7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 #define CLONE_NEWNET 0x40000000 /* New network namespace */
30 #define CLONE_IO 0x80000000 /* Clone io context */
35 #define SCHED_NORMAL 0
39 /* SCHED_ISO: reserved but not implemented yet */
41 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
42 #define SCHED_RESET_ON_FORK 0x40000000
50 #include <asm/param.h> /* for HZ */
52 #include <linux/capability.h>
53 #include <linux/threads.h>
54 #include <linux/kernel.h>
55 #include <linux/types.h>
56 #include <linux/timex.h>
57 #include <linux/jiffies.h>
58 #include <linux/rbtree.h>
59 #include <linux/thread_info.h>
60 #include <linux/cpumask.h>
61 #include <linux/errno.h>
62 #include <linux/nodemask.h>
63 #include <linux/mm_types.h>
65 #include <asm/system.h>
67 #include <asm/ptrace.h>
68 #include <asm/cputime.h>
70 #include <linux/smp.h>
71 #include <linux/sem.h>
72 #include <linux/signal.h>
73 #include <linux/path.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rculist.h>
83 #include <linux/rtmutex.h>
85 #include <linux/time.h>
86 #include <linux/param.h>
87 #include <linux/resource.h>
88 #include <linux/timer.h>
89 #include <linux/hrtimer.h>
90 #include <linux/task_io_accounting.h>
91 #include <linux/kobject.h>
92 #include <linux/latencytop.h>
93 #include <linux/cred.h>
95 #include <asm/processor.h>
98 struct futex_pi_state;
99 struct robust_list_head;
102 struct perf_event_context;
105 * List of flags we want to share for kernel threads,
106 * if only because they are not used by them anyway.
108 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
111 * These are the constant used to fake the fixed-point load-average
112 * counting. Some notes:
113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
114 * a load-average precision of 10 bits integer + 11 bits fractional
115 * - if you want to count load-averages more often, you need more
116 * precision, or rounding will get you. With 2-second counting freq,
117 * the EXP_n values would be 1981, 2034 and 2043 if still using only
120 extern unsigned long avenrun[]; /* Load averages */
121 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
123 #define FSHIFT 11 /* nr of bits of precision */
124 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
125 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
126 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
127 #define EXP_5 2014 /* 1/exp(5sec/5min) */
128 #define EXP_15 2037 /* 1/exp(5sec/15min) */
130 #define CALC_LOAD(load,exp,n) \
132 load += n*(FIXED_1-exp); \
135 extern unsigned long total_forks;
136 extern int nr_threads;
137 DECLARE_PER_CPU(unsigned long, process_counts);
138 extern int nr_processes(void);
139 extern unsigned long nr_running(void);
140 extern unsigned long nr_uninterruptible(void);
141 extern unsigned long nr_iowait(void);
142 extern unsigned long nr_iowait_cpu(int cpu);
143 extern unsigned long this_cpu_load(void);
146 extern void calc_global_load(unsigned long ticks);
148 extern unsigned long get_parent_ip(unsigned long addr);
153 #ifdef CONFIG_SCHED_DEBUG
154 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
155 extern void proc_sched_set_task(struct task_struct *p);
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
160 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
163 static inline void proc_sched_set_task(struct task_struct *p)
167 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
173 * Task state bitmask. NOTE! These bits are also
174 * encoded in fs/proc/array.c: get_task_state().
176 * We have two separate sets of flags: task->state
177 * is about runnability, while task->exit_state are
178 * about the task exiting. Confusing, but this way
179 * modifying one set can't modify the other one by
182 #define TASK_RUNNING 0
183 #define TASK_INTERRUPTIBLE 1
184 #define TASK_UNINTERRUPTIBLE 2
185 #define __TASK_STOPPED 4
186 #define __TASK_TRACED 8
187 /* in tsk->exit_state */
188 #define EXIT_ZOMBIE 16
190 /* in tsk->state again */
192 #define TASK_WAKEKILL 128
193 #define TASK_WAKING 256
194 #define TASK_STATE_MAX 512
196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
198 extern char ___assert_task_state[1 - 2*!!(
199 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
201 /* Convenience macros for the sake of set_task_state */
202 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
203 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
204 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
206 /* Convenience macros for the sake of wake_up */
207 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
208 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
210 /* get_task_state() */
211 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
215 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
216 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
217 #define task_is_dead(task) ((task)->exit_state != 0)
218 #define task_is_stopped_or_traced(task) \
219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
220 #define task_contributes_to_load(task) \
221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
222 (task->flags & PF_FREEZING) == 0)
224 #define __set_task_state(tsk, state_value) \
225 do { (tsk)->state = (state_value); } while (0)
226 #define set_task_state(tsk, state_value) \
227 set_mb((tsk)->state, (state_value))
230 * set_current_state() includes a barrier so that the write of current->state
231 * is correctly serialised wrt the caller's subsequent test of whether to
234 * set_current_state(TASK_UNINTERRUPTIBLE);
235 * if (do_i_need_to_sleep())
238 * If the caller does not need such serialisation then use __set_current_state()
240 #define __set_current_state(state_value) \
241 do { current->state = (state_value); } while (0)
242 #define set_current_state(state_value) \
243 set_mb(current->state, (state_value))
245 /* Task command name length */
246 #define TASK_COMM_LEN 16
248 #include <linux/spinlock.h>
251 * This serializes "schedule()" and also protects
252 * the run-queue from deletions/modifications (but
253 * _adding_ to the beginning of the run-queue has
256 extern rwlock_t tasklist_lock;
257 extern spinlock_t mmlist_lock;
261 #ifdef CONFIG_PROVE_RCU
262 extern int lockdep_tasklist_lock_is_held(void);
263 #endif /* #ifdef CONFIG_PROVE_RCU */
265 extern void sched_init(void);
266 extern void sched_init_smp(void);
267 extern asmlinkage void schedule_tail(struct task_struct *prev);
268 extern void init_idle(struct task_struct *idle, int cpu);
269 extern void init_idle_bootup_task(struct task_struct *idle);
271 extern int runqueue_is_locked(int cpu);
273 extern cpumask_var_t nohz_cpu_mask;
274 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
275 extern int select_nohz_load_balancer(int cpu);
276 extern int get_nohz_load_balancer(void);
278 static inline int select_nohz_load_balancer(int cpu)
285 * Only dump TASK_* tasks. (0 for all tasks)
287 extern void show_state_filter(unsigned long state_filter);
289 static inline void show_state(void)
291 show_state_filter(0);
294 extern void show_regs(struct pt_regs *);
297 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
298 * task), SP is the stack pointer of the first frame that should be shown in the back
299 * trace (or NULL if the entire call-chain of the task should be shown).
