linux/include/linux/sched.h
<<
>>
Prefs
   1#ifndef _LINUX_SCHED_H
   2#define _LINUX_SCHED_H
   3
   4#include <linux/auxvec.h>       /* For AT_VECTOR_SIZE */
   5
   6/*
   7 * cloning flags:
   8 */
   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
  30/*
  31 * Scheduling policies
  32 */
  33#define SCHED_NORMAL            0
  34#define SCHED_FIFO              1
  35#define SCHED_RR                2
  36#define SCHED_BATCH             3
  37
  38#ifdef __KERNEL__
  39
  40struct sched_param {
  41        int sched_priority;
  42};
  43
  44#include <asm/param.h>  /* for HZ */
  45
  46#include <linux/capability.h>
  47#include <linux/threads.h>
  48#include <linux/kernel.h>
  49#include <linux/types.h>
  50#include <linux/timex.h>
  51#include <linux/jiffies.h>
  52#include <linux/rbtree.h>
  53#include <linux/thread_info.h>
  54#include <linux/cpumask.h>
  55#include <linux/errno.h>
  56#include <linux/nodemask.h>
  57
  58#include <asm/system.h>
  59#include <asm/semaphore.h>
  60#include <asm/page.h>
  61#include <asm/ptrace.h>
  62#include <asm/mmu.h>
  63#include <asm/cputime.h>
  64
  65#include <linux/smp.h>
  66#include <linux/sem.h>
  67#include <linux/signal.h>
  68#include <linux/securebits.h>
  69#include <linux/fs_struct.h>
  70#include <linux/compiler.h>
  71#include <linux/completion.h>
  72#include <linux/pid.h>
  73#include <linux/percpu.h>
  74#include <linux/topology.h>
  75#include <linux/seccomp.h>
  76#include <linux/rcupdate.h>
  77#include <linux/futex.h>
  78#include <linux/rtmutex.h>
  79
  80#include <linux/time.h>
  81#include <linux/param.h>
  82#include <linux/resource.h>
  83#include <linux/timer.h>
  84#include <linux/hrtimer.h>
  85
  86#include <asm/processor.h>
  87
  88struct exec_domain;
  89struct futex_pi_state;
  90
  91/*
  92 * List of flags we want to share for kernel threads,
  93 * if only because they are not used by them anyway.
  94 */
  95#define CLONE_KERNEL    (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
  96
  97/*
  98 * These are the constant used to fake the fixed-point load-average
  99 * counting. Some notes:
 100 *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
 101 *    a load-average precision of 10 bits integer + 11 bits fractional
 102 *  - if you want to count load-averages more often, you need more
 103 *    precision, or rounding will get you. With 2-second counting freq,
 104 *    the EXP_n values would be 1981, 2034 and 2043 if still using only
 105 *    11 bit fractions.
 106 */
 107extern unsigned long avenrun[];         /* Load averages */
 108
 109#define FSHIFT          11              /* nr of bits of precision */
 110#define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
 111#define LOAD_FREQ       (5*HZ)          /* 5 sec intervals */
 112#define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
 113#define EXP_5           2014            /* 1/exp(5sec/5min) */
 114#define EXP_15          2037            /* 1/exp(5sec/15min) */
 115
 116#define CALC_LOAD(load,exp,n) \
 117        load *= exp; \
 118        load += n*(FIXED_1-exp); \
 119        load >>= FSHIFT;
 120
 121extern unsigned long total_forks;
 122extern int nr_threads;
 123DECLARE_PER_CPU(unsigned long, process_counts);
 124extern int nr_processes(void);
 125extern unsigned long nr_running(void);
 126extern unsigned long nr_uninterruptible(void);
 127extern unsigned long nr_active(void);
 128extern unsigned long nr_iowait(void);
 129extern unsigned long weighted_cpuload(const int cpu);
 130
 131
 132/*
 133 * Task state bitmask. NOTE! These bits are also
 134 * encoded in fs/proc/array.c: get_task_state().
 135 *
 136 * We have two separate sets of flags: task->state
 137 * is about runnability, while task->exit_state are
 138 * about the task exiting. Confusing, but this way
 139 * modifying one set can't modify the other one by
 140 * mistake.
 141 */
 142#define TASK_RUNNING            0
 143#define TASK_INTERRUPTIBLE      1
 144#define TASK_UNINTERRUPTIBLE    2
 145#define TASK_STOPPED            4
 146#define TASK_TRACED             8
 147/* in tsk->exit_state */
 148#define EXIT_ZOMBIE             16
 149#define EXIT_DEAD               32
 150/* in tsk->state again */
 151#define TASK_NONINTERACTIVE     64
 152#define TASK_DEAD               128
 153
 154#define __set_task_state(tsk, state_value)              \
 155        do { (tsk)->state = (state_value); } while (0)
 156#define set_task_state(tsk, state_value)                \
 157        set_mb((tsk)->state, (state_value))
 158
 159/*
 160 * set_current_state() includes a barrier so that the write of current->state
 161 * is correctly serialised wrt the caller's subsequent test of whether to
 162 * actually sleep:
 163 *
 164 *      set_current_state(TASK_UNINTERRUPTIBLE);
 165 *      if (do_i_need_to_sleep())
 166 *              schedule();
 167 *
 168 * If the caller does not need such serialisation then use __set_current_state()
 169 */
 170#define __set_current_state(state_value)                        \
 171        do { current->state = (state_value); } while (0)
 172#define set_current_state(state_value)          \
 173        set_mb(current->state, (state_value))
 174
 175/* Task command name length */
 176#define TASK_COMM_LEN 16
 177
 178#include <linux/spinlock.h>
 179
 180/*
 181 * This serializes "schedule()" and also protects
 182 * the run-queue from deletions/modifications (but
 183 * _adding_ to the beginning of the run-queue has
 184 * a separate lock).
 185 */
 186extern rwlock_t tasklist_lock;
 187extern spinlock_t mmlist_lock;
 188
 189struct task_struct;
 190
 191extern void sched_init(void);
 192extern void sched_init_smp(void);
 193extern void init_idle(struct task_struct *idle, int cpu);
 194
 195extern cpumask_t nohz_cpu_mask;
 196
 197extern void show_state(void);
 198extern void show_regs(struct pt_regs *);
 199
 200/*
 201 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 202 * task), SP is the stack pointer of the first frame that should be shown in the back
 203 * trace (or NULL if the entire call-chain of the task should be shown).
 204 */
 205extern void show_stack(struct task_struct *task, unsigned long *sp);
 206
 207void io_schedule(void);
 208long io_schedule_timeout(long timeout);
 209
 210extern void cpu_init (void);
 211extern void trap_init(void);
 212extern void update_process_times(int user);
 213extern void scheduler_tick(void);
 214
 215#ifdef CONFIG_DETECT_SOFTLOCKUP
 216extern void softlockup_tick(void);
 217extern void spawn_softlockup_task(void);
 218extern void touch_softlockup_watchdog(void);
 219#else
 220static inline void softlockup_tick(void)
 221{
 222}
 223static inline void spawn_softlockup_task(void)
 224{
 225}
 226static inline void touch_softlockup_watchdog(void)
 227{
 228}
 229#endif
 230
 231
 232/* Attach to any functions which should be ignored in wchan output. */
 233#define __sched         __attribute__((__section__(".sched.text")))
 234/* Is this address in the __sched functions? */
 235extern int in_sched_functions(unsigned long addr);
 236
 237#define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 238extern signed long FASTCALL(schedule_timeout(signed long timeout));
 239extern signed long schedule_timeout_interruptible(signed long timeout);
 240extern signed long schedule_timeout_uninterruptible(signed long timeout);
 241asmlinkage void schedule(void);
 242
 243struct nsproxy;
 244
 245/* Maximum number of active map areas.. This is a random (large) number */
 246#define DEFAULT_MAX_MAP_COUNT   65536
 247
 248extern int sysctl_max_map_count;
 249
 250#include <linux/aio.h>
 251
 252extern unsigned long
 253arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 254                       unsigned long, unsigned long);
 255extern unsigned long
 256arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 257                          unsigned long len, unsigned long pgoff,
 258                          unsigned long flags);
 259extern void arch_unmap_area(struct mm_struct *, unsigned long);
 260extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
 261
 262#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
 263/*
 264 * The mm counters are not protected by its page_table_lock,
 265 * so must be incremented atomically.
