4 #include <asm/param.h> /* for HZ */
6 #include <linux/config.h>
7 #include <linux/capability.h>
8 #include <linux/threads.h>
9 #include <linux/kernel.h>
10 #include <linux/types.h>
11 #include <linux/timex.h>
12 #include <linux/jiffies.h>
13 #include <linux/rbtree.h>
14 #include <linux/thread_info.h>
15 #include <linux/cpumask.h>
17 #include <asm/system.h>
18 #include <asm/semaphore.h>
20 #include <asm/ptrace.h>
23 #include <linux/smp.h>
24 #include <linux/sem.h>
25 #include <linux/signal.h>
26 #include <linux/securebits.h>
27 #include <linux/fs_struct.h>
28 #include <linux/compiler.h>
29 #include <linux/completion.h>
30 #include <linux/pid.h>
31 #include <linux/percpu.h>
34 extern int exec_shield;
35 extern int exec_shield_randomize;
36 extern int print_fatal_signals;
41 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
42 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
43 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
44 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
45 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
46 #define CLONE_IDLETASK 0x00001000 /* set if new pid should be 0 (kernel only)*/
47 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
48 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
49 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
50 #define CLONE_THREAD 0x00010000 /* Same thread group? */
51 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
52 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
53 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
54 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
55 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
56 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
57 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
58 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
59 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
62 * List of flags we want to share for kernel threads,
63 * if only because they are not used by them anyway.
65 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
68 * These are the constant used to fake the fixed-point load-average
69 * counting. Some notes:
70 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
71 * a load-average precision of 10 bits integer + 11 bits fractional
72 * - if you want to count load-averages more often, you need more
73 * precision, or rounding will get you. With 2-second counting freq,
74 * the EXP_n values would be 1981, 2034 and 2043 if still using only
77 extern unsigned long avenrun[]; /* Load averages */
79 #define FSHIFT 11 /* nr of bits of precision */
80 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
81 #define LOAD_FREQ (5*HZ) /* 5 sec intervals */
82 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
83 #define EXP_5 2014 /* 1/exp(5sec/5min) */
84 #define EXP_15 2037 /* 1/exp(5sec/15min) */
86 #define CALC_LOAD(load,exp,n) \
88 load += n*(FIXED_1-exp); \
91 #define CT_TO_SECS(x) ((x) / HZ)
92 #define CT_TO_USECS(x) (((x) % HZ) * 1000000/HZ)
94 extern int nr_threads;
96 DECLARE_PER_CPU(unsigned long, process_counts);
97 DECLARE_PER_CPU(struct runqueue, runqueues);
98 extern int nr_processes(void);
99 extern unsigned long nr_running(void);
100 extern unsigned long nr_uninterruptible(void);
101 extern unsigned long nr_iowait(void);
103 #include <linux/time.h>
104 #include <linux/param.h>
105 #include <linux/resource.h>
106 #include <linux/timer.h>
108 #include <asm/processor.h>
109 #include <linux/vserver/context.h>
111 #define TASK_RUNNING 0
112 #define TASK_INTERRUPTIBLE 1
113 #define TASK_UNINTERRUPTIBLE 2
114 #define TASK_STOPPED 4
115 #define TASK_ZOMBIE 8
117 #define TASK_ONHOLD 32
119 #define __set_task_state(tsk, state_value) \
120 do { (tsk)->state = (state_value); } while (0)
121 #define set_task_state(tsk, state_value) \
122 set_mb((tsk)->state, (state_value))
124 #define __set_current_state(state_value) \
125 do { current->state = (state_value); } while (0)
126 #define set_current_state(state_value) \
127 set_mb(current->state, (state_value))
130 * Scheduling policies
132 #define SCHED_NORMAL 0
142 #include <linux/taskdelays.h>
143 #include <linux/spinlock.h>
146 * This serializes "schedule()" and also protects
147 * the run-queue from deletions/modifications (but
148 * _adding_ to the beginning of the run-queue has
151 extern rwlock_t tasklist_lock;
152 extern spinlock_t mmlist_lock;
154 typedef struct task_struct task_t;
156 extern void sched_init(void);
157 extern void sched_init_smp(void);
158 extern void init_idle(task_t *idle, int cpu);
160 extern cpumask_t nohz_cpu_mask;
162 extern void show_state(void);
163 extern void show_regs(struct pt_regs *);
166 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
167 * task), SP is the stack pointer of the first frame that should be shown in the back
168 * trace (or NULL if the entire call-chain of the task should be shown).
