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/path.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 struct fs_struct;
 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 extern unsigned long get_parent_ip(unsigned long addr);
 142
 143 struct seq_file;
 144 struct cfs_rq;
 145 struct task_group;
 146 #ifdef CONFIG_SCHED_DEBUG
 147 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
 148 extern void proc_sched_set_task(struct task_struct *p);
 149 extern void
 150 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
 151 #else
 152 static inline void
 153 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
 154 {
 155 }
 156 static inline void proc_sched_set_task(struct task_struct *p)
 157 {
 158 }
 159 static inline void
 160 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
 161 {
 162 }
 163 #endif
 164
 165 extern unsigned long long time_sync_thresh;
 166
 167 /*
 168  * Task state bitmask. NOTE! These bits are also
 169  * encoded in fs/proc/array.c: get_task_state().
 170  *
 171  * We have two separate sets of flags: task->state
 172  * is about runnability, while task->exit_state are
 173  * about the task exiting. Confusing, but this way
 174  * modifying one set can't modify the other one by
 175  * mistake.
 176  */
 177 #define TASK_RUNNING            0
 178 #define TASK_INTERRUPTIBLE      1
 179 #define TASK_UNINTERRUPTIBLE    2
 180 #define __TASK_STOPPED          4
 181 #define __TASK_TRACED           8
 182 /* in tsk->exit_state */
 183 #define EXIT_ZOMBIE             16
 184 #define EXIT_DEAD               32
 185 /* in tsk->state again */
 186 #define TASK_DEAD               64
 187 #define TASK_WAKEKILL           128
 188
 189 /* Convenience macros for the sake of set_task_state */
 190 #define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
 191 #define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
 192 #define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
 193
 194 /* Convenience macros for the sake of wake_up */
 195 #define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 196 #define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
 197
 198 /* get_task_state() */
 199 #define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 200                                  TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 201                                  __TASK_TRACED)
 202
 203 #define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
 204 #define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
 205 #define task_is_stopped_or_traced(task) \
 206                         ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 207 #define task_contributes_to_load(task)  \
 208                                 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
 209                                  (task->flags & PF_FROZEN) == 0)
 210
 211 #define __set_task_state(tsk, state_value)              \
 212         do { (tsk)->state = (state_value); } while (0)
 213 #define set_task_state(tsk, state_value)                \
 214         set_mb((tsk)->state, (state_value))
 215
 216 /*
 217  * set_current_state() includes a barrier so that the write of current->state
 218  * is correctly serialised wrt the caller's subsequent test of whether to
 219  * actually sleep:
 220  *
 221  *      set_current_state(TASK_UNINTERRUPTIBLE);
 222  *      if (do_i_need_to_sleep())
 223  *              schedule();
 224  *
 225  * If the caller does not need such serialisation then use __set_current_state()
 226  */
 227 #define __set_current_state(state_value)                        \
 228         do { current->state = (state_value); } while (0)
 229 #define set_current_state(state_value)          \
 230         set_mb(current->state, (state_value))
 231
 232 /* Task command name length */
 233 #define TASK_COMM_LEN 16
 234
 235 #include <linux/spinlock.h>
 236
 237 /*
 238  * This serializes "schedule()" and also protects
 239  * the run-queue from deletions/modifications (but
 240  * _adding_ to the beginning of the run-queue has
 241  * a separate lock).
 242  */
 243 extern rwlock_t tasklist_lock;
 244 extern spinlock_t mmlist_lock;
 245
 246 struct task_struct;
 247
 248 extern void sched_init(void);
 249 extern void sched_init_smp(void);
 250 extern asmlinkage void schedule_tail(struct task_struct *prev);
 251 extern void init_idle(struct task_struct *idle, int cpu);
 252 extern void init_idle_bootup_task(struct task_struct *idle);
 253
 254 extern int runqueue_is_locked(void);
 255 extern void task_rq_unlock_wait(struct task_struct *p);
 256
 257 extern cpumask_var_t nohz_cpu_mask;
 258 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
 259 extern int select_nohz_load_balancer(int cpu);
 260 #else
 261 static inline int select_nohz_load_balancer(int cpu)
 262 {
 263         return 0;
 264 }
 265 #endif
 266
 267 /*
 268  * Only dump TASK_* tasks. (0 for all tasks)
 269  */
 270 extern void show_state_filter(unsigned long state_filter);
 271
 272 static inline void show_state(void)
 273 {
 274         show_state_filter(0);
 275 }
 276
 277 extern void show_regs(struct pt_regs *);
 278
 279 /*
 280  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 281  * task), SP is the stack pointer of the first frame that should be shown in the back
 282  * trace (or NULL if the entire call-chain of the task should be shown).