301 extern void show_stack(struct task_struct *task, unsigned long *sp);
303 void io_schedule(void);
304 long io_schedule_timeout(long timeout);
306 extern void cpu_init (void);
307 extern void trap_init(void);
308 extern void update_process_times(int user);
309 extern void scheduler_tick(void);
311 extern void sched_show_task(struct task_struct *p);
313 #ifdef CONFIG_DETECT_SOFTLOCKUP
314 extern void softlockup_tick(void);
315 extern void touch_softlockup_watchdog(void);
316 extern void touch_softlockup_watchdog_sync(void);
317 extern void touch_all_softlockup_watchdogs(void);
318 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
320 size_t *lenp, loff_t *ppos);
321 extern unsigned int softlockup_panic;
322 extern int softlockup_thresh;
324 static inline void softlockup_tick(void)
327 static inline void touch_softlockup_watchdog(void)
330 static inline void touch_softlockup_watchdog_sync(void)
333 static inline void touch_all_softlockup_watchdogs(void)
338 #ifdef CONFIG_DETECT_HUNG_TASK
339 extern unsigned int sysctl_hung_task_panic;
340 extern unsigned long sysctl_hung_task_check_count;
341 extern unsigned long sysctl_hung_task_timeout_secs;
342 extern unsigned long sysctl_hung_task_warnings;
343 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
345 size_t *lenp, loff_t *ppos);
348 /* Attach to any functions which should be ignored in wchan output. */
349 #define __sched __attribute__((__section__(".sched.text")))
351 /* Linker adds these: start and end of __sched functions */
352 extern char __sched_text_start[], __sched_text_end[];
354 /* Is this address in the __sched functions? */
355 extern int in_sched_functions(unsigned long addr);
357 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
358 extern signed long schedule_timeout(signed long timeout);
359 extern signed long schedule_timeout_interruptible(signed long timeout);
360 extern signed long schedule_timeout_killable(signed long timeout);
361 extern signed long schedule_timeout_uninterruptible(signed long timeout);
362 asmlinkage void schedule(void);
363 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
366 struct user_namespace;
369 * Default maximum number of active map areas, this limits the number of vmas
370 * per mm struct. Users can overwrite this number by sysctl but there is a
373 * When a program's coredump is generated as ELF format, a section is created
374 * per a vma. In ELF, the number of sections is represented in unsigned short.
375 * This means the number of sections should be smaller than 65535 at coredump.
376 * Because the kernel adds some informative sections to a image of program at
377 * generating coredump, we need some margin. The number of extra sections is
378 * 1-3 now and depends on arch. We use "5" as safe margin, here.
380 #define MAPCOUNT_ELF_CORE_MARGIN (5)
381 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
383 extern int sysctl_max_map_count;
385 #include <linux/aio.h>
388 extern void arch_pick_mmap_layout(struct mm_struct *mm);
390 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
391 unsigned long, unsigned long);
393 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
394 unsigned long len, unsigned long pgoff,
395 unsigned long flags);
396 extern void arch_unmap_area(struct mm_struct *, unsigned long);
397 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
399 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
403 extern void set_dumpable(struct mm_struct *mm, int value);
404 extern int get_dumpable(struct mm_struct *mm);
408 #define MMF_DUMPABLE 0 /* core dump is permitted */
409 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
411 #define MMF_DUMPABLE_BITS 2
412 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
414 /* coredump filter bits */
415 #define MMF_DUMP_ANON_PRIVATE 2
416 #define MMF_DUMP_ANON_SHARED 3
417 #define MMF_DUMP_MAPPED_PRIVATE 4
418 #define MMF_DUMP_MAPPED_SHARED 5
419 #define MMF_DUMP_ELF_HEADERS 6
420 #define MMF_DUMP_HUGETLB_PRIVATE 7
421 #define MMF_DUMP_HUGETLB_SHARED 8
423 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
424 #define MMF_DUMP_FILTER_BITS 7
425 #define MMF_DUMP_FILTER_MASK \
426 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
427 #define MMF_DUMP_FILTER_DEFAULT \
428 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
429 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
431 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
432 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
434 # define MMF_DUMP_MASK_DEFAULT_ELF 0
436 /* leave room for more dump flags */
437 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
439 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
441 struct sighand_struct {
443 struct k_sigaction action[_NSIG];
445 wait_queue_head_t signalfd_wqh;
448 struct pacct_struct {
451 unsigned long ac_mem;
452 cputime_t ac_utime, ac_stime;
453 unsigned long ac_minflt, ac_majflt;
464 * struct task_cputime - collected CPU time counts
465 * @utime: time spent in user mode, in &cputime_t units
466 * @stime: time spent in kernel mode, in &cputime_t units
467 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
469 * This structure groups together three kinds of CPU time that are
470 * tracked for threads and thread groups. Most things considering
471 * CPU time want to group these counts together and treat all three
472 * of them in parallel.
474 struct task_cputime {
477 unsigned long long sum_exec_runtime;
479 /* Alternate field names when used to cache expirations. */
480 #define prof_exp stime
481 #define virt_exp utime
482 #define sched_exp sum_exec_runtime
484 #define INIT_CPUTIME \
485 (struct task_cputime) { \
486 .utime = cputime_zero, \
487 .stime = cputime_zero, \
488 .sum_exec_runtime = 0, \
492 * Disable preemption until the scheduler is running.
493 * Reset by start_kernel()->sched_init()->init_idle().
495 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
496 * before the scheduler is active -- see should_resched().
498 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
501 * struct thread_group_cputimer - thread group interval timer counts
502 * @cputime: thread group interval timers.
503 * @running: non-zero when there are timers running and
504 * @cputime receives updates.
505 * @lock: lock for fields in this struct.
507 * This structure contains the version of task_cputime, above, that is
508 * used for thread group CPU timer calculations.
510 struct thread_group_cputimer {
511 struct task_cputime cputime;
517 * NOTE! "signal_struct" does not have it's own
518 * locking, because a shared signal_struct always
519 * implies a shared sighand_struct, so locking
520 * sighand_struct is always a proper superset of
521 * the locking of signal_struct.
523 struct signal_struct {
528 wait_queue_head_t wait_chldexit; /* for wait4() */
530 /* current thread group signal load-balancing target: */
531 struct task_struct *curr_target;
533 /* shared signal handling: */
534 struct sigpending shared_pending;
536 /* thread group exit support */
539 * - notify group_exit_task when ->count is equal to notify_count
540 * - everyone except group_exit_task is stopped during signal delivery
541 * of fatal signals, group_exit_task processes the signal.
544 struct task_struct *group_exit_task;
546 /* thread group stop support, overloads group_exit_code too */
547 int group_stop_count;
548 unsigned int flags; /* see SIGNAL_* flags below */
550 /* POSIX.1b Interval Timers */
551 struct list_head posix_timers;
553 /* ITIMER_REAL timer for the process */
554 struct hrtimer real_timer;
555 struct pid *leader_pid;
556 ktime_t it_real_incr;
559 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
560 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
561 * values are defined to 0 and 1 respectively
563 struct cpu_itimer it[2];
566 * Thread group totals for process CPU timers.
567 * See thread_group_cputimer(), et al, for details.
569 struct thread_group_cputimer cputimer;
571 /* Earliest-expiration cache. */
572 struct task_cputime cputime_expires;
574 struct list_head cpu_timers[3];
576 struct pid *tty_old_pgrp;
578 /* boolean value for session group leader */
581 struct tty_struct *tty; /* NULL if no tty */
584 * Cumulative resource counters for dead threads in the group,
585 * and for reaped dead child processes forked by this group.
586 * Live threads maintain their own counters and add to these
587 * in __exit_signal, except for the group leader.
589 cputime_t utime, stime, cutime, cstime;
592 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
593 cputime_t prev_utime, prev_stime;
595 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
596 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
597 unsigned long inblock, oublock, cinblock, coublock;
598 unsigned long maxrss, cmaxrss;
599 struct task_io_accounting ioac;
602 * Cumulative ns of schedule CPU time fo dead threads in the
603 * group, not including a zombie group leader, (This only differs
604 * from jiffies_to_ns(utime + stime) if sched_clock uses something
605 * other than jiffies.)
607 unsigned long long sum_sched_runtime;
610 * We don't bother to synchronize most readers of this at all,
611 * because there is no reader checking a limit that actually needs
612 * to get both rlim_cur and rlim_max atomically, and either one
613 * alone is a single word that can safely be read normally.
