when the parent dies */
1231 unsigned int personality;
1232 unsigned did_exec:1;
1233 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1235 unsigned in_iowait:1;
1238 /* Revert to default priority/policy when forking */
1239 unsigned sched_reset_on_fork:1;
1244 #ifdef CONFIG_CC_STACKPROTECTOR
1245 /* Canary value for the -fstack-protector gcc feature */
1246 unsigned long stack_canary;
1250 * pointers to (original) parent process, youngest child, younger sibling,
1251 * older sibling, respectively. (p->father can be replaced with
1252 * p->real_parent->pid)
1254 struct task_struct *real_parent; /* real parent process */
1255 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1257 * children/sibling forms the list of my natural children
1259 struct list_head children; /* list of my children */
1260 struct list_head sibling; /* linkage in my parent's children list */
1261 struct task_struct *group_leader; /* threadgroup leader */
1264 * ptraced is the list of tasks this task is using ptrace on.
1265 * This includes both natural children and PTRACE_ATTACH targets.
1266 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1268 struct list_head ptraced;
1269 struct list_head ptrace_entry;
1271 /* PID/PID hash table linkage. */
1272 struct pid_link pids[PIDTYPE_MAX];
1273 struct list_head thread_group;
1275 struct completion *vfork_done; /* for vfork() */
1276 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1277 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1279 cputime_t utime, stime, utimescaled, stimescaled;
1281 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1282 cputime_t prev_utime, prev_stime;
1284 unsigned long nvcsw, nivcsw; /* context switch counts */
1285 struct timespec start_time; /* monotonic time */
1286 struct timespec real_start_time; /* boot based time */
1287 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1288 unsigned long min_flt, maj_flt;
1290 struct task_cputime cputime_expires;
1291 struct list_head cpu_timers[3];
1293 /* process credentials */
1294 const struct cred *real_cred; /* objective and real subjective task
1295 * credentials (COW) */
1296 const struct cred *cred; /* effective (overridable) subjective task
1297 * credentials (COW) */
1298 struct mutex cred_guard_mutex; /* guard against foreign influences on
1299 * credential calculations
1300 * (notably. ptrace) */
1301 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1303 char comm[TASK_COMM_LEN]; /* executable name excluding path
1304 - access with [gs]et_task_comm (which lock
1305 it with task_lock())
1306 - initialized normally by setup_new_exec */
1307 /* file system info */
1308 int link_count, total_link_count;
1309 #ifdef CONFIG_SYSVIPC
1311 struct sysv_sem sysvsem;
1313 #ifdef CONFIG_DETECT_HUNG_TASK
1314 /* hung task detection */
1315 unsigned long last_switch_count;
1317 /* CPU-specific state of this task */
1318 struct thread_struct thread;
1319 /* filesystem information */
1320 struct fs_struct *fs;
1321 /* open file information */
1322 struct files_struct *files;
1324 struct nsproxy *nsproxy;
1325 /* signal handlers */
1326 struct signal_struct *signal;
1327 struct sighand_struct *sighand;
1329 sigset_t blocked, real_blocked;
1330 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1331 struct sigpending pending;
1333 unsigned long sas_ss_sp;
1335 int (*notifier)(void *priv);
1336 void *notifier_data;
1337 sigset_t *notifier_mask;
1338 struct audit_context *audit_context;
1339 #ifdef CONFIG_AUDITSYSCALL
1341 unsigned int sessionid;
1345 /* Thread group tracking */
1348 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1350 spinlock_t alloc_lock;
