4 #include <linux/auxvec.h> /* For AT_VECTOR_SIZE */
9 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
10 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
11 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
12 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
13 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
14 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
15 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
16 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
17 #define CLONE_THREAD 0x00010000 /* Same thread group? */
18 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
19 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
20 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
21 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
22 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
23 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
24 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
25 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
26 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
27 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
28 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
29 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
30 #define CLONE_NEWNET 0x40000000 /* New network namespace */
35 #define SCHED_NORMAL 0
39 /* SCHED_ISO: reserved but not implemented yet */
48 #include <asm/param.h> /* for HZ */
50 #include <linux/capability.h>
51 #include <linux/threads.h>
52 #include <linux/kernel.h>
53 #include <linux/types.h>
54 #include <linux/timex.h>
55 #include <linux/jiffies.h>
56 #include <linux/rbtree.h>
57 #include <linux/thread_info.h>
58 #include <linux/cpumask.h>
59 #include <linux/errno.h>
60 #include <linux/nodemask.h>
62 #include <asm/system.h>
63 #include <asm/semaphore.h>
65 #include <asm/ptrace.h>
67 #include <asm/cputime.h>
69 #include <linux/smp.h>
70 #include <linux/sem.h>
71 #include <linux/signal.h>
72 #include <linux/securebits.h>
73 #include <linux/fs_struct.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/seccomp.h>
80 #include <linux/rcupdate.h>
81 #include <linux/futex.h>
82 #include <linux/rtmutex.h>
84 #include <linux/time.h>
85 #include <linux/param.h>
86 #include <linux/resource.h>
87 #include <linux/timer.h>
88 #include <linux/hrtimer.h>
89 #include <linux/task_io_accounting.h>
91 #include <asm/processor.h>
94 struct futex_pi_state;
98 * List of flags we want to share for kernel threads,
99 * if only because they are not used by them anyway.
101 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
104 * These are the constant used to fake the fixed-point load-average
105 * counting. Some notes:
106 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
107 * a load-average precision of 10 bits integer + 11 bits fractional
108 * - if you want to count load-averages more often, you need more
109 * precision, or rounding will get you. With 2-second counting freq,
110 * the EXP_n values would be 1981, 2034 and 2043 if still using only
113 extern unsigned long avenrun[]; /* Load averages */
115 #define FSHIFT 11 /* nr of bits of precision */
116 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
117 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
118 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
119 #define EXP_5 2014 /* 1/exp(5sec/5min) */
120 #define EXP_15 2037 /* 1/exp(5sec/15min) */
122 #define CALC_LOAD(load,exp,n) \
124 load += n*(FIXED_1-exp); \
127 extern unsigned long total_forks;
128 extern int nr_threads;
129 DECLARE_PER_CPU(unsigned long, process_counts);
130 extern int nr_processes(void);
131 extern unsigned long nr_running(void);
132 extern unsigned long nr_uninterruptible(void);
133 extern unsigned long nr_active(void);
134 extern unsigned long nr_iowait(void);
135 extern unsigned long weighted_cpuload(const int cpu);
139 #ifdef CONFIG_SCHED_DEBUG
140 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
141 extern void proc_sched_set_task(struct task_struct *p);
143 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
146 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
149 static inline void proc_sched_set_task(struct task_struct *p)
153 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
159 * Task state bitmask. NOTE! These bits are also
160 * encoded in fs/proc/array.c: get_task_state().
162 * We have two separate sets of flags: task->state
163 * is about runnability, while task->exit_state are
164 * about the task exiting. Confusing, but this way
165 * modifying one set can't modify the other one by
168 #define TASK_RUNNING 0
169 #define TASK_INTERRUPTIBLE 1
170 #define TASK_UNINTERRUPTIBLE 2
171 #define TASK_STOPPED 4
172 #define TASK_TRACED 8
173 /* in tsk->exit_state */
174 #define EXIT_ZOMBIE 16
176 /* in tsk->state again */
177 #define TASK_NONINTERACTIVE 64
178 #define TASK_DEAD 128
180 #define __set_task_state(tsk, state_value) \
181 do { (tsk)->state = (state_value); } while (0)
182 #define set_task_state(tsk, state_value) \
183 set_mb((tsk)->state, (state_value))
186 * set_current_state() includes a barrier so that the write of current->state
187 * is correctly serialised wrt the caller's subsequent test of whether to
190 * set_current_state(TASK_UNINTERRUPTIBLE);
191 * if (do_i_need_to_sleep())
194 * If the caller does not need such serialisation then use __set_current_state()
196 #define __set_current_state(state_value) \
197 do { current->state = (state_value); } while (0)
198 #define set_current_state(state_value) \
199 set_mb(current->state, (state_value))
201 /* Task command name length */
202 #define TASK_COMM_LEN 16
204 #include <linux/spinlock.h>
207 * This serializes "schedule()" and also protects
208 * the run-queue from deletions/modifications (but
209 * _adding_ to the beginning of the run-queue has
212 extern rwlock_t tasklist_lock;
213 extern spinlock_t mmlist_lock;
217 extern void sched_init(void);
218 extern void sched_init_smp(void);
219 extern void init_idle(struct task_struct *idle, int cpu);
220 extern void init_idle_bootup_task(struct task_struct *idle);
222 extern cpumask_t nohz_cpu_mask;
223 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
224 extern int select_nohz_load_balancer(int cpu);
226 static inline int select_nohz_load_balancer(int cpu)
233 * Only dump TASK_* tasks. (0 for all tasks)
235 extern void show_state_filter(unsigned long state_filter);
237 static inline void show_state(void)
239 show_state_filter(0);
242 extern void show_regs(struct pt_regs *);
245 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
246 * task), SP is the stack pointer of the first frame that should be shown in the back
247 * trace (or NULL if the entire call-chain of the task should be shown).
