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/perf_counter.h>
  75 #include <linux/pid.h>
  76 #include <linux/percpu.h>
  77 #include <linux/topology.h>
  78 #include <linux/proportions.h>
  79 #include <linux/seccomp.h>
  80 #include <linux/rcupdate.h>
  81 #include <linux/rtmutex.h>
  82
  83 #include <linux/time.h>
  84 #include <linux/param.h>
  85 #include <linux/resource.h>
  86 #include <linux/timer.h>
  87 #include <linux/hrtimer.h>
  88 #include <linux/task_io_accounting.h>
  89 #include <linux/kobject.h>
  90 #include <linux/latencytop.h>
  91 #include <linux/cred.h>
  92
  93 #include <asm/processor.h>
  94
  95 struct mem_cgroup;
  96 struct exec_domain;
  97 struct futex_pi_state;
  98 struct robust_list_head;
  99 struct bio;
 100 struct bts_tracer;
 101
 102 /*
 103  * List of flags we want to share for kernel threads,
 104  * if only because they are not used by them anyway.
 105  */
 106 #define CLONE_KERNEL    (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
 107
 108 /*
 109  * These are the constant used to fake the fixed-point load-average
 110  * counting. Some notes:
 111  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
 112  *    a load-average precision of 10 bits integer + 11 bits fractional
 113  *  - if you want to count load-averages more often, you need more
 114  *    precision, or rounding will get you. With 2-second counting freq,
 115  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
 116  *    11 bit fractions.
 117  */
 118 extern unsigned long avenrun[];         /* Load averages */
 119
 120 #define FSHIFT          11              /* nr of bits of precision */
 121 #define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
 122 #define LOAD_FREQ       (5*HZ+1)        /* 5 sec intervals */
 123 #define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
 124 #define EXP_5           2014            /* 1/exp(5sec/5min) */
 125 #define EXP_15          2037            /* 1/exp(5sec/15min) */
 126
 127 #define CALC_LOAD(load,exp,n) \
 128         load *= exp; \
 129         load += n*(FIXED_1-exp); \
 130         load >>= FSHIFT;
 131
 132 extern unsigned long total_forks;
 133 extern int nr_threads;
 134 DECLARE_PER_CPU(unsigned long, process_counts);
 135 extern int nr_processes(void);
 136 extern unsigned long nr_running(void);
 137 extern unsigned long nr_uninterruptible(void);
 138 extern unsigned long nr_active(void);
 139 extern unsigned long nr_iowait(void);
 140
 141 struct seq_file;
 142 struct cfs_rq;
 143 struct task_group;
 144 #ifdef CONFIG_SCHED_DEBUG
 145 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
 146 extern void proc_sched_set_task(struct task_struct *p);
 147 extern void
 148 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
 149 #else
 150 static inline void
 151 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
 152 {
 153 }
 154 static inline void proc_sched_set_task(struct task_struct *p)
 155 {
 156 }
 157 static inline void
 158 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
 159 {
 160 }
 161 #endif
 162
 163 extern unsigned long long time_sync_thresh;
 164
 165 /*
 166  * Task state bitmask. NOTE! These bits are also
 167  * encoded in fs/proc/array.c: get_task_state().
 168  *
 169  * We have two separate sets of flags: task->state
 170  * is about runnability, while task->exit_state are
 171  * about the task exiting. Confusing, but this way
 172  * modifying one set can't modify the other one by
 173  * mistake.
 174  */
 175 #define TASK_RUNNING            0
 176 #define TASK_INTERRUPTIBLE      1
 177 #define TASK_UNINTERRUPTIBLE    2
 178 #define __TASK_STOPPED          4
 179 #define __TASK_TRACED           8
 180 /* in tsk->exit_state */
 181 #define EXIT_ZOMBIE             16
 182 #define EXIT_DEAD               32
 183 /* in tsk->state again */
 184 #define TASK_DEAD               64
 185 #define TASK_WAKEKILL           128
 186
 187 /* Convenience macros for the sake of set_task_state */
 188 #define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
 189 #define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
 190 #define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
 191
 192 /* Convenience macros for the sake of wake_up */
 193 #define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 194 #define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
 195
 196 /* get_task_state() */
 197 #define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 198                                  TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 199                                  __TASK_TRACED)
 200
 201 #define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
 202 #define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
 203 #define task_is_stopped_or_traced(task) \
 204                         ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 205 #define task_contributes_to_load(task)  \
 206                                 ((task->state & TASK_UNINTERRUPTIBLE) != 0)
 207
 208 #define __set_task_state(tsk, state_value)              \
 209         do { (tsk)->state = (state_value); } while (0)
 210 #define set_task_state(tsk, state_value)                \
 211         set_mb((tsk)->state, (state_value))
 212
 213 /*
 214  * set_current_state() includes a barrier so that the write of current->state
 215  * is correctly serialised wrt the caller's subsequent test of whether to
 216  * actually sleep:
 217  *
 218  *      set_current_state(TASK_UNINTERRUPTIBLE);
 219  *      if (do_i_need_to_sleep())
 220  *              schedule();
 221  *
 222  * If the caller does not need such serialisation then use __set_current_state()
 223  */
 224 #define __set_current_state(state_value)                        \
 225         do { current->state = (state_value); } while (0)
 226 #define set_current_state(state_value)          \
 227         set_mb(current->state, (state_value))
 228
 229 /* Task command name length */
 230 #define TASK_COMM_LEN 16
 231
 232 #include <linux/spinlock.h>
 233
 234 /*
 235  * This serializes "schedule()" and also protects
 236  * the run-queue from deletions/modifications (but
 237  * _adding_ to the beginning of the run-queue has
 238  * a separate lock).
 239  */
 240 extern rwlock_t tasklist_lock;
 241 extern spinlock_t mmlist_lock;
 242
 243 struct task_struct;
 244
 245 extern void sched_init(void);
 246 extern void sched_init_smp(void);
 247 extern asmlinkage void schedule_tail(struct task_struct *prev);
 248 extern void init_idle(struct task_struct *idle, int cpu);
 249 extern void init_idle_bootup_task(struct task_struct *idle);
 250
 251 extern int runqueue_is_locked(void);
 252 extern void task_rq_unlock_wait(struct task_struct *p);
 253
 254 extern cpumask_var_t nohz_cpu_mask;
 255 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
 256 extern int select_nohz_load_balancer(int cpu);
 257 #else
 258 static inline int select_nohz_load_balancer(int cpu)
 259 {
 260         return 0;
 261 }
 262 #endif
 263
 264 /*
 265  * Only dump TASK_* tasks. (0 for all tasks)
 266  */
 267 extern void show_state_filter(unsigned long state_filter);
 268
 269 static inline void show_state(void)
 270 {
 271         show_state_filter(0);
 272 }
 273
 274 extern void show_regs(struct pt_regs *);
 275
 276 /*
 277  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 278  * task), SP is the stack pointer of the first frame that should be shown in the back
 279  * trace (or NULL if the entire call-chain of the task should be shown).
 280  */
 281 extern void show_stack(struct task_struct *task, unsigned long *sp);
 282
 283 void io_schedule(void);
 284 long io_schedule_timeout(long timeout);
 285
 286 extern void cpu_init (void);
 287 extern void trap_init(void);
 288 extern void update_process_times(int user);
 289 extern void scheduler_tick(void);
 290
 291 extern void sched_show_task(struct task_struct *p);
 292
 293 #ifdef CONFIG_DETECT_SOFTLOCKUP
 294 extern void softlockup_tick(void);
 295 extern void touch_softlockup_watchdog(void);
 296 extern void touch_all_softlockup_watchdogs(void);
 297 extern unsigned int  softlockup_panic;
 298 extern unsigned long sysctl_hung_task_check_count;
 299 extern unsigned long sysctl_hung_task_timeout_secs;
 300 extern unsigned long sysctl_hung_task_warnings;
 301 extern int softlockup_thresh;
 302 #else
 303 static inline void softlockup_tick(void)
 304 {
 305 }
 306 static inline void spawn_softlockup_task(void)
 307 {
 308 }
 309 static inline void touch_softlockup_watchdog(void)
 310 {
 311 }
 312 static inline void touch_all_softlockup_watchdogs(void)
 313 {
 314 }
 315 #endif
 316
 317
 318 /* Attach to any functions which should be ignored in wchan output. */
 319 #define __sched         __attribute__((__section__(".sched.text")))
 320
 321 /* Linker adds these: start and end of __sched functions */
 322 extern char __sched_text_start[], __sched_text_end[];
 323
 324 /* Is this address in the __sched functions? */
 325 extern int in_sched_functions(unsigned long addr);
 326
 327 #define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 328 extern signed long schedule_timeout(signed long timeout);
 329 extern signed long schedule_timeout_interruptible(signed long timeout);
 330 extern signed long schedule_timeout_killable(signed long timeout);
 331 extern signed long schedule_timeout_uninterruptible(signed long timeout);
 332 asmlinkage void schedule(void);
 333
 334 struct nsproxy;
 335 struct user_namespace;
 336
 337 /* Maximum number of active map areas.. This is a random (large) number */
 338 #define DEFAULT_MAX_MAP_COUNT   65536
 339
 340 extern int sysctl_max_map_count;
 341
 342 #include <linux/aio.h>
 343
 344 extern unsigned long
 345 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 346                        unsigned long, unsigned long);
 347 extern unsigned long
 348 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 349                           unsigned long len, unsigned long pgoff,
 350                           unsigned long flags);
 351 extern void arch_unmap_area(struct mm_struct *, unsigned long);
 352 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
 353
 354 #if USE_SPLIT_PTLOCKS
 355 /*
 356  * The mm counters are not protected by its page_table_lock,
 357  * so must be incremented atomically.