614 * getrlimit/setrlimit use task_lock(current->group_leader) to
615 * protect this instead of the siglock, because they really
616 * have no need to disable irqs.
618 struct rlimit rlim[RLIM_NLIMITS];
620 #ifdef CONFIG_BSD_PROCESS_ACCT
621 struct pacct_struct pacct; /* per-process accounting information */
623 #ifdef CONFIG_TASKSTATS
624 struct taskstats *stats;
628 struct tty_audit_buf *tty_audit_buf;
631 int oom_adj; /* OOM kill score adjustment (bit shift) */
634 /* Context switch must be unlocked if interrupts are to be enabled */
635 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
636 # define __ARCH_WANT_UNLOCKED_CTXSW
640 * Bits in flags field of signal_struct.
642 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
643 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
644 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
645 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
647 * Pending notifications to parent.
649 #define SIGNAL_CLD_STOPPED 0x00000010
650 #define SIGNAL_CLD_CONTINUED 0x00000020
651 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
653 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
655 /* If true, all threads except ->group_exit_task have pending SIGKILL */
656 static inline int signal_group_exit(const struct signal_struct *sig)
658 return (sig->flags & SIGNAL_GROUP_EXIT) ||
659 (sig->group_exit_task != NULL);
663 * Some day this will be a full-fledged user tracking system..
666 atomic_t __count; /* reference count */
667 atomic_t processes; /* How many processes does this user have? */
668 atomic_t files; /* How many open files does this user have? */
669 atomic_t sigpending; /* How many pending signals does this user have? */
670 #ifdef CONFIG_INOTIFY_USER
671 atomic_t inotify_watches; /* How many inotify watches does this user have? */
672 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
675 atomic_t epoll_watches; /* The number of file descriptors currently watched */
677 #ifdef CONFIG_POSIX_MQUEUE
678 /* protected by mq_lock */
679 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
681 unsigned long locked_shm; /* How many pages of mlocked shm ? */
684 struct key *uid_keyring; /* UID specific keyring */
685 struct key *session_keyring; /* UID's default session keyring */
688 /* Hash table maintenance information */
689 struct hlist_node uidhash_node;
691 struct user_namespace *user_ns;
693 #ifdef CONFIG_PERF_EVENTS
694 atomic_long_t locked_vm;
698 extern int uids_sysfs_init(void);
700 extern struct user_struct *find_user(uid_t);
702 extern struct user_struct root_user;
703 #define INIT_USER (&root_user)
706 struct backing_dev_info;
707 struct reclaim_state;
709 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
711 /* cumulative counters */
712 unsigned long pcount; /* # of times run on this cpu */
713 unsigned long long run_delay; /* time spent waiting on a runqueue */
716 unsigned long long last_arrival,/* when we last ran on a cpu */
717 last_queued; /* when we were last queued to run */
718 #ifdef CONFIG_SCHEDSTATS
720 unsigned int bkl_count;
723 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
725 #ifdef CONFIG_TASK_DELAY_ACCT
726 struct task_delay_info {
728 unsigned int flags; /* Private per-task flags */
730 /* For each stat XXX, add following, aligned appropriately
732 * struct timespec XXX_start, XXX_end;
736 * Atomicity of updates to XXX_delay, XXX_count protected by
737 * single lock above (split into XXX_lock if contention is an issue).
741 * XXX_count is incremented on every XXX operation, the delay
742 * associated with the operation is added to XXX_delay.
743 * XXX_delay contains the accumulated delay time in nanoseconds.
745 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
746 u64 blkio_delay; /* wait for sync block io completion */
747 u64 swapin_delay; /* wait for swapin block io completion */
748 u32 blkio_count; /* total count of the number of sync block */
749 /* io operations performed */
750 u32 swapin_count; /* total count of the number of swapin block */
751 /* io operations performed */
753 struct timespec freepages_start, freepages_end;
754 u64 freepages_delay; /* wait for memory reclaim */
755 u32 freepages_count; /* total count of memory reclaim */
757 #endif /* CONFIG_TASK_DELAY_ACCT */
759 static inline int sched_info_on(void)
761 #ifdef CONFIG_SCHEDSTATS
763 #elif defined(CONFIG_TASK_DELAY_ACCT)
764 extern int delayacct_on;
779 * sched-domains (multiprocessor balancing) declarations:
783 * Increase resolution of nice-level calculations:
785 #define SCHED_LOAD_SHIFT 10
786 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
788 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
791 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
792 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
793 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
794 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
795 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
796 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
797 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
798 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
799 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
800 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
801 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
803 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
805 enum powersavings_balance_level {
806 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
807 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
808 * first for long running threads
810 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
811 * cpu package for power savings
813 MAX_POWERSAVINGS_BALANCE_LEVELS
816 extern int sched_mc_power_savings, sched_smt_power_savings;
818 static inline int sd_balance_for_mc_power(void)
820 if (sched_smt_power_savings)
821 return SD_POWERSAVINGS_BALANCE;
823 if (!sched_mc_power_savings)
824 return SD_PREFER_SIBLING;
829 static inline int sd_balance_for_package_power(void)
831 if (sched_mc_power_savings | sched_smt_power_savings)
832 return SD_POWERSAVINGS_BALANCE;
834 return SD_PREFER_SIBLING;
838 * Optimise SD flags for power savings:
839 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
840 * Keep default SD flags if sched_{smt,mc}_power_saving=0
843 static inline int sd_power_saving_flags(void)
845 if (sched_mc_power_savings | sched_smt_power_savings)
846 return SD_BALANCE_NEWIDLE;
852 struct sched_group *next; /* Must be a circular list */
855 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
858 unsigned int cpu_power;
859 unsigned int group_weight;
862 * The CPUs this group covers.
864 * NOTE: this field is variable length. (Allocated dynamically
865 * by attaching extra space to the end of the structure,
866 * depending on how many CPUs the kernel has booted up with)
868 * It is also be embedded into static data structures at build
869 * time. (See 'struct static_sched_group' in kernel/sched.c)
871 unsigned long cpumask[0];
874 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
876 return to_cpumask(sg->cpumask);
879 enum sched_domain_level {
889 struct sched_domain_attr {
890 int relax_domain_level;
893 #define SD_ATTR_INIT (struct sched_domain_attr) { \
894 .relax_domain_level = -1, \
897 struct sched_domain {
898 /* These fields must be setup */
899 struct sched_domain *parent; /* top domain must be null terminated */
900 struct sched_domain *child; /* bottom domain must be null terminated */
901 struct sched_group *groups; /* the balancing groups of the domain */
902 unsigned long min_interval; /* Minimum balance interval ms */
903 unsigned long max_interval; /* Maximum balance interval ms */
904 unsigned int busy_factor; /* less balancing by factor if busy */
905 unsigned int imbalance_pct; /* No balance until over watermark */
906 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
907 unsigned int busy_idx;
908 unsigned int idle_idx;
909 unsigned int newidle_idx;
910 unsigned int wake_idx;
911 unsigned int forkexec_idx;
912 unsigned int smt_gain;
913 int flags; /* See SD_* */
914 enum sched_domain_level level;
916 /* Runtime fields. */
917 unsigned long last_balance; /* init to jiffies. units in jiffies */
918 unsigned int balance_interval; /* initialise to 1. units in ms. */
919 unsigned int nr_balance_failed; /* initialise to 0 */
923 #ifdef CONFIG_SCHEDSTATS
924 /* load_balance() stats */
925 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
926 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
927 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
928 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
929 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
931 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
932 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
934 /* Active load balancing */
935 unsigned int alb_count;
936 unsigned int alb_failed;
937 unsigned int alb_pushed;
939 /* SD_BALANCE_EXEC stats */
940 unsigned int sbe_count;
941 unsigned int sbe_balanced;
942 unsigned int sbe_pushed;
944 /* SD_BALANCE_FORK stats */
945 unsigned int sbf_count;
946 unsigned int sbf_balanced;
947 unsigned int sbf_pushed;
949 /* try_to_wake_up() stats */
950 unsigned int ttwu_wake_remote;
951 unsigned int ttwu_move_affine;
952 unsigned int ttwu_move_balance;
954 #ifdef CONFIG_SCHED_DEBUG
958 unsigned int span_weight;
960 * Span of all CPUs in this domain.