249 extern void show_stack(struct task_struct *task, unsigned long *sp);
251 void io_schedule(void);
252 long io_schedule_timeout(long timeout);
254 extern void cpu_init (void);
255 extern void trap_init(void);
256 extern void update_process_times(int user);
257 extern void scheduler_tick(void);
259 #ifdef CONFIG_DETECT_SOFTLOCKUP
260 extern void softlockup_tick(void);
261 extern void spawn_softlockup_task(void);
262 extern void touch_softlockup_watchdog(void);
263 extern void touch_all_softlockup_watchdogs(void);
265 static inline void softlockup_tick(void)
268 static inline void spawn_softlockup_task(void)
271 static inline void touch_softlockup_watchdog(void)
274 static inline void touch_all_softlockup_watchdogs(void)
280 /* Attach to any functions which should be ignored in wchan output. */
281 #define __sched __attribute__((__section__(".sched.text")))
282 /* Is this address in the __sched functions? */
283 extern int in_sched_functions(unsigned long addr);
285 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
286 extern signed long FASTCALL(schedule_timeout(signed long timeout));
287 extern signed long schedule_timeout_interruptible(signed long timeout);
288 extern signed long schedule_timeout_uninterruptible(signed long timeout);
289 asmlinkage void schedule(void);
292 struct user_namespace;
294 /* Maximum number of active map areas.. This is a random (large) number */
295 #define DEFAULT_MAX_MAP_COUNT 65536
297 extern int sysctl_max_map_count;
299 #include <linux/aio.h>
302 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
303 unsigned long, unsigned long);
305 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
306 unsigned long len, unsigned long pgoff,
307 unsigned long flags);
308 extern void arch_unmap_area(struct mm_struct *, unsigned long);
309 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
311 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
313 * The mm counters are not protected by its page_table_lock,
314 * so must be incremented atomically.
316 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
317 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
318 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
319 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
320 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
321 typedef atomic_long_t mm_counter_t;
323 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
325 * The mm counters are protected by its page_table_lock,
326 * so can be incremented directly.
328 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
329 #define get_mm_counter(mm, member) ((mm)->_##member)
330 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
331 #define inc_mm_counter(mm, member) (mm)->_##member++
332 #define dec_mm_counter(mm, member) (mm)->_##member--
333 typedef unsigned long mm_counter_t;
335 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
337 #define get_mm_rss(mm) \
338 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
339 #define update_hiwater_rss(mm) do { \
340 unsigned long _rss = get_mm_rss(mm); \
341 if ((mm)->hiwater_rss < _rss) \
342 (mm)->hiwater_rss = _rss; \
344 #define update_hiwater_vm(mm) do { \
345 if ((mm)->hiwater_vm < (mm)->total_vm) \
346 (mm)->hiwater_vm = (mm)->total_vm; \
349 extern void set_dumpable(struct mm_struct *mm, int value);
350 extern int get_dumpable(struct mm_struct *mm);
354 #define MMF_DUMPABLE 0 /* core dump is permitted */
355 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
356 #define MMF_DUMPABLE_BITS 2
358 /* coredump filter bits */
359 #define MMF_DUMP_ANON_PRIVATE 2
360 #define MMF_DUMP_ANON_SHARED 3
361 #define MMF_DUMP_MAPPED_PRIVATE 4
362 #define MMF_DUMP_MAPPED_SHARED 5
363 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
364 #define MMF_DUMP_FILTER_BITS 4
365 #define MMF_DUMP_FILTER_MASK \
366 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
367 #define MMF_DUMP_FILTER_DEFAULT \
368 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
371 struct vm_area_struct * mmap; /* list of VMAs */
372 struct rb_root mm_rb;
373 struct vm_area_struct * mmap_cache; /* last find_vma result */
374 unsigned long (*get_unmapped_area) (struct file *filp,
375 unsigned long addr, unsigned long len,
376 unsigned long pgoff, unsigned long flags);
377 void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
378 unsigned long mmap_base; /* base of mmap area */
379 unsigned long task_size; /* size of task vm space */
380 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */
381 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */
383 atomic_t mm_users; /* How many users with user space? */
384 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
385 int map_count; /* number of VMAs */
386 struct rw_semaphore mmap_sem;
387 spinlock_t page_table_lock; /* Protects page tables and some counters */
389 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung
390 * together off init_mm.mmlist, and are protected
394 /* Special counters, in some configurations protected by the
395 * page_table_lock, in other configurations by being atomic.
397 mm_counter_t _file_rss;
398 mm_counter_t _anon_rss;
400 unsigned long hiwater_rss; /* High-watermark of RSS usage */
401 unsigned long hiwater_vm; /* High-water virtual memory usage */
403 unsigned long total_vm, locked_vm, shared_vm, exec_vm;
404 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
405 unsigned long start_code, end_code, start_data, end_data;
406 unsigned long start_brk, brk, start_stack;
407 unsigned long arg_start, arg_end, env_start, env_end;
409 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
411 cpumask_t cpu_vm_mask;
413 /* Architecture-specific MM context */
414 mm_context_t context;
416 /* Swap token stuff */
418 * Last value of global fault stamp as seen by this process.
419 * In other words, this value gives an indication of how long
420 * it has been since this task got the token.
421 * Look at mm/thrash.c
423 unsigned int faultstamp;
424 unsigned int token_priority;
425 unsigned int last_interval;
427 unsigned long flags; /* Must use atomic bitops to access the bits */
429 /* coredumping support */
431 struct completion *core_startup_done, core_done;
434 rwlock_t ioctx_list_lock;
435 struct kioctx *ioctx_list;
438 struct sighand_struct {
440 struct k_sigaction action[_NSIG];
442 wait_queue_head_t signalfd_wqh;
445 struct pacct_struct {
448 unsigned long ac_mem;
449 cputime_t ac_utime, ac_stime;
450 unsigned long ac_minflt, ac_majflt;
454 * NOTE! "signal_struct" does not have it's own
455 * locking, because a shared signal_struct always
456 * implies a shared sighand_struct, so locking
457 * sighand_struct is always a proper superset of
458 * the locking of signal_struct.
460 struct signal_struct {
464 wait_queue_head_t wait_chldexit; /* for wait4() */
466 /* current thread group signal load-balancing target: */
467 struct task_struct *curr_target;
469 /* shared signal handling: */
470 struct sigpending shared_pending;
472 /* thread group exit support */
475 * - notify group_exit_task when ->count is equal to notify_count
476 * - everyone except group_exit_task is stopped during signal delivery
477 * of fatal signals, group_exit_task processes the signal.