 283  */
 284 extern void show_stack(struct task_struct *task, unsigned long *sp);
 285
 286 void io_schedule(void);
 287 long io_schedule_timeout(long timeout);
 288
 289 extern void cpu_init (void);
 290 extern void trap_init(void);
 291 extern void update_process_times(int user);
 292 extern void scheduler_tick(void);
 293
 294 extern void sched_show_task(struct task_struct *p);
 295
 296 #ifdef CONFIG_DETECT_SOFTLOCKUP
 297 extern void softlockup_tick(void);
 298 extern void touch_softlockup_watchdog(void);
 299 extern void touch_all_softlockup_watchdogs(void);
 300 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
 301                                     struct file *filp, void __user *buffer,
 302                                     size_t *lenp, loff_t *ppos);
 303 extern unsigned int  softlockup_panic;
 304 extern int softlockup_thresh;
 305 #else
 306 static inline void softlockup_tick(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 #ifdef CONFIG_DETECT_HUNG_TASK
 318 extern unsigned int  sysctl_hung_task_panic;
 319 extern unsigned long sysctl_hung_task_check_count;
 320 extern unsigned long sysctl_hung_task_timeout_secs;
 321 extern unsigned long sysctl_hung_task_warnings;
 322 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
 323                                          struct file *filp, void __user *buffer,
 324                                          size_t *lenp, loff_t *ppos);
 325 #endif
 326
 327 /* Attach to any functions which should be ignored in wchan output. */
 328 #define __sched         __attribute__((__section__(".sched.text")))
 329
 330 /* Linker adds these: start and end of __sched functions */
 331 extern char __sched_text_start[], __sched_text_end[];
 332
 333 /* Is this address in the __sched functions? */
 334 extern int in_sched_functions(unsigned long addr);
 335
 336 #define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 337 extern signed long schedule_timeout(signed long timeout);
 338 extern signed long schedule_timeout_interruptible(signed long timeout);
 339 extern signed long schedule_timeout_killable(signed long timeout);
 340 extern signed long schedule_timeout_uninterruptible(signed long timeout);
 341 asmlinkage void __schedule(void);
 342 asmlinkage void schedule(void);
 343 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
 344
 345 struct nsproxy;
 346 struct user_namespace;
 347
 348 /* Maximum number of active map areas.. This is a random (large) number */
 349 #define DEFAULT_MAX_MAP_COUNT   65536
 350
 351 extern int sysctl_max_map_count;
 352
 353 #include <linux/aio.h>
 354
 355 extern unsigned long
 356 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 357                        unsigned long, unsigned long);
 358 extern unsigned long
 359 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 360                           unsigned long len, unsigned long pgoff,
 361                           unsigned long flags);
 362 extern void arch_unmap_area(struct mm_struct *, unsigned long);
 363 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
 364
 365 #if USE_SPLIT_PTLOCKS
 366 /*
 367  * The mm counters are not protected by its page_table_lock,
 368  * so must be incremented atomically.
 369  */
 370 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
 371 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
 372 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
 373 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
 374 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
 375
 376 #else  /* !USE_SPLIT_PTLOCKS */
 377 /*
 378  * The mm counters are protected by its page_table_lock,
 379  * so can be incremented directly.