 266 */
 267#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
 268#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
 269#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
 270#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
 271#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
 272typedef atomic_long_t mm_counter_t;
 273
 274#else  /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
 275/*
 276 * The mm counters are protected by its page_table_lock,
 277 * so can be incremented directly.
 278 */
 279#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
 280#define get_mm_counter(mm, member) ((mm)->_##member)
 281#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
 282#define inc_mm_counter(mm, member) (mm)->_##member++
 283#define dec_mm_counter(mm, member) (mm)->_##member--
 284typedef unsigned long mm_counter_t;
 285
 286#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
 287
 288#define get_mm_rss(mm)                                  \
 289        (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
 290#define update_hiwater_rss(mm)  do {                    \
 291        unsigned long _rss = get_mm_rss(mm);            \
 292        if ((mm)->hiwater_rss < _rss)                   \
 293                (mm)->hiwater_rss = _rss;               \
 294} while (0)
 295#define update_hiwater_vm(mm)   do {                    \
 296        if ((mm)->hiwater_vm < (mm)->total_vm)          \
 297                (mm)->hiwater_vm = (mm)->total_vm;      \
 298} while (0)
 299
 300struct mm_struct {
 301        struct vm_area_struct * mmap;           /* list of VMAs */
 302        struct rb_root mm_rb;
 303        struct vm_area_struct * mmap_cache;     /* last find_vma result */
 304        unsigned long (*get_unmapped_area) (struct file *filp,
 305                                unsigned long addr, unsigned long len,
 306                                unsigned long pgoff, unsigned long flags);
 307        void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
 308        unsigned long mmap_base;                /* base of mmap area */
 309        unsigned long task_size;                /* size of task vm space */
 310        unsigned long cached_hole_size;         /* if non-zero, the largest hole below free_area_cache */
 311        unsigned long free_area_cache;          /* first hole of size cached_hole_size or larger */
 312        pgd_t * pgd;
 313        atomic_t mm_users;                      /* How many users with user space? */
 314        atomic_t mm_count;                      /* How many references to "struct mm_struct" (users count as 1) */
 315        int map_count;                          /* number of VMAs */
 316        struct rw_semaphore mmap_sem;
 317        spinlock_t page_table_lock;             /* Protects page tables and some counters */
 318
 319        struct list_head mmlist;                /* List of maybe swapped mm's.  These are globally strung
 320                                                 * together off init_mm.mmlist, and are protected
 321                                                 * by mmlist_lock
 322                                                 */
 323
 324        /* Special counters, in some configurations protected by the
 325         * page_table_lock, in other configurations by being atomic.
 326         */
 327        mm_counter_t _file_rss;
 328        mm_counter_t _anon_rss;
 329
 330        unsigned long hiwater_rss;      /* High-watermark of RSS usage */
 331        unsigned long hiwater_vm;       /* High-water virtual memory usage */
 332
 333        unsigned long total_vm, locked_vm, shared_vm, exec_vm;
 334        unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
 335        unsigned long start_code, end_code, start_data, end_data;
 336        unsigned long start_brk, brk, start_stack;
 337        unsigned long arg_start, arg_end, env_start, env_end;
 338
 339        unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
 340
 341        unsigned dumpable:2;
 342        cpumask_t cpu_vm_mask;
 343
 344        /* Architecture-specific MM context */
 345        mm_context_t context;
 346
 347        /* Token based thrashing protection. */
 348        unsigned long swap_token_time;
 349        char recent_pagein;
 350
 351        /* coredumping support */
 352        int core_waiters;
 353        struct completion *core_startup_done, core_done;
 354
 355        /* aio bits */
 356        rwlock_t                ioctx_list_lock;
 357        struct kioctx           *ioctx_list;
 358};
 359
 360struct sighand_struct {
 361        atomic_t                count;
 362        struct k_sigaction      action[_NSIG];
 363        spinlock_t              siglock;
 364};
 365
 366struct pacct_struct {
 367        int                     ac_flag;
 368        long                    ac_exitcode;
 369        unsigned long           ac_mem;
 370        cputime_t               ac_utime, ac_stime;
 371        unsigned long           ac_minflt, ac_majflt;
 372};
 373
 374/*
 375 * NOTE! "signal_struct" does not have it's own
 376 * locking, because a shared signal_struct always
 377 * implies a shared sighand_struct, so locking
 378 * sighand_struct is always a proper superset of
 379 * the locking of signal_struct.
 380 */
 381struct signal_struct {
 382        atomic_t                count;
 383        atomic_t                live;
 384
 385        wait_queue_head_t       wait_chldexit;  /* for wait4() */
 386
 387        /* current thread group signal load-balancing target: */
 388        struct task_struct      *curr_target;
 389
 390        /* shared signal handling: */
 391        struct sigpending       shared_pending;
 392
 393        /* thread group exit support */
 394        int                     group_exit_code;
 395        /* overloaded:
 396         * - notify group_exit_task when ->count is equal to notify_count
 397         * - everyone except group_exit_task is stopped during signal delivery
 398         *   of fatal signals, group_exit_task processes the signal.
 399         */
 400        struct task_struct      *group_exit_task;
 401        int                     notify_count;
 402
 403        /* thread group stop support, overloads group_exit_code too */
 404        int                     group_stop_count;
 405        unsigned int            flags; /* see SIGNAL_* flags below */
 406
 407        /* POSIX.1b Interval Timers */
 408        struct list_head posix_timers;
 409
 410        /* ITIMER_REAL timer for the process */
 411        struct hrtimer real_timer;
 412        struct task_struct *tsk;
 413        ktime_t it_real_incr;
 414
 415        /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
 416        cputime_t it_prof_expires, it_virt_expires;
 417        cputime_t it_prof_incr, it_virt_incr;
 418
 419        /* job control IDs */
 420        pid_t pgrp;
 421        pid_t tty_old_pgrp;
 422        pid_t session;
 423        /* boolean value for session group leader */
 424        int leader;
 425
 426        struct tty_struct *tty; /* NULL if no tty */
 427
 428        /*
 429         * Cumulative resource counters for dead threads in the group,
 430         * and for reaped dead child processes forked by this group.
 431         * Live threads maintain their own counters and add to these
 432         * in __exit_signal, except for the group leader.
 433         */
 434        cputime_t utime, stime, cutime, cstime;
 435        unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 436        unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 437
 438        /*
 439         * Cumulative ns of scheduled CPU time for dead threads in the
 440         * group, not including a zombie group leader.  (This only differs
 441         * from jiffies_to_ns(utime + stime) if sched_clock uses something
 442         * other than jiffies.)