170 extern void show_stack(struct task_struct *task, unsigned long *sp);
172 void io_schedule(void);
173 long io_schedule_timeout(long timeout);
175 extern void cpu_init (void);
176 extern void trap_init(void);
177 extern void update_process_times(int user);
178 extern void scheduler_tick(int user_tick, int system);
179 extern unsigned long cache_decay_ticks;
181 /* Attach to any functions which should be ignored in wchan output. */
182 #define __sched __attribute__((__section__(".sched.text")))
183 /* Is this address in the __sched functions? */
184 extern int in_sched_functions(unsigned long addr);
186 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
187 extern signed long FASTCALL(schedule_timeout(signed long timeout));
188 asmlinkage void schedule(void);
192 /* Maximum number of active map areas.. This is a random (large) number */
193 #define DEFAULT_MAX_MAP_COUNT 65536
195 extern int sysctl_max_map_count;
197 #include <linux/aio.h>
200 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
201 unsigned long, unsigned long);
204 arch_get_unmapped_exec_area(struct file *, unsigned long, unsigned long,
205 unsigned long, unsigned long);
207 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
208 unsigned long len, unsigned long pgoff,
209 unsigned long flags);
210 extern void arch_unmap_area(struct vm_area_struct *area);
211 extern void arch_unmap_area_topdown(struct vm_area_struct *area);
215 struct vm_area_struct * mmap; /* list of VMAs */
216 struct rb_root mm_rb;
217 struct vm_area_struct * mmap_cache; /* last find_vma result */
218 unsigned long (*get_unmapped_area) (struct file *filp,
219 unsigned long addr, unsigned long len,
220 unsigned long pgoff, unsigned long flags);
221 unsigned long (*get_unmapped_exec_area) (struct file *filp,
222 unsigned long addr, unsigned long len,
223 unsigned long pgoff, unsigned long flags);
224 void (*unmap_area) (struct vm_area_struct *area);
225 unsigned long mmap_base; /* base of mmap area */
226 unsigned long free_area_cache; /* first hole */
228 atomic_t mm_users; /* How many users with user space? */
229 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */
230 int map_count; /* number of VMAs */
231 struct rw_semaphore mmap_sem;
232 spinlock_t page_table_lock; /* Protects task page tables and mm->rss */
234 struct list_head mmlist; /* List of all active mm's. These are globally strung
235 * together off init_mm.mmlist, and are protected
239 unsigned long start_code, end_code, start_data, end_data;
240 unsigned long start_brk, brk, start_stack;
241 unsigned long arg_start, arg_end, env_start, env_end;
242 unsigned long rss, total_vm, locked_vm;
243 unsigned long def_flags;
245 unsigned long saved_auxv[40]; /* for /proc/PID/auxv */
248 cpumask_t cpu_vm_mask;
250 /* Architecture-specific MM context */
251 mm_context_t context;
252 struct vx_info *mm_vx_info;
254 /* Token based thrashing protection. */
255 unsigned long swap_token_time;
258 /* coredumping support */
260 struct completion *core_startup_done, core_done;
263 rwlock_t ioctx_list_lock;
264 struct kioctx *ioctx_list;
266 struct kioctx default_kioctx;
267 #ifdef CONFIG_CKRM_RES_MEM
268 struct ckrm_mem_res *memclass;
269 struct list_head tasklist; /* list of all tasks sharing this address space */
270 spinlock_t peertask_lock; /* protect above tasklist */
274 extern int mmlist_nr;
276 struct sighand_struct {
278 struct k_sigaction action[_NSIG];
283 * NOTE! "signal_struct" does not have it's own
284 * locking, because a shared signal_struct always
285 * implies a shared sighand_struct, so locking
286 * sighand_struct is always a proper superset of
287 * the locking of signal_struct.
289 struct signal_struct {
292 /* current thread group signal load-balancing target: */
295 /* shared signal handling: */
296 struct sigpending shared_pending;
298 /* thread group exit support */
302 * - notify group_exit_task when ->count is equal to notify_count
303 * - everyone except group_exit_task is stopped during signal delivery
304 * of fatal signals, group_exit_task processes the signal.
306 struct task_struct *group_exit_task;
309 /* thread group stop support, overloads group_exit_code too */
310 int group_stop_count;
312 /* POSIX.1b Interval Timers */
313 struct list_head posix_timers;
315 /* job control IDs */
319 /* boolean value for session group leader */
322 struct tty_struct *tty; /* NULL if no tty */
326 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
327 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL tasks are
328 * in the range MAX_RT_PRIO..MAX_PRIO-1. Priority values
329 * are inverted: lower p->prio value means higher priority.
331 * The MAX_USER_RT_PRIO value allows the actual maximum
332 * RT priority to be separate from the value exported to
333 * user-space. This allows kernel threads to set their
334 * priority to a value higher than any user task. Note:
335 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
338 #define MAX_USER_RT_PRIO 100
339 #define MAX_RT_PRIO MAX_USER_RT_PRIO
341 #define MAX_PRIO (MAX_RT_PRIO + 40)
343 #define rt_task(p) ((p)->prio < MAX_RT_PRIO)
346 * Some day this will be a full-fledged user tracking system..
