include/linux/sched.h

   1 #ifndef _LINUX_SCHED_H
   2 #define _LINUX_SCHED_H
   3
   4 /*
   5  * cloning flags:
   6  */
   7 #define CSIGNAL         0x000000ff      /* signal mask to be sent at exit */
   8 #define CLONE_VM        0x00000100      /* set if VM shared between processes */
   9 #define CLONE_FS        0x00000200      /* set if fs info shared between processes */
  10 #define CLONE_FILES     0x00000400      /* set if open files shared between processes */
  11 #define CLONE_SIGHAND   0x00000800      /* set if signal handlers and blocked signals shared */
  12 #define CLONE_PTRACE    0x00002000      /* set if we want to let tracing continue on the child too */
  13 #define CLONE_VFORK     0x00004000      /* set if the parent wants the child to wake it up on mm_release */
  14 #define CLONE_PARENT    0x00008000      /* set if we want to have the same parent as the cloner */
  15 #define CLONE_THREAD    0x00010000      /* Same thread group? */
  16 #define CLONE_NEWNS     0x00020000      /* New namespace group? */
  17 #define CLONE_SYSVSEM   0x00040000      /* share system V SEM_UNDO semantics */
  18 #define CLONE_SETTLS    0x00080000      /* create a new TLS for the child */
  19 #define CLONE_PARENT_SETTID     0x00100000      /* set the TID in the parent */
  20 #define CLONE_CHILD_CLEARTID    0x00200000      /* clear the TID in the child */
  21 #define CLONE_DETACHED          0x00400000      /* Unused, ignored */
  22 #define CLONE_UNTRACED          0x00800000      /* set if the tracing process can't force CLONE_PTRACE on this clone */
  23 #define CLONE_CHILD_SETTID      0x01000000      /* set the TID in the child */
  24 #define CLONE_STOPPED           0x02000000      /* Start in stopped state */
  25 #define CLONE_NEWUTS            0x04000000      /* New utsname group? */
  26 #define CLONE_NEWIPC            0x08000000      /* New ipcs */
  27 #define CLONE_NEWUSER           0x10000000      /* New user namespace */
  28 #define CLONE_NEWPID            0x20000000      /* New pid namespace */
  29 #define CLONE_NEWNET            0x40000000      /* New network namespace */
  30 #define CLONE_IO                0x80000000      /* Clone io context */
  31
  32 /*
  33  * Scheduling policies
  34  */
  35 #define SCHED_NORMAL            0
  36 #define SCHED_FIFO              1
  37 #define SCHED_RR                2
  38 #define SCHED_BATCH             3
  39 /* SCHED_ISO: reserved but not implemented yet */
  40 #define SCHED_IDLE              5
  41
  42 #ifdef __KERNEL__
  43
  44 struct sched_param {
  45         int sched_priority;
  46 };
  47
  48 #include <asm/param.h>  /* for HZ */
  49
  50 #include <linux/capability.h>
  51 #include <linux/threads.h>
  52 #include <linux/kernel.h>
  53 #include <linux/types.h>
  54 #include <linux/timex.h>
  55 #include <linux/jiffies.h>
  56 #include <linux/rbtree.h>
  57 #include <linux/thread_info.h>
  58 #include <linux/cpumask.h>
  59 #include <linux/errno.h>
  60 #include <linux/nodemask.h>
  61 #include <linux/mm_types.h>
  62
  63 #include <asm/system.h>
  64 #include <asm/page.h>
  65 #include <asm/ptrace.h>
  66 #include <asm/cputime.h>
  67
  68 #include <linux/smp.h>
  69 #include <linux/sem.h>
  70 #include <linux/signal.h>
  71 #include <linux/fs_struct.h>
  72 #include <linux/compiler.h>
  73 #include <linux/completion.h>
  74 #include <linux/pid.h>
  75 #include <linux/percpu.h>
  76 #include <linux/topology.h>
  77 #include <linux/proportions.h>
  78 #include <linux/seccomp.h>
  79 #include <linux/rcupdate.h>
  80 #include <linux/rtmutex.h>
  81
  82 #include <linux/time.h>
  83 #include <linux/param.h>
  84 #include <linux/resource.h>
  85 #include <linux/timer.h>
  86 #include <linux/hrtimer.h>
  87 #include <linux/task_io_accounting.h>
  88 #include <linux/kobject.h>
  89 #include <linux/latencytop.h>
  90
  91 #include <asm/processor.h>
  92
  93 struct mem_cgroup;
  94 struct exec_domain;
  95 struct futex_pi_state;
  96 struct robust_list_head;
  97 struct bio;
  98
  99 /*
 100  * List of flags we want to share for kernel threads,
 101  * if only because they are not used by them anyway.
 102  */
 103 #define CLONE_KERNEL    (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
 104
 105 /*
 106  * These are the constant used to fake the fixed-point load-average
 107  * counting. Some notes:
 108  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
 109  *    a load-average precision of 10 bits integer + 11 bits fractional
 110  *  - if you want to count load-averages more often, you need more
 111  *    precision, or rounding will get you. With 2-second counting freq,
 112  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
 113  *    11 bit fractions.
 114  */
 115 extern unsigned long avenrun[];         /* Load averages */
 116
 117 #define FSHIFT          11              /* nr of bits of precision */
 118 #define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
 119 #define LOAD_FREQ       (5*HZ+1)        /* 5 sec intervals */
 120 #define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
 121 #define EXP_5           2014            /* 1/exp(5sec/5min) */
 122 #define EXP_15          2037            /* 1/exp(5sec/15min) */
 123
 124 #define CALC_LOAD(load,exp,n) \
 125         load *= exp; \
 126         load += n*(FIXED_1-exp); \
 127         load >>= FSHIFT;
 128
 129 extern unsigned long total_forks;
 130 extern int nr_threads;
 131 DECLARE_PER_CPU(unsigned long, process_counts);
 132 extern int nr_processes(void);
 133 extern unsigned long nr_running(void);
 134 extern unsigned long nr_uninterruptible(void);
 135 extern unsigned long nr_active(void);
 136 extern unsigned long nr_iowait(void);
 137 extern unsigned long weighted_cpuload(const int cpu);
 138
 139 struct seq_file;
 140 struct cfs_rq;
 141 struct task_group;
 142 #ifdef CONFIG_SCHED_DEBUG
 143 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
 144 extern void proc_sched_set_task(struct task_struct *p);
 145 extern void
 146 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
 147 #else
 148 static inline void
 149 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
 150 {
 151 }
 152 static inline void proc_sched_set_task(struct task_struct *p)
 153 {
 154 }
 155 static inline void
 156 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
 157 {
 158 }
 159 #endif
 160
 161 /*
 162  * Task state bitmask. NOTE! These bits are also
 163  * encoded in fs/proc/array.c: get_task_state().
 164  *
 165  * We have two separate sets of flags: task->state
 166  * is about runnability, while task->exit_state are
 167  * about the task exiting. Confusing, but this way
 168  * modifying one set can't modify the other one by
 169  * mistake.
 170  */
 171 #define TASK_RUNNING            0
 172 #define TASK_INTERRUPTIBLE      1
 173 #define TASK_UNINTERRUPTIBLE    2
 174 #define __TASK_STOPPED          4
 175 #define __TASK_TRACED           8
 176 /* in tsk->exit_state */
 177 #define EXIT_ZOMBIE             16
 178 #define EXIT_DEAD               32
 179 /* in tsk->state again */
 180 #define TASK_DEAD               64
 181 #define TASK_WAKEKILL           128
 182
 183 /* Convenience macros for the sake of set_task_state */
 184 #define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
 185 #define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
 186 #define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
 187
 188 /* Convenience macros for the sake of wake_up */
 189 #define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 190 #define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
 191
 192 /* get_task_state() */
 193 #define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 194                                  TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 195                                  __TASK_TRACED)
 196
 197 #define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
 198 #define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
 199 #define task_is_stopped_or_traced(task) \
 200                         ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 201 #define task_contributes_to_load(task)  \
 202                                 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
 203
 204 #define __set_task_state(tsk, state_value)              \
 205         do { (tsk)->state = (state_value); } while (0)
 206 #define set_task_state(tsk, state_value)                \
 207         set_mb((tsk)->state, (state_value))
 208
 209 /*
 210  * set_current_state() includes a barrier so that the write of current->state
 211  * is correctly serialised wrt the caller's subsequent test of whether to
 212  * actually sleep:
 213  *
 214  *      set_current_state(TASK_UNINTERRUPTIBLE);
 215  *      if (do_i_need_to_sleep())
 216  *              schedule();
 217  *
 218  * If the caller does not need such serialisation then use __set_current_state()
 219  */
 220 #define __set_current_state(state_value)                        \
 221         do { current->state = (state_value); } while (0)
 222 #define set_current_state(state_value)          \
 223         set_mb(current->state, (state_value))
 224
 225 /* Task command name length */
 226 #define TASK_COMM_LEN 16
 227
 228 #include <linux/spinlock.h>
 229
 230 /*
 231  * This serializes "schedule()" and also protects
 232  * the run-queue from deletions/modifications (but
 233  * _adding_ to the beginning of the run-queue has
 234  * a separate lock).