 443         */
 444        unsigned long long sched_time;
 445
 446        /*
 447         * We don't bother to synchronize most readers of this at all,
 448         * because there is no reader checking a limit that actually needs
 449         * to get both rlim_cur and rlim_max atomically, and either one
 450         * alone is a single word that can safely be read normally.
 451         * getrlimit/setrlimit use task_lock(current->group_leader) to
 452         * protect this instead of the siglock, because they really
 453         * have no need to disable irqs.
 454         */
 455        struct rlimit rlim[RLIM_NLIMITS];
 456
 457        struct list_head cpu_timers[3];
 458
 459        /* keep the process-shared keyrings here so that they do the right
 460         * thing in threads created with CLONE_THREAD */
 461#ifdef CONFIG_KEYS
 462        struct key *session_keyring;    /* keyring inherited over fork */
 463        struct key *process_keyring;    /* keyring private to this process */
 464#endif
 465#ifdef CONFIG_BSD_PROCESS_ACCT
 466        struct pacct_struct pacct;      /* per-process accounting information */
 467#endif
 468#ifdef CONFIG_TASKSTATS
 469        struct taskstats *stats;
 470#endif
 471};
 472
 473/* Context switch must be unlocked if interrupts are to be enabled */
 474#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
 475# define __ARCH_WANT_UNLOCKED_CTXSW
 476#endif
 477
 478/*
 479 * Bits in flags field of signal_struct.
 480 */
 481#define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 482#define SIGNAL_STOP_DEQUEUED    0x00000002 /* stop signal dequeued */
 483#define SIGNAL_STOP_CONTINUED   0x00000004 /* SIGCONT since WCONTINUED reap */
 484#define SIGNAL_GROUP_EXIT       0x00000008 /* group exit in progress */
 485
 486
 487/*
 488 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
 489 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
 490 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
 491 * values are inverted: lower p->prio value means higher priority.
 492 *
 493 * The MAX_USER_RT_PRIO value allows the actual maximum
 494 * RT priority to be separate from the value exported to
 495 * user-space.  This allows kernel threads to set their
 496 * priority to a value higher than any user task. Note:
 497 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
 498 */
 499
 500#define MAX_USER_RT_PRIO        100
 501#define MAX_RT_PRIO             MAX_USER_RT_PRIO
 502
 503#define MAX_PRIO                (MAX_RT_PRIO + 40)
 504
 505#define rt_prio(prio)           unlikely((prio) < MAX_RT_PRIO)
 506#define rt_task(p)              rt_prio((p)->prio)
 507#define batch_task(p)           (unlikely((p)->policy == SCHED_BATCH))
 508#define is_rt_policy(p)         ((p) != SCHED_NORMAL && (p) != SCHED_BATCH)
 509#define has_rt_policy(p)        unlikely(is_rt_policy((p)->policy))
 510
 511/*
 512 * Some day this will be a full-fledged user tracking system..
 513 */
 514struct user_struct {
 515        atomic_t __count;       /* reference count */
 516        atomic_t processes;     /* How many processes does this user have? */
 517        atomic_t files;         /* How many open files does this user have? */
 518        atomic_t sigpending;    /* How many pending signals does this user have? */
 519#ifdef CONFIG_INOTIFY_USER
 520        atomic_t inotify_watches; /* How many inotify watches does this user have? */
 521        atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 522#endif
 523        /* protected by mq_lock */
 524        unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 525        unsigned long locked_shm; /* How many pages of mlocked shm ? */
 526
 527#ifdef CONFIG_KEYS
 528        struct key *uid_keyring;        /* UID specific keyring */
 529        struct key *session_keyring;    /* UID's default session keyring */
 530#endif
 531
 532        /* Hash table maintenance information */
 533        struct list_head uidhash_list;
 534        uid_t uid;
 535};
 536
 537extern struct user_struct *find_user(uid_t);
 538
 539extern struct user_struct root_user;
 540#define INIT_USER (&root_user)
 541
 542struct backing_dev_info;
 543struct reclaim_state;
 544
 545#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 546struct sched_info {
 547        /* cumulative counters */
 548        unsigned long   cpu_time,       /* time spent on the cpu */
 549                        run_delay,      /* time spent waiting on a runqueue */
 550                        pcnt;           /* # of timeslices run on this cpu */
 551
 552        /* timestamps */
 553        unsigned long   last_arrival,   /* when we last ran on a cpu */
 554                        last_queued;    /* when we were last queued to run */
 555};
 556#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
 557
 558#ifdef CONFIG_SCHEDSTATS
 559extern struct file_operations proc_schedstat_operations;
 560#endif /* CONFIG_SCHEDSTATS */
 561
 562#ifdef CONFIG_TASK_DELAY_ACCT
 563struct task_delay_info {
 564        spinlock_t      lock;
 565        unsigned int    flags;  /* Private per-task flags */
 566
 567        /* For each stat XXX, add following, aligned appropriately
 568         *
 569         * struct timespec XXX_start, XXX_end;
 570         * u64 XXX_delay;
 571         * u32 XXX_count;
 572         *
 573         * Atomicity of updates to XXX_delay, XXX_count protected by
 574         * single lock above (split into XXX_lock if contention is an issue).
 575         */
 576
 577        /*
 578         * XXX_count is incremented on every XXX operation, the delay
 579         * associated with the operation is added to XXX_delay.
 580         * XXX_delay contains the accumulated delay time in nanoseconds.
 581         */
 582        struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
 583        u64 blkio_delay;        /* wait for sync block io completion */
 584        u64 swapin_delay;       /* wait for swapin block io completion */
 585        u32 blkio_count;        /* total count of the number of sync block */
 586                                /* io operations performed */
 587        u32 swapin_count;       /* total count of the number of swapin block */
 588                                /* io operations performed */
 589};
 590#endif  /* CONFIG_TASK_DELAY_ACCT */
 591
 592static inline int sched_info_on(void)
 593{
 594#ifdef CONFIG_SCHEDSTATS
 595        return 1;
 596#elif defined(CONFIG_TASK_DELAY_ACCT)
 597        extern int delayacct_on;
 598        return delayacct_on;
 599#else
 600        return 0;
 601#endif
 602}
 603
 604enum idle_type
 605{
 606        SCHED_IDLE,
 607        NOT_IDLE,
 608        NEWLY_IDLE,
 609        MAX_IDLE_TYPES
 610};
 611
 612/*
 613 * sched-domains (multiprocessor balancing) declarations:
 614 */
 615#define SCHED_LOAD_SCALE        128UL   /* increase resolution of load */
 616
 617#ifdef CONFIG_SMP
 618#define SD_LOAD_BALANCE         1       /* Do load balancing on this domain. */
 619#define SD_BALANCE_NEWIDLE      2       /* Balance when about to become idle */
 620#define SD_BALANCE_EXEC         4       /* Balance on exec */
 621#define SD_BALANCE_FORK         8       /* Balance on fork, clone */
 622#define SD_WAKE_IDLE            16      /* Wake to idle CPU on task wakeup */
 623#define SD_WAKE_AFFINE          32      /* Wake task to waking CPU */
 624#define SD_WAKE_BALANCE         64      /* Perform balancing at task wakeup */
 625#define SD_SHARE_CPUPOWER       128     /* Domain members share cpu power */
 626#define SD_POWERSAVINGS_BALANCE 256     /* Balance for power savings */
 627#define SD_SHARE_PKG_RESOURCES  512     /* Domain members share cpu pkg resources */
 628
 629#define BALANCE_FOR_MC_POWER    \
 630        (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
 631
 632#define BALANCE_FOR_PKG_POWER   \
 633        ((sched_mc_power_savings || sched_smt_power_savings) ?  \
 634         SD_POWERSAVINGS_BALANCE : 0)
 635
 636#define test_sd_parent(sd, flag)        ((sd->parent &&         \
 637                                         (sd->parent->flags & flag)) ? 1 : 0)
 638
 639
 640struct sched_group {
 641        struct sched_group *next;       /* Must be a circular list */
 642        cpumask_t cpumask;
 643
 644        /*
 645         * CPU power of this group, SCHED_LOAD_SCALE being max power for a
 646         * single CPU. This is read only (except for setup, hotplug CPU).