 380  */
 381 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
 382 #define get_mm_counter(mm, member) ((mm)->_##member)
 383 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
 384 #define inc_mm_counter(mm, member) (mm)->_##member++
 385 #define dec_mm_counter(mm, member) (mm)->_##member--
 386
 387 #endif /* !USE_SPLIT_PTLOCKS */
 388
 389 #define get_mm_rss(mm)                                  \
 390         (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
 391 #define update_hiwater_rss(mm)  do {                    \
 392         unsigned long _rss = get_mm_rss(mm);            \
 393         if ((mm)->hiwater_rss < _rss)                   \
 394                 (mm)->hiwater_rss = _rss;               \
 395 } while (0)
 396 #define update_hiwater_vm(mm)   do {                    \
 397         if ((mm)->hiwater_vm < (mm)->total_vm)          \
 398                 (mm)->hiwater_vm = (mm)->total_vm;      \
 399 } while (0)
 400
 401 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
 402 {
 403         return max(mm->hiwater_rss, get_mm_rss(mm));
 404 }
 405
 406 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
 407 {
 408         return max(mm->hiwater_vm, mm->total_vm);
 409 }
 410
 411 extern void set_dumpable(struct mm_struct *mm, int value);
 412 extern int get_dumpable(struct mm_struct *mm);
 413
 414 /* mm flags */
 415 /* dumpable bits */
 416 #define MMF_DUMPABLE      0  /* core dump is permitted */
 417 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
 418 #define MMF_DUMPABLE_BITS 2
 419
 420 /* coredump filter bits */
 421 #define MMF_DUMP_ANON_PRIVATE   2
 422 #define MMF_DUMP_ANON_SHARED    3
 423 #define MMF_DUMP_MAPPED_PRIVATE 4
 424 #define MMF_DUMP_MAPPED_SHARED  5
 425 #define MMF_DUMP_ELF_HEADERS    6
 426 #define MMF_DUMP_HUGETLB_PRIVATE 7
 427 #define MMF_DUMP_HUGETLB_SHARED  8
 428 #define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
 429 #define MMF_DUMP_FILTER_BITS    7
 430 #define MMF_DUMP_FILTER_MASK \
 431         (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
 432 #define MMF_DUMP_FILTER_DEFAULT \
 433         ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
 434          (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
 435
 436 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
 437 # define MMF_DUMP_MASK_DEFAULT_ELF      (1 << MMF_DUMP_ELF_HEADERS)
 438 #else
 439 # define MMF_DUMP_MASK_DEFAULT_ELF      0
 440 #endif
 441
 442 struct sighand_struct {
 443         atomic_t                count;
 444         struct k_sigaction      action[_NSIG];
 445         spinlock_t              siglock;
 446         wait_queue_head_t       signalfd_wqh;
 447 };
 448
 449 struct pacct_struct {
 450         int                     ac_flag;
 451         long                    ac_exitcode;
 452         unsigned long           ac_mem;
 453         cputime_t               ac_utime, ac_stime;
 454         unsigned long           ac_minflt, ac_majflt;
 455 };
 456
 457 /**
 458  * struct task_cputime - collected CPU time counts
 459  * @utime:              time spent in user mode, in &cputime_t units
 460  * @stime:              time spent in kernel mode, in &cputime_t units
 461  * @sum_exec_runtime:   total time spent on the CPU, in nanoseconds
 462  *
 463  * This structure groups together three kinds of CPU time that are
 464  * tracked for threads and thread groups.  Most things considering
 465  * CPU time want to group these counts together and treat all three
 466  * of them in parallel.
 467  */
 468 struct task_cputime {
 469         cputime_t utime;
 470         cputime_t stime;
 471         unsigned long long sum_exec_runtime;
 472 };
 473 /* Alternate field names when used to cache expirations. */
 474 #define prof_exp        stime
 475 #define virt_exp        utime
 476 #define sched_exp       sum_exec_runtime
 477
 478 #define INIT_CPUTIME    \
 479         (struct task_cputime) {                                 \
 480                 .utime = cputime_zero,                          \
 481                 .stime = cputime_zero,                          \
 482                 .sum_exec_runtime = 0,                          \
 483         }
 484
 485 /**
 486  * struct thread_group_cputimer - thread group interval timer counts
 487  * @cputime:            thread group interval timers.