349 atomic_t __count; /* reference count */
350 atomic_t processes; /* How many processes does this user have? */
351 atomic_t files; /* How many open files does this user have? */
352 atomic_t sigpending; /* How many pending signals does this user have? */
353 /* protected by mq_lock */
354 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
355 unsigned long locked_shm; /* How many pages of mlocked shm ? */
357 /* Hash table maintenance information */
358 struct list_head uidhash_list;
363 extern struct user_struct *find_user(xid_t, uid_t);
365 extern struct user_struct root_user;
366 #define INIT_USER (&root_user)
368 typedef struct prio_array prio_array_t;
369 struct backing_dev_info;
370 struct reclaim_state;
372 /* POSIX.1b interval timer structure. */
374 struct list_head list; /* free/ allocate list */
376 clockid_t it_clock; /* which timer type */
377 timer_t it_id; /* timer id */
378 int it_overrun; /* overrun on pending signal */
379 int it_overrun_last; /* overrun on last delivered signal */
380 int it_requeue_pending; /* waiting to requeue this timer */
381 int it_sigev_notify; /* notify word of sigevent struct */
382 int it_sigev_signo; /* signo word of sigevent struct */
383 sigval_t it_sigev_value; /* value word of sigevent struct */
384 unsigned long it_incr; /* interval specified in jiffies */
385 struct task_struct *it_process; /* process to send signal to */
386 struct timer_list it_timer;
387 struct sigqueue *sigq; /* signal queue entry. */
388 struct list_head abs_timer_entry; /* clock abs_timer_list */
389 struct timespec wall_to_prev; /* wall_to_monotonic used when set */
393 struct io_context; /* See blkdev.h */
394 void exit_io_context(void);
396 #define NGROUPS_SMALL 32
397 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t)))
401 gid_t small_block[NGROUPS_SMALL];
407 * get_group_info() must be called with the owning task locked (via task_lock())
408 * when task != current. The reason being that the vast majority of callers are
409 * looking at current->group_info, which can not be changed except by the
410 * current task. Changing current->group_info requires the task lock, too.
412 #define get_group_info(group_info) do { \
413 atomic_inc(&(group_info)->usage); \
416 #define put_group_info(group_info) do { \
417 if (atomic_dec_and_test(&(group_info)->usage)) \
418 groups_free(group_info); \
421 struct group_info *groups_alloc(int gidsetsize);
422 void groups_free(struct group_info *group_info);
423 int set_current_groups(struct group_info *group_info);
424 /* access the groups "array" with this macro */
425 #define GROUP_AT(gi, i) \
426 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
429 struct audit_context; /* See audit.c */
433 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
434 struct thread_info *thread_info;
436 unsigned long flags; /* per process flags, defined below */
437 unsigned long ptrace;
439 int lock_depth; /* Lock depth */
441 int prio, static_prio;
442 struct list_head run_list;
445 unsigned long sleep_avg;
446 long interactive_credit;
447 unsigned long long timestamp;
450 unsigned long policy;
451 cpumask_t cpus_allowed;
452 unsigned int time_slice, first_time_slice;
454 struct list_head tasks;
456 * ptrace_list/ptrace_children forms the list of my children
457 * that were stolen by a ptracer.
459 struct list_head ptrace_children;
460 struct list_head ptrace_list;
462 struct mm_struct *mm, *active_mm;
465 struct linux_binfmt *binfmt;
466 int exit_code, exit_signal;
467 int pdeath_signal; /* The signal sent when the parent dies */
469 unsigned long personality;
474 * pointers to (original) parent process, youngest child, younger sibling,
475 * older sibling, respectively. (p->father can be replaced with
478 struct task_struct *real_parent; /* real parent process (when being debugged) */
479 struct task_struct *parent; /* parent process */
481 * children/sibling forms the list of my children plus the
482 * tasks I'm ptracing.