 647         */
 648        unsigned long cpu_power;
 649};
 650
 651struct sched_domain {
 652        /* These fields must be setup */
 653        struct sched_domain *parent;    /* top domain must be null terminated */
 654        struct sched_domain *child;     /* bottom domain must be null terminated */
 655        struct sched_group *groups;     /* the balancing groups of the domain */
 656        cpumask_t span;                 /* span of all CPUs in this domain */
 657        unsigned long min_interval;     /* Minimum balance interval ms */
 658        unsigned long max_interval;     /* Maximum balance interval ms */
 659        unsigned int busy_factor;       /* less balancing by factor if busy */
 660        unsigned int imbalance_pct;     /* No balance until over watermark */
 661        unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
 662        unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
 663        unsigned int per_cpu_gain;      /* CPU % gained by adding domain cpus */
 664        unsigned int busy_idx;
 665        unsigned int idle_idx;
 666        unsigned int newidle_idx;
 667        unsigned int wake_idx;
 668        unsigned int forkexec_idx;
 669        int flags;                      /* See SD_* */
 670
 671        /* Runtime fields. */
 672        unsigned long last_balance;     /* init to jiffies. units in jiffies */
 673        unsigned int balance_interval;  /* initialise to 1. units in ms. */
 674        unsigned int nr_balance_failed; /* initialise to 0 */
 675
 676#ifdef CONFIG_SCHEDSTATS
 677        /* load_balance() stats */
 678        unsigned long lb_cnt[MAX_IDLE_TYPES];
 679        unsigned long lb_failed[MAX_IDLE_TYPES];
 680        unsigned long lb_balanced[MAX_IDLE_TYPES];
 681        unsigned long lb_imbalance[MAX_IDLE_TYPES];
 682        unsigned long lb_gained[MAX_IDLE_TYPES];
 683        unsigned long lb_hot_gained[MAX_IDLE_TYPES];
 684        unsigned long lb_nobusyg[MAX_IDLE_TYPES];
 685        unsigned long lb_nobusyq[MAX_IDLE_TYPES];
 686
 687        /* Active load balancing */
 688        unsigned long alb_cnt;
 689        unsigned long alb_failed;
 690        unsigned long alb_pushed;
 691
 692        /* SD_BALANCE_EXEC stats */
 693        unsigned long sbe_cnt;
 694        unsigned long sbe_balanced;
 695        unsigned long sbe_pushed;
 696
 697        /* SD_BALANCE_FORK stats */
 698        unsigned long sbf_cnt;
 699        unsigned long sbf_balanced;
 700        unsigned long sbf_pushed;
 701
 702        /* try_to_wake_up() stats */
 703        unsigned long ttwu_wake_remote;
 704        unsigned long ttwu_move_affine;
 705        unsigned long ttwu_move_balance;
 706#endif
 707};
 708
 709extern int partition_sched_domains(cpumask_t *partition1,
 710                                    cpumask_t *partition2);
 711
 712/*
 713 * Maximum cache size the migration-costs auto-tuning code will
 714 * search from:
 715 */
 716extern unsigned int max_cache_size;
 717
 718#endif  /* CONFIG_SMP */
 719
 720
 721struct io_context;                      /* See blkdev.h */
 722struct cpuset;
 723
 724#define NGROUPS_SMALL           32
 725#define NGROUPS_PER_BLOCK       ((int)(PAGE_SIZE / sizeof(gid_t)))
 726struct group_info {
 727        int ngroups;
 728        atomic_t usage;
 729        gid_t small_block[NGROUPS_SMALL];
 730        int nblocks;
 731        gid_t *blocks[0];
 732};
 733
 734/*
 735 * get_group_info() must be called with the owning task locked (via task_lock())
 736 * when task != current.  The reason being that the vast majority of callers are
 737 * looking at current->group_info, which can not be changed except by the
 738 * current task.  Changing current->group_info requires the task lock, too.
 739 */
 740#define get_group_info(group_info) do { \
 741        atomic_inc(&(group_info)->usage); \
 742} while (0)
 743
 744#define put_group_info(group_info) do { \
 745        if (atomic_dec_and_test(&(group_info)->usage)) \
 746                groups_free(group_info); \
 747} while (0)
 748
 749extern struct group_info *groups_alloc(int gidsetsize);
 750extern void groups_free(struct group_info *group_info);
 751extern int set_current_groups(struct group_info *group_info);
 752extern int groups_search(struct group_info *group_info, gid_t grp);
 753/* access the groups "array" with this macro */
 754#define GROUP_AT(gi, i) \
 755    ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
 756
 757#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
 758extern void prefetch_stack(struct task_struct *t);
 759#else
 760static inline void prefetch_stack(struct task_struct *t) { }
 761#endif
 762
 763struct audit_context;           /* See audit.c */
 764struct mempolicy;
 765struct pipe_inode_info;
 766struct uts_namespace;
 767
 768enum sleep_type {
 769        SLEEP_NORMAL,
 770        SLEEP_NONINTERACTIVE,
 771        SLEEP_INTERACTIVE,
 772        SLEEP_INTERRUPTED,
 773};
 774
 775struct prio_array;
 776
 777struct task_struct {
 778        volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
 779        struct thread_info *thread_info;
 780        atomic_t usage;
 781        unsigned long flags;    /* per process flags, defined below */
 782        unsigned long ptrace;
 783
 784        int lock_depth;         /* BKL lock depth */
 785
 786#ifdef CONFIG_SMP
 787#ifdef __ARCH_WANT_UNLOCKED_CTXSW
 788        int oncpu;
 789#endif
 790#endif
 791        int load_weight;        /* for niceness load balancing purposes */
 792        int prio, static_prio, normal_prio;
 793        struct list_head run_list;
 794        struct prio_array *array;
 795
 796        unsigned short ioprio;
 797#ifdef CONFIG_BLK_DEV_IO_TRACE
 798        unsigned int btrace_seq;
 799#endif
 800        unsigned long sleep_avg;
 801        unsigned long long timestamp, last_ran;
 802        unsigned long long sched_time; /* sched_clock time spent running */
 803        enum sleep_type sleep_type;
 804
 805        unsigned long policy;
 806        cpumask_t cpus_allowed;
 807        unsigned int time_slice, first_time_slice;
 808
 809#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 810        struct sched_info sched_info;
 811#endif
 812
 813        struct list_head tasks;
 814        /*
 815         * ptrace_list/ptrace_children forms the list of my children
 816         * that were stolen by a ptracer.