 235  */
 236 extern rwlock_t tasklist_lock;
 237 extern spinlock_t mmlist_lock;
 238
 239 struct task_struct;
 240
 241 extern void sched_init(void);
 242 extern void sched_init_smp(void);
 243 extern asmlinkage void schedule_tail(struct task_struct *prev);
 244 extern void init_idle(struct task_struct *idle, int cpu);
 245 extern void init_idle_bootup_task(struct task_struct *idle);
 246
 247 extern cpumask_t nohz_cpu_mask;
 248 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
 249 extern int select_nohz_load_balancer(int cpu);
 250 #else
 251 static inline int select_nohz_load_balancer(int cpu)
 252 {
 253         return 0;
 254 }
 255 #endif
 256
 257 extern unsigned long rt_needs_cpu(int cpu);
 258
 259 /*
 260  * Only dump TASK_* tasks. (0 for all tasks)
 261  */
 262 extern void show_state_filter(unsigned long state_filter);
 263
 264 static inline void show_state(void)
 265 {
 266         show_state_filter(0);
 267 }
 268
 269 extern void show_regs(struct pt_regs *);
 270
 271 /*
 272  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 273  * task), SP is the stack pointer of the first frame that should be shown in the back
 274  * trace (or NULL if the entire call-chain of the task should be shown).
 275  */
 276 extern void show_stack(struct task_struct *task, unsigned long *sp);
 277
 278 void io_schedule(void);
 279 long io_schedule_timeout(long timeout);
 280
 281 extern void cpu_init (void);
 282 extern void trap_init(void);
 283 extern void account_process_tick(struct task_struct *task, int user);
 284 extern void update_process_times(int user);
 285 extern void scheduler_tick(void);
 286 extern void hrtick_resched(void);
 287
 288 extern void sched_show_task(struct task_struct *p);
 289
 290 #ifdef CONFIG_DETECT_SOFTLOCKUP
 291 extern void softlockup_tick(void);
 292 extern void spawn_softlockup_task(void);
 293 extern void touch_softlockup_watchdog(void);
 294 extern void touch_all_softlockup_watchdogs(void);
 295 extern unsigned long  softlockup_thresh;
 296 extern unsigned long sysctl_hung_task_check_count;
 297 extern unsigned long sysctl_hung_task_timeout_secs;
 298 extern unsigned long sysctl_hung_task_warnings;
 299 #else
 300 static inline void softlockup_tick(void)
 301 {
 302 }
 303 static inline void spawn_softlockup_task(void)
 304 {
 305 }
 306 static inline void touch_softlockup_watchdog(void)
 307 {
 308 }
 309 static inline void touch_all_softlockup_watchdogs(void)
 310 {
 311 }
 312 #endif
 313
 314
 315 /* Attach to any functions which should be ignored in wchan output. */
 316 #define __sched         __attribute__((__section__(".sched.text")))
 317
 318 /* Linker adds these: start and end of __sched functions */
 319 extern char __sched_text_start[], __sched_text_end[];
 320
 321 /* Is this address in the __sched functions? */
 322 extern int in_sched_functions(unsigned long addr);
 323
 324 #define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 325 extern signed long schedule_timeout(signed long timeout);
 326 extern signed long schedule_timeout_interruptible(signed long timeout);
 327 extern signed long schedule_timeout_killable(signed long timeout);
 328 extern signed long schedule_timeout_uninterruptible(signed long timeout);
 329 asmlinkage void schedule(void);
 330
 331 struct nsproxy;
 332 struct user_namespace;
 333
 334 /* Maximum number of active map areas.. This is a random (large) number */
 335 #define DEFAULT_MAX_MAP_COUNT   65536
 336
 337 extern int sysctl_max_map_count;
 338
 339 #include <linux/aio.h>
 340
 341 extern unsigned long
 342 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 343                        unsigned long, unsigned long);
 344 extern unsigned long
 345 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 346                           unsigned long len, unsigned long pgoff,
 347                           unsigned long flags);
 348 extern void arch_unmap_area(struct mm_struct *, unsigned long);
 349 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
 350
 351 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
 352 /*
 353  * The mm counters are not protected by its page_table_lock,
 354  * so must be incremented atomically.
 355  */
 356 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
 357 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
 358 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
 359 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
 360 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
 361
 362 #else  /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
 363 /*
 364  * The mm counters are protected by its page_table_lock,
 365  * so can be incremented directly.
 366  */
 367 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
 368 #define get_mm_counter(mm, member) ((mm)->_##member)
 369 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
 370 #define inc_mm_counter(mm, member) (mm)->_##member++
 371 #define dec_mm_counter(mm, member) (mm)->_##member--
 372
 373 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
 374
 375 #define get_mm_rss(mm)                                  \
 376         (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
 377 #define update_hiwater_rss(mm)  do {                    \
 378         unsigned long _rss = get_mm_rss(mm);            \
 379         if ((mm)->hiwater_rss < _rss)                   \
 380                 (mm)->hiwater_rss = _rss;               \
 381 } while (0)
 382 #define update_hiwater_vm(mm)   do {                    \
 383         if ((mm)->hiwater_vm < (mm)->total_vm)          \
 384                 (mm)->hiwater_vm = (mm)->total_vm;      \
 385 } while (0)
 386
 387 extern void set_dumpable(struct mm_struct *mm, int value);
 388 extern int get_dumpable(struct mm_struct *mm);
 389
 390 /* mm flags */
 391 /* dumpable bits */
 392 #define MMF_DUMPABLE      0  /* core dump is permitted */
 393 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
 394 #define MMF_DUMPABLE_BITS 2
 395
 396 /* coredump filter bits */
 397 #define MMF_DUMP_ANON_PRIVATE   2
 398 #define MMF_DUMP_ANON_SHARED    3
 399 #define MMF_DUMP_MAPPED_PRIVATE 4
 400 #define MMF_DUMP_MAPPED_SHARED  5
 401 #define MMF_DUMP_ELF_HEADERS    6
 402 #define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
 403 #define MMF_DUMP_FILTER_BITS    5
 404 #define MMF_DUMP_FILTER_MASK \
 405         (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
 406 #define MMF_DUMP_FILTER_DEFAULT \
 407         ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED))
 408
 409 struct sighand_struct {
 410         atomic_t                count;
 411         struct k_sigaction      action[_NSIG];
 412         spinlock_t              siglock;
 413         wait_queue_head_t       signalfd_wqh;
 414 };
 415
 416 struct pacct_struct {
 417         int                     ac_flag;
 418         long                    ac_exitcode;
 419         unsigned long           ac_mem;
 420         cputime_t               ac_utime, ac_stime;
 421         unsigned long           ac_minflt, ac_majflt;
 422 };
 423
 424 /*
 425  * NOTE! "signal_struct" does not have it's own
 426  * locking, because a shared signal_struct always
 427  * implies a shared sighand_struct, so locking
 428  * sighand_struct is always a proper superset of
 429  * the locking of signal_struct.
 430  */
 431 struct signal_struct {
 432         atomic_t                count;
 433         atomic_t                live;
 434
 435         wait_queue_head_t       wait_chldexit;  /* for wait4() */
 436
 437         /* current thread group signal load-balancing target: */
 438         struct task_struct      *curr_target;
 439
 440         /* shared signal handling: */
 441         struct sigpending       shared_pending;
 442
 443         /* thread group exit support */
 444         int                     group_exit_code;
 445         /* overloaded:
 446          * - notify group_exit_task when ->count is equal to notify_count
 447          * - everyone except group_exit_task is stopped during signal delivery
 448          *   of fatal signals, group_exit_task processes the signal.
 449          */
 450         struct task_struct      *group_exit_task;
 451         int                     notify_count;
 452
 453         /* thread group stop support, overloads group_exit_code too */
 454         int                     group_stop_count;
 455         unsigned int            flags; /* see SIGNAL_* flags below */
 456
 457         /* POSIX.1b Interval Timers */
 458         struct list_head posix_timers;
 459
 460         /* ITIMER_REAL timer for the process */
 461         struct hrtimer real_timer;
 462         struct pid *leader_pid;
 463         ktime_t it_real_incr;
 464
 465         /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
 466         cputime_t it_prof_expires, it_virt_expires;
 467         cputime_t it_prof_incr, it_virt_incr;
 468
 469         /* job control IDs */
 470
 471         /*
 472          * pgrp and session fields are deprecated.
 473          * use the task_session_Xnr and task_pgrp_Xnr routines below
 474          */
 475
 476         union {
 477                 pid_t pgrp __deprecated;
 478                 pid_t __pgrp;
 479         };
 480
 481         struct pid *tty_old_pgrp;
 482
 483         union {
 484                 pid_t session __deprecated;
 485                 pid_t __session;
 486         };
 487
 488         /* boolean value for session group leader */
 489         int leader;
 490
 491         struct tty_struct *tty; /* NULL if no tty */
 492
 493         /*
 494          * Cumulative resource counters for dead threads in the group,
 495          * and for reaped dead child processes forked by this group.
 496          * Live threads maintain their own counters and add to these
 497          * in __exit_signal, except for the group leader.
 498          */
 499         cputime_t utime, stime, cutime, cstime;
 500         cputime_t gtime;
 501         cputime_t cgtime;
 502         unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 503         unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 504         unsigned long inblock, oublock, cinblock, coublock;
 505
 506         /*
 507          * Cumulative ns of scheduled CPU time for dead threads in the
 508          * group, not including a zombie group leader.  (This only differs
 509          * from jiffies_to_ns(utime + stime) if sched_clock uses something
 510          * other than jiffies.)
 511          */
 512         unsigned long long sum_sched_runtime;
 513
 514         /*
 515          * We don't bother to synchronize most readers of this at all,
 516          * because there is no reader checking a limit that actually needs
 517          * to get both rlim_cur and rlim_max atomically, and either one
 518          * alone is a single word that can safely be read normally.