1352 #ifdef CONFIG_GENERIC_HARDIRQS
1353 /* IRQ handler threads */
1354 struct irqaction *irqaction;
1357 /* Protection of the PI data structures: */
1358 raw_spinlock_t pi_lock;
1360 #ifdef CONFIG_RT_MUTEXES
1361 /* PI waiters blocked on a rt_mutex held by this task */
1362 struct plist_head pi_waiters;
1363 /* Deadlock detection and priority inheritance handling */
1364 struct rt_mutex_waiter *pi_blocked_on;
1367 #ifdef CONFIG_DEBUG_MUTEXES
1368 /* mutex deadlock detection */
1369 struct mutex_waiter *blocked_on;
1371 #ifdef CONFIG_TRACE_IRQFLAGS
1372 unsigned int irq_events;
1373 unsigned long hardirq_enable_ip;
1374 unsigned long hardirq_disable_ip;
1375 unsigned int hardirq_enable_event;
1376 unsigned int hardirq_disable_event;
1377 int hardirqs_enabled;
1378 int hardirq_context;
1379 unsigned long softirq_disable_ip;
1380 unsigned long softirq_enable_ip;
1381 unsigned int softirq_disable_event;
1382 unsigned int softirq_enable_event;
1383 int softirqs_enabled;
1384 int softirq_context;
1386 #ifdef CONFIG_LOCKDEP
1387 # define MAX_LOCK_DEPTH 48UL
1390 unsigned int lockdep_recursion;
1391 struct held_lock held_locks[MAX_LOCK_DEPTH];
1392 gfp_t lockdep_reclaim_gfp;
1395 /* journalling filesystem info */
1398 /* stacked block device info */
1399 struct bio_list *bio_list;
1402 struct reclaim_state *reclaim_state;
1404 struct backing_dev_info *backing_dev_info;
1406 struct io_context *io_context;
1408 unsigned long ptrace_message;
1409 siginfo_t *last_siginfo; /* For ptrace use. */
1410 struct task_io_accounting ioac;
1411 #if defined(CONFIG_TASK_XACCT)
1412 u64 acct_rss_mem1; /* accumulated rss usage */
1413 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1414 cputime_t acct_timexpd; /* stime + utime since last update */
1416 #ifdef CONFIG_CPUSETS
1417 nodemask_t mems_allowed; /* Protected by alloc_lock */
1418 int mems_allowed_change_disable;
1419 int cpuset_mem_spread_rotor;
1420 int cpuset_slab_spread_rotor;
1422 #ifdef CONFIG_CGROUPS
1423 /* Control Group info protected by css_set_lock */
1424 struct css_set *cgroups;
1425 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1426 struct list_head cg_list;
1429 struct robust_list_head __user *robust_list;
1430 #ifdef CONFIG_COMPAT
1431 struct compat_robust_list_head __user *compat_robust_list;
1433 struct list_head pi_state_list;
1434 struct futex_pi_state *pi_state_cache;
1436 #ifdef CONFIG_PERF_EVENTS
1437 struct perf_event_context *perf_event_ctxp;
1438 struct mutex perf_event_mutex;
1439 struct list_head perf_event_list;
1442 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1445 atomic_t fs_excl; /* holding fs exclusive resources */
1446 struct rcu_head rcu;
1449 * cache last used pipe for splice
1451 struct pipe_inode_info *splice_pipe;
1452 #ifdef CONFIG_TASK_DELAY_ACCT
1453 struct task_delay_info *delays;
1455 #ifdef CONFIG_FAULT_INJECTION
1458 struct prop_local_single dirties;
1459 #ifdef CONFIG_LATENCYTOP
1460 int latency_record_count;
1461 struct latency_record latency_record[LT_SAVECOUNT];
1464 * time slack values; these are used to round up poll() and
1465 * select() etc timeout values. These are in nanoseconds.
1467 unsigned long timer_slack_ns;
1468 unsigned long default_timer_slack_ns;
1470 struct list_head *scm_work_list;
1471 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1472 /* Index of current stored address in ret_stack */
1474 /* Stack of return addresses for return function tracing */
1475 struct ftrace_ret_stack *ret_stack;
1476 /* time stamp for last schedule */
1477 unsigned long long ftrace_timestamp;
1479 * Number of functions that haven't been traced
1480 * because of depth overrun.