484 struct list_head children; /* list of my children */
485 struct list_head sibling; /* linkage in my parent's children list */
486 struct task_struct *group_leader; /* threadgroup leader */
488 /* PID/PID hash table linkage. */
489 struct pid_link pids[PIDTYPE_MAX];
491 wait_queue_head_t wait_chldexit; /* for wait4() */
492 struct completion *vfork_done; /* for vfork() */
493 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
494 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
496 unsigned long rt_priority;
497 unsigned long it_real_value, it_prof_value, it_virt_value;
498 unsigned long it_real_incr, it_prof_incr, it_virt_incr;
499 struct timer_list real_timer;
500 unsigned long utime, stime, cutime, cstime;
501 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; /* context switch counts */
503 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
504 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
505 /* process credentials */
506 uid_t uid,euid,suid,fsuid;
507 gid_t gid,egid,sgid,fsgid;
508 struct group_info *group_info;
509 kernel_cap_t cap_effective, cap_inheritable, cap_permitted;
510 int keep_capabilities:1;
511 struct user_struct *user;
513 struct rlimit rlim[RLIM_NLIMITS];
514 unsigned short used_math;
516 /* file system info */
517 int link_count, total_link_count;
519 struct sysv_sem sysvsem;
520 /* CPU-specific state of this task */
521 struct thread_struct thread;
522 /* filesystem information */
523 struct fs_struct *fs;
524 /* open file information */
525 struct files_struct *files;
527 struct namespace *namespace;
528 /* signal handlers */
529 struct signal_struct *signal;
530 struct sighand_struct *sighand;
532 sigset_t blocked, real_blocked;
533 struct sigpending pending;
535 unsigned long sas_ss_sp;
537 int (*notifier)(void *priv);
539 sigset_t *notifier_mask;
543 void (*tux_exit)(void);
547 struct audit_context *audit_context;
549 /* vserver context data */
551 struct vx_info *vx_info;
553 /* vserver network data */
555 struct nx_info *nx_info;
557 /* Thread group tracking */
560 /* Protection of (de-)allocation: mm, files, fs, tty */
561 spinlock_t alloc_lock;
562 /* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
563 spinlock_t proc_lock;
564 /* context-switch lock */
565 spinlock_t switch_lock;
567 /* journalling filesystem info */
571 struct reclaim_state *reclaim_state;
573 struct dentry *proc_dentry;
574 struct backing_dev_info *backing_dev_info;
576 struct io_context *io_context;
580 unsigned long ptrace_message;
581 siginfo_t *last_siginfo; /* For ptrace use. */
584 struct mempolicy *mempolicy;
585 short il_next; /* could be shared with used_math */
589 spinlock_t ckrm_tsklock;
591 #ifdef CONFIG_CKRM_TYPE_TASKCLASS
592 // .. Hubertus should change to CONFIG_CKRM_TYPE_TASKCLASS
593 struct ckrm_task_class *taskclass;
594 struct list_head taskclass_link;
595 #ifdef CONFIG_CKRM_CPU_SCHEDULE
596 struct ckrm_cpu_class *cpu_class;
598 #endif // CONFIG_CKRM_TYPE_TASKCLASS
599 #ifdef CONFIG_CKRM_RES_MEM
600 struct list_head mm_peers; // list of tasks using same mm_struct
601 #endif // CONFIG_CKRM_RES_MEM
602 #endif // CONFIG_CKRM
603 struct task_delay_info delays;
606 static inline pid_t process_group(struct task_struct *tsk)
608 return tsk->signal->pgrp;
611 extern void __put_task_struct(struct task_struct *tsk);
612 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
613 #define put_task_struct(tsk) \
614 do { if (atomic_dec_and_test(&(tsk)->usage)) __put_task_struct(tsk); } while(0)
619 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
620 /* Not implemented yet, only for 486*/
621 #define PF_STARTING 0x00000002 /* being created */
622 #define PF_EXITING 0x00000004 /* getting shut down */
623 #define PF_DEAD 0x00000008 /* Dead */
624 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
625 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
626 #define PF_DUMPCORE 0x00000200 /* dumped core */
627 #define PF_SIGNALED 0x00000400 /* killed by a signal */
628 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
629 #define PF_MEMDIE 0x00001000 /* Killed for out-of-memory */
630 #define PF_FLUSHER 0x00002000 /* responsible for disk writeback */
632 #define PF_FREEZE 0x00004000 /* this task should be frozen for suspend */
633 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
634 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
635 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
636 #define PF_KSWAPD 0x00040000 /* I am kswapd */
637 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
638 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
639 #define PF_SYNCWRITE 0x00200000 /* I am doing a sync write */
640 #define PF_RELOCEXEC 0x00400000 /* relocate shared libraries */
643 #define PF_MEMIO 0x00400000 /* I am potentially doing I/O for mem */
644 #define PF_IOWAIT 0x00800000 /* I am waiting on disk I/O */
647 #define SCHED_LOAD_SCALE 128UL /* increase resolution of load */
649 #define SD_BALANCE_NEWIDLE 1 /* Balance when about to become idle */
650 #define SD_BALANCE_EXEC 2 /* Balance on exec */
651 #define SD_BALANCE_CLONE 4 /* Balance on clone */
652 #define SD_WAKE_IDLE 8 /* Wake to idle CPU on task wakeup */
653 #define SD_WAKE_AFFINE 16 /* Wake task to waking CPU */
654 #define SD_WAKE_BALANCE 32 /* Perform balancing at task wakeup */
655 #define SD_SHARE_CPUPOWER 64 /* Domain members share cpu power */
658 struct sched_group *next; /* Must be a circular list */
662 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
663 * single CPU. This should be read only (except for setup). Although
664 * it will need to be written to at cpu hot(un)plug time, perhaps the
665 * cpucontrol semaphore will provide enough exclusion?