479 struct task_struct *group_exit_task;
482 /* thread group stop support, overloads group_exit_code too */
483 int group_stop_count;
484 unsigned int flags; /* see SIGNAL_* flags below */
486 /* POSIX.1b Interval Timers */
487 struct list_head posix_timers;
489 /* ITIMER_REAL timer for the process */
490 struct hrtimer real_timer;
491 struct task_struct *tsk;
492 ktime_t it_real_incr;
494 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
495 cputime_t it_prof_expires, it_virt_expires;
496 cputime_t it_prof_incr, it_virt_incr;
498 /* job control IDs */
500 struct pid *tty_old_pgrp;
503 pid_t session __deprecated;
507 /* boolean value for session group leader */
510 struct tty_struct *tty; /* NULL if no tty */
513 * Cumulative resource counters for dead threads in the group,
514 * and for reaped dead child processes forked by this group.
515 * Live threads maintain their own counters and add to these
516 * in __exit_signal, except for the group leader.
518 cputime_t utime, stime, cutime, cstime;
519 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
520 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
521 unsigned long inblock, oublock, cinblock, coublock;
524 * Cumulative ns of scheduled CPU time for dead threads in the
525 * group, not including a zombie group leader. (This only differs
526 * from jiffies_to_ns(utime + stime) if sched_clock uses something
527 * other than jiffies.)
529 unsigned long long sum_sched_runtime;
532 * We don't bother to synchronize most readers of this at all,
533 * because there is no reader checking a limit that actually needs
534 * to get both rlim_cur and rlim_max atomically, and either one
535 * alone is a single word that can safely be read normally.
536 * getrlimit/setrlimit use task_lock(current->group_leader) to
537 * protect this instead of the siglock, because they really
538 * have no need to disable irqs.
540 struct rlimit rlim[RLIM_NLIMITS];
542 struct list_head cpu_timers[3];
544 /* keep the process-shared keyrings here so that they do the right
545 * thing in threads created with CLONE_THREAD */
547 struct key *session_keyring; /* keyring inherited over fork */
548 struct key *process_keyring; /* keyring private to this process */
550 #ifdef CONFIG_BSD_PROCESS_ACCT
551 struct pacct_struct pacct; /* per-process accounting information */
553 #ifdef CONFIG_TASKSTATS
554 struct taskstats *stats;
558 struct tty_audit_buf *tty_audit_buf;
562 /* Context switch must be unlocked if interrupts are to be enabled */
563 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
564 # define __ARCH_WANT_UNLOCKED_CTXSW
568 * Bits in flags field of signal_struct.
570 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
571 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
572 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
573 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
576 * Some day this will be a full-fledged user tracking system..
579 atomic_t __count; /* reference count */
580 atomic_t processes; /* How many processes does this user have? */
581 atomic_t files; /* How many open files does this user have? */
582 atomic_t sigpending; /* How many pending signals does this user have? */
583 #ifdef CONFIG_INOTIFY_USER
584 atomic_t inotify_watches; /* How many inotify watches does this user have? */
585 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
587 /* protected by mq_lock */
588 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
589 unsigned long locked_shm; /* How many pages of mlocked shm ? */
592 struct key *uid_keyring; /* UID specific keyring */
593 struct key *session_keyring; /* UID's default session keyring */
596 /* Hash table maintenance information */
597 struct hlist_node uidhash_node;
601 extern struct user_struct *find_user(uid_t);
603 extern struct user_struct root_user;
604 #define INIT_USER (&root_user)
606 struct backing_dev_info;
607 struct reclaim_state;
609 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
611 /* cumulative counters */
612 unsigned long pcnt; /* # of times run on this cpu */
613 unsigned long long cpu_time, /* time spent on the cpu */
614 run_delay; /* time spent waiting on a runqueue */
617 unsigned long long last_arrival,/* when we last ran on a cpu */
618 last_queued; /* when we were last queued to run */
620 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
622 #ifdef CONFIG_SCHEDSTATS
623 extern const struct file_operations proc_schedstat_operations;
624 #endif /* CONFIG_SCHEDSTATS */
626 #ifdef CONFIG_TASK_DELAY_ACCT
627 struct task_delay_info {
629 unsigned int flags; /* Private per-task flags */
631 /* For each stat XXX, add following, aligned appropriately
633 * struct timespec XXX_start, XXX_end;
637 * Atomicity of updates to XXX_delay, XXX_count protected by
638 * single lock above (split into XXX_lock if contention is an issue).
642 * XXX_count is incremented on every XXX operation, the delay
643 * associated with the operation is added to XXX_delay.
644 * XXX_delay contains the accumulated delay time in nanoseconds.
646 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
647 u64 blkio_delay; /* wait for sync block io completion */
648 u64 swapin_delay; /* wait for swapin block io completion */
649 u32 blkio_count; /* total count of the number of sync block */
650 /* io operations performed */
651 u32 swapin_count; /* total count of the number of swapin block */
652 /* io operations performed */
654 #endif /* CONFIG_TASK_DELAY_ACCT */
656 static inline int sched_info_on(void)
658 #ifdef CONFIG_SCHEDSTATS
660 #elif defined(CONFIG_TASK_DELAY_ACCT)
661 extern int delayacct_on;
676 * sched-domains (multiprocessor balancing) declarations:
680 * Increase resolution of nice-level calculations:
682 #define SCHED_LOAD_SHIFT 10
683 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
685 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
688 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
689 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
690 #define SD_BALANCE_EXEC 4 /* Balance on exec */
691 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
692 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
693 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
694 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
695 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
696 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
697 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
698 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
700 #define BALANCE_FOR_MC_POWER \
701 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
703 #define BALANCE_FOR_PKG_POWER \
704 ((sched_mc_power_savings || sched_smt_power_savings) ? \
705 SD_POWERSAVINGS_BALANCE : 0)
707 #define test_sd_parent(sd, flag) ((sd->parent && \
708 (sd->parent->flags & flag)) ? 1 : 0)
712 struct sched_group *next; /* Must be a circular list */
716 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
717 * single CPU. This is read only (except for setup, hotplug CPU).