 488  * @running:            non-zero when there are timers running and
 489  *                      @cputime receives updates.
 490  * @lock:               lock for fields in this struct.
 491  *
 492  * This structure contains the version of task_cputime, above, that is
 493  * used for thread group CPU timer calculations.
 494  */
 495 struct thread_group_cputimer {
 496         struct task_cputime cputime;
 497         int running;
 498         spinlock_t lock;
 499 };
 500
 501 /*
 502  * NOTE! "signal_struct" does not have it's own
 503  * locking, because a shared signal_struct always
 504  * implies a shared sighand_struct, so locking
 505  * sighand_struct is always a proper superset of
 506  * the locking of signal_struct.
 507  */
 508 struct signal_struct {
 509         atomic_t                count;
 510         atomic_t                live;
 511
 512         wait_queue_head_t       wait_chldexit;  /* for wait4() */
 513
 514         /* current thread group signal load-balancing target: */
 515         struct task_struct      *curr_target;
 516
 517         /* shared signal handling: */
 518         struct sigpending       shared_pending;
 519
 520         /* thread group exit support */
 521         int                     group_exit_code;
 522         /* overloaded:
 523          * - notify group_exit_task when ->count is equal to notify_count
 524          * - everyone except group_exit_task is stopped during signal delivery
 525          *   of fatal signals, group_exit_task processes the signal.
 526          */
 527         int                     notify_count;
 528         struct task_struct      *group_exit_task;
 529
 530         /* thread group stop support, overloads group_exit_code too */
 531         int                     group_stop_count;
 532         unsigned int            flags; /* see SIGNAL_* flags below */
 533
 534         /* POSIX.1b Interval Timers */
 535         struct list_head posix_timers;
 536
 537         /* ITIMER_REAL timer for the process */
 538         struct hrtimer real_timer;
 539         struct pid *leader_pid;
 540         ktime_t it_real_incr;
 541
 542         /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
 543         cputime_t it_prof_expires, it_virt_expires;
 544         cputime_t it_prof_incr, it_virt_incr;
 545
 546         /*
 547          * Thread group totals for process CPU timers.
 548          * See thread_group_cputimer(), et al, for details.
 549          */
 550         struct thread_group_cputimer cputimer;
 551
 552         /* Earliest-expiration cache. */
 553         struct task_cputime cputime_expires;
 554
 555         struct list_head cpu_timers[3];
 556
 557         struct pid *tty_old_pgrp;
 558
 559         /* boolean value for session group leader */
 560         int leader;
 561
 562         struct tty_struct *tty; /* NULL if no tty */
 563
 564         /*
 565          * Cumulative resource counters for dead threads in the group,
 566          * and for reaped dead child processes forked by this group.
 567          * Live threads maintain their own counters and add to these
 568          * in __exit_signal, except for the group leader.
 569          */
 570         cputime_t utime, stime, cutime, cstime;
 571         cputime_t gtime;
 572         cputime_t cgtime;
 573         unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 574         unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 575         unsigned long inblock, oublock, cinblock, coublock;
 576         struct task_io_accounting ioac;
 577
 578         /*
 579          * Cumulative ns of schedule CPU time fo dead threads in the
 580          * group, not including a zombie group leader, (This only differs
 581          * from jiffies_to_ns(utime + stime) if sched_clock uses something
 582          * other than jiffies.)
 583          */
 584         unsigned long long sum_sched_runtime;
 585
 586         /*
 587          * We don't bother to synchronize most readers of this at all,
 588          * because there is no reader checking a limit that actually needs
 589          * to get both rlim_cur and rlim_max atomically, and either one
 590          * alone is a single word that can safely be read normally.
 591          * getrlimit/setrlimit use task_lock(current->group_leader) to
 592          * protect this instead of the siglock, because they really
 593          * have no need to disable irqs.