667 unsigned long cpu_power;
670 struct sched_domain {
671 /* These fields must be setup */
672 struct sched_domain *parent; /* top domain must be null terminated */
673 struct sched_group *groups; /* the balancing groups of the domain */
674 cpumask_t span; /* span of all CPUs in this domain */
675 unsigned long min_interval; /* Minimum balance interval ms */
676 unsigned long max_interval; /* Maximum balance interval ms */
677 unsigned int busy_factor; /* less balancing by factor if busy */
678 unsigned int imbalance_pct; /* No balance until over watermark */
679 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
680 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
681 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */
682 int flags; /* See SD_* */
684 /* Runtime fields. */
685 unsigned long last_balance; /* init to jiffies. units in jiffies */
686 unsigned int balance_interval; /* initialise to 1. units in ms. */
687 unsigned int nr_balance_failed; /* initialise to 0 */
690 /* Common values for SMT siblings */
691 #define SD_SIBLING_INIT (struct sched_domain) { \
692 .span = CPU_MASK_NONE, \
698 .imbalance_pct = 110, \
699 .cache_hot_time = 0, \
700 .cache_nice_tries = 0, \
701 .per_cpu_gain = 15, \
702 .flags = SD_BALANCE_NEWIDLE \
707 | SD_SHARE_CPUPOWER, \
708 .last_balance = jiffies, \
709 .balance_interval = 1, \
710 .nr_balance_failed = 0, \
713 /* Common values for CPUs */
714 #define SD_CPU_INIT (struct sched_domain) { \
715 .span = CPU_MASK_NONE, \
721 .imbalance_pct = 125, \
722 .cache_hot_time = (5*1000000/2), \
723 .cache_nice_tries = 1, \
724 .per_cpu_gain = 100, \
725 .flags = SD_BALANCE_NEWIDLE \
730 .last_balance = jiffies, \
731 .balance_interval = 1, \
732 .nr_balance_failed = 0, \
736 /* Common values for NUMA nodes */
737 #define SD_NODE_INIT (struct sched_domain) { \
738 .span = CPU_MASK_NONE, \
742 .max_interval = 32, \
744 .imbalance_pct = 125, \
745 .cache_hot_time = (10*1000000), \
746 .cache_nice_tries = 1, \
747 .per_cpu_gain = 100, \
748 .flags = SD_BALANCE_EXEC \
751 .last_balance = jiffies, \
752 .balance_interval = 1, \
753 .nr_balance_failed = 0, \
757 extern void cpu_attach_domain(struct sched_domain *sd, int cpu);
759 extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
761 static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
767 extern unsigned long long sched_clock(void);
770 extern void sched_balance_exec(void);
772 #define sched_balance_exec() {}
775 extern void sched_idle_next(void);
776 extern void set_user_nice(task_t *p, long nice);
777 extern int task_prio(const task_t *p);
778 extern int task_nice(const task_t *p);
779 extern int task_curr(const task_t *p);
780 extern int idle_cpu(int cpu);
785 * These are the runqueue data structures:
787 typedef struct runqueue runqueue_t;
789 #ifdef CONFIG_CKRM_CPU_SCHEDULE
790 #include <linux/ckrm_classqueue.h>
793 #ifdef CONFIG_CKRM_CPU_SCHEDULE
796 * if belong to different class, compare class priority
797 * otherwise compare task priority
799 #define TASK_PREEMPTS_CURR(p, rq) \
800 (((p)->cpu_class != (rq)->curr->cpu_class) && ((rq)->curr != (rq)->idle))? class_preempts_curr((p),(rq)->curr) : ((p)->prio < (rq)->curr->prio)
802 #define BITMAP_SIZE ((((MAX_PRIO+1+7)/8)+sizeof(long)-1)/sizeof(long))
804 unsigned int nr_active;
805 unsigned long bitmap[BITMAP_SIZE];
806 struct list_head queue[MAX_PRIO];
808 #define rq_active(p,rq) (rq->active)
809 #define rq_expired(p,rq) (rq->expired)
810 #define ckrm_rebalance_tick(j,this_cpu) do {} while (0)
811 #define TASK_PREEMPTS_CURR(p, rq) \
812 ((p)->prio < (rq)->curr->prio)
816 * This is the main, per-CPU runqueue data structure.
818 * Locking rule: those places that want to lock multiple runqueues
819 * (such as the load balancing or the thread migration code), lock
820 * acquire operations must be ordered by ascending &runqueue.
826 * nr_running and cpu_load should be in the same cacheline because
827 * remote CPUs use both these fields when doing load calculation.