 519          * getrlimit/setrlimit use task_lock(current->group_leader) to
 520          * protect this instead of the siglock, because they really
 521          * have no need to disable irqs.
 522          */
 523         struct rlimit rlim[RLIM_NLIMITS];
 524
 525         struct list_head cpu_timers[3];
 526
 527         /* keep the process-shared keyrings here so that they do the right
 528          * thing in threads created with CLONE_THREAD */
 529 #ifdef CONFIG_KEYS
 530         struct key *session_keyring;    /* keyring inherited over fork */
 531         struct key *process_keyring;    /* keyring private to this process */
 532 #endif
 533 #ifdef CONFIG_BSD_PROCESS_ACCT
 534         struct pacct_struct pacct;      /* per-process accounting information */
 535 #endif
 536 #ifdef CONFIG_TASKSTATS
 537         struct taskstats *stats;
 538 #endif
 539 #ifdef CONFIG_AUDIT
 540         unsigned audit_tty;
 541         struct tty_audit_buf *tty_audit_buf;
 542 #endif
 543 };
 544
 545 /* Context switch must be unlocked if interrupts are to be enabled */
 546 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
 547 # define __ARCH_WANT_UNLOCKED_CTXSW
 548 #endif
 549
 550 /*
 551  * Bits in flags field of signal_struct.
 552  */
 553 #define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 554 #define SIGNAL_STOP_DEQUEUED    0x00000002 /* stop signal dequeued */
 555 #define SIGNAL_STOP_CONTINUED   0x00000004 /* SIGCONT since WCONTINUED reap */
 556 #define SIGNAL_GROUP_EXIT       0x00000008 /* group exit in progress */
 557 /*
 558  * Pending notifications to parent.
 559  */
 560 #define SIGNAL_CLD_STOPPED      0x00000010
 561 #define SIGNAL_CLD_CONTINUED    0x00000020
 562 #define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
 563
 564 #define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
 565
 566 /* If true, all threads except ->group_exit_task have pending SIGKILL */
 567 static inline int signal_group_exit(const struct signal_struct *sig)
 568 {
 569         return  (sig->flags & SIGNAL_GROUP_EXIT) ||
 570                 (sig->group_exit_task != NULL);
 571 }
 572
 573 /*
 574  * Some day this will be a full-fledged user tracking system..
 575  */
 576 struct user_struct {
 577         atomic_t __count;       /* reference count */
 578         atomic_t processes;     /* How many processes does this user have? */
 579         atomic_t files;         /* How many open files does this user have? */
 580         atomic_t sigpending;    /* How many pending signals does this user have? */
 581 #ifdef CONFIG_INOTIFY_USER
 582         atomic_t inotify_watches; /* How many inotify watches does this user have? */
 583         atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 584 #endif
 585 #ifdef CONFIG_POSIX_MQUEUE
 586         /* protected by mq_lock */
 587         unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 588 #endif
 589         unsigned long locked_shm; /* How many pages of mlocked shm ? */
 590
 591 #ifdef CONFIG_KEYS
 592         struct key *uid_keyring;        /* UID specific keyring */
 593         struct key *session_keyring;    /* UID's default session keyring */
 594 #endif
 595
 596         /* Hash table maintenance information */
 597         struct hlist_node uidhash_node;
 598         uid_t uid;
 599
 600 #ifdef CONFIG_USER_SCHED
 601         struct task_group *tg;
 602 #ifdef CONFIG_SYSFS
 603         struct kobject kobj;
 604         struct work_struct work;
 605 #endif
 606 #endif
 607 };
 608
 609 extern int uids_sysfs_init(void);
 610
 611 extern struct user_struct *find_user(uid_t);
 612
 613 extern struct user_struct root_user;
 614 #define INIT_USER (&root_user)
 615
 616 struct backing_dev_info;
 617 struct reclaim_state;
 618
 619 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 620 struct sched_info {
 621         /* cumulative counters */
 622         unsigned long pcount;         /* # of times run on this cpu */
 623         unsigned long long cpu_time,  /* time spent on the cpu */
 624                            run_delay; /* time spent waiting on a runqueue */
 625
 626         /* timestamps */
 627         unsigned long long last_arrival,/* when we last ran on a cpu */
 628                            last_queued; /* when we were last queued to run */
 629 #ifdef CONFIG_SCHEDSTATS
 630         /* BKL stats */
 631         unsigned int bkl_count;
 632 #endif
 633 };
 634 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
 635
 636 #ifdef CONFIG_SCHEDSTATS
 637 extern const struct file_operations proc_schedstat_operations;
 638 #endif /* CONFIG_SCHEDSTATS */
 639
 640 #ifdef CONFIG_TASK_DELAY_ACCT
 641 struct task_delay_info {
 642         spinlock_t      lock;
 643         unsigned int    flags;  /* Private per-task flags */
 644
 645         /* For each stat XXX, add following, aligned appropriately
 646          *
 647          * struct timespec XXX_start, XXX_end;
 648          * u64 XXX_delay;
 649          * u32 XXX_count;
 650          *
 651          * Atomicity of updates to XXX_delay, XXX_count protected by
 652          * single lock above (split into XXX_lock if contention is an issue).
 653          */
 654
 655         /*
 656          * XXX_count is incremented on every XXX operation, the delay
 657          * associated with the operation is added to XXX_delay.
 658          * XXX_delay contains the accumulated delay time in nanoseconds.
 659          */
 660         struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
 661         u64 blkio_delay;        /* wait for sync block io completion */
 662         u64 swapin_delay;       /* wait for swapin block io completion */
 663         u32 blkio_count;        /* total count of the number of sync block */
 664                                 /* io operations performed */
 665         u32 swapin_count;       /* total count of the number of swapin block */
 666                                 /* io operations performed */
 667 };
 668 #endif  /* CONFIG_TASK_DELAY_ACCT */
 669
 670 static inline int sched_info_on(void)
 671 {
 672 #ifdef CONFIG_SCHEDSTATS
 673         return 1;
 674 #elif defined(CONFIG_TASK_DELAY_ACCT)
 675         extern int delayacct_on;
 676         return delayacct_on;
 677 #else
 678         return 0;
 679 #endif
 680 }
 681
 682 enum cpu_idle_type {
 683         CPU_IDLE,
 684         CPU_NOT_IDLE,
 685         CPU_NEWLY_IDLE,
 686         CPU_MAX_IDLE_TYPES
 687 };
 688
 689 /*
 690  * sched-domains (multiprocessor balancing) declarations:
 691  */
 692
 693 /*
 694  * Increase resolution of nice-level calculations:
 695  */
 696 #define SCHED_LOAD_SHIFT        10
 697 #define SCHED_LOAD_SCALE        (1L << SCHED_LOAD_SHIFT)
 698
 699 #define SCHED_LOAD_SCALE_FUZZ   SCHED_LOAD_SCALE
 700
 701 #ifdef CONFIG_SMP
 702 #define SD_LOAD_BALANCE         1       /* Do load balancing on this domain. */
 703 #define SD_BALANCE_NEWIDLE      2       /* Balance when about to become idle */
 704 #define SD_BALANCE_EXEC         4       /* Balance on exec */
 705 #define SD_BALANCE_FORK         8       /* Balance on fork, clone */
 706 #define SD_WAKE_IDLE            16      /* Wake to idle CPU on task wakeup */
 707 #define SD_WAKE_AFFINE          32      /* Wake task to waking CPU */
 708 #define SD_WAKE_BALANCE         64      /* Perform balancing at task wakeup */
 709 #define SD_SHARE_CPUPOWER       128     /* Domain members share cpu power */
 710 #define SD_POWERSAVINGS_BALANCE 256     /* Balance for power savings */
 711 #define SD_SHARE_PKG_RESOURCES  512     /* Domain members share cpu pkg resources */
 712 #define SD_SERIALIZE            1024    /* Only a single load balancing instance */
 713 #define SD_WAKE_IDLE_FAR        2048    /* Gain latency sacrificing cache hit */
 714
 715 #define BALANCE_FOR_MC_POWER    \
 716         (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
 717
 718 #define BALANCE_FOR_PKG_POWER   \
 719         ((sched_mc_power_savings || sched_smt_power_savings) ?  \
 720          SD_POWERSAVINGS_BALANCE : 0)
 721
 722 #define test_sd_parent(sd, flag)        ((sd->parent &&         \
 723                                          (sd->parent->flags & flag)) ? 1 : 0)
 724
 725
 726 struct sched_group {
 727         struct sched_group *next;       /* Must be a circular list */
 728         cpumask_t cpumask;
 729
 730         /*
 731          * CPU power of this group, SCHED_LOAD_SCALE being max power for a
 732          * single CPU. This is read only (except for setup, hotplug CPU).