962 * NOTE: this field is variable length. (Allocated dynamically
963 * by attaching extra space to the end of the structure,
964 * depending on how many CPUs the kernel has booted up with)
966 * It is also be embedded into static data structures at build
967 * time. (See 'struct static_sched_domain' in kernel/sched.c)
969 unsigned long span[0];
972 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
974 return to_cpumask(sd->span);
977 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
978 struct sched_domain_attr *dattr_new);
980 /* Allocate an array of sched domains, for partition_sched_domains(). */
981 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
982 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
984 /* Test a flag in parent sched domain */
985 static inline int test_sd_parent(struct sched_domain *sd, int flag)
987 if (sd->parent && (sd->parent->flags & flag))
993 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
994 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
996 #else /* CONFIG_SMP */
998 struct sched_domain_attr;
1001 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1002 struct sched_domain_attr *dattr_new)
1005 #endif /* !CONFIG_SMP */
1008 struct io_context; /* See blkdev.h */
1011 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1012 extern void prefetch_stack(struct task_struct *t);
1014 static inline void prefetch_stack(struct task_struct *t) { }
1017 struct audit_context; /* See audit.c */
1019 struct pipe_inode_info;
1020 struct uts_namespace;
1023 struct sched_domain;
1028 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1029 #define WF_FORK 0x02 /* child wakeup after fork */
1031 #define ENQUEUE_WAKEUP 1
1032 #define ENQUEUE_WAKING 2
1033 #define ENQUEUE_HEAD 4
1035 #define DEQUEUE_SLEEP 1
1037 struct sched_class {
1038 const struct sched_class *next;
1040 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1041 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1042 void (*yield_task) (struct rq *rq);
1044 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1046 struct task_struct * (*pick_next_task) (struct rq *rq);
1047 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1050 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1051 int sd_flag, int flags);
1053 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1054 void (*post_schedule) (struct rq *this_rq);
1055 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1056 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1058 void (*set_cpus_allowed)(struct task_struct *p,
1059 const struct cpumask *newmask);
1061 void (*rq_online)(struct rq *rq);
1062 void (*rq_offline)(struct rq *rq);
1065 void (*set_curr_task) (struct rq *rq);
1066 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1067 void (*task_fork) (struct task_struct *p);
1069 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1071 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1073 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1074 int oldprio, int running);
1076 unsigned int (*get_rr_interval) (struct rq *rq,
1077 struct task_struct *task);
1079 #ifdef CONFIG_FAIR_GROUP_SCHED
1080 void (*task_move_group) (struct task_struct *p, int on_rq);
1084 struct load_weight {
1085 unsigned long weight, inv_weight;
1088 #ifdef CONFIG_SCHEDSTATS
1089 struct sched_statistics {
1099 s64 sum_sleep_runtime;
1106 u64 nr_migrations_cold;
1107 u64 nr_failed_migrations_affine;
1108 u64 nr_failed_migrations_running;
1109 u64 nr_failed_migrations_hot;
1110 u64 nr_forced_migrations;
1113 u64 nr_wakeups_sync;
1114 u64 nr_wakeups_migrate;
1115 u64 nr_wakeups_local;
1116 u64 nr_wakeups_remote;
1117 u64 nr_wakeups_affine;
1118 u64 nr_wakeups_affine_attempts;
1119 u64 nr_wakeups_passive;
1120 u64 nr_wakeups_idle;
1124 struct sched_entity {
1125 struct load_weight load; /* for load-balancing */
1126 struct rb_node run_node;
1127 struct list_head group_node;
1131 u64 sum_exec_runtime;
1133 u64 prev_sum_exec_runtime;
1137 #ifdef CONFIG_SCHEDSTATS
1138 struct sched_statistics statistics;
1141 #ifdef CONFIG_FAIR_GROUP_SCHED
1142 struct sched_entity *parent;
1143 /* rq on which this entity is (to be) queued: */
1144 struct cfs_rq *cfs_rq;
1145 /* rq "owned" by this entity/group: */
1146 struct cfs_rq *my_q;
1150 struct sched_rt_entity {
1151 struct list_head run_list;
1152 unsigned long timeout;
1153 unsigned int time_slice;
1154 int nr_cpus_allowed;
1156 struct sched_rt_entity *back;
1157 #ifdef CONFIG_RT_GROUP_SCHED
1158 struct sched_rt_entity *parent;
1159 /* rq on which this entity is (to be) queued: */
1160 struct rt_rq *rt_rq;
1161 /* rq "owned" by this entity/group: */
1168 struct task_struct {
1169 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1172 unsigned int flags; /* per process flags, defined below */
1173 unsigned int ptrace;
1175 int lock_depth; /* BKL lock depth */
1178 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1183 int prio, static_prio, normal_prio;
1184 unsigned int rt_priority;
1185 const struct sched_class *sched_class;
1186 struct sched_entity se;
1187 struct sched_rt_entity rt;
1189 #ifdef CONFIG_PREEMPT_NOTIFIERS
1190 /* list of struct preempt_notifier: */
1191 struct hlist_head preempt_notifiers;
1195 * fpu_counter contains the number of consecutive context switches
1196 * that the FPU is used. If this is over a threshold, the lazy fpu
1197 * saving becomes unlazy to save the trap. This is an unsigned char
1198 * so that after 256 times the counter wraps and the behavior turns
1199 * lazy again; this to deal with bursty apps that only use FPU for
1202 unsigned char fpu_counter;
1203 #ifdef CONFIG_BLK_DEV_IO_TRACE
1204 unsigned int btrace_seq;
1207 unsigned int policy;
1208 cpumask_t cpus_allowed;
1210 #ifdef CONFIG_TREE_PREEMPT_RCU
1211 int rcu_read_lock_nesting;
1212 char rcu_read_unlock_special;
1213 struct rcu_node *rcu_blocked_node;
1214 struct list_head rcu_node_entry;
1215 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1217 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1218 struct sched_info sched_info;
1221 struct list_head tasks;
1222 struct plist_node pushable_tasks;
1224 struct mm_struct *mm, *active_mm;
1225 #if defined(SPLIT_RSS_COUNTING)
1226 struct task_rss_stat rss_stat;
1230 int exit_code, exit_signal;
1231 int pdeath_signal; /* The signal sent when the parent dies */
1233 unsigned int personality;
1234 unsigned did_exec:1;
1235 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1237 unsigned in_iowait:1;
1240 /* Revert to default priority/policy when forking */
1241 unsigned sched_reset_on_fork:1;
1246 #ifdef CONFIG_CC_STACKPROTECTOR
1247 /* Canary value for the -fstack-protector gcc feature */
1248 unsigned long stack_canary;
1252 * pointers to (original) parent process, youngest child, younger sibling,
1253 * older sibling, respectively. (p->father can be replaced with
1254 * p->real_parent->pid)
1256 struct task_struct *real_parent; /* real parent process */
1257 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1259 * children/sibling forms the list of my natural children
1261 struct list_head children; /* list of my children */
1262 struct list_head sibling; /* linkage in my parent's children list */
1263 struct task_struct *group_leader; /* threadgroup leader */
1266 * ptraced is the list of tasks this task is using ptrace on.