 817         */
 818        struct list_head ptrace_children;
 819        struct list_head ptrace_list;
 820
 821        struct mm_struct *mm, *active_mm;
 822
 823/* task state */
 824        struct linux_binfmt *binfmt;
 825        long exit_state;
 826        int exit_code, exit_signal;
 827        int pdeath_signal;  /*  The signal sent when the parent dies  */
 828        /* ??? */
 829        unsigned long personality;
 830        unsigned did_exec:1;
 831        pid_t pid;
 832        pid_t tgid;
 833
 834#ifdef CONFIG_CC_STACKPROTECTOR
 835        /* Canary value for the -fstack-protector gcc feature */
 836        unsigned long stack_canary;
 837#endif
 838        /* 
 839         * pointers to (original) parent process, youngest child, younger sibling,
 840         * older sibling, respectively.  (p->father can be replaced with 
 841         * p->parent->pid)
 842         */
 843        struct task_struct *real_parent; /* real parent process (when being debugged) */
 844        struct task_struct *parent;     /* parent process */
 845        /*
 846         * children/sibling forms the list of my children plus the
 847         * tasks I'm ptracing.
 848         */
 849        struct list_head children;      /* list of my children */
 850        struct list_head sibling;       /* linkage in my parent's children list */
 851        struct task_struct *group_leader;       /* threadgroup leader */
 852
 853        /* PID/PID hash table linkage. */
 854        struct pid_link pids[PIDTYPE_MAX];
 855        struct list_head thread_group;
 856
 857        struct completion *vfork_done;          /* for vfork() */
 858        int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
 859        int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
 860
 861        unsigned long rt_priority;
 862        cputime_t utime, stime;
 863        unsigned long nvcsw, nivcsw; /* context switch counts */
 864        struct timespec start_time;
 865/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
 866        unsigned long min_flt, maj_flt;
 867
 868        cputime_t it_prof_expires, it_virt_expires;
 869        unsigned long long it_sched_expires;
 870        struct list_head cpu_timers[3];
 871
 872/* process credentials */
 873        uid_t uid,euid,suid,fsuid;
 874        gid_t gid,egid,sgid,fsgid;
 875        struct group_info *group_info;
 876        kernel_cap_t   cap_effective, cap_inheritable, cap_permitted;
 877        unsigned keep_capabilities:1;
 878        struct user_struct *user;
 879#ifdef CONFIG_KEYS
 880        struct key *request_key_auth;   /* assumed request_key authority */
 881        struct key *thread_keyring;     /* keyring private to this thread */
 882        unsigned char jit_keyring;      /* default keyring to attach requested keys to */
 883#endif
 884        /*
 885         * fpu_counter contains the number of consecutive context switches
 886         * that the FPU is used. If this is over a threshold, the lazy fpu
 887         * saving becomes unlazy to save the trap. This is an unsigned char
 888         * so that after 256 times the counter wraps and the behavior turns
 889         * lazy again; this to deal with bursty apps that only use FPU for
 890         * a short time
 891         */
 892        unsigned char fpu_counter;
 893        int oomkilladj; /* OOM kill score adjustment (bit shift). */
 894        char comm[TASK_COMM_LEN]; /* executable name excluding path
 895                                     - access with [gs]et_task_comm (which lock
 896                                       it with task_lock())
 897                                     - initialized normally by flush_old_exec */
 898/* file system info */
 899        int link_count, total_link_count;
 900#ifdef CONFIG_SYSVIPC
 901/* ipc stuff */
 902        struct sysv_sem sysvsem;
 903#endif
 904/* CPU-specific state of this task */
 905        struct thread_struct thread;
 906/* filesystem information */
 907        struct fs_struct *fs;
 908/* open file information */
 909        struct files_struct *files;
 910/* namespaces */
 911        struct nsproxy *nsproxy;
 912/* signal handlers */
 913        struct signal_struct *signal;
 914        struct sighand_struct *sighand;
 915
 916        sigset_t blocked, real_blocked;
 917        sigset_t saved_sigmask;         /* To be restored with TIF_RESTORE_SIGMASK */
 918        struct sigpending pending;
 919
 920        unsigned long sas_ss_sp;
 921        size_t sas_ss_size;
 922        int (*notifier)(void *priv);
 923        void *notifier_data;
 924        sigset_t *notifier_mask;
 925        
 926        void *security;
 927        struct audit_context *audit_context;
 928        seccomp_t seccomp;
 929
 930/* Thread group tracking */
 931        u32 parent_exec_id;
 932        u32 self_exec_id;
 933/* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
 934        spinlock_t alloc_lock;
 935
 936        /* Protection of the PI data structures: */
 937        spinlock_t pi_lock;
 938
 939#ifdef CONFIG_RT_MUTEXES
 940        /* PI waiters blocked on a rt_mutex held by this task */
 941        struct plist_head pi_waiters;
 942        /* Deadlock detection and priority inheritance handling */
 943        struct rt_mutex_waiter *pi_blocked_on;
 944#endif
 945
 946#ifdef CONFIG_DEBUG_MUTEXES
 947        /* mutex deadlock detection */
 948        struct mutex_waiter *blocked_on;
 949#endif
 950#ifdef CONFIG_TRACE_IRQFLAGS
 951        unsigned int irq_events;
 952        int hardirqs_enabled;
 953        unsigned long hardirq_enable_ip;
 954        unsigned int hardirq_enable_event;
 955        unsigned long hardirq_disable_ip;
 956        unsigned int hardirq_disable_event;
 957        int softirqs_enabled;
 958        unsigned long softirq_disable_ip;
 959        unsigned int softirq_disable_event;
 960        unsigned long softirq_enable_ip;
 961        unsigned int softirq_enable_event;
 962        int hardirq_context;
 963        int softirq_context;
 964#endif
 965#ifdef CONFIG_LOCKDEP
 966# define MAX_LOCK_DEPTH 30UL
 967        u64 curr_chain_key;
 968        int lockdep_depth;
 969        struct held_lock held_locks[MAX_LOCK_DEPTH];
 970        unsigned int lockdep_recursion;
 971#endif
 972
 973/* journalling filesystem info */
 974        void *journal_info;
 975
 976/* VM state */
 977        struct reclaim_state *reclaim_state;
 978
 979        struct backing_dev_info *backing_dev_info;
 980
 981        struct io_context *io_context;
 982
 983        unsigned long ptrace_message;
 984        siginfo_t *last_siginfo; /* For ptrace use.  */
 985/*
 986 * current io wait handle: wait queue entry to use for io waits
 987 * If this thread is processing aio, this points at the waitqueue
 988 * inside the currently handled kiocb. It may be NULL (i.e. default
 989 * to a stack based synchronous wait) if its doing sync IO.