 733          * Note : Never change cpu_power without recompute its reciprocal
 734          */
 735         unsigned int __cpu_power;
 736         /*
 737          * reciprocal value of cpu_power to avoid expensive divides
 738          * (see include/linux/reciprocal_div.h)
 739          */
 740         u32 reciprocal_cpu_power;
 741 };
 742
 743 enum sched_domain_level {
 744         SD_LV_NONE = 0,
 745         SD_LV_SIBLING,
 746         SD_LV_MC,
 747         SD_LV_CPU,
 748         SD_LV_NODE,
 749         SD_LV_ALLNODES,
 750         SD_LV_MAX
 751 };
 752
 753 struct sched_domain_attr {
 754         int relax_domain_level;
 755 };
 756
 757 #define SD_ATTR_INIT    (struct sched_domain_attr) {    \
 758         .relax_domain_level = -1,                       \
 759 }
 760
 761 struct sched_domain {
 762         /* These fields must be setup */
 763         struct sched_domain *parent;    /* top domain must be null terminated */
 764         struct sched_domain *child;     /* bottom domain must be null terminated */
 765         struct sched_group *groups;     /* the balancing groups of the domain */
 766         cpumask_t span;                 /* span of all CPUs in this domain */
 767         int first_cpu;                  /* cache of the first cpu in this domain */
 768         unsigned long min_interval;     /* Minimum balance interval ms */
 769         unsigned long max_interval;     /* Maximum balance interval ms */
 770         unsigned int busy_factor;       /* less balancing by factor if busy */
 771         unsigned int imbalance_pct;     /* No balance until over watermark */
 772         unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
 773         unsigned int busy_idx;
 774         unsigned int idle_idx;
 775         unsigned int newidle_idx;
 776         unsigned int wake_idx;
 777         unsigned int forkexec_idx;
 778         int flags;                      /* See SD_* */
 779         enum sched_domain_level level;
 780
 781         /* Runtime fields. */
 782         unsigned long last_balance;     /* init to jiffies. units in jiffies */
 783         unsigned int balance_interval;  /* initialise to 1. units in ms. */
 784         unsigned int nr_balance_failed; /* initialise to 0 */
 785
 786 #ifdef CONFIG_SCHEDSTATS
 787         /* load_balance() stats */
 788         unsigned int lb_count[CPU_MAX_IDLE_TYPES];
 789         unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
 790         unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
 791         unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
 792         unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
 793         unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
 794         unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
 795         unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
 796
 797         /* Active load balancing */
 798         unsigned int alb_count;
 799         unsigned int alb_failed;
 800         unsigned int alb_pushed;
 801
 802         /* SD_BALANCE_EXEC stats */
 803         unsigned int sbe_count;
 804         unsigned int sbe_balanced;
 805         unsigned int sbe_pushed;
 806
 807         /* SD_BALANCE_FORK stats */
 808         unsigned int sbf_count;
 809         unsigned int sbf_balanced;
 810         unsigned int sbf_pushed;
 811
 812         /* try_to_wake_up() stats */
 813         unsigned int ttwu_wake_remote;
 814         unsigned int ttwu_move_affine;
 815         unsigned int ttwu_move_balance;
 816 #endif
 817 };
 818
 819 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new,
 820                                     struct sched_domain_attr *dattr_new);
 821 extern int arch_reinit_sched_domains(void);
 822
 823 #endif  /* CONFIG_SMP */
 824
 825 /*
 826  * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
 827  * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
 828  * task of nice 0 or enough lower priority tasks to bring up the
 829  * weighted_cpuload
 830  */
 831 static inline int above_background_load(void)
 832 {
 833         unsigned long cpu;
 834
 835         for_each_online_cpu(cpu) {
 836                 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
 837                         return 1;
 838         }
 839         return 0;
 840 }
 841
 842 struct io_context;                      /* See blkdev.h */
 843 #define NGROUPS_SMALL           32
 844 #define NGROUPS_PER_BLOCK       ((unsigned int)(PAGE_SIZE / sizeof(gid_t)))
 845 struct group_info {
 846         int ngroups;
 847         atomic_t usage;
 848         gid_t small_block[NGROUPS_SMALL];
 849         int nblocks;
 850         gid_t *blocks[0];
 851 };
 852
 853 /*
 854  * get_group_info() must be called with the owning task locked (via task_lock())
 855  * when task != current.  The reason being that the vast majority of callers are
 856  * looking at current->group_info, which can not be changed except by the
 857  * current task.  Changing current->group_info requires the task lock, too.
 858  */
 859 #define get_group_info(group_info) do { \
 860         atomic_inc(&(group_info)->usage); \
 861 } while (0)
 862
 863 #define put_group_info(group_info) do { \
 864         if (atomic_dec_and_test(&(group_info)->usage)) \
 865                 groups_free(group_info); \
 866 } while (0)
 867
 868 extern struct group_info *groups_alloc(int gidsetsize);
 869 extern void groups_free(struct group_info *group_info);
 870 extern int set_current_groups(struct group_info *group_info);
 871 extern int groups_search(struct group_info *group_info, gid_t grp);
 872 /* access the groups "array" with this macro */
 873 #define GROUP_AT(gi, i) \
 874     ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
 875
 876 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
 877 extern void prefetch_stack(struct task_struct *t);
 878 #else
 879 static inline void prefetch_stack(struct task_struct *t) { }
 880 #endif
 881
 882 struct audit_context;           /* See audit.c */
 883 struct mempolicy;
 884 struct pipe_inode_info;
 885 struct uts_namespace;
 886
 887 struct rq;
 888 struct sched_domain;
 889
 890 struct sched_class {
 891         const struct sched_class *next;
 892
 893         void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
 894         void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
 895         void (*yield_task) (struct rq *rq);
 896         int  (*select_task_rq)(struct task_struct *p, int sync);
 897
 898         void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
 899
 900         struct task_struct * (*pick_next_task) (struct rq *rq);
 901         void (*put_prev_task) (struct rq *rq, struct task_struct *p);
 902
 903 #ifdef CONFIG_SMP
 904         unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
 905                         struct rq *busiest, unsigned long max_load_move,
 906                         struct sched_domain *sd, enum cpu_idle_type idle,
 907                         int *all_pinned, int *this_best_prio);
 908
 909         int (*move_one_task) (struct rq *this_rq, int this_cpu,
 910                               struct rq *busiest, struct sched_domain *sd,
 911                               enum cpu_idle_type idle);
 912         void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
 913         void (*post_schedule) (struct rq *this_rq);
 914         void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
 915 #endif
 916
 917         void (*set_curr_task) (struct rq *rq);
 918         void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
 919         void (*task_new) (struct rq *rq, struct task_struct *p);
 920         void (*set_cpus_allowed)(struct task_struct *p,
 921                                  const cpumask_t *newmask);
 922
 923         void (*join_domain)(struct rq *rq);
 924         void (*leave_domain)(struct rq *rq);
 925
 926         void (*switched_from) (struct rq *this_rq, struct task_struct *task,
 927                                int running);
 928         void (*switched_to) (struct rq *this_rq, struct task_struct *task,
 929                              int running);
 930         void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
 931                              int oldprio, int running);
 932
 933 #ifdef CONFIG_FAIR_GROUP_SCHED
 934         void (*moved_group) (struct task_struct *p);
 935 #endif
 936 };
 937
 938 struct load_weight {
 939         unsigned long weight, inv_weight;
 940 };
 941
 942 /*
 943  * CFS stats for a schedulable entity (task, task-group etc)
 944  *
 945  * Current field usage histogram:
 946  *
 947  *     4 se->block_start
 948  *     4 se->run_node
 949  *     4 se->sleep_start
 950  *     6 se->load.weight
 951  */
 952 struct sched_entity {
 953         struct load_weight      load;           /* for load-balancing */
 954         struct rb_node          run_node;
 955         struct list_head        group_node;
 956         unsigned int            on_rq;
 957
 958         u64                     exec_start;
 959         u64                     sum_exec_runtime;
 960         u64                     vruntime;
 961         u64                     prev_sum_exec_runtime;
 962
 963         u64                     last_wakeup;
 964         u64                     avg_overlap;
 965
 966 #ifdef CONFIG_SCHEDSTATS
 967         u64                     wait_start;
 968         u64                     wait_max;
 969         u64                     wait_count;
 970         u64                     wait_sum;
 971
 972         u64                     sleep_start;
 973         u64                     sleep_max;
 974         s64                     sum_sleep_runtime;
 975
 976         u64                     block_start;
 977         u64                     block_max;
 978         u64                     exec_max;
 979         u64                     slice_max;
 980
 981         u64                     nr_migrations;
 982         u64                     nr_migrations_cold;
 983         u64                     nr_failed_migrations_affine;
 984         u64                     nr_failed_migrations_running;
 985         u64                     nr_failed_migrations_hot;
 986         u64                     nr_forced_migrations;
 987         u64                     nr_forced2_migrations;
 988
 989         u64                     nr_wakeups;
 990         u64                     nr_wakeups_sync;
 991         u64                     nr_wakeups_migrate;
 992         u64                     nr_wakeups_local;
 993         u64                     nr_wakeups_remote;
 994         u64                     nr_wakeups_affine;
 995         u64                     nr_wakeups_affine_attempts;
 996         u64                     nr_wakeups_passive;
 997         u64                     nr_wakeups_idle;
 998 #endif
 999
1000 #ifdef CONFIG_FAIR_GROUP_SCHED
1001         struct sched_entity     *parent;
1002         /* rq on which this entity is (to be) queued: */
1003         struct cfs_rq           *cfs_rq;
1004         /* rq "owned" by this entity/group: */
1005         struct cfs_rq           *my_q;
1006 #endif
1007 };
1008
1009 struct sched_rt_entity {
1010         struct list_head run_list;
1011         unsigned int time_slice;
1012         unsigned long timeout;
1013         int nr_cpus_allowed;
1014
1015         struct sched_rt_entity *back;
1016 #ifdef CONFIG_RT_GROUP_SCHED
1017         struct sched_rt_entity  *parent;
1018         /* rq on which this entity is (to be) queued: */
1019         struct rt_rq            *rt_rq;
1020         /* rq "owned" by this entity/group: */
1021         struct rt_rq            *my_q;
1022 #endif
1023 };
1024
1025 struct task_struct {
1026         volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1027         void *stack;
1028         atomic_t usage;
1029         unsigned int flags;     /* per process flags, defined below */
1030         unsigned int ptrace;
1031
1032         int lock_depth;         /* BKL lock depth */
1033
1034 #ifdef CONFIG_SMP
1035 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1036         int oncpu;
1037 #endif
1038 #endif
1039
1040         int prio, static_prio, normal_prio;
1041         const struct sched_class *sched_class;
1042         struct sched_entity se;
1043         struct sched_rt_entity rt;
1044
1045 #ifdef CONFIG_PREEMPT_NOTIFIERS
1046         /* list of struct preempt_notifier: */
1047         struct hlist_head preempt_notifiers;
1048 #endif
1049
1050         /*
1051          * fpu_counter contains the number of consecutive context switches
1052          * that the FPU is used. If this is over a threshold, the lazy fpu
1053          * saving becomes unlazy to save the trap. This is an unsigned char
1054          * so that after 256 times the counter wraps and the behavior turns
1055          * lazy again; this to deal with bursty apps that only use FPU for
1056          * a short time
1057          */
1058         unsigned char fpu_counter;
1059         s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1060 #ifdef CONFIG_BLK_DEV_IO_TRACE
1061         unsigned int btrace_seq;
1062 #endif
1063
1064         unsigned int policy;
1065         cpumask_t cpus_allowed;
1066
1067 #ifdef CONFIG_PREEMPT_RCU
1068         int rcu_read_lock_nesting;
1069         int rcu_flipctr_idx;
1070 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1071
1072 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1073         struct sched_info sched_info;
1074 #endif
1075
1076         struct list_head tasks;
1077         /*
1078          * ptrace_list/ptrace_children forms the list of my children
1079          * that were stolen by a ptracer.