 358  */
 359 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
 360 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
 361 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
 362 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
 363 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
 364
 365 #else  /* !USE_SPLIT_PTLOCKS */
 366 /*
 367  * The mm counters are protected by its page_table_lock,
 368  * so can be incremented directly.
 369  */
 370 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
 371 #define get_mm_counter(mm, member) ((mm)->_##member)
 372 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
 373 #define inc_mm_counter(mm, member) (mm)->_##member++
 374 #define dec_mm_counter(mm, member) (mm)->_##member--
 375
 376 #endif /* !USE_SPLIT_PTLOCKS */
 377
 378 #define get_mm_rss(mm)                                  \
 379         (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
 380 #define update_hiwater_rss(mm)  do {                    \
 381         unsigned long _rss = get_mm_rss(mm);            \
 382         if ((mm)->hiwater_rss < _rss)                   \
 383                 (mm)->hiwater_rss = _rss;               \
 384 } while (0)
 385 #define update_hiwater_vm(mm)   do {                    \
 386         if ((mm)->hiwater_vm < (mm)->total_vm)          \
 387                 (mm)->hiwater_vm = (mm)->total_vm;      \
 388 } while (0)
 389
 390 #define get_mm_hiwater_rss(mm)  max((mm)->hiwater_rss, get_mm_rss(mm))
 391 #define get_mm_hiwater_vm(mm)   max((mm)->hiwater_vm, (mm)->total_vm)
 392
 393 extern void set_dumpable(struct mm_struct *mm, int value);
 394 extern int get_dumpable(struct mm_struct *mm);
 395
 396 /* mm flags */
 397 /* dumpable bits */
 398 #define MMF_DUMPABLE      0  /* core dump is permitted */
 399 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
 400 #define MMF_DUMPABLE_BITS 2
 401
 402 /* coredump filter bits */
 403 #define MMF_DUMP_ANON_PRIVATE   2
 404 #define MMF_DUMP_ANON_SHARED    3
 405 #define MMF_DUMP_MAPPED_PRIVATE 4
 406 #define MMF_DUMP_MAPPED_SHARED  5
 407 #define MMF_DUMP_ELF_HEADERS    6
 408 #define MMF_DUMP_HUGETLB_PRIVATE 7
 409 #define MMF_DUMP_HUGETLB_SHARED  8
 410 #define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
 411 #define MMF_DUMP_FILTER_BITS    7
 412 #define MMF_DUMP_FILTER_MASK \
 413         (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
 414 #define MMF_DUMP_FILTER_DEFAULT \
 415         ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
 416          (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
 417
 418 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
 419 # define MMF_DUMP_MASK_DEFAULT_ELF      (1 << MMF_DUMP_ELF_HEADERS)
 420 #else
 421 # define MMF_DUMP_MASK_DEFAULT_ELF      0
 422 #endif
 423
 424 struct sighand_struct {
 425         atomic_t                count;
 426         struct k_sigaction      action[_NSIG];
 427         spinlock_t              siglock;
 428         wait_queue_head_t       signalfd_wqh;
 429 };
 430
 431 struct pacct_struct {
 432         int                     ac_flag;
 433         long                    ac_exitcode;
 434         unsigned long           ac_mem;
 435         cputime_t               ac_utime, ac_stime;
 436         unsigned long           ac_minflt, ac_majflt;
 437 };
 438
 439 /**
 440  * struct task_cputime - collected CPU time counts
 441  * @utime:              time spent in user mode, in &cputime_t units
 442  * @stime:              time spent in kernel mode, in &cputime_t units
 443  * @sum_exec_runtime:   total time spent on the CPU, in nanoseconds
 444  *
 445  * This structure groups together three kinds of CPU time that are
 446  * tracked for threads and thread groups.  Most things considering
 447  * CPU time want to group these counts together and treat all three
 448  * of them in parallel.
 449  */
 450 struct task_cputime {
 451         cputime_t utime;
 452         cputime_t stime;
 453         unsigned long long sum_exec_runtime;
 454 };
 455 /* Alternate field names when used to cache expirations. */
 456 #define prof_exp        stime
 457 #define virt_exp        utime
 458 #define sched_exp       sum_exec_runtime
 459
 460 /**
 461  * struct thread_group_cputime - thread group interval timer counts
 462  * @totals:             thread group interval timers; substructure for
 463  *                      uniprocessor kernel, per-cpu for SMP kernel.
 464  *
 465  * This structure contains the version of task_cputime, above, that is
 466  * used for thread group CPU clock calculations.
 467  */
 468 struct thread_group_cputime {
 469         struct task_cputime *totals;
 470 };
 471
 472 /*
 473  * NOTE! "signal_struct" does not have it's own
 474  * locking, because a shared signal_struct always
 475  * implies a shared sighand_struct, so locking
 476  * sighand_struct is always a proper superset of
 477  * the locking of signal_struct.
 478  */
 479 struct signal_struct {
 480         atomic_t                count;
 481         atomic_t                live;
 482
 483         wait_queue_head_t       wait_chldexit;  /* for wait4() */
 484
 485         /* current thread group signal load-balancing target: */
 486         struct task_struct      *curr_target;
 487
 488         /* shared signal handling: */
 489         struct sigpending       shared_pending;
 490
 491         /* thread group exit support */
 492         int                     group_exit_code;
 493         /* overloaded:
 494          * - notify group_exit_task when ->count is equal to notify_count
 495          * - everyone except group_exit_task is stopped during signal delivery
 496          *   of fatal signals, group_exit_task processes the signal.
 497          */
 498         int                     notify_count;
 499         struct task_struct      *group_exit_task;
 500
 501         /* thread group stop support, overloads group_exit_code too */
 502         int                     group_stop_count;
 503         unsigned int            flags; /* see SIGNAL_* flags below */
 504
 505         /* POSIX.1b Interval Timers */
 506         struct list_head posix_timers;
 507
 508         /* ITIMER_REAL timer for the process */
 509         struct hrtimer real_timer;
 510         struct pid *leader_pid;
 511         ktime_t it_real_incr;
 512
 513         /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
 514         cputime_t it_prof_expires, it_virt_expires;
 515         cputime_t it_prof_incr, it_virt_incr;
 516
 517         /*
 518          * Thread group totals for process CPU clocks.
 519          * See thread_group_cputime(), et al, for details.
 520          */
 521         struct thread_group_cputime cputime;
 522
 523         /* Earliest-expiration cache. */
 524         struct task_cputime cputime_expires;
 525
 526         struct list_head cpu_timers[3];
 527
 528         /* job control IDs */
 529
 530         /*
 531          * pgrp and session fields are deprecated.
 532          * use the task_session_Xnr and task_pgrp_Xnr routines below
 533          */
 534
 535         union {
 536                 pid_t pgrp __deprecated;
 537                 pid_t __pgrp;
 538         };
 539
 540         struct pid *tty_old_pgrp;
 541
 542         union {
 543                 pid_t session __deprecated;
 544                 pid_t __session;
 545         };
 546
 547         /* boolean value for session group leader */
 548         int leader;
 549
 550         struct tty_struct *tty; /* NULL if no tty */
 551
 552         /*
 553          * Cumulative resource counters for dead threads in the group,
 554          * and for reaped dead child processes forked by this group.
 555          * Live threads maintain their own counters and add to these
 556          * in __exit_signal, except for the group leader.
 557          */
 558         cputime_t cutime, cstime;
 559         cputime_t gtime;
 560         cputime_t cgtime;
 561         unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 562         unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 563         unsigned long inblock, oublock, cinblock, coublock;
 564         struct task_io_accounting ioac;
 565
 566         /*
 567          * We don't bother to synchronize most readers of this at all,
 568          * because there is no reader checking a limit that actually needs
 569          * to get both rlim_cur and rlim_max atomically, and either one
 570          * alone is a single word that can safely be read normally.
 571          * getrlimit/setrlimit use task_lock(current->group_leader) to
 572          * protect this instead of the siglock, because they really
 573          * have no need to disable irqs.
 574          */
 575         struct rlimit rlim[RLIM_NLIMITS];
 576
 577 #ifdef CONFIG_BSD_PROCESS_ACCT
 578         struct pacct_struct pacct;      /* per-process accounting information */
 579 #endif
 580 #ifdef CONFIG_TASKSTATS
 581         struct taskstats *stats;
 582 #endif
 583 #ifdef CONFIG_AUDIT
 584         unsigned audit_tty;
 585         struct tty_audit_buf *tty_audit_buf;
 586 #endif
 587 };
 588
 589 /* Context switch must be unlocked if interrupts are to be enabled */
 590 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
 591 # define __ARCH_WANT_UNLOCKED_CTXSW
 592 #endif
 593
 594 /*
 595  * Bits in flags field of signal_struct.
 596  */
 597 #define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 598 #define SIGNAL_STOP_DEQUEUED    0x00000002 /* stop signal dequeued */
 599 #define SIGNAL_STOP_CONTINUED   0x00000004 /* SIGCONT since WCONTINUED reap */
 600 #define SIGNAL_GROUP_EXIT       0x00000008 /* group exit in progress */
 601 /*
 602  * Pending notifications to parent.
 603  */
 604 #define SIGNAL_CLD_STOPPED      0x00000010
 605 #define SIGNAL_CLD_CONTINUED    0x00000020
 606 #define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
 607
 608 #define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
 609
 610 /* If true, all threads except ->group_exit_task have pending SIGKILL */
 611 static inline int signal_group_exit(const struct signal_struct *sig)
 612 {
 613         return  (sig->flags & SIGNAL_GROUP_EXIT) ||
 614                 (sig->group_exit_task != NULL);
 615 }
 616
 617 /*
 618  * Some day this will be a full-fledged user tracking system..