718 * Note : Never change cpu_power without recompute its reciprocal
720 unsigned int __cpu_power;
722 * reciprocal value of cpu_power to avoid expensive divides
723 * (see include/linux/reciprocal_div.h)
725 u32 reciprocal_cpu_power;
728 struct sched_domain {
729 /* These fields must be setup */
730 struct sched_domain *parent; /* top domain must be null terminated */
731 struct sched_domain *child; /* bottom domain must be null terminated */
732 struct sched_group *groups; /* the balancing groups of the domain */
733 cpumask_t span; /* span of all CPUs in this domain */
734 unsigned long min_interval; /* Minimum balance interval ms */
735 unsigned long max_interval; /* Maximum balance interval ms */
736 unsigned int busy_factor; /* less balancing by factor if busy */
737 unsigned int imbalance_pct; /* No balance until over watermark */
738 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
739 unsigned int busy_idx;
740 unsigned int idle_idx;
741 unsigned int newidle_idx;
742 unsigned int wake_idx;
743 unsigned int forkexec_idx;
744 int flags; /* See SD_* */
746 /* Runtime fields. */
747 unsigned long last_balance; /* init to jiffies. units in jiffies */
748 unsigned int balance_interval; /* initialise to 1. units in ms. */
749 unsigned int nr_balance_failed; /* initialise to 0 */
751 #ifdef CONFIG_SCHEDSTATS
752 /* load_balance() stats */
753 unsigned long lb_cnt[CPU_MAX_IDLE_TYPES];
754 unsigned long lb_failed[CPU_MAX_IDLE_TYPES];
755 unsigned long lb_balanced[CPU_MAX_IDLE_TYPES];
756 unsigned long lb_imbalance[CPU_MAX_IDLE_TYPES];
757 unsigned long lb_gained[CPU_MAX_IDLE_TYPES];
758 unsigned long lb_hot_gained[CPU_MAX_IDLE_TYPES];
759 unsigned long lb_nobusyg[CPU_MAX_IDLE_TYPES];
760 unsigned long lb_nobusyq[CPU_MAX_IDLE_TYPES];
762 /* Active load balancing */
763 unsigned long alb_cnt;
764 unsigned long alb_failed;
765 unsigned long alb_pushed;
767 /* SD_BALANCE_EXEC stats */
768 unsigned long sbe_cnt;
769 unsigned long sbe_balanced;
770 unsigned long sbe_pushed;
772 /* SD_BALANCE_FORK stats */
773 unsigned long sbf_cnt;
774 unsigned long sbf_balanced;
775 unsigned long sbf_pushed;
777 /* try_to_wake_up() stats */
778 unsigned long ttwu_wake_remote;
779 unsigned long ttwu_move_affine;
780 unsigned long ttwu_move_balance;
784 extern int partition_sched_domains(cpumask_t *partition1,
785 cpumask_t *partition2);
787 #endif /* CONFIG_SMP */
790 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
791 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
792 * task of nice 0 or enough lower priority tasks to bring up the
795 static inline int above_background_load(void)
799 for_each_online_cpu(cpu) {
800 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
806 struct io_context; /* See blkdev.h */
809 #define NGROUPS_SMALL 32
810 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
814 gid_t small_block[NGROUPS_SMALL];
820 * get_group_info() must be called with the owning task locked (via task_lock())
821 * when task != current. The reason being that the vast majority of callers are
822 * looking at current->group_info, which can not be changed except by the
823 * current task. Changing current->group_info requires the task lock, too.
825 #define get_group_info(group_info) do { \
826 atomic_inc(&(group_info)->usage); \
829 #define put_group_info(group_info) do { \
830 if (atomic_dec_and_test(&(group_info)->usage)) \
831 groups_free(group_info); \
834 extern struct group_info *groups_alloc(int gidsetsize);
835 extern void groups_free(struct group_info *group_info);
836 extern int set_current_groups(struct group_info *group_info);
837 extern int groups_search(struct group_info *group_info, gid_t grp);
838 /* access the groups "array" with this macro */
839 #define GROUP_AT(gi, i) \
840 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
842 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
843 extern void prefetch_stack(struct task_struct *t);
845 static inline void prefetch_stack(struct task_struct *t) { }
848 struct audit_context; /* See audit.c */
850 struct pipe_inode_info;
851 struct uts_namespace;
857 struct sched_class *next;
859 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
860 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
861 void (*yield_task) (struct rq *rq, struct task_struct *p);
863 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
865 struct task_struct * (*pick_next_task) (struct rq *rq);
866 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
868 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
870 unsigned long max_nr_move, unsigned long max_load_move,
871 struct sched_domain *sd, enum cpu_idle_type idle,
872 int *all_pinned, int *this_best_prio);
874 void (*set_curr_task) (struct rq *rq);
875 void (*task_tick) (struct rq *rq, struct task_struct *p);
876 void (*task_new) (struct rq *rq, struct task_struct *p);
880 unsigned long weight, inv_weight;
884 * CFS stats for a schedulable entity (task, task-group etc)
886 * Current field usage histogram:
893 struct sched_entity {
895 struct load_weight load; /* for load-balancing */
896 struct rb_node run_node;
900 u64 sum_exec_runtime;
902 u64 prev_sum_exec_runtime;
904 #ifdef CONFIG_SCHEDSTATS
910 s64 sum_sleep_runtime;
918 #ifdef CONFIG_FAIR_GROUP_SCHED
919 struct sched_entity *parent;
920 /* rq on which this entity is (to be) queued: */
921 struct cfs_rq *cfs_rq;
922 /* rq "owned" by this entity/group: */
928 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
931 unsigned int flags; /* per process flags, defined below */
934 int lock_depth; /* BKL lock depth */
937 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
942 int prio, static_prio, normal_prio;
943 struct list_head run_list;
944 struct sched_class *sched_class;
945 struct sched_entity se;
947 #ifdef CONFIG_PREEMPT_NOTIFIERS
948 /* list of struct preempt_notifier: */
949 struct hlist_head preempt_notifiers;
952 unsigned short ioprio;
953 #ifdef CONFIG_BLK_DEV_IO_TRACE
954 unsigned int btrace_seq;
958 cpumask_t cpus_allowed;
959 unsigned int time_slice;
961 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
962 struct sched_info sched_info;
965 struct list_head tasks;
967 * ptrace_list/ptrace_children forms the list of my children
968 * that were stolen by a ptracer.