829 unsigned long nr_running;
830 #if defined(CONFIG_SMP)
831 unsigned long cpu_load;
833 unsigned long long nr_switches, nr_preempt;
834 unsigned long expired_timestamp, nr_uninterruptible;
835 unsigned long long timestamp_last_tick;
837 struct mm_struct *prev_mm;
838 #ifdef CONFIG_CKRM_CPU_SCHEDULE
839 unsigned long ckrm_cpu_load;
840 struct classqueue_struct classqueue;
842 prio_array_t *active, *expired, arrays[2];
844 int best_expired_prio;
848 struct sched_domain *sd;
850 /* For active balancing */
854 task_t *migration_thread;
855 struct list_head migration_queue;
857 struct list_head hold_queue;
862 * The default (Linux) execution domain.
864 extern struct exec_domain default_exec_domain;
867 struct thread_info thread_info;
868 unsigned long stack[THREAD_SIZE/sizeof(long)];
871 #ifndef __HAVE_ARCH_KSTACK_END
872 static inline int kstack_end(void *addr)
874 /* Reliable end of stack detection:
875 * Some APM bios versions misalign the stack
877 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
881 extern union thread_union init_thread_union;
882 extern struct task_struct init_task;
884 extern struct mm_struct init_mm;
886 extern struct task_struct *find_task_by_pid(int pid);
887 extern void set_special_pids(pid_t session, pid_t pgrp);
888 extern void __set_special_pids(pid_t session, pid_t pgrp);
890 /* per-UID process charging. */
891 extern struct user_struct * alloc_uid(xid_t, uid_t);
892 static inline struct user_struct *get_uid(struct user_struct *u)
894 atomic_inc(&u->__count);
897 extern void free_uid(struct user_struct *);
898 extern void switch_uid(struct user_struct *);
900 #include <asm/current.h>
902 extern unsigned long itimer_ticks;
903 extern unsigned long itimer_next;
904 extern void do_timer(struct pt_regs *);
906 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
907 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
908 extern void FASTCALL(wake_up_forked_process(struct task_struct * tsk));
910 extern void kick_process(struct task_struct *tsk);
911 extern void FASTCALL(wake_up_forked_thread(struct task_struct * tsk));
913 static inline void kick_process(struct task_struct *tsk) { }
914 static inline void wake_up_forked_thread(struct task_struct * tsk)
916 wake_up_forked_process(tsk);
919 extern void FASTCALL(sched_fork(task_t * p));
920 extern void FASTCALL(sched_exit(task_t * p));
922 extern int in_group_p(gid_t);
923 extern int in_egroup_p(gid_t);
925 extern void proc_caches_init(void);
926 extern void flush_signals(struct task_struct *);
927 extern void flush_signal_handlers(struct task_struct *, int force_default);
928 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
930 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
935 spin_lock_irqsave(&tsk->sighand->siglock, flags);
936 ret = dequeue_signal(tsk, mask, info);
937 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
942 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
944 extern void unblock_all_signals(void);
945 extern void release_task(struct task_struct * p);
946 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
947 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
948 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
949 extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
950 extern int kill_pg_info(int, struct siginfo *, pid_t);
951 extern int kill_sl_info(int, struct siginfo *, pid_t);
952 extern int kill_proc_info(int, struct siginfo *, pid_t);
953 extern void notify_parent(struct task_struct *, int);
954 extern void do_notify_parent(struct task_struct *, int);
955 extern void force_sig(int, struct task_struct *);
956 extern void force_sig_specific(int, struct task_struct *);
957 extern int send_sig(int, struct task_struct *, int);
958 extern void zap_other_threads(struct task_struct *p);
959 extern int kill_pg(pid_t, int, int);
960 extern int kill_sl(pid_t, int, int);
961 extern int kill_proc(pid_t, int, int);
962 extern struct sigqueue *sigqueue_alloc(void);
963 extern void sigqueue_free(struct sigqueue *);
964 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *);
965 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *);
966 extern int do_sigaction(int, const struct k_sigaction *, struct k_sigaction *);
967 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
969 /* These can be the second arg to send_sig_info/send_group_sig_info. */
970 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
971 #define SEND_SIG_PRIV ((struct siginfo *) 1)
972 #define SEND_SIG_FORCED ((struct siginfo *) 2)
974 /* True if we are on the alternate signal stack. */
976 static inline int on_sig_stack(unsigned long sp)
978 return (sp - current->sas_ss_sp < current->sas_ss_size);
981 static inline int sas_ss_flags(unsigned long sp)
983 return (current->sas_ss_size == 0 ? SS_DISABLE
984 : on_sig_stack(sp) ? SS_ONSTACK : 0);
988 #ifdef CONFIG_SECURITY
989 /* code is in security.c */
990 extern int capable(int cap);
992 static inline int capable(int cap)
994 if (cap_raised(current->cap_effective, cap)) {
995 current->flags |= PF_SUPERPRIV;
1003 * Routines for handling mm_structs
1005 extern struct mm_struct * mm_alloc(void);