 594          */
 595         struct rlimit rlim[RLIM_NLIMITS];
 596
 597 #ifdef CONFIG_BSD_PROCESS_ACCT
 598         struct pacct_struct pacct;      /* per-process accounting information */
 599 #endif
 600 #ifdef CONFIG_TASKSTATS
 601         struct taskstats *stats;
 602 #endif
 603 #ifdef CONFIG_AUDIT
 604         unsigned audit_tty;
 605         struct tty_audit_buf *tty_audit_buf;
 606 #endif
 607 };
 608
 609 /* Context switch must be unlocked if interrupts are to be enabled */
 610 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
 611 # define __ARCH_WANT_UNLOCKED_CTXSW
 612 #endif
 613
 614 /*
 615  * Bits in flags field of signal_struct.
 616  */
 617 #define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 618 #define SIGNAL_STOP_DEQUEUED    0x00000002 /* stop signal dequeued */
 619 #define SIGNAL_STOP_CONTINUED   0x00000004 /* SIGCONT since WCONTINUED reap */
 620 #define SIGNAL_GROUP_EXIT       0x00000008 /* group exit in progress */
 621 /*
 622  * Pending notifications to parent.
 623  */
 624 #define SIGNAL_CLD_STOPPED      0x00000010
 625 #define SIGNAL_CLD_CONTINUED    0x00000020
 626 #define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
 627
 628 #define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
 629
 630 /* If true, all threads except ->group_exit_task have pending SIGKILL */
 631 static inline int signal_group_exit(const struct signal_struct *sig)
 632 {
 633         return  (sig->flags & SIGNAL_GROUP_EXIT) ||
 634                 (sig->group_exit_task != NULL);
 635 }
 636
 637 /*
 638  * Some day this will be a full-fledged user tracking system..
 639  */
 640 struct user_struct {
 641         atomic_t __count;       /* reference count */
 642         atomic_t processes;     /* How many processes does this user have? */
 643         atomic_t files;         /* How many open files does this user have? */
 644         atomic_t sigpending;    /* How many pending signals does this user have? */
 645 #ifdef CONFIG_INOTIFY_USER
 646         atomic_t inotify_watches; /* How many inotify watches does this user have? */
 647         atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 648 #endif
 649 #ifdef CONFIG_EPOLL
 650         atomic_t epoll_watches; /* The number of file descriptors currently watched */
 651 #endif
 652 #ifdef CONFIG_POSIX_MQUEUE
 653         /* protected by mq_lock */
 654         unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 655 #endif
 656         unsigned long locked_shm; /* How many pages of mlocked shm ? */
 657
 658 #ifdef CONFIG_KEYS
 659         struct key *uid_keyring;        /* UID specific keyring */
 660         struct key *session_keyring;    /* UID's default session keyring */
 661 #endif
 662
 663         /* Hash table maintenance information */
 664         struct hlist_node uidhash_node;
 665         uid_t uid;
 666         struct user_namespace *user_ns;
 667
 668 #ifdef CONFIG_USER_SCHED
 669         struct task_group *tg;
 670 #ifdef CONFIG_SYSFS
 671         struct kobject kobj;
 672         struct work_struct work;
 673 #endif
 674 #endif
 675 };
 676
 677 extern int uids_sysfs_init(void);
 678
 679 extern struct user_struct *find_user(uid_t);
 680
 681 extern struct user_struct root_user;
 682 #define INIT_USER (&root_user)
 683
 684
 685 struct backing_dev_info;
 686 struct reclaim_state;
 687
 688 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 689 struct sched_info {
 690         /* cumulative counters */
 691         unsigned long pcount;         /* # of times run on this cpu */
 692         unsigned long long run_delay; /* time spent waiting on a runqueue */
 693
 694         /* timestamps */
 695         unsigned long long last_arrival,/* when we last ran on a cpu */
 696                            last_queued; /* when we were last queued to run */
 697 #ifdef CONFIG_SCHEDSTATS
 698         /* BKL stats */
 699         unsigned int bkl_count;
 700 #endif
 701 };
 702 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
 703
 704 #ifdef CONFIG_TASK_DELAY_ACCT
 705 struct task_delay_info {
 706         spinlock_t      lock;
 707         unsigned int    flags;  /* Private per-task flags */
 708
 709         /* For each stat XXX, add following, aligned appropriately
 710          *
 711          * struct timespec XXX_start, XXX_end;
 712          * u64 XXX_delay;
 713          * u32 XXX_count;
 714          *
 715          * Atomicity of updates to XXX_delay, XXX_count protected by
 716          * single lock above (split into XXX_lock if contention is an issue).