970 struct list_head ptrace_children;
971 struct list_head ptrace_list;
973 struct mm_struct *mm, *active_mm;
976 struct linux_binfmt *binfmt;
978 int exit_code, exit_signal;
979 int pdeath_signal; /* The signal sent when the parent dies */
981 unsigned int personality;
986 #ifdef CONFIG_CC_STACKPROTECTOR
987 /* Canary value for the -fstack-protector gcc feature */
988 unsigned long stack_canary;
991 * pointers to (original) parent process, youngest child, younger sibling,
992 * older sibling, respectively. (p->father can be replaced with
995 struct task_struct *real_parent; /* real parent process (when being debugged) */
996 struct task_struct *parent; /* parent process */
998 * children/sibling forms the list of my children plus the
999 * tasks I'm ptracing.
1001 struct list_head children; /* list of my children */
1002 struct list_head sibling; /* linkage in my parent's children list */
1003 struct task_struct *group_leader; /* threadgroup leader */
1005 /* PID/PID hash table linkage. */
1006 struct pid_link pids[PIDTYPE_MAX];
1007 struct list_head thread_group;
1009 struct completion *vfork_done; /* for vfork() */
1010 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1011 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1013 unsigned int rt_priority;
1014 cputime_t utime, stime;
1015 unsigned long nvcsw, nivcsw; /* context switch counts */
1016 struct timespec start_time; /* monotonic time */
1017 struct timespec real_start_time; /* boot based time */
1018 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1019 unsigned long min_flt, maj_flt;
1021 cputime_t it_prof_expires, it_virt_expires;
1022 unsigned long long it_sched_expires;
1023 struct list_head cpu_timers[3];
1025 /* process credentials */
1026 uid_t uid,euid,suid,fsuid;
1027 gid_t gid,egid,sgid,fsgid;
1028 struct group_info *group_info;
1029 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
1030 unsigned keep_capabilities:1;
1031 struct user_struct *user;
1033 struct key *request_key_auth; /* assumed request_key authority */
1034 struct key *thread_keyring; /* keyring private to this thread */
1035 unsigned char jit_keyring; /* default keyring to attach requested keys to */
1038 * fpu_counter contains the number of consecutive context switches
1039 * that the FPU is used. If this is over a threshold, the lazy fpu
1040 * saving becomes unlazy to save the trap. This is an unsigned char
1041 * so that after 256 times the counter wraps and the behavior turns
1042 * lazy again; this to deal with bursty apps that only use FPU for
1045 unsigned char fpu_counter;
1046 int oomkilladj; /* OOM kill score adjustment (bit shift). */
1047 char comm[TASK_COMM_LEN]; /* executable name excluding path
1048 - access with [gs]et_task_comm (which lock
1049 it with task_lock())
1050 - initialized normally by flush_old_exec */
1051 /* file system info */
1052 int link_count, total_link_count;
1053 #ifdef CONFIG_SYSVIPC
1055 struct sysv_sem sysvsem;
1057 /* CPU-specific state of this task */
1058 struct thread_struct thread;
1059 /* filesystem information */
1060 struct fs_struct *fs;
1061 /* open file information */
1062 struct files_struct *files;
1064 struct nsproxy *nsproxy;
1065 /* signal handlers */
1066 struct signal_struct *signal;
1067 struct sighand_struct *sighand;
1069 sigset_t blocked, real_blocked;
1070 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */
1071 struct sigpending pending;
1073 unsigned long sas_ss_sp;
1075 int (*notifier)(void *priv);
1076 void *notifier_data;
1077 sigset_t *notifier_mask;
1080 struct audit_context *audit_context;
1083 /* Thread group tracking */
1086 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1087 spinlock_t alloc_lock;
1089 /* Protection of the PI data structures: */
1092 #ifdef CONFIG_RT_MUTEXES
1093 /* PI waiters blocked on a rt_mutex held by this task */
1094 struct plist_head pi_waiters;
1095 /* Deadlock detection and priority inheritance handling */
1096 struct rt_mutex_waiter *pi_blocked_on;
1099 #ifdef CONFIG_DEBUG_MUTEXES
1100 /* mutex deadlock detection */
1101 struct mutex_waiter *blocked_on;
1103 #ifdef CONFIG_TRACE_IRQFLAGS
1104 unsigned int irq_events;
1105 int hardirqs_enabled;
1106 unsigned long hardirq_enable_ip;
1107 unsigned int hardirq_enable_event;
1108 unsigned long hardirq_disable_ip;
1109 unsigned int hardirq_disable_event;
1110 int softirqs_enabled;
1111 unsigned long softirq_disable_ip;
1112 unsigned int softirq_disable_event;
1113 unsigned long softirq_enable_ip;
1114 unsigned int softirq_enable_event;
1115 int hardirq_context;
1116 int softirq_context;
1118 #ifdef CONFIG_LOCKDEP
1119 # define MAX_LOCK_DEPTH 30UL
1122 struct held_lock held_locks[MAX_LOCK_DEPTH];
1123 unsigned int lockdep_recursion;
1126 /* journalling filesystem info */
1129 /* stacked block device info */
1130 struct bio *bio_list, **bio_tail;
1133 struct reclaim_state *reclaim_state;
1135 struct backing_dev_info *backing_dev_info;
1137 struct io_context *io_context;
1139 unsigned long ptrace_message;
1140 siginfo_t *last_siginfo; /* For ptrace use. */
1142 * current io wait handle: wait queue entry to use for io waits
1143 * If this thread is processing aio, this points at the waitqueue
1144 * inside the currently handled kiocb. It may be NULL (i.e. default
1145 * to a stack based synchronous wait) if its doing sync IO.