 619  */
 620 struct user_struct {
 621         atomic_t __count;       /* reference count */
 622         atomic_t processes;     /* How many processes does this user have? */
 623         atomic_t files;         /* How many open files does this user have? */
 624         atomic_t sigpending;    /* How many pending signals does this user have? */
 625 #ifdef CONFIG_INOTIFY_USER
 626         atomic_t inotify_watches; /* How many inotify watches does this user have? */
 627         atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 628 #endif
 629 #ifdef CONFIG_EPOLL
 630         atomic_t epoll_devs;    /* The number of epoll descriptors currently open */
 631         atomic_t epoll_watches; /* The number of file descriptors currently watched */
 632 #endif
 633 #ifdef CONFIG_POSIX_MQUEUE
 634         /* protected by mq_lock */
 635         unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 636 #endif
 637         unsigned long locked_shm; /* How many pages of mlocked shm ? */
 638
 639 #ifdef CONFIG_KEYS
 640         struct key *uid_keyring;        /* UID specific keyring */
 641         struct key *session_keyring;    /* UID's default session keyring */
 642 #endif
 643
 644         /* Hash table maintenance information */
 645         struct hlist_node uidhash_node;
 646         uid_t uid;
 647         struct user_namespace *user_ns;
 648
 649 #ifdef CONFIG_USER_SCHED
 650         struct task_group *tg;
 651 #ifdef CONFIG_SYSFS
 652         struct kobject kobj;
 653         struct work_struct work;
 654 #endif
 655 #endif
 656 };
 657
 658 extern int uids_sysfs_init(void);
 659
 660 extern struct user_struct *find_user(uid_t);
 661
 662 extern struct user_struct root_user;
 663 #define INIT_USER (&root_user)
 664
 665
 666 struct backing_dev_info;
 667 struct reclaim_state;
 668
 669 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 670 struct sched_info {
 671         /* cumulative counters */
 672         unsigned long pcount;         /* # of times run on this cpu */
 673         unsigned long long run_delay; /* time spent waiting on a runqueue */
 674
 675         /* timestamps */
 676         unsigned long long last_arrival,/* when we last ran on a cpu */
 677                            last_queued; /* when we were last queued to run */
 678 #ifdef CONFIG_SCHEDSTATS
 679         /* BKL stats */
 680         unsigned int bkl_count;
 681 #endif
 682 };
 683 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
 684
 685 #ifdef CONFIG_TASK_DELAY_ACCT
 686 struct task_delay_info {
 687         spinlock_t      lock;
 688         unsigned int    flags;  /* Private per-task flags */
 689
 690         /* For each stat XXX, add following, aligned appropriately
 691          *
 692          * struct timespec XXX_start, XXX_end;
 693          * u64 XXX_delay;
 694          * u32 XXX_count;
 695          *
 696          * Atomicity of updates to XXX_delay, XXX_count protected by
 697          * single lock above (split into XXX_lock if contention is an issue).
 698          */
 699
 700         /*
 701          * XXX_count is incremented on every XXX operation, the delay
 702          * associated with the operation is added to XXX_delay.
 703          * XXX_delay contains the accumulated delay time in nanoseconds.
 704          */
 705         struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
 706         u64 blkio_delay;        /* wait for sync block io completion */
 707         u64 swapin_delay;       /* wait for swapin block io completion */
 708         u32 blkio_count;        /* total count of the number of sync block */
 709                                 /* io operations performed */
 710         u32 swapin_count;       /* total count of the number of swapin block */
 711                                 /* io operations performed */
 712
 713         struct timespec freepages_start, freepages_end;
 714         u64 freepages_delay;    /* wait for memory reclaim */
 715         u32 freepages_count;    /* total count of memory reclaim */
 716 };
 717 #endif  /* CONFIG_TASK_DELAY_ACCT */
 718
 719 static inline int sched_info_on(void)
 720 {
 721 #ifdef CONFIG_SCHEDSTATS
 722         return 1;
 723 #elif defined(CONFIG_TASK_DELAY_ACCT)
 724         extern int delayacct_on;
 725         return delayacct_on;
 726 #else
 727         return 0;
 728 #endif
 729 }
 730
 731 enum cpu_idle_type {
 732         CPU_IDLE,
 733         CPU_NOT_IDLE,
 734         CPU_NEWLY_IDLE,
 735         CPU_MAX_IDLE_TYPES
 736 };
 737
 738 /*
 739  * sched-domains (multiprocessor balancing) declarations:
 740  */
 741
 742 /*
 743  * Increase resolution of nice-level calculations:
 744  */
 745 #define SCHED_LOAD_SHIFT        10
 746 #define SCHED_LOAD_SCALE        (1L << SCHED_LOAD_SHIFT)
 747
 748 #define SCHED_LOAD_SCALE_FUZZ   SCHED_LOAD_SCALE
 749
 750 #ifdef CONFIG_SMP
 751 #define SD_LOAD_BALANCE         1       /* Do load balancing on this domain. */
 752 #define SD_BALANCE_NEWIDLE      2       /* Balance when about to become idle */
 753 #define SD_BALANCE_EXEC         4       /* Balance on exec */
 754 #define SD_BALANCE_FORK         8       /* Balance on fork, clone */
 755 #define SD_WAKE_IDLE            16      /* Wake to idle CPU on task wakeup */
 756 #define SD_WAKE_AFFINE          32      /* Wake task to waking CPU */
 757 #define SD_WAKE_BALANCE         64      /* Perform balancing at task wakeup */
 758 #define SD_SHARE_CPUPOWER       128     /* Domain members share cpu power */
 759 #define SD_POWERSAVINGS_BALANCE 256     /* Balance for power savings */
 760 #define SD_SHARE_PKG_RESOURCES  512     /* Domain members share cpu pkg resources */
 761 #define SD_SERIALIZE            1024    /* Only a single load balancing instance */
 762 #define SD_WAKE_IDLE_FAR        2048    /* Gain latency sacrificing cache hit */
 763
 764 enum powersavings_balance_level {
 765         POWERSAVINGS_BALANCE_NONE = 0,  /* No power saving load balance */
 766         POWERSAVINGS_BALANCE_BASIC,     /* Fill one thread/core/package
 767                                          * first for long running threads
 768                                          */
 769         POWERSAVINGS_BALANCE_WAKEUP,    /* Also bias task wakeups to semi-idle
 770                                          * cpu package for power savings
 771                                          */
 772         MAX_POWERSAVINGS_BALANCE_LEVELS
 773 };
 774
 775 extern int sched_mc_power_savings, sched_smt_power_savings;
 776
 777 static inline int sd_balance_for_mc_power(void)
 778 {
 779         if (sched_smt_power_savings)
 780                 return SD_POWERSAVINGS_BALANCE;
 781
 782         return 0;
 783 }
 784
 785 static inline int sd_balance_for_package_power(void)
 786 {
 787         if (sched_mc_power_savings | sched_smt_power_savings)
 788                 return SD_POWERSAVINGS_BALANCE;
 789
 790         return 0;
 791 }
 792
 793 /*
 794  * Optimise SD flags for power savings:
 795  * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
 796  * Keep default SD flags if sched_{smt,mc}_power_saving=0
 797  */
 798
 799 static inline int sd_power_saving_flags(void)
 800 {
 801         if (sched_mc_power_savings | sched_smt_power_savings)
 802                 return SD_BALANCE_NEWIDLE;
 803
 804         return 0;
 805 }
 806
 807 struct sched_group {
 808         struct sched_group *next;       /* Must be a circular list */
 809
 810         /*
 811          * CPU power of this group, SCHED_LOAD_SCALE being max power for a
 812          * single CPU. This is read only (except for setup, hotplug CPU).