1482 atomic_t trace_overrun;
1483 /* Pause for the tracing */
1484 atomic_t tracing_graph_pause;
1486 #ifdef CONFIG_TRACING
1487 /* state flags for use by tracers */
1488 unsigned long trace;
1489 /* bitmask of trace recursion */
1490 unsigned long trace_recursion;
1491 #endif /* CONFIG_TRACING */
1492 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1493 struct memcg_batch_info {
1494 int do_batch; /* incremented when batch uncharge started */
1495 struct mem_cgroup *memcg; /* target memcg of uncharge */
1496 unsigned long bytes; /* uncharged usage */
1497 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1502 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1503 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1506 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1507 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1508 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1509 * values are inverted: lower p->prio value means higher priority.
1511 * The MAX_USER_RT_PRIO value allows the actual maximum
1512 * RT priority to be separate from the value exported to
1513 * user-space. This allows kernel threads to set their
1514 * priority to a value higher than any user task. Note:
1515 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1518 #define MAX_USER_RT_PRIO 100
1519 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1521 #define MAX_PRIO (MAX_RT_PRIO + 40)
1522 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1524 static inline int rt_prio(int prio)
1526 if (unlikely(prio < MAX_RT_PRIO))
1531 static inline int rt_task(struct task_struct *p)
1533 return rt_prio(p->prio);
1536 static inline struct pid *task_pid(struct task_struct *task)
1538 return task->pids[PIDTYPE_PID].pid;
1541 static inline struct pid *task_tgid(struct task_struct *task)
1543 return task->group_leader->pids[PIDTYPE_PID].pid;
1547 * Without tasklist or rcu lock it is not safe to dereference
1548 * the result of task_pgrp/task_session even if task == current,
1549 * we can race with another thread doing sys_setsid/sys_setpgid.
1551 static inline struct pid *task_pgrp(struct task_struct *task)
1553 return task->group_leader->pids[PIDTYPE_PGID].pid;
1556 static inline struct pid *task_session(struct task_struct *task)
1558 return task->group_leader->pids[PIDTYPE_SID].pid;
1561 struct pid_namespace;
1564 * the helpers to get the task's different pids as they are seen
1565 * from various namespaces
1567 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1568 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1570 * task_xid_nr_ns() : id seen from the ns specified;
1572 * set_task_vxid() : assigns a virtual id to a task;
1574 * see also pid_nr() etc in include/linux/pid.h
1576 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1577 struct pid_namespace *ns);
1579 static inline pid_t task_pid_nr(struct task_struct *tsk)
1584 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1585 struct pid_namespace *ns)
1587 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1590 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1592 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1596 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1601 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1603 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1605 return pid_vnr(task_tgid(tsk));
1609 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1610 struct pid_namespace *ns)
1612 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1615 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1617 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1621 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1622 struct pid_namespace *ns)
1624 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1627 static inline pid_t task_session_vnr(struct task_struct *tsk)
1629 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1632 /* obsolete, do not use */
1633 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1635 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1639 * pid_alive - check that a task structure is not stale
1640 * @p: Task structure to be checked.
1642 * Test if a process is not yet dead (at most zombie state)
1643 * If pid_alive fails, then pointers within the task structure
1644 * can be stale and must not be dereferenced.
1646 static inline int pid_alive(struct task_struct *p)
1648 return p->pids[PIDTYPE_PID].pid != NULL;
1652 * is_global_init - check if a task structure is init
1653 * @tsk: Task structure to be checked.
1655 * Check if a task structure is the first user space task the kernel created.
1657 static inline int is_global_init(struct task_struct *tsk)
1659 return tsk->pid == 1;
1663 * is_container_init:
1664 * check whether in the task is init in its own pid namespace.
1666 extern int is_container_init(struct task_struct *tsk);
1668 extern struct pid *cad_pid;
1670 extern void free_task(struct task_struct *tsk);
1671 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1673 extern void __put_task_struct(struct task_struct *t);
1675 static inline void put_task_struct(struct task_struct *t)
1677 if (atomic_dec_and_test(&t->usage))
1678 __put_task_struct(t);
1681 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1682 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1687 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1688 /* Not implemented yet, only for 486*/
1689 #define PF_STARTING 0x00000002 /* being created */
1690 #define PF_EXITING 0x00000004 /* getting shut down */
1691 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1692 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1693 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1694 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1695 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1696 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1697 #define PF_DUMPCORE 0x00000200 /* dumped core */
1698 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1699 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1700 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1701 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1702 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1703 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1704 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1705 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1706 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1707 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1708 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1709 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1710 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1711 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1712 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1713 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1714 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1715 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1716 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1717 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1718 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1719 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1722 * Only the _current_ task can read/write to tsk->flags, but other
1723 * tasks can access tsk->flags in readonly mode for example
1724 * with tsk_used_math (like during threaded core dumping).