 990 */
 991        wait_queue_t *io_wait;
 992/* i/o counters(bytes read/written, #syscalls */
 993        u64 rchar, wchar, syscr, syscw;
 994#if defined(CONFIG_TASK_XACCT)
 995        u64 acct_rss_mem1;      /* accumulated rss usage */
 996        u64 acct_vm_mem1;       /* accumulated virtual memory usage */
 997        cputime_t acct_stimexpd;/* stime since last update */
 998#endif
 999#ifdef CONFIG_NUMA
1000        struct mempolicy *mempolicy;
1001        short il_next;
1002#endif
1003#ifdef CONFIG_CPUSETS
1004        struct cpuset *cpuset;
1005        nodemask_t mems_allowed;
1006        int cpuset_mems_generation;
1007        int cpuset_mem_spread_rotor;
1008#endif
1009        struct robust_list_head __user *robust_list;
1010#ifdef CONFIG_COMPAT
1011        struct compat_robust_list_head __user *compat_robust_list;
1012#endif
1013        struct list_head pi_state_list;
1014        struct futex_pi_state *pi_state_cache;
1015
1016        atomic_t fs_excl;       /* holding fs exclusive resources */
1017        struct rcu_head rcu;
1018
1019        /*
1020         * cache last used pipe for splice
1021         */
1022        struct pipe_inode_info *splice_pipe;
1023#ifdef  CONFIG_TASK_DELAY_ACCT
1024        struct task_delay_info *delays;
1025#endif
1026};
1027
1028static inline pid_t process_group(struct task_struct *tsk)
1029{
1030        return tsk->signal->pgrp;
1031}
1032
1033static inline struct pid *task_pid(struct task_struct *task)
1034{
1035        return task->pids[PIDTYPE_PID].pid;
1036}
1037
1038static inline struct pid *task_tgid(struct task_struct *task)
1039{
1040        return task->group_leader->pids[PIDTYPE_PID].pid;
1041}
1042
1043static inline struct pid *task_pgrp(struct task_struct *task)
1044{
1045        return task->group_leader->pids[PIDTYPE_PGID].pid;
1046}
1047
1048static inline struct pid *task_session(struct task_struct *task)
1049{
1050        return task->group_leader->pids[PIDTYPE_SID].pid;
1051}
1052
1053/**
1054 * pid_alive - check that a task structure is not stale
1055 * @p: Task structure to be checked.
1056 *
1057 * Test if a process is not yet dead (at most zombie state)
1058 * If pid_alive fails, then pointers within the task structure
1059 * can be stale and must not be dereferenced.
1060 */
1061static inline int pid_alive(struct task_struct *p)
1062{
1063        return p->pids[PIDTYPE_PID].pid != NULL;
1064}
1065
1066/**
1067 * is_init - check if a task structure is init
1068 * @tsk: Task structure to be checked.
1069 *
1070 * Check if a task structure is the first user space task the kernel created.
1071 */
1072static inline int is_init(struct task_struct *tsk)
1073{
1074        return tsk->pid == 1;
1075}
1076
1077extern struct pid *cad_pid;
1078
1079extern void free_task(struct task_struct *tsk);
1080#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1081
1082extern void __put_task_struct(struct task_struct *t);
1083
1084static inline void put_task_struct(struct task_struct *t)
1085{
1086        if (atomic_dec_and_test(&t->usage))
1087                __put_task_struct(t);
1088}
1089
1090/*
1091 * Per process flags
1092 */
1093#define PF_ALIGNWARN    0x00000001      /* Print alignment warning msgs */
1094                                        /* Not implemented yet, only for 486*/
1095#define PF_STARTING     0x00000002      /* being created */
1096#define PF_EXITING      0x00000004      /* getting shut down */
1097#define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
1098#define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
1099#define PF_DUMPCORE     0x00000200      /* dumped core */
1100#define PF_SIGNALED     0x00000400      /* killed by a signal */
1101#define PF_MEMALLOC     0x00000800      /* Allocating memory */
1102#define PF_FLUSHER      0x00001000      /* responsible for disk writeback */
1103#define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
1104#define PF_FREEZE       0x00004000      /* this task is being frozen for suspend now */
1105#define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
1106#define PF_FROZEN       0x00010000      /* frozen for system suspend */
1107#define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
1108#define PF_KSWAPD       0x00040000      /* I am kswapd */
1109#define PF_SWAPOFF      0x00080000      /* I am in swapoff */
1110#define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
1111#define PF_BORROWED_MM  0x00200000      /* I am a kthread doing use_mm */
1112#define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
1113#define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
1114#define PF_SPREAD_PAGE  0x01000000      /* Spread page cache over cpuset */
1115#define PF_SPREAD_SLAB  0x02000000      /* Spread some slab caches over cpuset */
1116#define PF_MEMPOLICY    0x10000000      /* Non-default NUMA mempolicy */
1117#define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
1118
1119/*
1120 * Only the _current_ task can read/write to tsk->flags, but other
1121 * tasks can access tsk->flags in readonly mode for example
1122 * with tsk_used_math (like during threaded core dumping).
1123 * There is however an exception to this rule during ptrace
1124 * or during fork: the ptracer task is allowed to write to the
1125 * child->flags of its traced child (same goes for fork, the parent
1126 * can write to the child->flags), because we're guaranteed the
1127 * child is not running and in turn not changing child->flags
1128 * at the same time the parent does it.
1129 */
1130#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1131#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1132#define clear_used_math() clear_stopped_child_used_math(current)
1133#define set_used_math() set_stopped_child_used_math(current)
1134#define conditional_stopped_child_used_math(condition, child) \
1135        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1136#define conditional_used_math(condition) \
1137        conditional_stopped_child_used_math(condition, current)
1138#define copy_to_stopped_child_used_math(child) \
1139        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1140/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1141#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1142#define used_math() tsk_used_math(current)
1143
1144#ifdef CONFIG_SMP
1145extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1146#else
1147static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1148{
1149        if (!cpu_isset(0, new_mask))
1150                return -EINVAL;
1151        return 0;
1152}
1153#endif
1154
1155extern unsigned long long sched_clock(void);
1156extern unsigned long long
1157current_sched_time(const struct task_struct *current_task);
1158
1159/* sched_exec is called by processes performing an exec */
1160#ifdef CONFIG_SMP
1161extern void sched_exec(void);
1162#else
1163#define sched_exec()   {}
1164#endif
1165
1166#ifdef CONFIG_HOTPLUG_CPU
1167extern void idle_task_exit(void);
1168#else
1169static inline void idle_task_exit(void) {}
1170#endif
1171
1172extern void sched_idle_next(void);
1173
1174#ifdef CONFIG_RT_MUTEXES
1175extern int rt_mutex_getprio(struct task_struct *p);
1176extern void rt_mutex_setprio(struct task_struct *p, int prio);
1177extern void rt_mutex_adjust_pi(struct task_struct *p);
1178#else
1179static inline int rt_mutex_getprio(struct task_struct *p)
1180{
1181        return p->normal_prio;
1182}
1183# define rt_mutex_adjust_pi(p)          do { } while (0)
1184#endif
1185
1186extern void set_user_nice(struct task_struct *p, long nice);
1187extern int task_prio(const struct task_struct *p);
1188extern int task_nice(const struct task_struct *p);
1189extern int can_nice(const struct task_struct *p, const int nice);
1190extern int task_curr(const struct task_struct *p);
1191extern int idle_cpu(int cpu);
1192extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1193extern struct task_struct *idle_task(int cpu);
1194extern struct task_struct *curr_task(int cpu);
1195extern void set_curr_task(int cpu, struct task_struct *p);
1196
1197void yield(void);
1198
1199/*
1200 * The default (Linux) execution domain.