1147 wait_queue_t *io_wait;
1148 #ifdef CONFIG_TASK_XACCT
1149 /* i/o counters(bytes read/written, #syscalls */
1150 u64 rchar, wchar, syscr, syscw;
1152 struct task_io_accounting ioac;
1153 #if defined(CONFIG_TASK_XACCT)
1154 u64 acct_rss_mem1; /* accumulated rss usage */
1155 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1156 cputime_t acct_stimexpd;/* stime since last update */
1159 struct mempolicy *mempolicy;
1162 #ifdef CONFIG_CPUSETS
1163 struct cpuset *cpuset;
1164 nodemask_t mems_allowed;
1165 int cpuset_mems_generation;
1166 int cpuset_mem_spread_rotor;
1168 struct robust_list_head __user *robust_list;
1169 #ifdef CONFIG_COMPAT
1170 struct compat_robust_list_head __user *compat_robust_list;
1172 struct list_head pi_state_list;
1173 struct futex_pi_state *pi_state_cache;
1175 atomic_t fs_excl; /* holding fs exclusive resources */
1176 struct rcu_head rcu;
1179 * cache last used pipe for splice
1181 struct pipe_inode_info *splice_pipe;
1182 #ifdef CONFIG_TASK_DELAY_ACCT
1183 struct task_delay_info *delays;
1185 #ifdef CONFIG_FAULT_INJECTION
1191 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1192 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1193 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1194 * values are inverted: lower p->prio value means higher priority.
1196 * The MAX_USER_RT_PRIO value allows the actual maximum
1197 * RT priority to be separate from the value exported to
1198 * user-space. This allows kernel threads to set their
1199 * priority to a value higher than any user task. Note:
1200 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1203 #define MAX_USER_RT_PRIO 100
1204 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1206 #define MAX_PRIO (MAX_RT_PRIO + 40)
1207 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1209 static inline int rt_prio(int prio)
1211 if (unlikely(prio < MAX_RT_PRIO))
1216 static inline int rt_task(struct task_struct *p)
1218 return rt_prio(p->prio);
1221 static inline pid_t process_group(struct task_struct *tsk)
1223 return tsk->signal->pgrp;
1226 static inline pid_t signal_session(struct signal_struct *sig)
1228 return sig->__session;
1231 static inline pid_t process_session(struct task_struct *tsk)
1233 return signal_session(tsk->signal);
1236 static inline void set_signal_session(struct signal_struct *sig, pid_t session)
1238 sig->__session = session;
1241 static inline struct pid *task_pid(struct task_struct *task)
1243 return task->pids[PIDTYPE_PID].pid;
1246 static inline struct pid *task_tgid(struct task_struct *task)
1248 return task->group_leader->pids[PIDTYPE_PID].pid;
1251 static inline struct pid *task_pgrp(struct task_struct *task)
1253 return task->group_leader->pids[PIDTYPE_PGID].pid;
1256 static inline struct pid *task_session(struct task_struct *task)
1258 return task->group_leader->pids[PIDTYPE_SID].pid;
1262 * pid_alive - check that a task structure is not stale
1263 * @p: Task structure to be checked.
1265 * Test if a process is not yet dead (at most zombie state)
1266 * If pid_alive fails, then pointers within the task structure
1267 * can be stale and must not be dereferenced.
1269 static inline int pid_alive(struct task_struct *p)
1271 return p->pids[PIDTYPE_PID].pid != NULL;
1275 * is_init - check if a task structure is init
1276 * @tsk: Task structure to be checked.
1278 * Check if a task structure is the first user space task the kernel created.
1280 static inline int is_init(struct task_struct *tsk)
1282 return tsk->pid == 1;
1285 extern struct pid *cad_pid;
1287 extern void free_task(struct task_struct *tsk);
1288 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1290 extern void __put_task_struct(struct task_struct *t);
1292 static inline void put_task_struct(struct task_struct *t)
1294 if (atomic_dec_and_test(&t->usage))
1295 __put_task_struct(t);
1301 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1302 /* Not implemented yet, only for 486*/
1303 #define PF_STARTING 0x00000002 /* being created */
1304 #define PF_EXITING 0x00000004 /* getting shut down */
1305 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1306 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1307 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1308 #define PF_DUMPCORE 0x00000200 /* dumped core */
1309 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1310 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1311 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1312 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1313 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1314 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1315 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1316 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1317 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1318 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1319 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */
1320 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1321 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1322 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1323 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1324 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1325 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1326 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1329 * Only the _current_ task can read/write to tsk->flags, but other
1330 * tasks can access tsk->flags in readonly mode for example
1331 * with tsk_used_math (like during threaded core dumping).
1332 * There is however an exception to this rule during ptrace
1333 * or during fork: the ptracer task is allowed to write to the
1334 * child->flags of its traced child (same goes for fork, the parent
1335 * can write to the child->flags), because we're guaranteed the
1336 * child is not running and in turn not changing child->flags
1337 * at the same time the parent does it.