1725 * There is however an exception to this rule during ptrace
1726 * or during fork: the ptracer task is allowed to write to the
1727 * child->flags of its traced child (same goes for fork, the parent
1728 * can write to the child->flags), because we're guaranteed the
1729 * child is not running and in turn not changing child->flags
1730 * at the same time the parent does it.
1732 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1733 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1734 #define clear_used_math() clear_stopped_child_used_math(current)
1735 #define set_used_math() set_stopped_child_used_math(current)
1736 #define conditional_stopped_child_used_math(condition, child) \
1737 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1738 #define conditional_used_math(condition) \
1739 conditional_stopped_child_used_math(condition, current)
1740 #define copy_to_stopped_child_used_math(child) \
1741 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1742 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1743 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1744 #define used_math() tsk_used_math(current)
1746 #ifdef CONFIG_TREE_PREEMPT_RCU
1748 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1749 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1751 static inline void rcu_copy_process(struct task_struct *p)
1753 p->rcu_read_lock_nesting = 0;
1754 p->rcu_read_unlock_special = 0;
1755 p->rcu_blocked_node = NULL;
1756 INIT_LIST_HEAD(&p->rcu_node_entry);
1761 static inline void rcu_copy_process(struct task_struct *p)
1768 extern int set_cpus_allowed_ptr(struct task_struct *p,
1769 const struct cpumask *new_mask);
1771 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1772 const struct cpumask *new_mask)
1774 if (!cpumask_test_cpu(0, new_mask))
1780 #ifndef CONFIG_CPUMASK_OFFSTACK
1781 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1783 return set_cpus_allowed_ptr(p, &new_mask);
1788 * Do not use outside of architecture code which knows its limitations.
1790 * sched_clock() has no promise of monotonicity or bounded drift between
1791 * CPUs, use (which you should not) requires disabling IRQs.
1793 * Please use one of the three interfaces below.
1795 extern unsigned long long notrace sched_clock(void);
1797 * See the comment in kernel/sched_clock.c
1799 extern u64 cpu_clock(int cpu);
1800 extern u64 local_clock(void);
1801 extern u64 sched_clock_cpu(int cpu);
1804 extern void sched_clock_init(void);
1806 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1807 static inline void sched_clock_tick(void)
1811 static inline void sched_clock_idle_sleep_event(void)
1815 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1820 * Architectures can set this to 1 if they have specified
1821 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1822 * but then during bootup it turns out that sched_clock()
1823 * is reliable after all:
1825 extern int sched_clock_stable;
1827 extern void sched_clock_tick(void);
1828 extern void sched_clock_idle_sleep_event(void);
1829 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1832 extern unsigned long long
1833 task_sched_runtime(struct task_struct *task);
1834 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1836 /* sched_exec is called by processes performing an exec */
1838 extern void sched_exec(void);
1840 #define sched_exec() {}
1843 extern void sched_clock_idle_sleep_event(void);
1844 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1846 #ifdef CONFIG_HOTPLUG_CPU
1847 extern void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p);
1848 extern void idle_task_exit(void);
1850 static inline void idle_task_exit(void) {}
1853 extern void sched_idle_next(void);
1855 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1856 extern void wake_up_idle_cpu(int cpu);
1858 static inline void wake_up_idle_cpu(int cpu) { }
1861 extern unsigned int sysctl_sched_latency;
1862 extern unsigned int sysctl_sched_min_granularity;
1863 extern unsigned int sysctl_sched_wakeup_granularity;
1864 extern unsigned int sysctl_sched_shares_ratelimit;
1865 extern unsigned int sysctl_sched_shares_thresh;
1866 extern unsigned int sysctl_sched_child_runs_first;
1868 enum sched_tunable_scaling {
1869 SCHED_TUNABLESCALING_NONE,
1870 SCHED_TUNABLESCALING_LOG,
1871 SCHED_TUNABLESCALING_LINEAR,
1872 SCHED_TUNABLESCALING_END,
1874 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1876 #ifdef CONFIG_SCHED_DEBUG
1877 extern unsigned int sysctl_sched_migration_cost;