1201 */
1202extern struct exec_domain       default_exec_domain;
1203
1204union thread_union {
1205        struct thread_info thread_info;
1206        unsigned long stack[THREAD_SIZE/sizeof(long)];
1207};
1208
1209#ifndef __HAVE_ARCH_KSTACK_END
1210static inline int kstack_end(void *addr)
1211{
1212        /* Reliable end of stack detection:
1213         * Some APM bios versions misalign the stack
1214         */
1215        return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1216}
1217#endif
1218
1219extern union thread_union init_thread_union;
1220extern struct task_struct init_task;
1221
1222extern struct   mm_struct init_mm;
1223
1224#define find_task_by_pid(nr)    find_task_by_pid_type(PIDTYPE_PID, nr)
1225extern struct task_struct *find_task_by_pid_type(int type, int pid);
1226extern void set_special_pids(pid_t session, pid_t pgrp);
1227extern void __set_special_pids(pid_t session, pid_t pgrp);
1228
1229/* per-UID process charging. */
1230extern struct user_struct * alloc_uid(uid_t);
1231static inline struct user_struct *get_uid(struct user_struct *u)
1232{
1233        atomic_inc(&u->__count);
1234        return u;
1235}
1236extern void free_uid(struct user_struct *);
1237extern void switch_uid(struct user_struct *);
1238
1239#include <asm/current.h>
1240
1241extern void do_timer(unsigned long ticks);
1242
1243extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1244extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1245extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1246                                                unsigned long clone_flags));
1247#ifdef CONFIG_SMP
1248 extern void kick_process(struct task_struct *tsk);
1249#else
1250 static inline void kick_process(struct task_struct *tsk) { }
1251#endif
1252extern void FASTCALL(sched_fork(struct task_struct * p, int clone_flags));
1253extern void FASTCALL(sched_exit(struct task_struct * p));
1254
1255extern int in_group_p(gid_t);
1256extern int in_egroup_p(gid_t);
1257
1258extern void proc_caches_init(void);
1259extern void flush_signals(struct task_struct *);
1260extern void flush_signal_handlers(struct task_struct *, int force_default);
1261extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1262
1263static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1264{
1265        unsigned long flags;
1266        int ret;
1267
1268        spin_lock_irqsave(&tsk->sighand->siglock, flags);
1269        ret = dequeue_signal(tsk, mask, info);
1270        spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1271
1272        return ret;
1273}       
1274
1275extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1276                              sigset_t *mask);
1277extern void unblock_all_signals(void);
1278extern void release_task(struct task_struct * p);
1279extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1280extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1281extern int force_sigsegv(int, struct task_struct *);
1282extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1283extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1284extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1285extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1286extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1287extern int kill_pgrp(struct pid *pid, int sig, int priv);
1288extern int kill_pid(struct pid *pid, int sig, int priv);
1289extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1290extern int kill_pg_info(int, struct siginfo *, pid_t);
1291extern int kill_proc_info(int, struct siginfo *, pid_t);
1292extern void do_notify_parent(struct task_struct *, int);
1293extern void force_sig(int, struct task_struct *);
1294extern void force_sig_specific(int, struct task_struct *);
1295extern int send_sig(int, struct task_struct *, int);
1296extern void zap_other_threads(struct task_struct *p);
1297extern int kill_pg(pid_t, int, int);
1298extern int kill_proc(pid_t, int, int);
1299extern struct sigqueue *sigqueue_alloc(void);
1300extern void sigqueue_free(struct sigqueue *);
1301extern int send_sigqueue(int, struct sigqueue *,  struct task_struct *);
1302extern int send_group_sigqueue(int, struct sigqueue *,  struct task_struct *);
1303extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1304extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1305
1306static inline int kill_cad_pid(int sig, int priv)
1307{
1308        return kill_pid(cad_pid, sig, priv);
1309}
1310
1311/* These can be the second arg to send_sig_info/send_group_sig_info.  */
1312#define SEND_SIG_NOINFO ((struct siginfo *) 0)
1313#define SEND_SIG_PRIV   ((struct siginfo *) 1)
1314#define SEND_SIG_FORCED ((struct siginfo *) 2)
1315
1316static inline int is_si_special(const struct siginfo *info)
1317{
1318        return info <= SEND_SIG_FORCED;
1319}
1320
1321/* True if we are on the alternate signal stack.  */
1322
1323static inline int on_sig_stack(unsigned long sp)
1324{
1325        return (sp - current->sas_ss_sp < current->sas_ss_size);
1326}
1327
1328static inline int sas_ss_flags(unsigned long sp)
1329{
1330        return (current->sas_ss_size == 0 ? SS_DISABLE
1331                : on_sig_stack(sp) ? SS_ONSTACK : 0);
1332}
1333
1334/*
1335 * Routines for handling mm_structs
1336 */
1337extern struct mm_struct * mm_alloc(void);
1338
1339/* mmdrop drops the mm and the page tables */
1340extern void FASTCALL(__mmdrop(struct mm_struct *));
1341static inline void mmdrop(struct mm_struct * mm)
1342{
1343        if (atomic_dec_and_test(&mm->mm_count))
1344                __mmdrop(mm);
1345}
1346
1347/* mmput gets rid of the mappings and all user-space */
1348extern void mmput(struct mm_struct *);
1349/* Grab a reference to a task's mm, if it is not already going away */
1350extern struct mm_struct *get_task_mm(struct task_struct *task);
1351/* Remove the current tasks stale references to the old mm_struct */
1352extern void mm_release(struct task_struct *, struct mm_struct *);
1353
1354extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1355extern void flush_thread(void);
1356extern void exit_thread(void);
1357
1358extern void exit_files(struct task_struct *);
1359extern void __cleanup_signal(struct signal_struct *);
1360extern void __cleanup_sighand(struct sighand_struct *);
1361extern void exit_itimers(struct signal_struct *);
1362
1363extern NORET_TYPE void do_group_exit(int);
1364
1365extern void daemonize(const char *, ...);
1366extern int allow_signal(int);
1367extern int disallow_signal(int);
1368extern struct task_struct *child_reaper;
1369
1370extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1371extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1372struct task_struct *fork_idle(int);
1373
1374extern void set_task_comm(struct task_struct *tsk, char *from);
1375extern void get_task_comm(char *to, struct task_struct *tsk);
1376
1377#ifdef CONFIG_SMP
1378extern void wait_task_inactive(struct task_struct * p);
1379#else
1380#define wait_task_inactive(p)   do { } while (0)
1381#endif
1382
1383#define remove_parent(p)        list_del_init(&(p)->sibling)
1384#define add_parent(p)           list_add_tail(&(p)->sibling,&(p)->parent->children)
1385
1386#define next_task(p)    list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1387
1388#define for_each_process(p) \
1389        for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1390
1391/*
1392 * Careful: do_each_thread/while_each_thread is a double loop so
1393 *          'break' will not work as expected - use goto instead.
1394 */
1395#define do_each_thread(g, t) \
1396        for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1397
1398#define while_each_thread(g, t) \
1399        while ((t = next_thread(t)) != g)
1400
1401/* de_thread depends on thread_group_leader not being a pid based check */
1402#define thread_group_leader(p)  (p == p->group_leader)
1403
1404/* Do to the insanities of de_thread it is possible for a process
1405 * to have the pid of the thread group leader without actually being
1406 * the thread group leader.  For iteration through the pids in proc
1407 * all we care about is that we have a task with the appropriate
1408 * pid, we don't actually care if we have the right task.
1409 */
1410static inline int has_group_leader_pid(struct task_struct *p)
1411{
1412        return p->pid == p->tgid;
1413}
1414
1415static inline struct task_struct *next_thread(const struct task_struct *p)
1416{
1417        return list_entry(rcu_dereference(p->thread_group.next),
1418                          struct task_struct, thread_group);
1419}
1420
1421static inline int thread_group_empty(struct task_struct *p)
1422{
1423        return list_empty(&p->thread_group);
1424}
1425
1426#define delay_group_leader(p) \
1427                (thread_group_leader(p) && !thread_group_empty(p))
1428
1429/*
1430 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1431 * subscriptions and synchronises with wait4().  Also used in procfs.  Also
1432 * pins the final release of task.io_context.  Also protects ->cpuset.