1267 * This includes both natural children and PTRACE_ATTACH targets.
1268 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1270 struct list_head ptraced;
1271 struct list_head ptrace_entry;
1273 /* PID/PID hash table linkage. */
1274 struct pid_link pids[PIDTYPE_MAX];
1275 struct list_head thread_group;
1277 struct completion *vfork_done; /* for vfork() */
1278 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1279 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1281 cputime_t utime, stime, utimescaled, stimescaled;
1283 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1284 cputime_t prev_utime, prev_stime;
1286 unsigned long nvcsw, nivcsw; /* context switch counts */
1287 struct timespec start_time; /* monotonic time */
1288 struct timespec real_start_time; /* boot based time */
1289 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1290 unsigned long min_flt, maj_flt;
1292 struct task_cputime cputime_expires;
1293 struct list_head cpu_timers[3];
1295 /* process credentials */
1296 const struct cred *real_cred; /* objective and real subjective task
1297 * credentials (COW) */
1298 const struct cred *cred; /* effective (overridable) subjective task
1299 * credentials (COW) */
1300 struct mutex cred_guard_mutex; /* guard against foreign influences on
1301 * credential calculations
1302 * (notably. ptrace) */
1303 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1305 char comm[TASK_COMM_LEN]; /* executable name excluding path
1306 - access with [gs]et_task_comm (which lock
1307 it with task_lock())
1308 - initialized normally by setup_new_exec */
1309 /* file system info */
1310 int link_count, total_link_count;
1311 #ifdef CONFIG_SYSVIPC
1313 struct sysv_sem sysvsem;
1315 #ifdef CONFIG_DETECT_HUNG_TASK
1316 /* hung task detection */
1317 unsigned long last_switch_count;
1319 /* CPU-specific state of this task */
1320 struct thread_struct thread;
1321 /* filesystem information */
1322 struct fs_struct *fs;
1323 /* open file information */
1324 struct files_struct *files;
1326 struct nsproxy *nsproxy;
1327 /* signal handlers */
1328 struct signal_struct *signal;
1329 struct sighand_struct *sighand;
1331 sigset_t blocked, real_blocked;
1332 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1333 struct sigpending pending;
1335 unsigned long sas_ss_sp;
1337 int (*notifier)(void *priv);
1338 void *notifier_data;
1339 sigset_t *notifier_mask;
1340 struct audit_context *audit_context;
1341 #ifdef CONFIG_AUDITSYSCALL
1343 unsigned int sessionid;
1347 /* Thread group tracking */
1350 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1352 spinlock_t alloc_lock;
1354 #ifdef CONFIG_GENERIC_HARDIRQS
1355 /* IRQ handler threads */
1356 struct irqaction *irqaction;
1359 /* Protection of the PI data structures: */
1360 raw_spinlock_t pi_lock;
1362 #ifdef CONFIG_RT_MUTEXES
1363 /* PI waiters blocked on a rt_mutex held by this task */
1364 struct plist_head pi_waiters;
1365 /* Deadlock detection and priority inheritance handling */
1366 struct rt_mutex_waiter *pi_blocked_on;
1369 #ifdef CONFIG_DEBUG_MUTEXES
1370 /* mutex deadlock detection */
1371 struct mutex_waiter *blocked_on;
1373 #ifdef CONFIG_TRACE_IRQFLAGS
1374 unsigned int irq_events;
1375 unsigned long hardirq_enable_ip;
1376 unsigned long hardirq_disable_ip;
1377 unsigned int hardirq_enable_event;
1378 unsigned int hardirq_disable_event;
1379 int hardirqs_enabled;
1380 int hardirq_context;
1381 unsigned long softirq_disable_ip;
1382 unsigned long softirq_enable_ip;
1383 unsigned int softirq_disable_event;
1384 unsigned int softirq_enable_event;
1385 int softirqs_enabled;
1386 int softirq_context;
1388 #ifdef CONFIG_LOCKDEP
1389 # define MAX_LOCK_DEPTH 48UL
1392 unsigned int lockdep_recursion;
1393 struct held_lock held_locks[MAX_LOCK_DEPTH];
1394 gfp_t lockdep_reclaim_gfp;
1397 /* journalling filesystem info */
1400 /* stacked block device info */
1401 struct bio_list *bio_list;
1404 struct reclaim_state *reclaim_state;
1406 struct backing_dev_info *backing_dev_info;
1408 struct io_context *io_context;
1410 unsigned long ptrace_message;
1411 siginfo_t *last_siginfo; /* For ptrace use. */
1412 struct task_io_accounting ioac;
1413 #if defined(CONFIG_TASK_XACCT)
1414 u64 acct_rss_mem1; /* accumulated rss usage */
1415 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1416 cputime_t acct_timexpd; /* stime + utime since last update */
1418 #ifdef CONFIG_CPUSETS
1419 nodemask_t mems_allowed; /* Protected by alloc_lock */
1420 int mems_allowed_change_disable;
1421 int cpuset_mem_spread_rotor;
1422 int cpuset_slab_spread_rotor;
1424 #ifdef CONFIG_CGROUPS
1425 /* Control Group info protected by css_set_lock */
1426 struct css_set *cgroups;
1427 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1428 struct list_head cg_list;
1431 struct robust_list_head __user *robust_list;
1432 #ifdef CONFIG_COMPAT
1433 struct compat_robust_list_head __user *compat_robust_list;
1435 struct list_head pi_state_list;
1436 struct futex_pi_state *pi_state_cache;
1438 #ifdef CONFIG_PERF_EVENTS
1439 struct perf_event_context *perf_event_ctxp;
1440 struct mutex perf_event_mutex;
1441 struct list_head perf_event_list;
1444 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1447 atomic_t fs_excl; /* holding fs exclusive resources */
1448 struct rcu_head rcu;
1451 * cache last used pipe for splice
1453 struct pipe_inode_info *splice_pipe;
1454 #ifdef CONFIG_TASK_DELAY_ACCT
1455 struct task_delay_info *delays;
1457 #ifdef CONFIG_FAULT_INJECTION
1460 struct prop_local_single dirties;
1461 #ifdef CONFIG_LATENCYTOP
1462 int latency_record_count;
1463 struct latency_record latency_record[LT_SAVECOUNT];
1466 * time slack values; these are used to round up poll() and
1467 * select() etc timeout values. These are in nanoseconds.
1469 unsigned long timer_slack_ns;
1470 unsigned long default_timer_slack_ns;
1472 struct list_head *scm_work_list;
1473 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1474 /* Index of current stored address in ret_stack */
1476 /* Stack of return addresses for return function tracing */
1477 struct ftrace_ret_stack *ret_stack;
1478 /* time stamp for last schedule */
1479 unsigned long long ftrace_timestamp;
1481 * Number of functions that haven't been traced
1482 * because of depth overrun.