1080          */
1081         struct list_head ptrace_children;
1082         struct list_head ptrace_list;
1083
1084         struct mm_struct *mm, *active_mm;
1085
1086 /* task state */
1087         struct linux_binfmt *binfmt;
1088         int exit_state;
1089         int exit_code, exit_signal;
1090         int pdeath_signal;  /*  The signal sent when the parent dies  */
1091         /* ??? */
1092         unsigned int personality;
1093         unsigned did_exec:1;
1094         pid_t pid;
1095         pid_t tgid;
1096
1097 #ifdef CONFIG_CC_STACKPROTECTOR
1098         /* Canary value for the -fstack-protector gcc feature */
1099         unsigned long stack_canary;
1100 #endif
1101         /*
1102          * pointers to (original) parent process, youngest child, younger sibling,
1103          * older sibling, respectively.  (p->father can be replaced with
1104          * p->parent->pid)
1105          */
1106         struct task_struct *real_parent; /* real parent process (when being debugged) */
1107         struct task_struct *parent;     /* parent process */
1108         /*
1109          * children/sibling forms the list of my children plus the
1110          * tasks I'm ptracing.
1111          */
1112         struct list_head children;      /* list of my children */
1113         struct list_head sibling;       /* linkage in my parent's children list */
1114         struct task_struct *group_leader;       /* threadgroup leader */
1115
1116         /* PID/PID hash table linkage. */
1117         struct pid_link pids[PIDTYPE_MAX];
1118         struct list_head thread_group;
1119
1120         struct completion *vfork_done;          /* for vfork() */
1121         int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1122         int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1123
1124         unsigned int rt_priority;
1125         cputime_t utime, stime, utimescaled, stimescaled;
1126         cputime_t gtime;
1127         cputime_t prev_utime, prev_stime;
1128         unsigned long nvcsw, nivcsw; /* context switch counts */
1129         struct timespec start_time;             /* monotonic time */
1130         struct timespec real_start_time;        /* boot based time */
1131 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1132         unsigned long min_flt, maj_flt;
1133
1134         cputime_t it_prof_expires, it_virt_expires;
1135         unsigned long long it_sched_expires;
1136         struct list_head cpu_timers[3];
1137
1138 /* process credentials */
1139         uid_t uid,euid,suid,fsuid;
1140         gid_t gid,egid,sgid,fsgid;
1141         struct group_info *group_info;
1142         kernel_cap_t   cap_effective, cap_inheritable, cap_permitted, cap_bset;
1143         unsigned securebits;
1144         struct user_struct *user;
1145 #ifdef CONFIG_KEYS
1146         struct key *request_key_auth;   /* assumed request_key authority */
1147         struct key *thread_keyring;     /* keyring private to this thread */
1148         unsigned char jit_keyring;      /* default keyring to attach requested keys to */
1149 #endif
1150         char comm[TASK_COMM_LEN]; /* executable name excluding path
1151                                      - access with [gs]et_task_comm (which lock
1152                                        it with task_lock())
1153                                      - initialized normally by flush_old_exec */
1154 /* file system info */
1155         int link_count, total_link_count;
1156 #ifdef CONFIG_SYSVIPC
1157 /* ipc stuff */
1158         struct sysv_sem sysvsem;
1159 #endif
1160 #ifdef CONFIG_DETECT_SOFTLOCKUP
1161 /* hung task detection */
1162         unsigned long last_switch_timestamp;
1163         unsigned long last_switch_count;
1164 #endif
1165 /* CPU-specific state of this task */
1166         struct thread_struct thread;
1167 /* filesystem information */
1168         struct fs_struct *fs;
1169 /* open file information */
1170         struct files_struct *files;
1171 /* namespaces */
1172         struct nsproxy *nsproxy;
1173 /* signal handlers */
1174         struct signal_struct *signal;
1175         struct sighand_struct *sighand;
1176
1177         sigset_t blocked, real_blocked;
1178         sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1179         struct sigpending pending;
1180
1181         unsigned long sas_ss_sp;
1182         size_t sas_ss_size;
1183         int (*notifier)(void *priv);
1184         void *notifier_data;
1185         sigset_t *notifier_mask;
1186 #ifdef CONFIG_SECURITY
1187         void *security;
1188 #endif
1189         struct audit_context *audit_context;
1190 #ifdef CONFIG_AUDITSYSCALL
1191         uid_t loginuid;
1192         unsigned int sessionid;
1193 #endif
1194         seccomp_t seccomp;
1195
1196 /* Thread group tracking */
1197         u32 parent_exec_id;
1198         u32 self_exec_id;
1199 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1200         spinlock_t alloc_lock;
1201
1202         /* Protection of the PI data structures: */
1203         spinlock_t pi_lock;
1204
1205 #ifdef CONFIG_RT_MUTEXES
1206         /* PI waiters blocked on a rt_mutex held by this task */
1207         struct plist_head pi_waiters;
1208         /* Deadlock detection and priority inheritance handling */
1209         struct rt_mutex_waiter *pi_blocked_on;
1210 #endif
1211
1212 #ifdef CONFIG_DEBUG_MUTEXES
1213         /* mutex deadlock detection */
1214         struct mutex_waiter *blocked_on;
1215 #endif
1216 #ifdef CONFIG_TRACE_IRQFLAGS
1217         unsigned int irq_events;
1218         int hardirqs_enabled;
1219         unsigned long hardirq_enable_ip;
1220         unsigned int hardirq_enable_event;
1221         unsigned long hardirq_disable_ip;
1222         unsigned int hardirq_disable_event;
1223         int softirqs_enabled;
1224         unsigned long softirq_disable_ip;
1225         unsigned int softirq_disable_event;
1226         unsigned long softirq_enable_ip;
1227         unsigned int softirq_enable_event;
1228         int hardirq_context;
1229         int softirq_context;
1230 #endif
1231 #ifdef CONFIG_LOCKDEP
1232 # define MAX_LOCK_DEPTH 48UL
1233         u64 curr_chain_key;
1234         int lockdep_depth;
1235         struct held_lock held_locks[MAX_LOCK_DEPTH];
1236         unsigned int lockdep_recursion;
1237 #endif
1238
1239 /* journalling filesystem info */
1240         void *journal_info;
1241
1242 /* stacked block device info */
1243         struct bio *bio_list, **bio_tail;
1244
1245 /* VM state */
1246         struct reclaim_state *reclaim_state;
1247
1248         struct backing_dev_info *backing_dev_info;
1249
1250         struct io_context *io_context;
1251
1252         unsigned long ptrace_message;
1253         siginfo_t *last_siginfo; /* For ptrace use.  */
1254 #ifdef CONFIG_TASK_XACCT
1255 /* i/o counters(bytes read/written, #syscalls */
1256         u64 rchar, wchar, syscr, syscw;
1257 #endif
1258         struct task_io_accounting ioac;
1259 #if defined(CONFIG_TASK_XACCT)
1260         u64 acct_rss_mem1;      /* accumulated rss usage */
1261         u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1262         cputime_t acct_stimexpd;/* stime since last update */
1263 #endif
1264 #ifdef CONFIG_NUMA
1265         struct mempolicy *mempolicy;
1266         short il_next;
1267 #endif
1268 #ifdef CONFIG_CPUSETS
1269         nodemask_t mems_allowed;
1270         int cpuset_mems_generation;
1271         int cpuset_mem_spread_rotor;
1272 #endif
1273 #ifdef CONFIG_CGROUPS
1274         /* Control Group info protected by css_set_lock */
1275         struct css_set *cgroups;
1276         /* cg_list protected by css_set_lock and tsk->alloc_lock */
1277         struct list_head cg_list;
1278 #endif
1279 #ifdef CONFIG_FUTEX
1280         struct robust_list_head __user *robust_list;
1281 #ifdef CONFIG_COMPAT
1282         struct compat_robust_list_head __user *compat_robust_list;
1283 #endif
1284         struct list_head pi_state_list;
1285         struct futex_pi_state *pi_state_cache;
1286 #endif
1287         atomic_t fs_excl;       /* holding fs exclusive resources */
1288         struct rcu_head rcu;
1289
1290         /*
1291          * cache last used pipe for splice
1292          */
1293         struct pipe_inode_info *splice_pipe;
1294 #ifdef  CONFIG_TASK_DELAY_ACCT
1295         struct task_delay_info *delays;
1296 #endif
1297 #ifdef CONFIG_FAULT_INJECTION
1298         int make_it_fail;
1299 #endif
1300         struct prop_local_single dirties;
1301 #ifdef CONFIG_LATENCYTOP
1302         int latency_record_count;
1303         struct latency_record latency_record[LT_SAVECOUNT];
1304 #endif
1305 };
1306
1307 /*
1308  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1309  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1310  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1311  * values are inverted: lower p->prio value means higher priority.
