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