1878 extern unsigned int sysctl_sched_nr_migrate;
1879 extern unsigned int sysctl_sched_time_avg;
1880 extern unsigned int sysctl_timer_migration;
1882 int sched_proc_update_handler(struct ctl_table *table, int write,
1883 void __user *buffer, size_t *length,
1886 #ifdef CONFIG_SCHED_DEBUG
1887 static inline unsigned int get_sysctl_timer_migration(void)
1889 return sysctl_timer_migration;
1892 static inline unsigned int get_sysctl_timer_migration(void)
1897 extern unsigned int sysctl_sched_rt_period;
1898 extern int sysctl_sched_rt_runtime;
1900 int sched_rt_handler(struct ctl_table *table, int write,
1901 void __user *buffer, size_t *lenp,
1904 extern unsigned int sysctl_sched_compat_yield;
1906 #ifdef CONFIG_RT_MUTEXES
1907 extern int rt_mutex_getprio(struct task_struct *p);
1908 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1909 extern void rt_mutex_adjust_pi(struct task_struct *p);
1911 static inline int rt_mutex_getprio(struct task_struct *p)
1913 return p->normal_prio;
1915 # define rt_mutex_adjust_pi(p) do { } while (0)
1918 extern void set_user_nice(struct task_struct *p, long nice);
1919 extern int task_prio(const struct task_struct *p);
1920 extern int task_nice(const struct task_struct *p);
1921 extern int can_nice(const struct task_struct *p, const int nice);
1922 extern int task_curr(const struct task_struct *p);
1923 extern int idle_cpu(int cpu);
1924 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1925 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1926 struct sched_param *);
1927 extern struct task_struct *idle_task(int cpu);
1928 extern struct task_struct *curr_task(int cpu);
1929 extern void set_curr_task(int cpu, struct task_struct *p);
1934 * The default (Linux) execution domain.
1936 extern struct exec_domain default_exec_domain;
1938 union thread_union {
1939 struct thread_info thread_info;
1940 unsigned long stack[THREAD_SIZE/sizeof(long)];
1943 #ifndef __HAVE_ARCH_KSTACK_END
1944 static inline int kstack_end(void *addr)
1946 /* Reliable end of stack detection:
1947 * Some APM bios versions misalign the stack
1949 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1953 extern union thread_union init_thread_union;
1954 extern struct task_struct init_task;
1956 extern struct mm_struct init_mm;
1958 extern struct pid_namespace init_pid_ns;
1961 * find a task by one of its numerical ids
1963 * find_task_by_pid_ns():
1964 * finds a task by its pid in the specified namespace
1965 * find_task_by_vpid():
1966 * finds a task by its virtual pid
1968 * see also find_vpid() etc in include/linux/pid.h
1971 extern struct task_struct *find_task_by_vpid(pid_t nr);
1972 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1973 struct pid_namespace *ns);
1975 extern void __set_special_pids(struct pid *pid);
1977 /* per-UID process charging. */
1978 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1979 static inline struct user_struct *get_uid(struct user_struct *u)
1981 atomic_inc(&u->__count);
1984 extern void free_uid(struct user_struct *);
1985 extern void release_uids(struct user_namespace *ns);
1987 #include <asm/current.h>
1989 extern void do_timer(unsigned long ticks);
1991 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1992 extern int wake_up_process(struct task_struct *tsk);
1993 extern void wake_up_new_task(struct task_struct *tsk,
1994 unsigned long clone_flags);
1996 extern void kick_process(struct task_struct *tsk);
1998 static inline void kick_process(struct task_struct *tsk) { }
2000 extern void sched_fork(struct task_struct *p, int clone_flags);
2001 extern void sched_dead(struct task_struct *p);
2003 extern void proc_caches_init(void);
2004 extern void flush_signals(struct task_struct *);
2005 extern void __flush_signals(struct task_struct *);
2006 extern void ignore_signals(struct task_struct *);
2007 extern void flush_signal_handlers(struct task_struct *, int force_default);
2008 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2010 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2012 unsigned long flags;
2015 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2016 ret = dequeue_signal(tsk, mask, info);
2017 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2022 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2024 extern void unblock_all_signals(void);
2025 extern void release_task(struct task_struct * p);
2026 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2027 extern int force_sigsegv(int, struct task_struct *);