1312  *
1313  * The MAX_USER_RT_PRIO value allows the actual maximum
1314  * RT priority to be separate from the value exported to
1315  * user-space.  This allows kernel threads to set their
1316  * priority to a value higher than any user task. Note:
1317  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1318  */
1319
1320 #define MAX_USER_RT_PRIO        100
1321 #define MAX_RT_PRIO             MAX_USER_RT_PRIO
1322
1323 #define MAX_PRIO                (MAX_RT_PRIO + 40)
1324 #define DEFAULT_PRIO            (MAX_RT_PRIO + 20)
1325
1326 static inline int rt_prio(int prio)
1327 {
1328         if (unlikely(prio < MAX_RT_PRIO))
1329                 return 1;
1330         return 0;
1331 }
1332
1333 static inline int rt_task(struct task_struct *p)
1334 {
1335         return rt_prio(p->prio);
1336 }
1337
1338 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1339 {
1340         tsk->signal->__session = session;
1341 }
1342
1343 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1344 {
1345         tsk->signal->__pgrp = pgrp;
1346 }
1347
1348 static inline struct pid *task_pid(struct task_struct *task)
1349 {
1350         return task->pids[PIDTYPE_PID].pid;
1351 }
1352
1353 static inline struct pid *task_tgid(struct task_struct *task)
1354 {
1355         return task->group_leader->pids[PIDTYPE_PID].pid;
1356 }
1357
1358 static inline struct pid *task_pgrp(struct task_struct *task)
1359 {
1360         return task->group_leader->pids[PIDTYPE_PGID].pid;
1361 }
1362
1363 static inline struct pid *task_session(struct task_struct *task)
1364 {
1365         return task->group_leader->pids[PIDTYPE_SID].pid;
1366 }
1367
1368 struct pid_namespace;
1369
1370 /*
1371  * the helpers to get the task's different pids as they are seen
1372  * from various namespaces
1373  *
1374  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1375  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1376  *                     current.
1377  * task_xid_nr_ns()  : id seen from the ns specified;
1378  *
1379  * set_task_vxid()   : assigns a virtual id to a task;
1380  *
1381  * see also pid_nr() etc in include/linux/pid.h
1382  */
1383
1384 static inline pid_t task_pid_nr(struct task_struct *tsk)
1385 {
1386         return tsk->pid;
1387 }
1388
1389 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1390
1391 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1392 {
1393         return pid_vnr(task_pid(tsk));
1394 }
1395
1396
1397 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1398 {
1399         return tsk->tgid;
1400 }
1401
1402 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1403
1404 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1405 {
1406         return pid_vnr(task_tgid(tsk));
1407 }
1408
1409
1410 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1411 {
1412         return tsk->signal->__pgrp;
1413 }
1414
1415 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1416
1417 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1418 {
1419         return pid_vnr(task_pgrp(tsk));
1420 }
1421
1422
1423 static inline pid_t task_session_nr(struct task_struct *tsk)
1424 {
1425         return tsk->signal->__session;
1426 }
1427
1428 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1429
1430 static inline pid_t task_session_vnr(struct task_struct *tsk)
1431 {
1432         return pid_vnr(task_session(tsk));
1433 }
1434
1435
1436 /**
1437  * pid_alive - check that a task structure is not stale
1438  * @p: Task structure to be checked.
1439  *
1440  * Test if a process is not yet dead (at most zombie state)
1441  * If pid_alive fails, then pointers within the task structure
1442  * can be stale and must not be dereferenced.
1443  */
1444 static inline int pid_alive(struct task_struct *p)
1445 {
1446         return p->pids[PIDTYPE_PID].pid != NULL;
1447 }
1448
1449 /**
1450  * is_global_init - check if a task structure is init
1451  * @tsk: Task structure to be checked.
1452  *
1453  * Check if a task structure is the first user space task the kernel created.
1454  */
1455 static inline int is_global_init(struct task_struct *tsk)
1456 {
1457         return tsk->pid == 1;
1458 }
1459
1460 /*
1461  * is_container_init:
1462  * check whether in the task is init in its own pid namespace.
1463  */
1464 extern int is_container_init(struct task_struct *tsk);
1465
1466 extern struct pid *cad_pid;
1467
1468 extern void free_task(struct task_struct *tsk);
1469 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1470
1471 extern void __put_task_struct(struct task_struct *t);
1472
1473 static inline void put_task_struct(struct task_struct *t)
1474 {
1475         if (atomic_dec_and_test(&t->usage))
1476                 __put_task_struct(t);
1477 }
1478
1479 /*
1480  * Per process flags
1481  */
1482 #define PF_ALIGNWARN    0x00000001      /* Print alignment warning msgs */
1483                                         /* Not implemented yet, only for 486*/
1484 #define PF_STARTING     0x00000002      /* being created */
1485 #define PF_EXITING      0x00000004      /* getting shut down */
1486 #define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
1487 #define PF_VCPU         0x00000010      /* I'm a virtual CPU */
1488 #define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
1489 #define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
1490 #define PF_DUMPCORE     0x00000200      /* dumped core */
1491 #define PF_SIGNALED     0x00000400      /* killed by a signal */
1492 #define PF_MEMALLOC     0x00000800      /* Allocating memory */
1493 #define PF_FLUSHER      0x00001000      /* responsible for disk writeback */
1494 #define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
1495 #define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
1496 #define PF_FROZEN       0x00010000      /* frozen for system suspend */
1497 #define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
1498 #define PF_KSWAPD       0x00040000      /* I am kswapd */
1499 #define PF_SWAPOFF      0x00080000      /* I am in swapoff */
1500 #define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
1501 #define PF_BORROWED_MM  0x00200000      /* I am a kthread doing use_mm */
1502 #define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
1503 #define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
1504 #define PF_SPREAD_PAGE  0x01000000      /* Spread page cache over cpuset */
1505 #define PF_SPREAD_SLAB  0x02000000      /* Spread some slab caches over cpuset */
1506 #define PF_MEMPOLICY    0x10000000      /* Non-default NUMA mempolicy */
1507 #define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
1508 #define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezeable */
1509
1510 /*
1511  * Only the _current_ task can read/write to tsk->flags, but other
1512  * tasks can access tsk->flags in readonly mode for example
1513  * with tsk_used_math (like during threaded core dumping).
1514  * There is however an exception to this rule during ptrace
1515  * or during fork: the ptracer task is allowed to write to the
1516  * child->flags of its traced child (same goes for fork, the parent
1517  * can write to the child->flags), because we're guaranteed the
1518  * child is not running and in turn not changing child->flags
1519  * at the same time the parent does it.