1484 atomic_t trace_overrun;
1485 /* Pause for the tracing */
1486 atomic_t tracing_graph_pause;
1488 #ifdef CONFIG_TRACING
1489 /* state flags for use by tracers */
1490 unsigned long trace;
1491 /* bitmask of trace recursion */
1492 unsigned long trace_recursion;
1493 #endif /* CONFIG_TRACING */
1494 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1495 struct memcg_batch_info {
1496 int do_batch; /* incremented when batch uncharge started */
1497 struct mem_cgroup *memcg; /* target memcg of uncharge */
1498 unsigned long bytes; /* uncharged usage */
1499 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1504 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1505 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1508 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1509 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1510 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1511 * values are inverted: lower p->prio value means higher priority.
1513 * The MAX_USER_RT_PRIO value allows the actual maximum
1514 * RT priority to be separate from the value exported to
1515 * user-space. This allows kernel threads to set their
1516 * priority to a value higher than any user task. Note:
1517 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1520 #define MAX_USER_RT_PRIO 100
1521 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1523 #define MAX_PRIO (MAX_RT_PRIO + 40)
1524 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1526 static inline int rt_prio(int prio)
1528 if (unlikely(prio < MAX_RT_PRIO))
1533 static inline int rt_task(struct task_struct *p)
1535 return rt_prio(p->prio);
1538 static inline struct pid *task_pid(struct task_struct *task)
1540 return task->pids[PIDTYPE_PID].pid;
1543 static inline struct pid *task_tgid(struct task_struct *task)
1545 return task->group_leader->pids[PIDTYPE_PID].pid;
1549 * Without tasklist or rcu lock it is not safe to dereference
1550 * the result of task_pgrp/task_session even if task == current,
1551 * we can race with another thread doing sys_setsid/sys_setpgid.
1553 static inline struct pid *task_pgrp(struct task_struct *task)
1555 return task->group_leader->pids[PIDTYPE_PGID].pid;
1558 static inline struct pid *task_session(struct task_struct *task)
1560 return task->group_leader->pids[PIDTYPE_SID].pid;
1563 struct pid_namespace;
1566 * the helpers to get the task's different pids as they are seen
1567 * from various namespaces
1569 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1570 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1572 * task_xid_nr_ns() : id seen from the ns specified;
1574 * set_task_vxid() : assigns a virtual id to a task;
1576 * see also pid_nr() etc in include/linux/pid.h
1578 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1579 struct pid_namespace *ns);
1581 static inline pid_t task_pid_nr(struct task_struct *tsk)
1586 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1587 struct pid_namespace *ns)
1589 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1592 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1594 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1598 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1603 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1605 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1607 return pid_vnr(task_tgid(tsk));
1611 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1612 struct pid_namespace *ns)
1614 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1617 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1619 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1623 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1624 struct pid_namespace *ns)
1626 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1629 static inline pid_t task_session_vnr(struct task_struct *tsk)
1631 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1634 /* obsolete, do not use */
1635 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1637 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1641 * pid_alive - check that a task structure is not stale
1642 * @p: Task structure to be checked.
1644 * Test if a process is not yet dead (at most zombie state)
1645 * If pid_alive fails, then pointers within the task structure
1646 * can be stale and must not be dereferenced.
1648 static inline int pid_alive(struct task_struct *p)
1650 return p->pids[PIDTYPE_PID].pid != NULL;
1654 * is_global_init - check if a task structure is init
1655 * @tsk: Task structure to be checked.
1657 * Check if a task structure is the first user space task the kernel created.
1659 static inline int is_global_init(struct task_struct *tsk)
1661 return tsk->pid == 1;
1665 * is_container_init:
1666 * check whether in the task is init in its own pid namespace.
1668 extern int is_container_init(struct task_struct *tsk);
1670 extern struct pid *cad_pid;
1672 extern void free_task(struct task_struct *tsk);
1673 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1675 extern void __put_task_struct(struct task_struct *t);
1677 static inline void put_task_struct(struct task_struct *t)
1679 if (atomic_dec_and_test(&t->usage))
1680 __put_task_struct(t);
1683 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1684 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1689 #define PF_KSOFTIRQD 0x00000001 /* I am ksoftirqd */
1690 #define PF_STARTING 0x00000002 /* being created */
1691 #define PF_EXITING 0x00000004 /* getting shut down */
1692 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1693 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
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 * Architectures can set this to 1 if they have specified
1789 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1790 * but then during bootup it turns out that sched_clock()
1791 * is reliable after all:
1793 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1794 extern int sched_clock_stable;
1797 /* ftrace calls sched_clock() directly */
1798 extern unsigned long long notrace sched_clock(void);
1800 extern void sched_clock_init(void);
1801 extern u64 sched_clock_cpu(int cpu);
1803 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1804 static inline void sched_clock_tick(void)
1808 static inline void sched_clock_idle_sleep_event(void)
1812 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1816 extern void sched_clock_tick(void);
1817 extern void sched_clock_idle_sleep_event(void);
1818 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1822 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1823 * clock constructed from sched_clock():
1825 extern unsigned long long cpu_clock(int cpu);
1827 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1829 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1830 * The reason for this explicit opt-in is not to have perf penalty with
1831 * slow sched_clocks.
1833 extern void enable_sched_clock_irqtime(void);
1834 extern void disable_sched_clock_irqtime(void);
1836 static inline void enable_sched_clock_irqtime(void) {}
1837 static inline void disable_sched_clock_irqtime(void) {}
1840 extern unsigned long long
1841 task_sched_runtime(struct task_struct *task);
1842 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1844 /* sched_exec is called by processes performing an exec */
1846 extern void sched_exec(void);
1848 #define sched_exec() {}
1851 extern void sched_clock_idle_sleep_event(void);
1852 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1854 #ifdef CONFIG_HOTPLUG_CPU
1855 extern void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p);
1856 extern void idle_task_exit(void);
1858 static inline void idle_task_exit(void) {}
1861 extern void sched_idle_next(void);
1863 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1864 extern void wake_up_idle_cpu(int cpu);
1866 static inline void wake_up_idle_cpu(int cpu) { }
1869 extern unsigned int sysctl_sched_latency;
1870 extern unsigned int sysctl_sched_min_granularity;
1871 extern unsigned int sysctl_sched_wakeup_granularity;
1872 extern unsigned int sysctl_sched_shares_ratelimit;
1873 extern unsigned int sysctl_sched_shares_thresh;
1874 extern unsigned int sysctl_sched_child_runs_first;
1876 enum sched_tunable_scaling {
1877 SCHED_TUNABLESCALING_NONE,
1878 SCHED_TUNABLESCALING_LOG,
1879 SCHED_TUNABLESCALING_LINEAR,
1880 SCHED_TUNABLESCALING_END,
1882 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1884 #ifdef CONFIG_SCHED_DEBUG
1885 extern unsigned int sysctl_sched_migration_cost;
1886 extern unsigned int sysctl_sched_nr_migrate;
1887 extern unsigned int sysctl_sched_time_avg;
1888 extern unsigned int sysctl_timer_migration;
1890 int sched_proc_update_handler(struct ctl_table *table, int write,
1891 void __user *buffer, size_t *length,
1894 #ifdef CONFIG_SCHED_DEBUG
1895 static inline unsigned int get_sysctl_timer_migration(void)
1897 return sysctl_timer_migration;
1900 static inline unsigned int get_sysctl_timer_migration(void)
1905 extern unsigned int sysctl_sched_rt_period;
1906 extern int sysctl_sched_rt_runtime;
1908 int sched_rt_handler(struct ctl_table *table, int write,
1909 void __user *buffer, size_t *lenp,
1912 extern unsigned int sysctl_sched_compat_yield;
1914 #ifdef CONFIG_RT_MUTEXES
1915 extern int rt_mutex_getprio(struct task_struct *p);
1916 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1917 extern void rt_mutex_adjust_pi(struct task_struct *p);
1919 static inline int rt_mutex_getprio(struct task_struct *p)
1921 return p->normal_prio;
1923 # define rt_mutex_adjust_pi(p) do { } while (0)
1926 extern void set_user_nice(struct task_struct *p, long nice);
1927 extern int task_prio(const struct task_struct *p);
1928 extern int task_nice(const struct task_struct *p);
1929 extern int can_nice(const struct task_struct *p, const int nice);
1930 extern int task_curr(const struct task_struct *p);
1931 extern int idle_cpu(int cpu);
1932 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1933 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1934 struct sched_param *);
1935 extern struct task_struct *idle_task(int cpu);
1936 extern struct task_struct *curr_task(int cpu);
1937 extern void set_curr_task(int cpu, struct task_struct *p);
1942 * The default (Linux) execution domain.