 813          * Note : Never change cpu_power without recompute its reciprocal
 814          */
 815         unsigned int __cpu_power;
 816         /*
 817          * reciprocal value of cpu_power to avoid expensive divides
 818          * (see include/linux/reciprocal_div.h)
 819          */
 820         u32 reciprocal_cpu_power;
 821
 822         unsigned long cpumask[];
 823 };
 824
 825 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
 826 {
 827         return to_cpumask(sg->cpumask);
 828 }
 829
 830 enum sched_domain_level {
 831         SD_LV_NONE = 0,
 832         SD_LV_SIBLING,
 833         SD_LV_MC,
 834         SD_LV_CPU,
 835         SD_LV_NODE,
 836         SD_LV_ALLNODES,
 837         SD_LV_MAX
 838 };
 839
 840 struct sched_domain_attr {
 841         int relax_domain_level;
 842 };
 843
 844 #define SD_ATTR_INIT    (struct sched_domain_attr) {    \
 845         .relax_domain_level = -1,                       \
 846 }
 847
 848 struct sched_domain {
 849         /* These fields must be setup */
 850         struct sched_domain *parent;    /* top domain must be null terminated */
 851         struct sched_domain *child;     /* bottom domain must be null terminated */
 852         struct sched_group *groups;     /* the balancing groups of the domain */
 853         unsigned long min_interval;     /* Minimum balance interval ms */
 854         unsigned long max_interval;     /* Maximum balance interval ms */
 855         unsigned int busy_factor;       /* less balancing by factor if busy */
 856         unsigned int imbalance_pct;     /* No balance until over watermark */
 857         unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
 858         unsigned int busy_idx;
 859         unsigned int idle_idx;
 860         unsigned int newidle_idx;
 861         unsigned int wake_idx;
 862         unsigned int forkexec_idx;
 863         int flags;                      /* See SD_* */
 864         enum sched_domain_level level;
 865
 866         /* Runtime fields. */
 867         unsigned long last_balance;     /* init to jiffies. units in jiffies */
 868         unsigned int balance_interval;  /* initialise to 1. units in ms. */
 869         unsigned int nr_balance_failed; /* initialise to 0 */
 870
 871         u64 last_update;
 872
 873 #ifdef CONFIG_SCHEDSTATS
 874         /* load_balance() stats */
 875         unsigned int lb_count[CPU_MAX_IDLE_TYPES];
 876         unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
 877         unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
 878         unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
 879         unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
 880         unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
 881         unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
 882         unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
 883
 884         /* Active load balancing */
 885         unsigned int alb_count;
 886         unsigned int alb_failed;
 887         unsigned int alb_pushed;
 888
 889         /* SD_BALANCE_EXEC stats */
 890         unsigned int sbe_count;
 891         unsigned int sbe_balanced;
 892         unsigned int sbe_pushed;
 893
 894         /* SD_BALANCE_FORK stats */
 895         unsigned int sbf_count;
 896         unsigned int sbf_balanced;
 897         unsigned int sbf_pushed;
 898
 899         /* try_to_wake_up() stats */
 900         unsigned int ttwu_wake_remote;
 901         unsigned int ttwu_move_affine;
 902         unsigned int ttwu_move_balance;
 903 #endif
 904 #ifdef CONFIG_SCHED_DEBUG
 905         char *name;
 906 #endif
 907
 908         /* span of all CPUs in this domain */
 909         unsigned long span[];
 910 };
 911
 912 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
 913 {
 914         return to_cpumask(sd->span);
 915 }
 916
 917 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
 918                                     struct sched_domain_attr *dattr_new);
 919
 920 /* Test a flag in parent sched domain */
 921 static inline int test_sd_parent(struct sched_domain *sd, int flag)
 922 {
 923         if (sd->parent && (sd->parent->flags & flag))
 924                 return 1;
 925
 926         return 0;
 927 }
 928
 929 #else /* CONFIG_SMP */
 930
 931 struct sched_domain_attr;
 932
 933 static inline void
 934 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
 935                         struct sched_domain_attr *dattr_new)
 936 {
 937 }
 938 #endif  /* !CONFIG_SMP */
 939
 940 struct io_context;                      /* See blkdev.h */
 941
 942
 943 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
 944 extern void prefetch_stack(struct task_struct *t);
 945 #else
 946 static inline void prefetch_stack(struct task_struct *t) { }
 947 #endif
 948
 949 struct audit_context;           /* See audit.c */
 950 struct mempolicy;
 951 struct pipe_inode_info;
 952 struct uts_namespace;
 953
 954 struct rq;
 955 struct sched_domain;
 956
 957 struct sched_class {
 958         const struct sched_class *next;
 959
 960         void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
 961         void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
 962         void (*yield_task) (struct rq *rq);
 963
 964         void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
 965
 966         struct task_struct * (*pick_next_task) (struct rq *rq);
 967         void (*put_prev_task) (struct rq *rq, struct task_struct *p);
 968
 969 #ifdef CONFIG_SMP
 970         int  (*select_task_rq)(struct task_struct *p, int sync);
 971
 972         unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
 973                         struct rq *busiest, unsigned long max_load_move,
 974                         struct sched_domain *sd, enum cpu_idle_type idle,
 975                         int *all_pinned, int *this_best_prio);
 976
 977         int (*move_one_task) (struct rq *this_rq, int this_cpu,
 978                               struct rq *busiest, struct sched_domain *sd,
 979                               enum cpu_idle_type idle);
 980         void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
 981         void (*post_schedule) (struct rq *this_rq);
 982         void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
 983
 984         void (*set_cpus_allowed)(struct task_struct *p,
 985                                  const struct cpumask *newmask);
 986
 987         void (*rq_online)(struct rq *rq);
 988         void (*rq_offline)(struct rq *rq);
 989 #endif
 990
 991         void (*set_curr_task) (struct rq *rq);
 992         void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
 993         void (*task_new) (struct rq *rq, struct task_struct *p);
 994
 995         void (*switched_from) (struct rq *this_rq, struct task_struct *task,
 996                                int running);
 997         void (*switched_to) (struct rq *this_rq, struct task_struct *task,
 998                              int running);
 999         void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1000                              int oldprio, int running);
1001
1002 #ifdef CONFIG_FAIR_GROUP_SCHED
1003         void (*moved_group) (struct task_struct *p);
1004 #endif
1005 };
1006
1007 struct load_weight {
1008         unsigned long weight, inv_weight;
1009 };
1010
1011 /*
1012  * CFS stats for a schedulable entity (task, task-group etc)
1013  *
1014  * Current field usage histogram:
1015  *
1016  *     4 se->block_start
1017  *     4 se->run_node
1018  *     4 se->sleep_start
1019  *     6 se->load.weight
1020  */
1021 struct sched_entity {
1022         struct load_weight      load;           /* for load-balancing */
1023         struct rb_node          run_node;
1024         struct list_head        group_node;
1025         unsigned int            on_rq;
1026
1027         u64                     exec_start;
1028         u64                     sum_exec_runtime;
1029         u64                     vruntime;
1030         u64                     prev_sum_exec_runtime;
1031
1032         u64                     last_wakeup;
1033         u64                     avg_overlap;
1034
1035         u64                     nr_migrations;
1036
1037 #ifdef CONFIG_SCHEDSTATS
1038         u64                     wait_start;
1039         u64                     wait_max;
1040         u64                     wait_count;
1041         u64                     wait_sum;
1042
1043         u64                     sleep_start;
1044         u64                     sleep_max;
1045         s64                     sum_sleep_runtime;
1046
1047         u64                     block_start;
1048         u64                     block_max;
1049         u64                     exec_max;
1050         u64                     slice_max;
1051
1052         u64                     nr_migrations_cold;
1053         u64                     nr_failed_migrations_affine;
1054         u64                     nr_failed_migrations_running;
1055         u64                     nr_failed_migrations_hot;
1056         u64                     nr_forced_migrations;
1057         u64                     nr_forced2_migrations;
1058
1059         u64                     nr_wakeups;
1060         u64                     nr_wakeups_sync;
1061         u64                     nr_wakeups_migrate;
1062         u64                     nr_wakeups_local;
1063         u64                     nr_wakeups_remote;
1064         u64                     nr_wakeups_affine;
1065         u64                     nr_wakeups_affine_attempts;
1066         u64                     nr_wakeups_passive;
1067         u64                     nr_wakeups_idle;
1068 #endif
1069
1070 #ifdef CONFIG_FAIR_GROUP_SCHED
1071         struct sched_entity     *parent;
1072         /* rq on which this entity is (to be) queued: */
1073         struct cfs_rq           *cfs_rq;
1074         /* rq "owned" by this entity/group: */
1075         struct cfs_rq           *my_q;
1076 #endif
1077 };
1078
1079 struct sched_rt_entity {
1080         struct list_head run_list;
1081         unsigned long timeout;
1082         unsigned int time_slice;
1083         int nr_cpus_allowed;
1084
1085         struct sched_rt_entity *back;
1086 #ifdef CONFIG_RT_GROUP_SCHED
1087         struct sched_rt_entity  *parent;
1088         /* rq on which this entity is (to be) queued: */
1089         struct rt_rq            *rt_rq;
1090         /* rq "owned" by this entity/group: */
1091         struct rt_rq            *my_q;
1092 #endif
1093 };
1094
1095 struct task_struct {
1096         volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1097         void *stack;
1098         atomic_t usage;
1099         unsigned int flags;     /* per process flags, defined below */
1100         unsigned int ptrace;
1101
1102         int lock_depth;         /* BKL lock depth */
1103
1104 #ifdef CONFIG_SMP
1105 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1106         int oncpu;
1107 #endif
1108 #endif
1109
1110         int prio, static_prio, normal_prio;
1111         unsigned int rt_priority;
1112         const struct sched_class *sched_class;
1113         struct sched_entity se;
1114         struct sched_rt_entity rt;
1115
1116 #ifdef CONFIG_PREEMPT_NOTIFIERS
1117         /* list of struct preempt_notifier: */
1118         struct hlist_head preempt_notifiers;
1119 #endif
1120
1121         /*
1122          * fpu_counter contains the number of consecutive context switches
1123          * that the FPU is used. If this is over a threshold, the lazy fpu
1124          * saving becomes unlazy to save the trap. This is an unsigned char
1125          * so that after 256 times the counter wraps and the behavior turns
1126          * lazy again; this to deal with bursty apps that only use FPU for
1127          * a short time
1128          */
1129         unsigned char fpu_counter;
1130         s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1131 #ifdef CONFIG_BLK_DEV_IO_TRACE
1132         unsigned int btrace_seq;
1133 #endif
1134
1135         unsigned int policy;
1136         cpumask_t cpus_allowed;
1137
1138 #ifdef CONFIG_PREEMPT_RCU
1139         int rcu_read_lock_nesting;
1140         int rcu_flipctr_idx;
1141 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1142
1143 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1144         struct sched_info sched_info;
1145 #endif
1146
1147         struct list_head tasks;
1148
1149         struct mm_struct *mm, *active_mm;
1150
1151 /* task state */
1152         struct linux_binfmt *binfmt;
1153         int exit_state;
1154         int exit_code, exit_signal;
1155         int pdeath_signal;  /*  The signal sent when the parent dies  */
1156         /* ??? */
1157         unsigned int personality;
1158         unsigned did_exec:1;
1159         pid_t pid;
1160         pid_t tgid;
1161
1162         /* Canary value for the -fstack-protector gcc feature */
1163         unsigned long stack_canary;
1164
1165         /*
1166          * pointers to (original) parent process, youngest child, younger sibling,
1167          * older sibling, respectively.  (p->father can be replaced with
1168          * p->real_parent->pid)
1169          */
1170         struct task_struct *real_parent; /* real parent process */
1171         struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1172         /*
1173          * children/sibling forms the list of my natural children
1174          */
1175         struct list_head children;      /* list of my children */
1176         struct list_head sibling;       /* linkage in my parent's children list */
1177         struct task_struct *group_leader;       /* threadgroup leader */
1178
1179         /*
1180          * ptraced is the list of tasks this task is using ptrace on.