1520  */
1521 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1522 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1523 #define clear_used_math() clear_stopped_child_used_math(current)
1524 #define set_used_math() set_stopped_child_used_math(current)
1525 #define conditional_stopped_child_used_math(condition, child) \
1526         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1527 #define conditional_used_math(condition) \
1528         conditional_stopped_child_used_math(condition, current)
1529 #define copy_to_stopped_child_used_math(child) \
1530         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1531 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1532 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1533 #define used_math() tsk_used_math(current)
1534
1535 #ifdef CONFIG_SMP
1536 extern int set_cpus_allowed_ptr(struct task_struct *p,
1537                                 const cpumask_t *new_mask);
1538 #else
1539 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1540                                        const cpumask_t *new_mask)
1541 {
1542         if (!cpu_isset(0, *new_mask))
1543                 return -EINVAL;
1544         return 0;
1545 }
1546 #endif
1547 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1548 {
1549         return set_cpus_allowed_ptr(p, &new_mask);
1550 }
1551
1552 extern unsigned long long sched_clock(void);
1553
1554 /*
1555  * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1556  * clock constructed from sched_clock():
1557  */
1558 extern unsigned long long cpu_clock(int cpu);
1559
1560 extern unsigned long long
1561 task_sched_runtime(struct task_struct *task);
1562
1563 /* sched_exec is called by processes performing an exec */
1564 #ifdef CONFIG_SMP
1565 extern void sched_exec(void);
1566 #else
1567 #define sched_exec()   {}
1568 #endif
1569
1570 extern void sched_clock_idle_sleep_event(void);
1571 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1572
1573 #ifdef CONFIG_HOTPLUG_CPU
1574 extern void idle_task_exit(void);
1575 #else
1576 static inline void idle_task_exit(void) {}
1577 #endif
1578
1579 extern void sched_idle_next(void);
1580
1581 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1582 extern void wake_up_idle_cpu(int cpu);
1583 #else
1584 static inline void wake_up_idle_cpu(int cpu) { }
1585 #endif
1586
1587 #ifdef CONFIG_SCHED_DEBUG
1588 extern unsigned int sysctl_sched_latency;
1589 extern unsigned int sysctl_sched_min_granularity;
1590 extern unsigned int sysctl_sched_wakeup_granularity;
1591 extern unsigned int sysctl_sched_child_runs_first;
1592 extern unsigned int sysctl_sched_features;
1593 extern unsigned int sysctl_sched_migration_cost;
1594 extern unsigned int sysctl_sched_nr_migrate;
1595
1596 int sched_nr_latency_handler(struct ctl_table *table, int write,
1597                 struct file *file, void __user *buffer, size_t *length,
1598                 loff_t *ppos);
1599 #endif
1600 extern unsigned int sysctl_sched_rt_period;
1601 extern int sysctl_sched_rt_runtime;
1602
1603 int sched_rt_handler(struct ctl_table *table, int write,
1604                 struct file *filp, void __user *buffer, size_t *lenp,
1605                 loff_t *ppos);
1606
1607 extern unsigned int sysctl_sched_compat_yield;
1608
1609 #ifdef CONFIG_RT_MUTEXES
1610 extern int rt_mutex_getprio(struct task_struct *p);
1611 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1612 extern void rt_mutex_adjust_pi(struct task_struct *p);
1613 #else
1614 static inline int rt_mutex_getprio(struct task_struct *p)
1615 {
1616         return p->normal_prio;
1617 }
1618 # define rt_mutex_adjust_pi(p)          do { } while (0)
1619 #endif
1620
1621 extern void set_user_nice(struct task_struct *p, long nice);
1622 extern int task_prio(const struct task_struct *p);
1623 extern int task_nice(const struct task_struct *p);
1624 extern int can_nice(const struct task_struct *p, const int nice);
1625 extern int task_curr(const struct task_struct *p);
1626 extern int idle_cpu(int cpu);
1627 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1628 extern struct task_struct *idle_task(int cpu);
1629 extern struct task_struct *curr_task(int cpu);
1630 extern void set_curr_task(int cpu, struct task_struct *p);
1631
1632 void yield(void);
1633
1634 /*
1635  * The default (Linux) execution domain.
1636  */
1637 extern struct exec_domain       default_exec_domain;
1638
1639 union thread_union {
1640         struct thread_info thread_info;
1641         unsigned long stack[THREAD_SIZE/sizeof(long)];
1642 };
1643
1644 #ifndef __HAVE_ARCH_KSTACK_END
1645 static inline int kstack_end(void *addr)
1646 {
1647         /* Reliable end of stack detection:
1648          * Some APM bios versions misalign the stack
1649          */
1650         return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1651 }
1652 #endif
1653
1654 extern union thread_union init_thread_union;
1655 extern struct task_struct init_task;
1656
1657 extern struct   mm_struct init_mm;
1658
1659 extern struct pid_namespace init_pid_ns;
1660
1661 /*
1662  * find a task by one of its numerical ids
1663  *
1664  * find_task_by_pid_type_ns():
1665  *      it is the most generic call - it finds a task by all id,
1666  *      type and namespace specified
1667  * find_task_by_pid_ns():
1668  *      finds a task by its pid in the specified namespace
1669  * find_task_by_vpid():
1670  *      finds a task by its virtual pid
1671  * find_task_by_pid():
1672  *      finds a task by its global pid
1673  *
1674  * see also find_pid() etc in include/linux/pid.h
1675  */
1676
1677 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1678                 struct pid_namespace *ns);
1679
1680 static inline struct task_struct *__deprecated find_task_by_pid(pid_t nr)
1681 {
1682         return find_task_by_pid_type_ns(PIDTYPE_PID, nr, &init_pid_ns);
1683 }
1684 extern struct task_struct *find_task_by_vpid(pid_t nr);
1685 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1686                 struct pid_namespace *ns);
1687
1688 extern void __set_special_pids(struct pid *pid);
1689
1690 /* per-UID process charging. */
1691 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1692 static inline struct user_struct *get_uid(struct user_struct *u)
1693 {
1694         atomic_inc(&u->__count);
1695         return u;
1696 }
1697 extern void free_uid(struct user_struct *);
1698 extern void switch_uid(struct user_struct *);
1699 extern void release_uids(struct user_namespace *ns);
1700
1701 #include <asm/current.h>
1702
1703 extern void do_timer(unsigned long ticks);
1704
1705 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1706 extern int wake_up_process(struct task_struct *tsk);
1707 extern void wake_up_new_task(struct task_struct *tsk,
1708                                 unsigned long clone_flags);
1709 #ifdef CONFIG_SMP
1710  extern void kick_process(struct task_struct *tsk);
1711 #else
1712  static inline void kick_process(struct task_struct *tsk) { }
1713 #endif
1714 extern void sched_fork(struct task_struct *p, int clone_flags);
1715 extern void sched_dead(struct task_struct *p);
1716
1717 extern int in_group_p(gid_t);
1718 extern int in_egroup_p(gid_t);
1719
1720 extern void proc_caches_init(void);
1721 extern void flush_signals(struct task_struct *);
1722 extern void ignore_signals(struct task_struct *);
1723 extern void flush_signal_handlers(struct task_struct *, int force_default);
1724 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1725
1726 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1727 {
1728         unsigned long flags;
1729         int ret;
1730
1731         spin_lock_irqsave(&tsk->sighand->siglock, flags);
1732         ret = dequeue_signal(tsk, mask, info);
1733         spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1734
1735         return ret;
1736 }
1737
1738 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1739                               sigset_t *mask);
1740 extern void unblock_all_signals(void);
1741 extern void release_task(struct task_struct * p);
1742 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1743 extern int force_sigsegv(int, struct task_struct *);
1744 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1745 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1746 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1747 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1748 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1749 extern int kill_pid(struct pid *pid, int sig, int priv);
1750 extern int kill_proc_info(int, struct siginfo *, pid_t);
1751 extern void do_notify_parent(struct task_struct *, int);
1752 extern void force_sig(int, struct task_struct *);
1753 extern void force_sig_specific(int, struct task_struct *);
1754 extern int send_sig(int, struct task_struct *, int);
1755 extern void zap_other_threads(struct task_struct *p);
1756 extern int kill_proc(pid_t, int, int);
1757 extern struct sigqueue *sigqueue_alloc(void);
1758 extern void sigqueue_free(struct sigqueue *);
1759 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
1760 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1761 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1762
1763 static inline int kill_cad_pid(int sig, int priv)
1764 {
1765         return kill_pid(cad_pid, sig, priv);
1766 }
1767
1768 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1769 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1770 #define SEND_SIG_PRIV   ((struct siginfo *) 1)
1771 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1772
1773 static inline int is_si_special(const struct siginfo *info)
1774 {
1775         return info <= SEND_SIG_FORCED;
1776 }
1777
1778 /* True if we are on the alternate signal stack.  */
1779
1780 static inline int on_sig_stack(unsigned long sp)
1781 {
1782         return (sp - current->sas_ss_sp < current->sas_ss_size);
1783 }
1784
1785 static inline int sas_ss_flags(unsigned long sp)
1786 {
1787         return (current->sas_ss_size == 0 ? SS_DISABLE
1788                 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1789 }
1790
1791 /*
1792  * Routines for handling mm_structs
1793  */
1794 extern struct mm_struct * mm_alloc(void);
1795
1796 /* mmdrop drops the mm and the page tables */
1797 extern void __mmdrop(struct mm_struct *);
1798 static inline void mmdrop(struct mm_struct * mm)
1799 {
1800         if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1801                 __mmdrop(mm);
1802 }
1803
1804 /* mmput gets rid of the mappings and all user-space */
1805 extern void mmput(struct mm_struct *);
1806 /* Grab a reference to a task's mm, if it is not already going away */
1807 extern struct mm_struct *get_task_mm(struct task_struct *task);
1808 /* Remove the current tasks stale references to the old mm_struct */
1809 extern void mm_release(struct task_struct *, struct mm_struct *);
1810 /* Allocate a new mm structure and copy contents from tsk->mm */
1811 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1812
1813 extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1814 extern void flush_thread(void);
1815 extern void exit_thread(void);
1816
1817 extern void exit_files(struct task_struct *);
1818 extern void __cleanup_signal(struct signal_struct *);
1819 extern void __cleanup_sighand(struct sighand_struct *);
1820 extern void exit_itimers(struct signal_struct *);
1821
1822 extern NORET_TYPE void do_group_exit(int);
1823
1824 extern void daemonize(const char *, ...);
1825 extern int allow_signal(int);
1826 extern int disallow_signal(int);
1827
1828 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1829 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1830 struct task_struct *fork_idle(int);
1831
1832 extern void set_task_comm(struct task_struct *tsk, char *from);
1833 extern char *get_task_comm(char *to, struct task_struct *tsk);
1834
1835 #ifdef CONFIG_SMP
1836 extern void wait_task_inactive(struct task_struct * p);
1837 #else
1838 #define wait_task_inactive(p)   do { } while (0)
1839 #endif
1840
1841 #define remove_parent(p)        list_del_init(&(p)->sibling)
1842 #define add_parent(p)           list_add_tail(&(p)->sibling,&(p)->parent->children)
1843
1844 #define next_task(p)    list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1845
1846 #define for_each_process(p) \
1847         for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1848
1849 /*
1850  * Careful: do_each_thread/while_each_thread is a double loop so
1851  *          'break' will not work as expected - use goto instead.