1944 extern struct exec_domain default_exec_domain;
1946 union thread_union {
1947 struct thread_info thread_info;
1948 unsigned long stack[THREAD_SIZE/sizeof(long)];
1951 #ifndef __HAVE_ARCH_KSTACK_END
1952 static inline int kstack_end(void *addr)
1954 /* Reliable end of stack detection:
1955 * Some APM bios versions misalign the stack
1957 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1961 extern union thread_union init_thread_union;
1962 extern struct task_struct init_task;
1964 extern struct mm_struct init_mm;
1966 extern struct pid_namespace init_pid_ns;
1969 * find a task by one of its numerical ids
1971 * find_task_by_pid_ns():
1972 * finds a task by its pid in the specified namespace
1973 * find_task_by_vpid():
1974 * finds a task by its virtual pid
1976 * see also find_vpid() etc in include/linux/pid.h
1979 extern struct task_struct *find_task_by_vpid(pid_t nr);
1980 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1981 struct pid_namespace *ns);
1983 extern void __set_special_pids(struct pid *pid);
1985 /* per-UID process charging. */
1986 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1987 static inline struct user_struct *get_uid(struct user_struct *u)
1989 atomic_inc(&u->__count);
1992 extern void free_uid(struct user_struct *);
1993 extern void release_uids(struct user_namespace *ns);
1995 #include <asm/current.h>
1997 extern void do_timer(unsigned long ticks);
1999 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2000 extern int wake_up_process(struct task_struct *tsk);
2001 extern void wake_up_new_task(struct task_struct *tsk,
2002 unsigned long clone_flags);
2004 extern void kick_process(struct task_struct *tsk);
2006 static inline void kick_process(struct task_struct *tsk) { }
2008 extern void sched_fork(struct task_struct *p, int clone_flags);
2009 extern void sched_dead(struct task_struct *p);
2011 extern void proc_caches_init(void);
2012 extern void flush_signals(struct task_struct *);
2013 extern void __flush_signals(struct task_struct *);
2014 extern void ignore_signals(struct task_struct *);
2015 extern void flush_signal_handlers(struct task_struct *, int force_default);
2016 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2018 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2020 unsigned long flags;
2023 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2024 ret = dequeue_signal(tsk, mask, info);
2025 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2030 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2032 extern void unblock_all_signals(void);
2033 extern void release_task(struct task_struct * p);
2034 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2035 extern int force_sigsegv(int, struct task_struct *);
2036 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2037 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2038 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2039 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2040 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2041 extern int kill_pid(struct pid *pid, int sig, int priv);
2042 extern int kill_proc_info(int, struct siginfo *, pid_t);
2043 extern int do_notify_parent(struct task_struct *, int);
2044 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2045 extern void force_sig(int, struct task_struct *);
2046 extern int send_sig(int, struct task_struct *, int);
2047 extern int zap_other_threads(struct task_struct *p);
2048 extern struct sigqueue *sigqueue_alloc(void);
2049 extern void sigqueue_free(struct sigqueue *);
2050 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2051 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2052 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2054 static inline int kill_cad_pid(int sig, int priv)
2056 return kill_pid(cad_pid, sig, priv);
2059 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2060 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2061 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2062 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2065 * True if we are on the alternate signal stack.
2067 static inline int on_sig_stack(unsigned long sp)
2069 #ifdef CONFIG_STACK_GROWSUP
2070 return sp >= current->sas_ss_sp &&
2071 sp - current->sas_ss_sp < current->sas_ss_size;
2073 return sp > current->sas_ss_sp &&
2074 sp - current->sas_ss_sp <= current->sas_ss_size;
2078 static inline int sas_ss_flags(unsigned long sp)
2080 return (current->sas_ss_size == 0 ? SS_DISABLE
2081 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2085 * Routines for handling mm_structs
2087 extern struct mm_struct * mm_alloc(void);
2089 /* mmdrop drops the mm and the page tables */
2090 extern void __mmdrop(struct mm_struct *);
2091 static inline void mmdrop(struct mm_struct * mm)
2093 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2097 /* mmput gets rid of the mappings and all user-space */
2098 extern void mmput(struct mm_struct *);
2099 /* Grab a reference to a task's mm, if it is not already going away */
2100 extern struct mm_struct *get_task_mm(struct task_struct *task);
2101 /* Remove the current tasks stale references to the old mm_struct */
2102 extern void mm_release(struct task_struct *, struct mm_struct *);
2103 /* Allocate a new mm structure and copy contents from tsk->mm */
2104 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2106 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2107 struct task_struct *, struct pt_regs *);
2108 extern void flush_thread(void);
2109 extern void exit_thread(void);
2111 extern void exit_files(struct task_struct *);
2112 extern void __cleanup_sighand(struct sighand_struct *);
2114 extern void exit_itimers(struct signal_struct *);
2115 extern void flush_itimer_signals(void);
2117 extern NORET_TYPE void do_group_exit(int);
2119 extern void daemonize(const char *, ...);
2120 extern int allow_signal(int);
2121 extern int disallow_signal(int);
2123 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2124 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2125 struct task_struct *fork_idle(int);
2127 extern void set_task_comm(struct task_struct *tsk, char *from);
2128 extern char *get_task_comm(char *to, struct task_struct *tsk);
2131 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2133 static inline unsigned long wait_task_inactive(struct task_struct *p,
2140 #define next_task(p) \
2141 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2143 #define for_each_process(p) \
2144 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2146 extern bool current_is_single_threaded(void);
2149 * Careful: do_each_thread/while_each_thread is a double loop so
2150 * 'break' will not work as expected - use goto instead.
2152 #define do_each_thread(g, t) \
2153 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2155 #define while_each_thread(g, t) \
2156 while ((t = next_thread(t)) != g)
2158 static inline int get_nr_threads(struct task_struct *tsk)
2160 return tsk->signal->nr_threads;
2163 /* de_thread depends on thread_group_leader not being a pid based check */
2164 #define thread_group_leader(p) (p == p->group_leader)
2166 /* Do to the insanities of de_thread it is possible for a process
2167 * to have the pid of the thread group leader without actually being
2168 * the thread group leader. For iteration through the pids in proc
2169 * all we care about is that we have a task with the appropriate
2170 * pid, we don't actually care if we have the right task.