1181          * This includes both natural children and PTRACE_ATTACH targets.
1182          * p->ptrace_entry is p's link on the p->parent->ptraced list.
1183          */
1184         struct list_head ptraced;
1185         struct list_head ptrace_entry;
1186
1187 #ifdef CONFIG_X86_PTRACE_BTS
1188         /*
1189          * This is the tracer handle for the ptrace BTS extension.
1190          * This field actually belongs to the ptracer task.
1191          */
1192         struct bts_tracer *bts;
1193         /*
1194          * The buffer to hold the BTS data.
1195          */
1196         void *bts_buffer;
1197         size_t bts_size;
1198 #endif /* CONFIG_X86_PTRACE_BTS */
1199
1200         /* PID/PID hash table linkage. */
1201         struct pid_link pids[PIDTYPE_MAX];
1202         struct list_head thread_group;
1203
1204         struct completion *vfork_done;          /* for vfork() */
1205         int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1206         int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1207
1208         cputime_t utime, stime, utimescaled, stimescaled;
1209         cputime_t gtime;
1210         cputime_t prev_utime, prev_stime;
1211         unsigned long nvcsw, nivcsw; /* context switch counts */
1212         struct timespec start_time;             /* monotonic time */
1213         struct timespec real_start_time;        /* boot based time */
1214 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1215         unsigned long min_flt, maj_flt;
1216
1217         struct task_cputime cputime_expires;
1218         struct list_head cpu_timers[3];
1219
1220 /* process credentials */
1221         const struct cred *real_cred;   /* objective and real subjective task
1222                                          * credentials (COW) */
1223         const struct cred *cred;        /* effective (overridable) subjective task
1224                                          * credentials (COW) */
1225         struct mutex cred_exec_mutex;   /* execve vs ptrace cred calculation mutex */
1226
1227         char comm[TASK_COMM_LEN]; /* executable name excluding path
1228                                      - access with [gs]et_task_comm (which lock
1229                                        it with task_lock())
1230                                      - initialized normally by flush_old_exec */
1231 /* file system info */
1232         int link_count, total_link_count;
1233 #ifdef CONFIG_SYSVIPC
1234 /* ipc stuff */
1235         struct sysv_sem sysvsem;
1236 #endif
1237 #ifdef CONFIG_DETECT_SOFTLOCKUP
1238 /* hung task detection */
1239         unsigned long last_switch_timestamp;
1240         unsigned long last_switch_count;
1241 #endif
1242 /* CPU-specific state of this task */
1243         struct thread_struct thread;
1244 /* filesystem information */
1245         struct fs_struct *fs;
1246 /* open file information */
1247         struct files_struct *files;
1248 /* namespaces */
1249         struct nsproxy *nsproxy;
1250 /* signal handlers */
1251         struct signal_struct *signal;
1252         struct sighand_struct *sighand;
1253
1254         sigset_t blocked, real_blocked;
1255         sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1256         struct sigpending pending;
1257
1258         unsigned long sas_ss_sp;
1259         size_t sas_ss_size;
1260         int (*notifier)(void *priv);
1261         void *notifier_data;
1262         sigset_t *notifier_mask;
1263         struct audit_context *audit_context;
1264 #ifdef CONFIG_AUDITSYSCALL
1265         uid_t loginuid;
1266         unsigned int sessionid;
1267 #endif
1268         seccomp_t seccomp;
1269
1270 /* Thread group tracking */
1271         u32 parent_exec_id;
1272         u32 self_exec_id;
1273 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1274         spinlock_t alloc_lock;
1275
1276         /* Protection of the PI data structures: */
1277         spinlock_t pi_lock;
1278
1279 #ifdef CONFIG_RT_MUTEXES
1280         /* PI waiters blocked on a rt_mutex held by this task */
1281         struct plist_head pi_waiters;
1282         /* Deadlock detection and priority inheritance handling */
1283         struct rt_mutex_waiter *pi_blocked_on;
1284 #endif
1285
1286 #ifdef CONFIG_DEBUG_MUTEXES
1287         /* mutex deadlock detection */
1288         struct mutex_waiter *blocked_on;
1289 #endif
1290 #ifdef CONFIG_TRACE_IRQFLAGS
1291         unsigned int irq_events;
1292         int hardirqs_enabled;
1293         unsigned long hardirq_enable_ip;
1294         unsigned int hardirq_enable_event;
1295         unsigned long hardirq_disable_ip;
1296         unsigned int hardirq_disable_event;
1297         int softirqs_enabled;
1298         unsigned long softirq_disable_ip;
1299         unsigned int softirq_disable_event;
1300         unsigned long softirq_enable_ip;
1301         unsigned int softirq_enable_event;
1302         int hardirq_context;
1303         int softirq_context;
1304 #endif
1305 #ifdef CONFIG_LOCKDEP
1306 # define MAX_LOCK_DEPTH 48UL
1307         u64 curr_chain_key;
1308         int lockdep_depth;
1309         unsigned int lockdep_recursion;
1310         struct held_lock held_locks[MAX_LOCK_DEPTH];
1311 #endif
1312
1313 /* journalling filesystem info */
1314         void *journal_info;
1315
1316 /* stacked block device info */
1317         struct bio *bio_list, **bio_tail;
1318
1319 /* VM state */
1320         struct reclaim_state *reclaim_state;
1321
1322         struct backing_dev_info *backing_dev_info;
1323
1324         struct io_context *io_context;
1325
1326         unsigned long ptrace_message;
1327         siginfo_t *last_siginfo; /* For ptrace use.  */
1328         struct task_io_accounting ioac;
1329 #if defined(CONFIG_TASK_XACCT)
1330         u64 acct_rss_mem1;      /* accumulated rss usage */
1331         u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1332         cputime_t acct_timexpd; /* stime + utime since last update */
1333 #endif
1334 #ifdef CONFIG_CPUSETS
1335         nodemask_t mems_allowed;
1336         int cpuset_mems_generation;
1337         int cpuset_mem_spread_rotor;
1338 #endif
1339 #ifdef CONFIG_CGROUPS
1340         /* Control Group info protected by css_set_lock */
1341         struct css_set *cgroups;
1342         /* cg_list protected by css_set_lock and tsk->alloc_lock */
1343         struct list_head cg_list;
1344 #endif
1345 #ifdef CONFIG_FUTEX
1346         struct robust_list_head __user *robust_list;
1347 #ifdef CONFIG_COMPAT
1348         struct compat_robust_list_head __user *compat_robust_list;
1349 #endif
1350         struct list_head pi_state_list;
1351         struct futex_pi_state *pi_state_cache;
1352 #endif
1353         struct perf_counter_context perf_counter_ctx;
1354 #ifdef CONFIG_NUMA
1355         struct mempolicy *mempolicy;
1356         short il_next;
1357 #endif
1358         atomic_t fs_excl;       /* holding fs exclusive resources */
1359         struct rcu_head rcu;
1360
1361         /*
1362          * cache last used pipe for splice
1363          */
1364         struct pipe_inode_info *splice_pipe;
1365 #ifdef  CONFIG_TASK_DELAY_ACCT
1366         struct task_delay_info *delays;
1367 #endif
1368 #ifdef CONFIG_FAULT_INJECTION
1369         int make_it_fail;
1370 #endif
1371         struct prop_local_single dirties;
1372 #ifdef CONFIG_LATENCYTOP
1373         int latency_record_count;
1374         struct latency_record latency_record[LT_SAVECOUNT];
1375 #endif
1376         /*
1377          * time slack values; these are used to round up poll() and
1378          * select() etc timeout values. These are in nanoseconds.
1379          */
1380         unsigned long timer_slack_ns;
1381         unsigned long default_timer_slack_ns;
1382
1383         struct list_head        *scm_work_list;
1384 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1385         /* Index of current stored adress in ret_stack */
1386         int curr_ret_stack;
1387         /* Stack of return addresses for return function tracing */
1388         struct ftrace_ret_stack *ret_stack;
1389         /*
1390          * Number of functions that haven't been traced
1391          * because of depth overrun.
1392          */
1393         atomic_t trace_overrun;
1394         /* Pause for the tracing */
1395         atomic_t tracing_graph_pause;
1396 #endif
1397 #ifdef CONFIG_TRACING
1398         /* state flags for use by tracers */
1399         unsigned long trace;
1400 #endif
1401 };
1402
1403 /*
1404  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1405  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1406  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1407  * values are inverted: lower p->prio value means higher priority.