1007 /* mmdrop drops the mm and the page tables */
1008 extern void FASTCALL(__mmdrop(struct mm_struct *));
1009 static inline void mmdrop(struct mm_struct * mm)
1011 if (atomic_dec_and_test(&mm->mm_count))
1015 /* mmput gets rid of the mappings and all user-space */
1016 extern void mmput(struct mm_struct *);
1017 /* Grab a reference to the mm if its not already going away */
1018 extern struct mm_struct *mmgrab(struct mm_struct *);
1019 /* Remove the current tasks stale references to the old mm_struct */
1020 extern void mm_release(struct task_struct *, struct mm_struct *);
1022 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1023 extern void flush_thread(void);
1024 extern void exit_thread(void);
1026 extern void exit_mm(struct task_struct *);
1027 extern void exit_files(struct task_struct *);
1028 extern void exit_signal(struct task_struct *);
1029 extern void __exit_signal(struct task_struct *);
1030 extern void exit_sighand(struct task_struct *);
1031 extern void __exit_sighand(struct task_struct *);
1032 extern void exit_itimers(struct signal_struct *);
1034 extern NORET_TYPE void do_group_exit(int);
1036 extern void reparent_to_init(void);
1037 extern void daemonize(const char *, ...);
1038 extern int allow_signal(int);
1039 extern int disallow_signal(int);
1040 extern task_t *child_reaper;
1042 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1043 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1044 extern struct task_struct * copy_process(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1047 extern void wait_task_inactive(task_t * p);
1049 #define wait_task_inactive(p) do { } while (0)
1052 #define remove_parent(p) list_del_init(&(p)->sibling)
1053 #define add_parent(p, parent) list_add_tail(&(p)->sibling,&(parent)->children)
1055 #define REMOVE_LINKS(p) do { \
1056 if (thread_group_leader(p)) \
1057 list_del_init(&(p)->tasks); \
1061 #define SET_LINKS(p) do { \
1062 if (thread_group_leader(p)) \
1063 list_add_tail(&(p)->tasks,&init_task.tasks); \
1064 add_parent(p, (p)->parent); \
1067 #define next_task(p) list_entry((p)->tasks.next, struct task_struct, tasks)
1068 #define prev_task(p) list_entry((p)->tasks.prev, struct task_struct, tasks)
1070 #define for_each_process(p) \
1071 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1074 * Careful: do_each_thread/while_each_thread is a double loop so
1075 * 'break' will not work as expected - use goto instead.
1077 #define do_each_thread(g, t) \
1078 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1080 #define while_each_thread(g, t) \
1081 while ((t = next_thread(t)) != g)
1083 extern task_t * FASTCALL(next_thread(const task_t *p));
1085 #define thread_group_leader(p) (p->pid == p->tgid)
1087 static inline int thread_group_empty(task_t *p)
1089 struct pid *pid = p->pids[PIDTYPE_TGID].pidptr;
1091 return pid->task_list.next->next == &pid->task_list;
1094 #define delay_group_leader(p) \
1095 (thread_group_leader(p) && !thread_group_empty(p))
1097 extern void unhash_process(struct task_struct *p);
1100 * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info and synchronises with
1103 * Nests both inside and outside of read_lock(&tasklist_lock).
1104 * It must not be nested with write_lock_irq(&tasklist_lock),
1105 * neither inside nor outside.
1107 static inline void task_lock(struct task_struct *p)
1109 spin_lock(&p->alloc_lock);
1112 static inline void task_unlock(struct task_struct *p)
1114 spin_unlock(&p->alloc_lock);
1118 * get_task_mm - acquire a reference to the task's mm
1120 * Returns %NULL if the task has no mm. User must release
1121 * the mm via mmput() after use.
1123 static inline struct mm_struct * get_task_mm(struct task_struct * task)
1125 struct mm_struct * mm;
1136 /* set thread flags in other task's structures
1137 * - see asm/thread_info.h for TIF_xxxx flags available
1139 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1141 set_ti_thread_flag(tsk->thread_info,flag);
1144 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1146 clear_ti_thread_flag(tsk->thread_info,flag);
1149 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1151 return test_and_set_ti_thread_flag(tsk->thread_info,flag);
1154 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1156 return test_and_clear_ti_thread_flag(tsk->thread_info,flag);
1159 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1161 return test_ti_thread_flag(tsk->thread_info,flag);
1164 static inline void set_tsk_need_resched(struct task_struct *tsk)
1166 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1169 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1171 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1174 static inline int signal_pending(struct task_struct *p)
1176 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1179 static inline int need_resched(void)
1181 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1184 extern void __cond_resched(void);
1186 static inline void cond_resched(void)
1188 #ifdef CONFIG_DEBUG_SPINLOCK_SLEEP
1189 __might_sleep(__FILE__, __LINE__, 0);
1195 * cond_resched_lock() - if a reschedule is pending, drop the given lock,
1196 * call schedule, and on return reacquire the lock.