2028 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2029 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2030 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2031 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2032 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2033 extern int kill_pid(struct pid *pid, int sig, int priv);
2034 extern int kill_proc_info(int, struct siginfo *, pid_t);
2035 extern int do_notify_parent(struct task_struct *, int);
2036 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2037 extern void force_sig(int, struct task_struct *);
2038 extern int send_sig(int, struct task_struct *, int);
2039 extern int zap_other_threads(struct task_struct *p);
2040 extern struct sigqueue *sigqueue_alloc(void);
2041 extern void sigqueue_free(struct sigqueue *);
2042 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2043 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2044 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2046 static inline int kill_cad_pid(int sig, int priv)
2048 return kill_pid(cad_pid, sig, priv);
2051 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2052 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2053 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2054 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2057 * True if we are on the alternate signal stack.
2059 static inline int on_sig_stack(unsigned long sp)
2061 #ifdef CONFIG_STACK_GROWSUP
2062 return sp >= current->sas_ss_sp &&
2063 sp - current->sas_ss_sp < current->sas_ss_size;
2065 return sp > current->sas_ss_sp &&
2066 sp - current->sas_ss_sp <= current->sas_ss_size;
2070 static inline int sas_ss_flags(unsigned long sp)
2072 return (current->sas_ss_size == 0 ? SS_DISABLE
2073 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2077 * Routines for handling mm_structs
2079 extern struct mm_struct * mm_alloc(void);
2081 /* mmdrop drops the mm and the page tables */
2082 extern void __mmdrop(struct mm_struct *);
2083 static inline void mmdrop(struct mm_struct * mm)
2085 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2089 /* mmput gets rid of the mappings and all user-space */
2090 extern void mmput(struct mm_struct *);
2091 /* Grab a reference to a task's mm, if it is not already going away */
2092 extern struct mm_struct *get_task_mm(struct task_struct *task);
2093 /* Remove the current tasks stale references to the old mm_struct */
2094 extern void mm_release(struct task_struct *, struct mm_struct *);
2095 /* Allocate a new mm structure and copy contents from tsk->mm */
2096 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2098 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2099 struct task_struct *, struct pt_regs *);
2100 extern void flush_thread(void);
2101 extern void exit_thread(void);
2103 extern void exit_files(struct task_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(const char *,
2116 const char __user * const __user *,
2117 const char __user * const __user *, struct pt_regs *);
2118 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2119 struct task_struct *fork_idle(int);
2121 extern void set_task_comm(struct task_struct *tsk, char *from);
2122 extern char *get_task_comm(char *to, struct task_struct *tsk);
2125 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2127 static inline unsigned long wait_task_inactive(struct task_struct *p,
2134 #define next_task(p) \
2135 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2137 #define for_each_process(p) \
2138 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2140 extern bool current_is_single_threaded(void);
2143 * Careful: do_each_thread/while_each_thread is a double loop so
2144 * 'break' will not work as expected - use goto instead.
2146 #define do_each_thread(g, t) \
2147 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2149 #define while_each_thread(g, t) \
2150 while ((t = next_thread(t)) != g)
2152 static inline int get_nr_threads(struct task_struct *tsk)
2154 return tsk->signal->nr_threads;
2157 /* de_thread depends on thread_group_leader not being a pid based check */
2158 #define thread_group_leader(p) (p == p->group_leader)
2160 /* Do to the insanities of de_thread it is possible for a process
2161 * to have the pid of the thread group leader without actually being
2162 * the thread group leader. For iteration through the pids in proc
2163 * all we care about is that we have a task with the appropriate
2164 * pid, we don't actually care if we have the right task.