1408  *
1409  * The MAX_USER_RT_PRIO value allows the actual maximum
1410  * RT priority to be separate from the value exported to
1411  * user-space.  This allows kernel threads to set their
1412  * priority to a value higher than any user task. Note:
1413  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1414  */
1415
1416 #define MAX_USER_RT_PRIO        100
1417 #define MAX_RT_PRIO             MAX_USER_RT_PRIO
1418
1419 #define MAX_PRIO                (MAX_RT_PRIO + 40)
1420 #define DEFAULT_PRIO            (MAX_RT_PRIO + 20)
1421
1422 static inline int rt_prio(int prio)
1423 {
1424         if (unlikely(prio < MAX_RT_PRIO))
1425                 return 1;
1426         return 0;
1427 }
1428
1429 static inline int rt_task(struct task_struct *p)
1430 {
1431         return rt_prio(p->prio);
1432 }
1433
1434 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1435 {
1436         tsk->signal->__session = session;
1437 }
1438
1439 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1440 {
1441         tsk->signal->__pgrp = pgrp;
1442 }
1443
1444 static inline struct pid *task_pid(struct task_struct *task)
1445 {
1446         return task->pids[PIDTYPE_PID].pid;
1447 }
1448
1449 static inline struct pid *task_tgid(struct task_struct *task)
1450 {
1451         return task->group_leader->pids[PIDTYPE_PID].pid;
1452 }
1453
1454 static inline struct pid *task_pgrp(struct task_struct *task)
1455 {
1456         return task->group_leader->pids[PIDTYPE_PGID].pid;
1457 }
1458
1459 static inline struct pid *task_session(struct task_struct *task)
1460 {
1461         return task->group_leader->pids[PIDTYPE_SID].pid;
1462 }
1463
1464 struct pid_namespace;
1465
1466 /*
1467  * the helpers to get the task's different pids as they are seen
1468  * from various namespaces
1469  *
1470  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1471  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1472  *                     current.
1473  * task_xid_nr_ns()  : id seen from the ns specified;
1474  *
1475  * set_task_vxid()   : assigns a virtual id to a task;
1476  *
1477  * see also pid_nr() etc in include/linux/pid.h
1478  */
1479
1480 static inline pid_t task_pid_nr(struct task_struct *tsk)
1481 {
1482         return tsk->pid;
1483 }
1484
1485 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1486
1487 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1488 {
1489         return pid_vnr(task_pid(tsk));
1490 }
1491
1492
1493 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1494 {
1495         return tsk->tgid;
1496 }
1497
1498 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1499
1500 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1501 {
1502         return pid_vnr(task_tgid(tsk));
1503 }
1504
1505
1506 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1507 {
1508         return tsk->signal->__pgrp;
1509 }
1510
1511 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1512
1513 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1514 {
1515         return pid_vnr(task_pgrp(tsk));
1516 }
1517
1518
1519 static inline pid_t task_session_nr(struct task_struct *tsk)
1520 {
1521         return tsk->signal->__session;
1522 }
1523
1524 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1525
1526 static inline pid_t task_session_vnr(struct task_struct *tsk)
1527 {
1528         return pid_vnr(task_session(tsk));
1529 }
1530
1531
1532 /**
1533  * pid_alive - check that a task structure is not stale
1534  * @p: Task structure to be checked.
1535  *
1536  * Test if a process is not yet dead (at most zombie state)
1537  * If pid_alive fails, then pointers within the task structure
1538  * can be stale and must not be dereferenced.
1539  */
1540 static inline int pid_alive(struct task_struct *p)
1541 {
1542         return p->pids[PIDTYPE_PID].pid != NULL;
1543 }
1544
1545 /**
1546  * is_global_init - check if a task structure is init
1547  * @tsk: Task structure to be checked.
1548  *
1549  * Check if a task structure is the first user space task the kernel created.
1550  */
1551 static inline int is_global_init(struct task_struct *tsk)
1552 {
1553         return tsk->pid == 1;
1554 }
1555
1556 /*
1557  * is_container_init:
1558  * check whether in the task is init in its own pid namespace.
1559  */
1560 extern int is_container_init(struct task_struct *tsk);
1561
1562 extern struct pid *cad_pid;
1563
1564 extern void free_task(struct task_struct *tsk);
1565 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1566
1567 extern void __put_task_struct(struct task_struct *t);
1568
1569 static inline void put_task_struct(struct task_struct *t)
1570 {
1571         if (atomic_dec_and_test(&t->usage))
1572                 __put_task_struct(t);
1573 }
1574
1575 extern cputime_t task_utime(struct task_struct *p);
1576 extern cputime_t task_stime(struct task_struct *p);
1577 extern cputime_t task_gtime(struct task_struct *p);
1578
1579 /*
1580  * Per process flags
1581  */
1582 #define PF_ALIGNWARN    0x00000001      /* Print alignment warning msgs */
1583                                         /* Not implemented yet, only for 486*/
1584 #define PF_STARTING     0x00000002      /* being created */
1585 #define PF_EXITING      0x00000004      /* getting shut down */
1586 #define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
1587 #define PF_VCPU         0x00000010      /* I'm a virtual CPU */
1588 #define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
1589 #define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
1590 #define PF_DUMPCORE     0x00000200      /* dumped core */
1591 #define PF_SIGNALED     0x00000400      /* killed by a signal */
1592 #define PF_MEMALLOC     0x00000800      /* Allocating memory */
1593 #define PF_FLUSHER      0x00001000      /* responsible for disk writeback */
1594 #define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
1595 #define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
1596 #define PF_FROZEN       0x00010000      /* frozen for system suspend */
1597 #define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
1598 #define PF_KSWAPD       0x00040000      /* I am kswapd */
1599 #define PF_SWAPOFF      0x00080000      /* I am in swapoff */
1600 #define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
1601 #define PF_KTHREAD      0x00200000      /* I am a kernel thread */
1602 #define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
1603 #define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
1604 #define PF_SPREAD_PAGE  0x01000000      /* Spread page cache over cpuset */
1605 #define PF_SPREAD_SLAB  0x02000000      /* Spread some slab caches over cpuset */
1606 #define PF_THREAD_BOUND 0x04000000      /* Thread bound to specific cpu */
1607 #define PF_MEMPOLICY    0x10000000      /* Non-default NUMA mempolicy */
1608 #define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
1609 #define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezeable */
1610 #define PF_FREEZER_NOSIG 0x80000000     /* Freezer won't send signals to it */
1611
1612 /*
1613  * Only the _current_ task can read/write to tsk->flags, but other
1614  * tasks can access tsk->flags in readonly mode for example
1615  * with tsk_used_math (like during threaded core dumping).
1616  * There is however an exception to this rule during ptrace
1617  * or during fork: the ptracer task is allowed to write to the
1618  * child->flags of its traced child (same goes for fork, the parent
1619  * can write to the child->flags), because we're guaranteed the
1620  * child is not running and in turn not changing child->flags
1621  * at the same time the parent does it.
1622  */
1623 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1624 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1625 #define clear_used_math() clear_stopped_child_used_math(current)
1626 #define set_used_math() set_stopped_child_used_math(current)
1627 #define conditional_stopped_child_used_math(condition, child) \
1628         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1629 #define conditional_used_math(condition) \
1630         conditional_stopped_child_used_math(condition, current)
1631 #define copy_to_stopped_child_used_math(child) \
1632         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1633 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1634 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1635 #define used_math() tsk_used_math(current)
1636
1637 #ifdef CONFIG_SMP
1638 extern int set_cpus_allowed_ptr(struct task_struct *p,
1639                                 const struct cpumask *new_mask);
1640 #else
1641 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1642                                        const struct cpumask *new_mask)
1643 {
1644         if (!cpumask_test_cpu(0, new_mask))
1645                 return -EINVAL;
1646         return 0;
1647 }
1648 #endif
1649 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1650 {
1651         return set_cpus_allowed_ptr(p, &new_mask);
1652 }
1653
1654 extern unsigned long long sched_clock(void);
1655
1656 extern void sched_clock_init(void);
1657 extern u64 sched_clock_cpu(int cpu);
1658
1659 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1660 static inline void sched_clock_tick(void)
1661 {
1662 }
1663
1664 static inline void sched_clock_idle_sleep_event(void)
1665 {
1666 }
1667
1668 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1669 {
1670 }
1671 #else
1672 extern void sched_clock_tick(void);
1673 extern void sched_clock_idle_sleep_event(void);
1674 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1675 #endif
1676
1677 /*
1678  * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1679  * clock constructed from sched_clock():
1680  */
1681 extern unsigned long long cpu_clock(int cpu);
1682
1683 extern unsigned long long
1684 task_sched_runtime(struct task_struct *task);
1685 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1686
1687 /* sched_exec is called by processes performing an exec */
1688 #ifdef CONFIG_SMP
1689 extern void sched_exec(void);
1690 #else
1691 #define sched_exec()   {}
1692 #endif
1693
1694 extern void sched_clock_idle_sleep_event(void);
1695 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1696
1697 #ifdef CONFIG_HOTPLUG_CPU
1698 extern void idle_task_exit(void);
1699 #else
1700 static inline void idle_task_exit(void) {}
1701 #endif
1702
1703 extern void sched_idle_next(void);
1704
1705 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1706 extern void wake_up_idle_cpu(int cpu);
1707 #else
1708 static inline void wake_up_idle_cpu(int cpu) { }
1709 #endif
1710
1711 extern unsigned int sysctl_sched_latency;
1712 extern unsigned int sysctl_sched_min_granularity;
1713 extern unsigned int sysctl_sched_wakeup_granularity;
1714 extern unsigned int sysctl_sched_shares_ratelimit;
1715 extern unsigned int sysctl_sched_shares_thresh;
1716 #ifdef CONFIG_SCHED_DEBUG
1717 extern unsigned int sysctl_sched_child_runs_first;
1718 extern unsigned int sysctl_sched_features;
1719 extern unsigned int sysctl_sched_migration_cost;
1720 extern unsigned int sysctl_sched_nr_migrate;
1721
1722 int sched_nr_latency_handler(struct ctl_table *table, int write,
1723                 struct file *file, void __user *buffer, size_t *length,
1724                 loff_t *ppos);
1725 #endif
1726 extern unsigned int sysctl_sched_rt_period;
1727 extern int sysctl_sched_rt_runtime;
1728
1729 int sched_rt_handler(struct ctl_table *table, int write,
1730                 struct file *filp, void __user *buffer, size_t *lenp,
1731                 loff_t *ppos);
1732
1733 extern unsigned int sysctl_sched_compat_yield;
1734
1735 #ifdef CONFIG_RT_MUTEXES
1736 extern int rt_mutex_getprio(struct task_struct *p);
1737 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1738 extern void rt_mutex_adjust_pi(struct task_struct *p);
1739 #else
1740 static inline int rt_mutex_getprio(struct task_struct *p)
1741 {
1742         return p->normal_prio;
1743 }
1744 # define rt_mutex_adjust_pi(p)          do { } while (0)
1745 #endif
1746
1747 extern void set_user_nice(struct task_struct *p, long nice);
1748 extern int task_prio(const struct task_struct *p);
1749 extern int task_nice(const struct task_struct *p);
1750 extern int can_nice(const struct task_struct *p, const int nice);
1751 extern int task_curr(const struct task_struct *p);
1752 extern int idle_cpu(int cpu);
1753 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1754 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1755                                       struct sched_param *);
1756 extern struct task_struct *idle_task(int cpu);
1757 extern struct task_struct *curr_task(int cpu);
1758 extern void set_curr_task(int cpu, struct task_struct *p);
1759
1760 void yield(void);
1761
1762 /*
1763  * The default (Linux) execution domain.