1198 * This works OK both with and without CONFIG_PREEMPT. We do strange low-level
1199 * operations here to prevent schedule() from being called twice (once via
1200 * spin_unlock(), once by hand).
1202 extern void __cond_resched_lock(spinlock_t * lock);
1204 static inline void cond_resched_lock(spinlock_t * lock)
1206 #ifdef CONFIG_DEBUG_SPINLOCK_SLEEP
1207 __might_sleep(__FILE__, __LINE__, 1);
1209 __cond_resched_lock(lock);
1212 /* Reevaluate whether the task has signals pending delivery.
1213 This is required every time the blocked sigset_t changes.
1214 callers must hold sighand->siglock. */
1216 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1217 extern void recalc_sigpending(void);
1219 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1222 * Wrappers for p->thread_info->cpu access. No-op on UP.
1226 static inline unsigned int task_cpu(const struct task_struct *p)
1228 return p->thread_info->cpu;
1231 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1233 p->thread_info->cpu = cpu;
1238 static inline unsigned int task_cpu(const struct task_struct *p)
1243 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1247 #endif /* CONFIG_SMP */
1250 /* API for registering delay info */
1251 #ifdef CONFIG_DELAY_ACCT
1253 #define test_delay_flag(tsk,flg) ((tsk)->flags & (flg))
1254 #define set_delay_flag(tsk,flg) ((tsk)->flags |= (flg))
1255 #define clear_delay_flag(tsk,flg) ((tsk)->flags &= ~(flg))
1257 #define def_delay_var(var) unsigned long long var
1258 #define get_delay(tsk,field) ((tsk)->delays.field)
1260 #define start_delay(var) ((var) = sched_clock())
1261 #define start_delay_set(var,flg) (set_delay_flag(current,flg),(var) = sched_clock())
1263 #define inc_delay(tsk,field) (((tsk)->delays.field)++)
1265 /* because of hardware timer drifts in SMPs and task continue on different cpu
1266 * then where the start_ts was taken there is a possibility that
1267 * end_ts < start_ts by some usecs. In this case we ignore the diff
1268 * and add nothing to the total.
1271 #define test_ts_integrity(start_ts,end_ts) (likely((end_ts) > (start_ts)))
1273 #define test_ts_integrity(start_ts,end_ts) (1)
1276 #define add_delay_ts(tsk,field,start_ts,end_ts) \
1277 do { if (test_ts_integrity(start_ts,end_ts)) (tsk)->delays.field += ((end_ts)-(start_ts)); } while (0)
1279 #define add_delay_clear(tsk,field,start_ts,flg) \
1281 unsigned long long now = sched_clock();\
1282 add_delay_ts(tsk,field,start_ts,now); \
1283 clear_delay_flag(tsk,flg); \
1286 static inline void add_io_delay(unsigned long long dstart)
1288 struct task_struct * tsk = current;
1289 unsigned long long now = sched_clock();
1290 unsigned long long val;
1292 if (test_ts_integrity(dstart,now))
1296 if (test_delay_flag(tsk,PF_MEMIO)) {
1297 tsk->delays.mem_iowait_total += val;
1298 tsk->delays.num_memwaits++;
1300 tsk->delays.iowait_total += val;
1301 tsk->delays.num_iowaits++;
1303 clear_delay_flag(tsk,PF_IOWAIT);
1306 inline static void init_delays(struct task_struct *tsk)
1308 memset((void*)&tsk->delays,0,sizeof(tsk->delays));
1313 #define test_delay_flag(tsk,flg) (0)
1314 #define set_delay_flag(tsk,flg) do { } while (0)
1315 #define clear_delay_flag(tsk,flg) do { } while (0)
1317 #define def_delay_var(var)
1318 #define get_delay(tsk,field) (0)
1320 #define start_delay(var) do { } while (0)
1321 #define start_delay_set(var,flg) do { } while (0)
1323 #define inc_delay(tsk,field) do { } while (0)
1324 #define add_delay_ts(tsk,field,start_ts,now) do { } while (0)
1325 #define add_delay_clear(tsk,field,start_ts,flg) do { } while (0)
1326 #define add_io_delay(dstart) do { } while (0)
1327 #define init_delays(tsk) do { } while (0)
1332 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1333 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1335 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1337 mm->mmap_base = TASK_UNMAPPED_BASE;
1338 mm->get_unmapped_area = arch_get_unmapped_area;
1339 mm->unmap_area = arch_unmap_area;
1343 #endif /* __KERNEL__ */