2166 static inline int has_group_leader_pid(struct task_struct *p)
2168 return p->pid == p->tgid;
2172 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2174 return p1->tgid == p2->tgid;
2177 static inline struct task_struct *next_thread(const struct task_struct *p)
2179 return list_entry_rcu(p->thread_group.next,
2180 struct task_struct, thread_group);
2183 static inline int thread_group_empty(struct task_struct *p)
2185 return list_empty(&p->thread_group);
2188 #define delay_group_leader(p) \
2189 (thread_group_leader(p) && !thread_group_empty(p))
2191 static inline int task_detached(struct task_struct *p)
2193 return p->exit_signal == -1;
2197 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2198 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2199 * pins the final release of task.io_context. Also protects ->cpuset and
2200 * ->cgroup.subsys[].
2202 * Nests both inside and outside of read_lock(&tasklist_lock).
2203 * It must not be nested with write_lock_irq(&tasklist_lock),
2204 * neither inside nor outside.
2206 static inline void task_lock(struct task_struct *p)
2208 spin_lock(&p->alloc_lock);
2211 static inline void task_unlock(struct task_struct *p)
2213 spin_unlock(&p->alloc_lock);
2216 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2217 unsigned long *flags);
2219 static inline void unlock_task_sighand(struct task_struct *tsk,
2220 unsigned long *flags)
2222 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2225 #ifndef __HAVE_THREAD_FUNCTIONS
2227 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2228 #define task_stack_page(task) ((task)->stack)
2230 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2232 *task_thread_info(p) = *task_thread_info(org);
2233 task_thread_info(p)->task = p;
2236 static inline unsigned long *end_of_stack(struct task_struct *p)
2238 return (unsigned long *)(task_thread_info(p) + 1);
2243 static inline int object_is_on_stack(void *obj)
2245 void *stack = task_stack_page(current);
2247 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2250 extern void thread_info_cache_init(void);
2252 #ifdef CONFIG_DEBUG_STACK_USAGE
2253 static inline unsigned long stack_not_used(struct task_struct *p)
2255 unsigned long *n = end_of_stack(p);
2257 do { /* Skip over canary */
2261 return (unsigned long)n - (unsigned long)end_of_stack(p);
2265 /* set thread flags in other task's structures
2266 * - see asm/thread_info.h for TIF_xxxx flags available
2268 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2270 set_ti_thread_flag(task_thread_info(tsk), flag);
2273 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2275 clear_ti_thread_flag(task_thread_info(tsk), flag);
2278 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2280 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2283 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2285 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2288 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2290 return test_ti_thread_flag(task_thread_info(tsk), flag);
2293 static inline void set_tsk_need_resched(struct task_struct *tsk)
2295 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2298 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2300 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2303 static inline int test_tsk_need_resched(struct task_struct *tsk)
2305 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2308 static inline int restart_syscall(void)
2310 set_tsk_thread_flag(current, TIF_SIGPENDING);
2311 return -ERESTARTNOINTR;
2314 static inline int signal_pending(struct task_struct *p)
2316 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2319 static inline int __fatal_signal_pending(struct task_struct *p)
2321 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2324 static inline int fatal_signal_pending(struct task_struct *p)
2326 return signal_pending(p) && __fatal_signal_pending(p);
2329 static inline int signal_pending_state(long state, struct task_struct *p)
2331 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2333 if (!signal_pending(p))
2336 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2339 static inline int need_resched(void)
2341 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2345 * cond_resched() and cond_resched_lock(): latency reduction via
2346 * explicit rescheduling in places that are safe. The return
2347 * value indicates whether a reschedule was done in fact.
2348 * cond_resched_lock() will drop the spinlock before scheduling,
2349 * cond_resched_softirq() will enable bhs before scheduling.