1339 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1340 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1341 #define clear_used_math() clear_stopped_child_used_math(current)
1342 #define set_used_math() set_stopped_child_used_math(current)
1343 #define conditional_stopped_child_used_math(condition, child) \
1344 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1345 #define conditional_used_math(condition) \
1346 conditional_stopped_child_used_math(condition, current)
1347 #define copy_to_stopped_child_used_math(child) \
1348 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1349 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1350 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1351 #define used_math() tsk_used_math(current)
1354 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1356 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1358 if (!cpu_isset(0, new_mask))
1364 extern unsigned long long sched_clock(void);
1367 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1368 * clock constructed from sched_clock():
1370 extern unsigned long long cpu_clock(int cpu);
1372 extern unsigned long long
1373 task_sched_runtime(struct task_struct *task);
1375 /* sched_exec is called by processes performing an exec */
1377 extern void sched_exec(void);
1379 #define sched_exec() {}
1382 extern void sched_clock_idle_sleep_event(void);
1383 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1385 #ifdef CONFIG_HOTPLUG_CPU
1386 extern void idle_task_exit(void);
1388 static inline void idle_task_exit(void) {}
1391 extern void sched_idle_next(void);
1393 #ifdef CONFIG_SCHED_DEBUG
1394 extern unsigned int sysctl_sched_latency;
1395 extern unsigned int sysctl_sched_min_granularity;
1396 extern unsigned int sysctl_sched_wakeup_granularity;
1397 extern unsigned int sysctl_sched_batch_wakeup_granularity;
1398 extern unsigned int sysctl_sched_stat_granularity;
1399 extern unsigned int sysctl_sched_runtime_limit;
1400 extern unsigned int sysctl_sched_child_runs_first;
1401 extern unsigned int sysctl_sched_features;
1404 extern unsigned int sysctl_sched_compat_yield;
1406 #ifdef CONFIG_RT_MUTEXES
1407 extern int rt_mutex_getprio(struct task_struct *p);
1408 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1409 extern void rt_mutex_adjust_pi(struct task_struct *p);
1411 static inline int rt_mutex_getprio(struct task_struct *p)
1413 return p->normal_prio;
1415 # define rt_mutex_adjust_pi(p) do { } while (0)
1418 extern void set_user_nice(struct task_struct *p, long nice);
1419 extern int task_prio(const struct task_struct *p);
1420 extern int task_nice(const struct task_struct *p);
1421 extern int can_nice(const struct task_struct *p, const int nice);
1422 extern int task_curr(const struct task_struct *p);
1423 extern int idle_cpu(int cpu);
1424 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1425 extern struct task_struct *idle_task(int cpu);
1426 extern struct task_struct *curr_task(int cpu);
1427 extern void set_curr_task(int cpu, struct task_struct *p);
1432 * The default (Linux) execution domain.
1434 extern struct exec_domain default_exec_domain;
1436 union thread_union {
1437 struct thread_info thread_info;
1438 unsigned long stack[THREAD_SIZE/sizeof(long)];
1441 #ifndef __HAVE_ARCH_KSTACK_END
1442 static inline int kstack_end(void *addr)
1444 /* Reliable end of stack detection:
1445 * Some APM bios versions misalign the stack
1447 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1451 extern union thread_union init_thread_union;
1452 extern struct task_struct init_task;
1454 extern struct mm_struct init_mm;
1456 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr)
1457 extern struct task_struct *find_task_by_pid_type(int type, int pid);
1458 extern void __set_special_pids(pid_t session, pid_t pgrp);
1460 /* per-UID process charging. */
1461 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1462 static inline struct user_struct *get_uid(struct user_struct *u)
1464 atomic_inc(&u->__count);
1467 extern void free_uid(struct user_struct *);
1468 extern void switch_uid(struct user_struct *);
1469 extern void release_uids(struct user_namespace *ns);
1471 #include <asm/current.h>
1473 extern void do_timer(unsigned long ticks);
1475 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1476 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1477 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1478 unsigned long clone_flags));
1480 extern void kick_process(struct task_struct *tsk);
1482 static inline void kick_process(struct task_struct *tsk) { }
1484 extern void sched_fork(struct task_struct *p, int clone_flags);
1485 extern void sched_dead(struct task_struct *p);
1487 extern int in_group_p(gid_t);
1488 extern int in_egroup_p(gid_t);
1490 extern void proc_caches_init(void);
1491 extern void flush_signals(struct task_struct *);
1492 extern void ignore_signals(struct task_struct *);
1493 extern void flush_signal_handlers(struct task_struct *, int force_default);
1494 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1496 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1498 unsigned long flags;
1501 spin_lock_irqsave(&tsk->sighand->siglock, flags);
1502 ret = dequeue_signal(tsk, mask, info);
1503 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1508 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1510 extern void unblock_all_signals(void);
1511 extern void release_task(struct task_struct * p);
1512 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1513 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1514 extern int force_sigsegv(int, struct task_struct *);
1515 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1516 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1517 extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1518 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1519 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1520 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1521 extern int kill_pid(struct pid *pid, int sig, int priv);
1522 extern int kill_proc_info(int, struct siginfo *, pid_t);
1523 extern void do_notify_parent(struct task_struct *, int);
1524 extern void force_sig(int, struct task_struct *);
1525 extern void force_sig_specific(int, struct task_struct *);
1526 extern int send_sig(int, struct task_struct *, int);
1527 extern void zap_other_threads(struct task_struct *p);
1528 extern int kill_proc(pid_t, int, int);
1529 extern struct sigqueue *sigqueue_alloc(void);
1530 extern void sigqueue_free(struct sigqueue *);
1531 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
1532 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
1533 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1534 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1536 static inline int kill_cad_pid(int sig, int priv)
1538 return kill_pid(cad_pid, sig, priv);
1541 /* These can be the second arg to send_sig_info/send_group_sig_info. */
1542 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1543 #define SEND_SIG_PRIV ((struct siginfo *) 1)
1544 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1546 static inline int is_si_special(const struct siginfo *info)
1548 return info <= SEND_SIG_FORCED;
1551 /* True if we are on the alternate signal stack. */
1553 static inline int on_sig_stack(unsigned long sp)
1555 return (sp - current->sas_ss_sp < current->sas_ss_size);
1558 static inline int sas_ss_flags(unsigned long sp)
1560 return (current->sas_ss_size == 0 ? SS_DISABLE
1561 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1565 * Routines for handling mm_structs
1567 extern struct mm_struct * mm_alloc(void);
1569 /* mmdrop drops the mm and the page tables */
1570 extern void FASTCALL(__mmdrop(struct mm_struct *));
1571 static inline void mmdrop(struct mm_struct * mm)
1573 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1577 /* mmput gets rid of the mappings and all user-space */
1578 extern void mmput(struct mm_struct *);
1579 /* Grab a reference to a task's mm, if it is not already going away */
1580 extern struct mm_struct *get_task_mm(struct task_struct *task);
1581 /* Remove the current tasks stale references to the old mm_struct */
1582 extern void mm_release(struct task_struct *, struct mm_struct *);
1584 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1585 extern void flush_thread(void);
1586 extern void exit_thread(void);
1588 extern void exit_files(struct task_struct *);
1589 extern void __cleanup_signal(struct signal_struct *);
1590 extern void __cleanup_sighand(struct sighand_struct *);
1591 extern void exit_itimers(struct signal_struct *);
1593 extern NORET_TYPE void do_group_exit(int);
1595 extern void daemonize(const char *, ...);
1596 extern int allow_signal(int);
1597 extern int disallow_signal(int);
1599 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1600 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1601 struct task_struct *fork_idle(int);
1603 extern void set_task_comm(struct task_struct *tsk, char *from);
1604 extern void get_task_comm(char *to, struct task_struct *tsk);
1607 extern void wait_task_inactive(struct task_struct * p);
1609 #define wait_task_inactive(p) do { } while (0)
1612 #define remove_parent(p) list_del_init(&(p)->sibling)
1613 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children)
1615 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1617 #define for_each_process(p) \
1618 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1621 * Careful: do_each_thread/while_each_thread is a double loop so
1622 * 'break' will not work as expected - use goto instead.