1764  */
1765 extern struct exec_domain       default_exec_domain;
1766
1767 union thread_union {
1768         struct thread_info thread_info;
1769         unsigned long stack[THREAD_SIZE/sizeof(long)];
1770 };
1771
1772 #ifndef __HAVE_ARCH_KSTACK_END
1773 static inline int kstack_end(void *addr)
1774 {
1775         /* Reliable end of stack detection:
1776          * Some APM bios versions misalign the stack
1777          */
1778         return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1779 }
1780 #endif
1781
1782 extern union thread_union init_thread_union;
1783 extern struct task_struct init_task;
1784
1785 extern struct   mm_struct init_mm;
1786
1787 extern struct pid_namespace init_pid_ns;
1788
1789 /*
1790  * find a task by one of its numerical ids
1791  *
1792  * find_task_by_pid_type_ns():
1793  *      it is the most generic call - it finds a task by all id,
1794  *      type and namespace specified
1795  * find_task_by_pid_ns():
1796  *      finds a task by its pid in the specified namespace
1797  * find_task_by_vpid():
1798  *      finds a task by its virtual pid
1799  *
1800  * see also find_vpid() etc in include/linux/pid.h
1801  */
1802
1803 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1804                 struct pid_namespace *ns);
1805
1806 extern struct task_struct *find_task_by_vpid(pid_t nr);
1807 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1808                 struct pid_namespace *ns);
1809
1810 extern void __set_special_pids(struct pid *pid);
1811
1812 /* per-UID process charging. */
1813 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1814 static inline struct user_struct *get_uid(struct user_struct *u)
1815 {
1816         atomic_inc(&u->__count);
1817         return u;
1818 }
1819 extern void free_uid(struct user_struct *);
1820 extern void release_uids(struct user_namespace *ns);
1821
1822 #include <asm/current.h>
1823
1824 extern void do_timer(unsigned long ticks);
1825
1826 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1827 extern int wake_up_process(struct task_struct *tsk);
1828 extern void wake_up_new_task(struct task_struct *tsk,
1829                                 unsigned long clone_flags);
1830 #ifdef CONFIG_SMP
1831  extern void kick_process(struct task_struct *tsk);
1832 #else
1833  static inline void kick_process(struct task_struct *tsk) { }
1834 #endif
1835 extern void sched_fork(struct task_struct *p, int clone_flags);
1836 extern void sched_dead(struct task_struct *p);
1837
1838 extern void proc_caches_init(void);
1839 extern void flush_signals(struct task_struct *);
1840 extern void ignore_signals(struct task_struct *);
1841 extern void flush_signal_handlers(struct task_struct *, int force_default);
1842 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1843
1844 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1845 {
1846         unsigned long flags;
1847         int ret;
1848
1849         spin_lock_irqsave(&tsk->sighand->siglock, flags);
1850         ret = dequeue_signal(tsk, mask, info);
1851         spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1852
1853         return ret;
1854 }
1855
1856 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1857                               sigset_t *mask);
1858 extern void unblock_all_signals(void);
1859 extern void release_task(struct task_struct * p);
1860 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1861 extern int force_sigsegv(int, struct task_struct *);
1862 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1863 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1864 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1865 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1866 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1867 extern int kill_pid(struct pid *pid, int sig, int priv);
1868 extern int kill_proc_info(int, struct siginfo *, pid_t);
1869 extern int do_notify_parent(struct task_struct *, int);
1870 extern void force_sig(int, struct task_struct *);
1871 extern void force_sig_specific(int, struct task_struct *);
1872 extern int send_sig(int, struct task_struct *, int);
1873 extern void zap_other_threads(struct task_struct *p);
1874 extern struct sigqueue *sigqueue_alloc(void);
1875 extern void sigqueue_free(struct sigqueue *);
1876 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
1877 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1878 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1879
1880 static inline int kill_cad_pid(int sig, int priv)
1881 {
1882         return kill_pid(cad_pid, sig, priv);
1883 }
1884
1885 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1886 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1887 #define SEND_SIG_PRIV   ((struct siginfo *) 1)
1888 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1889
1890 static inline int is_si_special(const struct siginfo *info)
1891 {
1892         return info <= SEND_SIG_FORCED;
1893 }
1894
1895 /* True if we are on the alternate signal stack.  */
1896
1897 static inline int on_sig_stack(unsigned long sp)
1898 {
1899         return (sp - current->sas_ss_sp < current->sas_ss_size);
1900 }
1901
1902 static inline int sas_ss_flags(unsigned long sp)
1903 {
1904         return (current->sas_ss_size == 0 ? SS_DISABLE
1905                 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1906 }
1907
1908 /*
1909  * Routines for handling mm_structs
1910  */
1911 extern struct mm_struct * mm_alloc(void);
1912
1913 /* mmdrop drops the mm and the page tables */
1914 extern void __mmdrop(struct mm_struct *);
1915 static inline void mmdrop(struct mm_struct * mm)
1916 {
1917         if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1918                 __mmdrop(mm);
1919 }
1920
1921 /* mmput gets rid of the mappings and all user-space */
1922 extern void mmput(struct mm_struct *);
1923 /* Grab a reference to a task's mm, if it is not already going away */
1924 extern struct mm_struct *get_task_mm(struct task_struct *task);
1925 /* Remove the current tasks stale references to the old mm_struct */
1926 extern void mm_release(struct task_struct *, struct mm_struct *);
1927 /* Allocate a new mm structure and copy contents from tsk->mm */
1928 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1929
1930 extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1931 extern void flush_thread(void);
1932 extern void exit_thread(void);
1933
1934 extern void exit_files(struct task_struct *);
1935 extern void __cleanup_signal(struct signal_struct *);
1936 extern void __cleanup_sighand(struct sighand_struct *);
1937
1938 extern void exit_itimers(struct signal_struct *);
1939 extern void flush_itimer_signals(void);
1940
1941 extern NORET_TYPE void do_group_exit(int);
1942
1943 extern void daemonize(const char *, ...);
1944 extern int allow_signal(int);
1945 extern int disallow_signal(int);
1946
1947 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1948 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1949 struct task_struct *fork_idle(int);
1950
1951 extern void set_task_comm(struct task_struct *tsk, char *from);
1952 extern char *get_task_comm(char *to, struct task_struct *tsk);
1953
1954 #ifdef CONFIG_SMP
1955 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1956 #else
1957 static inline unsigned long wait_task_inactive(struct task_struct *p,
1958                                                long match_state)
1959 {
1960         return 1;
1961 }
1962 #endif
1963
1964 #define next_task(p)    list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1965
1966 #define for_each_process(p) \
1967         for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1968
1969 extern bool is_single_threaded(struct task_struct *);
1970
1971 /*
1972  * Careful: do_each_thread/while_each_thread is a double loop so
1973  *          'break' will not work as expected - use goto instead.
1974  */
1975 #define do_each_thread(g, t) \
1976         for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1977
1978 #define while_each_thread(g, t) \
1979         while ((t = next_thread(t)) != g)
1980
1981 /* de_thread depends on thread_group_leader not being a pid based check */
1982 #define thread_group_leader(p)  (p == p->group_leader)
1983
1984 /* Do to the insanities of de_thread it is possible for a process
1985  * to have the pid of the thread group leader without actually being
1986  * the thread group leader.  For iteration through the pids in proc
1987  * all we care about is that we have a task with the appropriate
1988  * pid, we don't actually care if we have the right task.
1989  */
1990 static inline int has_group_leader_pid(struct task_struct *p)
1991 {
1992         return p->pid == p->tgid;
1993 }
1994
1995 static inline
1996 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1997 {
1998         return p1->tgid == p2->tgid;
1999 }
2000
2001 static inline struct task_struct *next_thread(const struct task_struct *p)
2002 {
2003         return list_entry(rcu_dereference(p->thread_group.next),
2004                           struct task_struct, thread_group);
2005 }
2006
2007 static inline int thread_group_empty(struct task_struct *p)
2008 {
2009         return list_empty(&p->thread_group);
2010 }
2011
2012 #define delay_group_leader(p) \
2013                 (thread_group_leader(p) && !thread_group_empty(p))
2014
2015 /*
2016  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2017  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2018  * pins the final release of task.io_context.  Also protects ->cpuset and
2019  * ->cgroup.subsys[].