2351 extern int _cond_resched(void);
2353 #define cond_resched() ({ \
2354 __might_sleep(__FILE__, __LINE__, 0); \
2358 extern int __cond_resched_lock(spinlock_t *lock);
2360 #ifdef CONFIG_PREEMPT
2361 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2363 #define PREEMPT_LOCK_OFFSET 0
2366 #define cond_resched_lock(lock) ({ \
2367 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2368 __cond_resched_lock(lock); \
2371 extern int __cond_resched_softirq(void);
2373 #define cond_resched_softirq() ({ \
2374 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2375 __cond_resched_softirq(); \
2379 * Does a critical section need to be broken due to another
2380 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2381 * but a general need for low latency)
2383 static inline int spin_needbreak(spinlock_t *lock)
2385 #ifdef CONFIG_PREEMPT
2386 return spin_is_contended(lock);
2393 * Thread group CPU time accounting.
2395 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2396 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2398 static inline void thread_group_cputime_init(struct signal_struct *sig)
2400 spin_lock_init(&sig->cputimer.lock);
2404 * Reevaluate whether the task has signals pending delivery.
2405 * Wake the task if so.
2406 * This is required every time the blocked sigset_t changes.
2407 * callers must hold sighand->siglock.
2409 extern void recalc_sigpending_and_wake(struct task_struct *t);
2410 extern void recalc_sigpending(void);
2412 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2415 * Wrappers for p->thread_info->cpu access. No-op on UP.
2419 static inline unsigned int task_cpu(const struct task_struct *p)
2421 return task_thread_info(p)->cpu;
2424 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2428 static inline unsigned int task_cpu(const struct task_struct *p)
2433 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2437 #endif /* CONFIG_SMP */
2439 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2440 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2442 extern void normalize_rt_tasks(void);
2444 #ifdef CONFIG_CGROUP_SCHED
2446 extern struct task_group init_task_group;
2448 extern struct task_group *sched_create_group(struct task_group *parent);
2449 extern void sched_destroy_group(struct task_group *tg);
2450 extern void sched_move_task(struct task_struct *tsk);
2451 #ifdef CONFIG_FAIR_GROUP_SCHED
2452 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2453 extern unsigned long sched_group_shares(struct task_group *tg);
2455 #ifdef CONFIG_RT_GROUP_SCHED
2456 extern int sched_group_set_rt_runtime(struct task_group *tg,
2457 long rt_runtime_us);
2458 extern long sched_group_rt_runtime(struct task_group *tg);
2459 extern int sched_group_set_rt_period(struct task_group *tg,
2461 extern long sched_group_rt_period(struct task_group *tg);
2462 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2466 extern int task_can_switch_user(struct user_struct *up,
2467 struct task_struct *tsk);
2469 #ifdef CONFIG_TASK_XACCT
2470 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2472 tsk->ioac.rchar += amt;
2475 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2477 tsk->ioac.wchar += amt;
2480 static inline void inc_syscr(struct task_struct *tsk)
2485 static inline void inc_syscw(struct task_struct *tsk)
2490 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2494 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2498 static inline void inc_syscr(struct task_struct *tsk)
2502 static inline void inc_syscw(struct task_struct *tsk)
2507 #ifndef TASK_SIZE_OF
2508 #define TASK_SIZE_OF(tsk) TASK_SIZE
2512 * Call the function if the target task is executing on a CPU right now:
2514 extern void task_oncpu_function_call(struct task_struct *p,
2515 void (*func) (void *info), void *info);
2518 #ifdef CONFIG_MM_OWNER
2519 extern void mm_update_next_owner(struct mm_struct *mm);
2520 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2522 static inline void mm_update_next_owner(struct mm_struct *mm)
2526 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2529 #endif /* CONFIG_MM_OWNER */
2531 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2534 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2537 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2540 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2543 static inline unsigned long rlimit(unsigned int limit)
2545 return task_rlimit(current, limit);
2548 static inline unsigned long rlimit_max(unsigned int limit)
2550 return task_rlimit_max(current, limit);
2553 #endif /* __KERNEL__ */