1624 #define do_each_thread(g, t) \
1625 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1627 #define while_each_thread(g, t) \
1628 while ((t = next_thread(t)) != g)
1630 /* de_thread depends on thread_group_leader not being a pid based check */
1631 #define thread_group_leader(p) (p == p->group_leader)
1633 /* Do to the insanities of de_thread it is possible for a process
1634 * to have the pid of the thread group leader without actually being
1635 * the thread group leader. For iteration through the pids in proc
1636 * all we care about is that we have a task with the appropriate
1637 * pid, we don't actually care if we have the right task.
1639 static inline int has_group_leader_pid(struct task_struct *p)
1641 return p->pid == p->tgid;
1644 static inline struct task_struct *next_thread(const struct task_struct *p)
1646 return list_entry(rcu_dereference(p->thread_group.next),
1647 struct task_struct, thread_group);
1650 static inline int thread_group_empty(struct task_struct *p)
1652 return list_empty(&p->thread_group);
1655 #define delay_group_leader(p) \
1656 (thread_group_leader(p) && !thread_group_empty(p))
1659 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1660 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1661 * pins the final release of task.io_context. Also protects ->cpuset.
1663 * Nests both inside and outside of read_lock(&tasklist_lock).
1664 * It must not be nested with write_lock_irq(&tasklist_lock),
1665 * neither inside nor outside.
1667 static inline void task_lock(struct task_struct *p)
1669 spin_lock(&p->alloc_lock);
1672 static inline void task_unlock(struct task_struct *p)
1674 spin_unlock(&p->alloc_lock);
1677 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1678 unsigned long *flags);
1680 static inline void unlock_task_sighand(struct task_struct *tsk,
1681 unsigned long *flags)
1683 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1686 #ifndef __HAVE_THREAD_FUNCTIONS
1688 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1689 #define task_stack_page(task) ((task)->stack)
1691 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1693 *task_thread_info(p) = *task_thread_info(org);
1694 task_thread_info(p)->task = p;
1697 static inline unsigned long *end_of_stack(struct task_struct *p)
1699 return (unsigned long *)(task_thread_info(p) + 1);
1704 /* set thread flags in other task's structures
1705 * - see asm/thread_info.h for TIF_xxxx flags available
1707 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1709 set_ti_thread_flag(task_thread_info(tsk), flag);
1712 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1714 clear_ti_thread_flag(task_thread_info(tsk), flag);
1717 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1719 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1722 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1724 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1727 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1729 return test_ti_thread_flag(task_thread_info(tsk), flag);
1732 static inline void set_tsk_need_resched(struct task_struct *tsk)
1734 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1737 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1739 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1742 static inline int signal_pending(struct task_struct *p)
1744 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1747 static inline int need_resched(void)
1749 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1753 * cond_resched() and cond_resched_lock(): latency reduction via
1754 * explicit rescheduling in places that are safe. The return
1755 * value indicates whether a reschedule was done in fact.
1756 * cond_resched_lock() will drop the spinlock before scheduling,
1757 * cond_resched_softirq() will enable bhs before scheduling.
1759 extern int cond_resched(void);
1760 extern int cond_resched_lock(spinlock_t * lock);
1761 extern int cond_resched_softirq(void);
1764 * Does a critical section need to be broken due to another
1767 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1768 # define need_lockbreak(lock) ((lock)->break_lock)
1770 # define need_lockbreak(lock) 0
1774 * Does a critical section need to be broken due to another
1775 * task waiting or preemption being signalled:
1777 static inline int lock_need_resched(spinlock_t *lock)
1779 if (need_lockbreak(lock) || need_resched())
1785 * Reevaluate whether the task has signals pending delivery.
1786 * Wake the task if so.
1787 * This is required every time the blocked sigset_t changes.
1788 * callers must hold sighand->siglock.
1790 extern void recalc_sigpending_and_wake(struct task_struct *t);
1791 extern void recalc_sigpending(void);
1793 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1796 * Wrappers for p->thread_info->cpu access. No-op on UP.
1800 static inline unsigned int task_cpu(const struct task_struct *p)
1802 return task_thread_info(p)->cpu;
1805 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1809 static inline unsigned int task_cpu(const struct task_struct *p)
1814 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1818 #endif /* CONFIG_SMP */
1820 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1821 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1823 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1825 mm->mmap_base = TASK_UNMAPPED_BASE;
1826 mm->get_unmapped_area = arch_get_unmapped_area;
1827 mm->unmap_area = arch_unmap_area;
1831 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1832 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1834 extern int sched_mc_power_savings, sched_smt_power_savings;
1836 extern void normalize_rt_tasks(void);
1838 #ifdef CONFIG_TASK_XACCT
1839 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1844 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1849 static inline void inc_syscr(struct task_struct *tsk)
1854 static inline void inc_syscw(struct task_struct *tsk)
1859 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1863 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1867 static inline void inc_syscr(struct task_struct *tsk)
1871 static inline void inc_syscw(struct task_struct *tsk)
1876 #endif /* __KERNEL__ */