2020  *
2021  * Nests both inside and outside of read_lock(&tasklist_lock).
2022  * It must not be nested with write_lock_irq(&tasklist_lock),
2023  * neither inside nor outside.
2024  */
2025 static inline void task_lock(struct task_struct *p)
2026 {
2027         spin_lock(&p->alloc_lock);
2028 }
2029
2030 static inline void task_unlock(struct task_struct *p)
2031 {
2032         spin_unlock(&p->alloc_lock);
2033 }
2034
2035 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2036                                                         unsigned long *flags);
2037
2038 static inline void unlock_task_sighand(struct task_struct *tsk,
2039                                                 unsigned long *flags)
2040 {
2041         spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2042 }
2043
2044 #ifndef __HAVE_THREAD_FUNCTIONS
2045
2046 #define task_thread_info(task)  ((struct thread_info *)(task)->stack)
2047 #define task_stack_page(task)   ((task)->stack)
2048
2049 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2050 {
2051         *task_thread_info(p) = *task_thread_info(org);
2052         task_thread_info(p)->task = p;
2053 }
2054
2055 static inline unsigned long *end_of_stack(struct task_struct *p)
2056 {
2057         return (unsigned long *)(task_thread_info(p) + 1);
2058 }
2059
2060 #endif
2061
2062 static inline int object_is_on_stack(void *obj)
2063 {
2064         void *stack = task_stack_page(current);
2065
2066         return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2067 }
2068
2069 extern void thread_info_cache_init(void);
2070
2071 #ifdef CONFIG_DEBUG_STACK_USAGE
2072 static inline unsigned long stack_not_used(struct task_struct *p)
2073 {
2074         unsigned long *n = end_of_stack(p);
2075
2076         do {    /* Skip over canary */
2077                 n++;
2078         } while (!*n);
2079
2080         return (unsigned long)n - (unsigned long)end_of_stack(p);
2081 }
2082 #endif
2083
2084 /* set thread flags in other task's structures
2085  * - see asm/thread_info.h for TIF_xxxx flags available
2086  */
2087 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2088 {
2089         set_ti_thread_flag(task_thread_info(tsk), flag);
2090 }
2091
2092 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2093 {
2094         clear_ti_thread_flag(task_thread_info(tsk), flag);
2095 }
2096
2097 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2098 {
2099         return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2100 }
2101
2102 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2103 {
2104         return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2105 }
2106
2107 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2108 {
2109         return test_ti_thread_flag(task_thread_info(tsk), flag);
2110 }
2111
2112 static inline void set_tsk_need_resched(struct task_struct *tsk)
2113 {
2114         set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2115 }
2116
2117 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2118 {
2119         clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2120 }
2121
2122 static inline int test_tsk_need_resched(struct task_struct *tsk)
2123 {
2124         return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2125 }
2126
2127 static inline int signal_pending(struct task_struct *p)
2128 {
2129         return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2130 }
2131
2132 extern int __fatal_signal_pending(struct task_struct *p);
2133
2134 static inline int fatal_signal_pending(struct task_struct *p)
2135 {
2136         return signal_pending(p) && __fatal_signal_pending(p);
2137 }
2138
2139 static inline int signal_pending_state(long state, struct task_struct *p)
2140 {
2141         if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2142                 return 0;
2143         if (!signal_pending(p))
2144                 return 0;
2145
2146         return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2147 }
2148
2149 static inline int need_resched(void)
2150 {
2151         return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2152 }
2153
2154 /*
2155  * cond_resched() and cond_resched_lock(): latency reduction via
2156  * explicit rescheduling in places that are safe. The return
2157  * value indicates whether a reschedule was done in fact.
2158  * cond_resched_lock() will drop the spinlock before scheduling,
2159  * cond_resched_softirq() will enable bhs before scheduling.
2160  */
2161 extern int _cond_resched(void);
2162 #ifdef CONFIG_PREEMPT_BKL
2163 static inline int cond_resched(void)
2164 {
2165         return 0;
2166 }
2167 #else
2168 static inline int cond_resched(void)
2169 {
2170         return _cond_resched();
2171 }
2172 #endif
2173 extern int cond_resched_lock(spinlock_t * lock);
2174 extern int cond_resched_softirq(void);
2175 static inline int cond_resched_bkl(void)
2176 {
2177         return _cond_resched();
2178 }
2179
2180 /*
2181  * Does a critical section need to be broken due to another
2182  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2183  * but a general need for low latency)
2184  */
2185 static inline int spin_needbreak(spinlock_t *lock)
2186 {
2187 #ifdef CONFIG_PREEMPT
2188         return spin_is_contended(lock);
2189 #else
2190         return 0;
2191 #endif
2192 }
2193
2194 /*
2195  * Thread group CPU time accounting.
2196  */
2197
2198 extern int thread_group_cputime_alloc(struct task_struct *);
2199 extern void thread_group_cputime(struct task_struct *, struct task_cputime *);
2200
2201 static inline void thread_group_cputime_init(struct signal_struct *sig)
2202 {
2203         sig->cputime.totals = NULL;
2204 }
2205
2206 static inline int thread_group_cputime_clone_thread(struct task_struct *curr)
2207 {
2208         if (curr->signal->cputime.totals)
2209                 return 0;
2210         return thread_group_cputime_alloc(curr);
2211 }
2212
2213 static inline void thread_group_cputime_free(struct signal_struct *sig)
2214 {
2215         free_percpu(sig->cputime.totals);
2216 }
2217
2218 /*
2219  * Reevaluate whether the task has signals pending delivery.
2220  * Wake the task if so.
2221  * This is required every time the blocked sigset_t changes.
2222  * callers must hold sighand->siglock.
2223  */
2224 extern void recalc_sigpending_and_wake(struct task_struct *t);
2225 extern void recalc_sigpending(void);
2226
2227 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2228
2229 /*
2230  * Wrappers for p->thread_info->cpu access. No-op on UP.
2231  */
2232 #ifdef CONFIG_SMP
2233
2234 static inline unsigned int task_cpu(const struct task_struct *p)
2235 {
2236         return task_thread_info(p)->cpu;
2237 }
2238
2239 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2240
2241 #else
2242
2243 static inline unsigned int task_cpu(const struct task_struct *p)
2244 {
2245         return 0;
2246 }
2247
2248 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2249 {
2250 }
2251
2252 #endif /* CONFIG_SMP */
2253
2254 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2255
2256 #ifdef CONFIG_TRACING
2257 extern void
2258 __trace_special(void *__tr, void *__data,
2259                 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2260 #else
2261 static inline void
2262 __trace_special(void *__tr, void *__data,
2263                 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2264 {
2265 }
2266 #endif
2267
2268 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2269 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2270
2271 extern void normalize_rt_tasks(void);
2272
2273 #ifdef CONFIG_GROUP_SCHED
2274
2275 extern struct task_group init_task_group;
2276 #ifdef CONFIG_USER_SCHED
2277 extern struct task_group root_task_group;
2278 extern void set_tg_uid(struct user_struct *user);
2279 #endif
2280
2281 extern struct task_group *sched_create_group(struct task_group *parent);
2282 extern void sched_destroy_group(struct task_group *tg);
2283 extern void sched_move_task(struct task_struct *tsk);
2284 #ifdef CONFIG_FAIR_GROUP_SCHED
2285 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2286 extern unsigned long sched_group_shares(struct task_group *tg);
2287 #endif
2288 #ifdef CONFIG_RT_GROUP_SCHED
2289 extern int sched_group_set_rt_runtime(struct task_group *tg,
2290                                       long rt_runtime_us);
2291 extern long sched_group_rt_runtime(struct task_group *tg);
2292 extern int sched_group_set_rt_period(struct task_group *tg,
2293                                       long rt_period_us);
2294 extern long sched_group_rt_period(struct task_group *tg);
2295 #endif
2296 #endif
2297
2298 #ifdef CONFIG_TASK_XACCT
2299 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2300 {
2301         tsk->ioac.rchar += amt;
2302 }
2303
2304 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2305 {
2306         tsk->ioac.wchar += amt;
2307 }
2308
2309 static inline void inc_syscr(struct task_struct *tsk)
2310 {
2311         tsk->ioac.syscr++;
2312 }
2313
2314 static inline void inc_syscw(struct task_struct *tsk)
2315 {
2316         tsk->ioac.syscw++;
2317 }
2318 #else
2319 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2320 {
2321 }
2322
2323 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2324 {
2325 }
2326
2327 static inline void inc_syscr(struct task_struct *tsk)
2328 {
2329 }
2330
2331 static inline void inc_syscw(struct task_struct *tsk)
2332 {
2333 }
2334 #endif
2335
2336 #ifndef TASK_SIZE_OF
2337 #define TASK_SIZE_OF(tsk)       TASK_SIZE
2338 #endif
2339
2340 /*
2341  * Call the function if the target task is executing on a CPU right now:
2342  */
2343 extern void task_oncpu_function_call(struct task_struct *p,
2344                                      void (*func) (void *info), void *info);
2345
2346
2347 #ifdef CONFIG_MM_OWNER
2348 extern void mm_update_next_owner(struct mm_struct *mm);
2349 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2350 #else
2351 static inline void mm_update_next_owner(struct mm_struct *mm)
2352 {
2353 }
2354
2355 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2356 {
2357 }
2358 #endif /* CONFIG_MM_OWNER */
2359
2360 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2361
2362 #endif /* __KERNEL__ */
2363
2364 #endif