 717          */
 718
 719         /*
 720          * XXX_count is incremented on every XXX operation, the delay
 721          * associated with the operation is added to XXX_delay.
 722          * XXX_delay contains the accumulated delay time in nanoseconds.
 723          */
 724         struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
 725         u64 blkio_delay;        /* wait for sync block io completion */
 726         u64 swapin_delay;       /* wait for swapin block io completion */
 727         u32 blkio_count;        /* total count of the number of sync block */
 728                                 /* io operations performed */
 729         u32 swapin_count;       /* total count of the number of swapin block */
 730                                 /* io operations performed */
 731
 732         struct timespec freepages_start, freepages_end;
 733         u64 freepages_delay;    /* wait for memory reclaim */
 734         u32 freepages_count;    /* total count of memory reclaim */
 735 };
 736 #endif  /* CONFIG_TASK_DELAY_ACCT */
 737
 738 static inline int sched_info_on(void)
 739 {
 740 #ifdef CONFIG_SCHEDSTATS
 741         return 1;
 742 #elif defined(CONFIG_TASK_DELAY_ACCT)
 743         extern int delayacct_on;
 744         return delayacct_on;
 745 #else
 746         return 0;
 747 #endif
 748 }
 749
 750 enum cpu_idle_type {
 751         CPU_IDLE,
 752         CPU_NOT_IDLE,
 753         CPU_NEWLY_IDLE,
 754         CPU_MAX_IDLE_TYPES
 755 };
 756
 757 /*
 758  * sched-domains (multiprocessor balancing) declarations:
 759  */
 760
 761 /*
 762  * Increase resolution of nice-level calculations:
 763  */
 764 #define SCHED_LOAD_SHIFT        10
 765 #define SCHED_LOAD_SCALE        (1L << SCHED_LOAD_SHIFT)
 766
 767 #define SCHED_LOAD_SCALE_FUZZ   SCHED_LOAD_SCALE
 768
 769 #ifdef CONFIG_SMP
 770 #define SD_LOAD_BALANCE         1       /* Do load balancing on this domain. */
 771 #define SD_BALANCE_NEWIDLE      2       /* Balance when about to become idle */
 772 #define SD_BALANCE_EXEC         4       /* Balance on exec */
 773 #define SD_BALANCE_FORK         8       /* Balance on fork, clone */
 774 #define SD_WAKE_IDLE            16      /* Wake to idle CPU on task wakeup */
 775 #define SD_WAKE_AFFINE          32      /* Wake task to waking CPU */
 776 #define SD_WAKE_BALANCE         64      /* Perform balancing at task wakeup */
 777 #define SD_SHARE_CPUPOWER       128     /* Domain members share cpu power */
 778 #define SD_POWERSAVINGS_BALANCE 256     /* Balance for power savings */
 779 #define SD_SHARE_PKG_RESOURCES  512     /* Domain members share cpu pkg resources */
 780 #define SD_SERIALIZE            1024    /* Only a single load balancing instance */
 781 #define SD_WAKE_IDLE_FAR        2048    /* Gain latency sacrificing cache hit */
 782
 783 enum powersavings_balance_level {
 784         POWERSAVINGS_BALANCE_NONE = 0,  /* No power saving load balance */
 785         POWERSAVINGS_BALANCE_BASIC,     /* Fill one thread/core/package
 786                                          * first for long running threads
 787                                          */
 788         POWERSAVINGS_BALANCE_WAKEUP,    /* Also bias task wakeups to semi-idle
 789                                          * cpu package for power savings
 790                                          */
 791         MAX_POWERSAVINGS_BALANCE_LEVELS
 792 };
 793
 794 extern int sched_mc_power_savings, sched_smt_power_savings;
 795
 796 static inline int sd_balance_for_mc_power(void)
 797 {
 798         if (sched_smt_power_savings)
 799                 return SD_POWERSAVINGS_BALANCE;
 800
 801         return 0;
 802 }
 803
 804 static inline int sd_balance_for_package_power(void)
 805 {
 806         if (sched_mc_power_savings | sched_smt_power_savings)
 807                 return SD_POWERSAVINGS_BALANCE;