1852  */
1853 #define do_each_thread(g, t) \
1854         for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1855
1856 #define while_each_thread(g, t) \
1857         while ((t = next_thread(t)) != g)
1858
1859 /* de_thread depends on thread_group_leader not being a pid based check */
1860 #define thread_group_leader(p)  (p == p->group_leader)
1861
1862 /* Do to the insanities of de_thread it is possible for a process
1863  * to have the pid of the thread group leader without actually being
1864  * the thread group leader.  For iteration through the pids in proc
1865  * all we care about is that we have a task with the appropriate
1866  * pid, we don't actually care if we have the right task.
1867  */
1868 static inline int has_group_leader_pid(struct task_struct *p)
1869 {
1870         return p->pid == p->tgid;
1871 }
1872
1873 static inline
1874 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1875 {
1876         return p1->tgid == p2->tgid;
1877 }
1878
1879 static inline struct task_struct *next_thread(const struct task_struct *p)
1880 {
1881         return list_entry(rcu_dereference(p->thread_group.next),
1882                           struct task_struct, thread_group);
1883 }
1884
1885 static inline int thread_group_empty(struct task_struct *p)
1886 {
1887         return list_empty(&p->thread_group);
1888 }
1889
1890 #define delay_group_leader(p) \
1891                 (thread_group_leader(p) && !thread_group_empty(p))
1892
1893 /*
1894  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1895  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
1896  * pins the final release of task.io_context.  Also protects ->cpuset and
1897  * ->cgroup.subsys[].
1898  *
1899  * Nests both inside and outside of read_lock(&tasklist_lock).
1900  * It must not be nested with write_lock_irq(&tasklist_lock),
1901  * neither inside nor outside.
1902  */
1903 static inline void task_lock(struct task_struct *p)
1904 {
1905         spin_lock(&p->alloc_lock);
1906 }
1907
1908 static inline void task_unlock(struct task_struct *p)
1909 {
1910         spin_unlock(&p->alloc_lock);
1911 }
1912
1913 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1914                                                         unsigned long *flags);
1915
1916 static inline void unlock_task_sighand(struct task_struct *tsk,
1917                                                 unsigned long *flags)
1918 {
1919         spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1920 }
1921
1922 #ifndef __HAVE_THREAD_FUNCTIONS
1923
1924 #define task_thread_info(task)  ((struct thread_info *)(task)->stack)
1925 #define task_stack_page(task)   ((task)->stack)
1926
1927 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1928 {
1929         *task_thread_info(p) = *task_thread_info(org);
1930         task_thread_info(p)->task = p;
1931 }
1932
1933 static inline unsigned long *end_of_stack(struct task_struct *p)
1934 {
1935         return (unsigned long *)(task_thread_info(p) + 1);
1936 }
1937
1938 #endif
1939
1940 extern void thread_info_cache_init(void);
1941
1942 /* set thread flags in other task's structures
1943  * - see asm/thread_info.h for TIF_xxxx flags available
1944  */
1945 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1946 {
1947         set_ti_thread_flag(task_thread_info(tsk), flag);
1948 }
1949
1950 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1951 {
1952         clear_ti_thread_flag(task_thread_info(tsk), flag);
1953 }
1954
1955 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1956 {
1957         return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1958 }
1959
1960 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1961 {
1962         return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1963 }
1964
1965 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1966 {
1967         return test_ti_thread_flag(task_thread_info(tsk), flag);
1968 }
1969
1970 static inline void set_tsk_need_resched(struct task_struct *tsk)
1971 {
1972         set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1973 }
1974
1975 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1976 {
1977         clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1978 }
1979
1980 static inline int signal_pending(struct task_struct *p)
1981 {
1982         return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1983 }
1984
1985 extern int __fatal_signal_pending(struct task_struct *p);
1986
1987 static inline int fatal_signal_pending(struct task_struct *p)
1988 {
1989         return signal_pending(p) && __fatal_signal_pending(p);
1990 }
1991
1992 static inline int need_resched(void)
1993 {
1994         return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1995 }
1996
1997 /*
1998  * cond_resched() and cond_resched_lock(): latency reduction via
1999  * explicit rescheduling in places that are safe. The return
2000  * value indicates whether a reschedule was done in fact.
2001  * cond_resched_lock() will drop the spinlock before scheduling,
2002  * cond_resched_softirq() will enable bhs before scheduling.
2003  */
2004 #ifdef CONFIG_PREEMPT
2005 static inline int cond_resched(void)
2006 {
2007         return 0;
2008 }
2009 #else
2010 extern int _cond_resched(void);
2011 static inline int cond_resched(void)
2012 {
2013         return _cond_resched();
2014 }
2015 #endif
2016 extern int cond_resched_lock(spinlock_t * lock);
2017 extern int cond_resched_softirq(void);
2018
2019 /*
2020  * Does a critical section need to be broken due to another
2021  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2022  * but a general need for low latency)
2023  */
2024 static inline int spin_needbreak(spinlock_t *lock)
2025 {
2026 #ifdef CONFIG_PREEMPT
2027         return spin_is_contended(lock);
2028 #else
2029         return 0;
2030 #endif
2031 }
2032
2033 /*
2034  * Reevaluate whether the task has signals pending delivery.
2035  * Wake the task if so.
2036  * This is required every time the blocked sigset_t changes.
2037  * callers must hold sighand->siglock.
2038  */
2039 extern void recalc_sigpending_and_wake(struct task_struct *t);
2040 extern void recalc_sigpending(void);
2041
2042 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2043
2044 /*
2045  * Wrappers for p->thread_info->cpu access. No-op on UP.
2046  */
2047 #ifdef CONFIG_SMP
2048
2049 static inline unsigned int task_cpu(const struct task_struct *p)
2050 {
2051         return task_thread_info(p)->cpu;
2052 }
2053
2054 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2055
2056 #else
2057
2058 static inline unsigned int task_cpu(const struct task_struct *p)
2059 {
2060         return 0;
2061 }
2062
2063 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2064 {
2065 }
2066
2067 #endif /* CONFIG_SMP */
2068
2069 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2070 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2071 #else
2072 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
2073 {
2074         mm->mmap_base = TASK_UNMAPPED_BASE;
2075         mm->get_unmapped_area = arch_get_unmapped_area;
2076         mm->unmap_area = arch_unmap_area;
2077 }
2078 #endif
2079
2080 extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask);
2081 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2082
2083 extern int sched_mc_power_savings, sched_smt_power_savings;
2084
2085 extern void normalize_rt_tasks(void);
2086
2087 #ifdef CONFIG_GROUP_SCHED
2088
2089 extern struct task_group init_task_group;
2090 #ifdef CONFIG_USER_SCHED
2091 extern struct task_group root_task_group;
2092 #endif
2093
2094 extern struct task_group *sched_create_group(struct task_group *parent);
2095 extern void sched_destroy_group(struct task_group *tg);
2096 extern void sched_move_task(struct task_struct *tsk);
2097 #ifdef CONFIG_FAIR_GROUP_SCHED
2098 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2099 extern unsigned long sched_group_shares(struct task_group *tg);
2100 #endif
2101 #ifdef CONFIG_RT_GROUP_SCHED
2102 extern int sched_group_set_rt_runtime(struct task_group *tg,
2103                                       long rt_runtime_us);
2104 extern long sched_group_rt_runtime(struct task_group *tg);
2105 extern int sched_group_set_rt_period(struct task_group *tg,
2106                                       long rt_period_us);
2107 extern long sched_group_rt_period(struct task_group *tg);
2108 #endif
2109 #endif
2110
2111 #ifdef CONFIG_TASK_XACCT
2112 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2113 {
2114         tsk->rchar += amt;
2115 }
2116
2117 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2118 {
2119         tsk->wchar += amt;
2120 }
2121
2122 static inline void inc_syscr(struct task_struct *tsk)
2123 {
2124         tsk->syscr++;
2125 }
2126
2127 static inline void inc_syscw(struct task_struct *tsk)
2128 {
2129         tsk->syscw++;
2130 }
2131 #else
2132 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2133 {
2134 }
2135
2136 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2137 {
2138 }
2139
2140 static inline void inc_syscr(struct task_struct *tsk)
2141 {
2142 }
2143
2144 static inline void inc_syscw(struct task_struct *tsk)
2145 {
2146 }
2147 #endif
2148
2149 #ifdef CONFIG_SMP
2150 void migration_init(void);
2151 #else
2152 static inline void migration_init(void)
2153 {
2154 }
2155 #endif
2156
2157 #ifndef TASK_SIZE_OF
2158 #define TASK_SIZE_OF(tsk)       TASK_SIZE
2159 #endif
2160
2161 #ifdef CONFIG_MM_OWNER
2162 extern void mm_update_next_owner(struct mm_struct *mm);
2163 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2164 #else
2165 static inline void mm_update_next_owner(struct mm_struct *mm)
2166 {
2167 }
2168
2169 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2170 {
2171 }
2172 #endif /* CONFIG_MM_OWNER */
2173
2174 #endif /* __KERNEL__ */
2175
2176 #endif