2172 static inline int has_group_leader_pid(struct task_struct *p)
2174 return p->pid == p->tgid;
2178 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2180 return p1->tgid == p2->tgid;
2183 static inline struct task_struct *next_thread(const struct task_struct *p)
2185 return list_entry_rcu(p->thread_group.next,
2186 struct task_struct, thread_group);
2189 static inline int thread_group_empty(struct task_struct *p)
2191 return list_empty(&p->thread_group);
2194 #define delay_group_leader(p) \
2195 (thread_group_leader(p) && !thread_group_empty(p))
2197 static inline int task_detached(struct task_struct *p)
2199 return p->exit_signal == -1;
2203 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2204 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2205 * pins the final release of task.io_context. Also protects ->cpuset and
2206 * ->cgroup.subsys[].
2208 * Nests both inside and outside of read_lock(&tasklist_lock).
2209 * It must not be nested with write_lock_irq(&tasklist_lock),
2210 * neither inside nor outside.
2212 static inline void task_lock(struct task_struct *p)
2214 spin_lock(&p->alloc_lock);
2217 static inline void task_unlock(struct task_struct *p)
2219 spin_unlock(&p->alloc_lock);
2222 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2223 unsigned long *flags);
2225 static inline void unlock_task_sighand(struct task_struct *tsk,
2226 unsigned long *flags)
2228 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2231 #ifndef __HAVE_THREAD_FUNCTIONS
2233 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2234 #define task_stack_page(task) ((task)->stack)
2236 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2238 *task_thread_info(p) = *task_thread_info(org);
2239 task_thread_info(p)->task = p;
2242 static inline unsigned long *end_of_stack(struct task_struct *p)
2244 return (unsigned long *)(task_thread_info(p) + 1);
2249 static inline int object_is_on_stack(void *obj)
2251 void *stack = task_stack_page(current);
2253 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2256 extern void thread_info_cache_init(void);
2258 #ifdef CONFIG_DEBUG_STACK_USAGE
2259 static inline unsigned long stack_not_used(struct task_struct *p)
2261 unsigned long *n = end_of_stack(p);
2263 do { /* Skip over canary */
2267 return (unsigned long)n - (unsigned long)end_of_stack(p);
2271 /* set thread flags in other task's structures
2272 * - see asm/thread_info.h for TIF_xxxx flags available
2274 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2276 set_ti_thread_flag(task_thread_info(tsk), flag);
2279 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2281 clear_ti_thread_flag(task_thread_info(tsk), flag);
2284 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2286 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2289 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2291 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2294 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2296 return test_ti_thread_flag(task_thread_info(tsk), flag);
2299 static inline void set_tsk_need_resched(struct task_struct *tsk)
2301 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2304 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2306 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2309 static inline int test_tsk_need_resched(struct task_struct *tsk)
2311 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2314 static inline int restart_syscall(void)
2316 set_tsk_thread_flag(current, TIF_SIGPENDING);
2317 return -ERESTARTNOINTR;
2320 static inline int signal_pending(struct task_struct *p)
2322 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2325 static inline int __fatal_signal_pending(struct task_struct *p)
2327 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2330 static inline int fatal_signal_pending(struct task_struct *p)
2332 return signal_pending(p) && __fatal_signal_pending(p);
2335 static inline int signal_pending_state(long state, struct task_struct *p)
2337 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2339 if (!signal_pending(p))
2342 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2345 static inline int need_resched(void)
2347 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2351 * cond_resched() and cond_resched_lock(): latency reduction via
2352 * explicit rescheduling in places that are safe. The return
2353 * value indicates whether a reschedule was done in fact.
2354 * cond_resched_lock() will drop the spinlock before scheduling,
2355 * cond_resched_softirq() will enable bhs before scheduling.
2357 extern int _cond_resched(void);
2359 #define cond_resched() ({ \
2360 __might_sleep(__FILE__, __LINE__, 0); \
2364 extern int __cond_resched_lock(spinlock_t *lock);
2366 #ifdef CONFIG_PREEMPT
2367 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2369 #define PREEMPT_LOCK_OFFSET 0
2372 #define cond_resched_lock(lock) ({ \
2373 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2374 __cond_resched_lock(lock); \
2377 extern int __cond_resched_softirq(void);
2379 #define cond_resched_softirq() ({ \
2380 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2381 __cond_resched_softirq(); \
2385 * Does a critical section need to be broken due to another
2386 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2387 * but a general need for low latency)
2389 static inline int spin_needbreak(spinlock_t *lock)
2391 #ifdef CONFIG_PREEMPT
2392 return spin_is_contended(lock);
2399 * Thread group CPU time accounting.
2401 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2402 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2404 static inline void thread_group_cputime_init(struct signal_struct *sig)
2406 spin_lock_init(&sig->cputimer.lock);
2410 * Reevaluate whether the task has signals pending delivery.
2411 * Wake the task if so.
2412 * This is required every time the blocked sigset_t changes.
2413 * callers must hold sighand->siglock.
2415 extern void recalc_sigpending_and_wake(struct task_struct *t);
2416 extern void recalc_sigpending(void);
2418 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2421 * Wrappers for p->thread_info->cpu access. No-op on UP.
2425 static inline unsigned int task_cpu(const struct task_struct *p)
2427 return task_thread_info(p)->cpu;
2430 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2434 static inline unsigned int task_cpu(const struct task_struct *p)
2439 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2443 #endif /* CONFIG_SMP */
2445 #ifdef CONFIG_TRACING
2447 __trace_special(void *__tr, void *__data,
2448 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2451 __trace_special(void *__tr, void *__data,
2452 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2457 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2458 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2460 extern void normalize_rt_tasks(void);
2462 #ifdef CONFIG_CGROUP_SCHED
2464 extern struct task_group init_task_group;
2466 extern struct task_group *sched_create_group(struct task_group *parent);
2467 extern void sched_destroy_group(struct task_group *tg);
2468 extern void sched_move_task(struct task_struct *tsk);
2469 #ifdef CONFIG_FAIR_GROUP_SCHED
2470 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2471 extern unsigned long sched_group_shares(struct task_group *tg);
2473 #ifdef CONFIG_RT_GROUP_SCHED
2474 extern int sched_group_set_rt_runtime(struct task_group *tg,
2475 long rt_runtime_us);
2476 extern long sched_group_rt_runtime(struct task_group *tg);
2477 extern int sched_group_set_rt_period(struct task_group *tg,
2479 extern long sched_group_rt_period(struct task_group *tg);
2480 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2484 extern int task_can_switch_user(struct user_struct *up,
2485 struct task_struct *tsk);
2487 #ifdef CONFIG_TASK_XACCT
2488 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2490 tsk->ioac.rchar += amt;
2493 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2495 tsk->ioac.wchar += amt;
2498 static inline void inc_syscr(struct task_struct *tsk)
2503 static inline void inc_syscw(struct task_struct *tsk)
2508 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2512 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2516 static inline void inc_syscr(struct task_struct *tsk)
2520 static inline void inc_syscw(struct task_struct *tsk)
2525 #ifndef TASK_SIZE_OF
2526 #define TASK_SIZE_OF(tsk) TASK_SIZE
2530 * Call the function if the target task is executing on a CPU right now:
2532 extern void task_oncpu_function_call(struct task_struct *p,
2533 void (*func) (void *info), void *info);
2536 #ifdef CONFIG_MM_OWNER
2537 extern void mm_update_next_owner(struct mm_struct *mm);
2538 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2540 static inline void mm_update_next_owner(struct mm_struct *mm)
2544 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2547 #endif /* CONFIG_MM_OWNER */
2549 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2552 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2555 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2558 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2561 static inline unsigned long rlimit(unsigned int limit)
2563 return task_rlimit(current, limit);
2566 static inline unsigned long rlimit_max(unsigned int limit)
2568 return task_rlimit_max(current, limit);
2571 #endif /* __KERNEL__ */