1433 *
1434 * Nests both inside and outside of read_lock(&tasklist_lock).
1435 * It must not be nested with write_lock_irq(&tasklist_lock),
1436 * neither inside nor outside.
1437 */
1438static inline void task_lock(struct task_struct *p)
1439{
1440        spin_lock(&p->alloc_lock);
1441}
1442
1443static inline void task_unlock(struct task_struct *p)
1444{
1445        spin_unlock(&p->alloc_lock);
1446}
1447
1448extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1449                                                        unsigned long *flags);
1450
1451static inline void unlock_task_sighand(struct task_struct *tsk,
1452                                                unsigned long *flags)
1453{
1454        spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1455}
1456
1457#ifndef __HAVE_THREAD_FUNCTIONS
1458
1459#define task_thread_info(task) (task)->thread_info
1460#define task_stack_page(task) ((void*)((task)->thread_info))
1461
1462static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1463{
1464        *task_thread_info(p) = *task_thread_info(org);
1465        task_thread_info(p)->task = p;
1466}
1467
1468static inline unsigned long *end_of_stack(struct task_struct *p)
1469{
1470        return (unsigned long *)(p->thread_info + 1);
1471}
1472
1473#endif
1474
1475/* set thread flags in other task's structures
1476 * - see asm/thread_info.h for TIF_xxxx flags available
1477 */
1478static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1479{
1480        set_ti_thread_flag(task_thread_info(tsk), flag);
1481}
1482
1483static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1484{
1485        clear_ti_thread_flag(task_thread_info(tsk), flag);
1486}
1487
1488static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1489{
1490        return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1491}
1492
1493static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1494{
1495        return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1496}
1497
1498static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1499{
1500        return test_ti_thread_flag(task_thread_info(tsk), flag);
1501}
1502
1503static inline void set_tsk_need_resched(struct task_struct *tsk)
1504{
1505        set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1506}
1507
1508static inline void clear_tsk_need_resched(struct task_struct *tsk)
1509{
1510        clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1511}
1512
1513static inline int signal_pending(struct task_struct *p)
1514{
1515        return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1516}
1517  
1518static inline int need_resched(void)
1519{
1520        return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1521}
1522
1523/*
1524 * cond_resched() and cond_resched_lock(): latency reduction via
1525 * explicit rescheduling in places that are safe. The return
1526 * value indicates whether a reschedule was done in fact.
1527 * cond_resched_lock() will drop the spinlock before scheduling,
1528 * cond_resched_softirq() will enable bhs before scheduling.
1529 */
1530extern int cond_resched(void);
1531extern int cond_resched_lock(spinlock_t * lock);
1532extern int cond_resched_softirq(void);
1533
1534/*
1535 * Does a critical section need to be broken due to another
1536 * task waiting?:
1537 */
1538#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1539# define need_lockbreak(lock) ((lock)->break_lock)
1540#else
1541# define need_lockbreak(lock) 0
1542#endif
1543
1544/*
1545 * Does a critical section need to be broken due to another
1546 * task waiting or preemption being signalled:
1547 */
1548static inline int lock_need_resched(spinlock_t *lock)
1549{
1550        if (need_lockbreak(lock) || need_resched())
1551                return 1;
1552        return 0;
1553}
1554
1555/* Reevaluate whether the task has signals pending delivery.
1556   This is required every time the blocked sigset_t changes.
1557   callers must hold sighand->siglock.  */
1558
1559extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1560extern void recalc_sigpending(void);
1561
1562extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1563
1564/*
1565 * Wrappers for p->thread_info->cpu access. No-op on UP.
1566 */
1567#ifdef CONFIG_SMP
1568
1569static inline unsigned int task_cpu(const struct task_struct *p)
1570{
1571        return task_thread_info(p)->cpu;
1572}
1573
1574static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1575{
1576        task_thread_info(p)->cpu = cpu;
1577}
1578
1579#else
1580
1581static inline unsigned int task_cpu(const struct task_struct *p)
1582{
1583        return 0;
1584}
1585
1586static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1587{
1588}
1589
1590#endif /* CONFIG_SMP */
1591
1592#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1593extern void arch_pick_mmap_layout(struct mm_struct *mm);
1594#else
1595static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1596{
1597        mm->mmap_base = TASK_UNMAPPED_BASE;
1598        mm->get_unmapped_area = arch_get_unmapped_area;
1599        mm->unmap_area = arch_unmap_area;
1600}
1601#endif
1602
1603extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1604extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1605
1606#include <linux/sysdev.h>
1607extern int sched_mc_power_savings, sched_smt_power_savings;
1608extern struct sysdev_attribute attr_sched_mc_power_savings, attr_sched_smt_power_savings;
1609extern int sched_create_sysfs_power_savings_entries(struct sysdev_class *cls);
1610
1611extern void normalize_rt_tasks(void);
1612
1613#ifdef CONFIG_PM
1614/*
1615 * Check if a process has been frozen
1616 */
1617static inline int frozen(struct task_struct *p)
1618{
1619        return p->flags & PF_FROZEN;
1620}
1621
1622/*
1623 * Check if there is a request to freeze a process
1624 */
1625static inline int freezing(struct task_struct *p)
1626{
1627        return p->flags & PF_FREEZE;
1628}
1629
1630/*
1631 * Request that a process be frozen
1632 * FIXME: SMP problem. We may not modify other process' flags!
1633 */
1634static inline void freeze(struct task_struct *p)
1635{
1636        p->flags |= PF_FREEZE;
1637}
1638
1639/*
1640 * Sometimes we may need to cancel the previous 'freeze' request
1641 */
1642static inline void do_not_freeze(struct task_struct *p)
1643{
1644        p->flags &= ~PF_FREEZE;
1645}
1646
1647/*
1648 * Wake up a frozen process
1649 */
1650static inline int thaw_process(struct task_struct *p)
1651{
1652        if (frozen(p)) {
1653                p->flags &= ~PF_FROZEN;
1654                wake_up_process(p);
1655                return 1;
1656        }
1657        return 0;
1658}
1659
1660/*
1661 * freezing is complete, mark process as frozen
1662 */
1663static inline void frozen_process(struct task_struct *p)
1664{
1665        p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1666}
1667
1668extern void refrigerator(void);
1669extern int freeze_processes(void);
1670extern void thaw_processes(void);
1671
1672static inline int try_to_freeze(void)
1673{
1674        if (freezing(current)) {
1675                refrigerator();
1676                return 1;
1677        } else
1678                return 0;
1679}
1680#else
1681static inline int frozen(struct task_struct *p) { return 0; }
1682static inline int freezing(struct task_struct *p) { return 0; }
1683static inline void freeze(struct task_struct *p) { BUG(); }
1684static inline int thaw_process(struct task_struct *p) { return 1; }
1685static inline void frozen_process(struct task_struct *p) { BUG(); }
1686
1687static inline void refrigerator(void) {}
1688static inline int freeze_processes(void) { BUG(); return 0; }
1689static inline void thaw_processes(void) {}
1690
1691static inline int try_to_freeze(void) { return 0; }
1692
1693#endif /* CONFIG_PM */
1694#endif /* __KERNEL__ */
1695
1696#endif
1697
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.