 808
 809         return 0;
 810 }
 811
 812 /*
 813  * Optimise SD flags for power savings:
 814  * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
 815  * Keep default SD flags if sched_{smt,mc}_power_saving=0
 816  */
 817
 818 static inline int sd_power_saving_flags(void)
 819 {
 820         if (sched_mc_power_savings | sched_smt_power_savings)
 821                 return SD_BALANCE_NEWIDLE;
 822
 823         return 0;
 824 }
 825
 826 struct sched_group {
 827         struct sched_group *next;       /* Must be a circular list */
 828
 829         /*
 830          * CPU power of this group, SCHED_LOAD_SCALE being max power for a
 831          * single CPU. This is read only (except for setup, hotplug CPU).
 832          * Note : Never change cpu_power without recompute its reciprocal
 833          */
 834         unsigned int __cpu_power;
 835         /*
 836          * reciprocal value of cpu_power to avoid expensive divides
 837          * (see include/linux/reciprocal_div.h)
 838          */
 839         u32 reciprocal_cpu_power;
 840
 841         unsigned long cpumask[];
 842 };
 843
 844 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
 845 {
 846         return to_cpumask(sg->cpumask);
 847 }
 848
 849 enum sched_domain_level {
 850         SD_LV_NONE = 0,
 851         SD_LV_SIBLING,
 852         SD_LV_MC,
 853         SD_LV_CPU,
 854         SD_LV_NODE,
 855         SD_LV_ALLNODES,
 856         SD_LV_MAX
 857 };
 858
 859 struct sched_domain_attr {
 860         int relax_domain_level;
 861 };
 862
 863 #define SD_ATTR_INIT    (struct sched_domain_attr) {    \
 864         .relax_domain_level = -1,                       \
 865 }
 866
 867 struct sched_domain {
 868         /* These fields must be setup */
 869         struct sched_domain *parent;    /* top domain must be null terminated */
 870         struct sched_domain *child;     /* bottom domain must be null terminated */
 871         struct sched_group *groups;     /* the balancing groups of the domain */
 872         unsigned long min_interval;     /* Minimum balance interval ms */
 873         unsigned long max_interval;     /* Maximum balance interval ms */
 874         unsigned int busy_factor;       /* less balancing by factor if busy */
 875         unsigned int imbalance_pct;     /* No balance until over watermark */
 876         unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
 877         unsigned int busy_idx;
 878         unsigned int idle_idx;
 879         unsigned int newidle_idx;
 880         unsigned int wake_idx;
 881         unsigned int forkexec_idx;
 882         int flags;                      /* See SD_* */
 883         enum sched_domain_level level;
 884
 885         /* Runtime fields. */
 886         unsigned long last_balance;     /* init to jiffies. units in jiffies */
 887         unsigned int balance_interval;  /* initialise to 1. units in ms. */
 888         unsigned int nr_balance_failed; /* initialise to 0 */
 889
 890         u64 last_update;
 891
 892 #ifdef CONFIG_SCHEDSTATS
 893         /* load_balance() stats */
 894         unsigned int lb_count[CPU_MAX_IDLE_TYPES];
 895         unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
 896         unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
 897         unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
 898         unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
 899         unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
 900         unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
 901         unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
 902
 903         /* Active load balancing */
 904         unsigned int alb_count;
 905         unsigned int alb_failed;
 906         unsigned int alb_pushed;
 907
 908         /* SD_BALANCE_EXEC stats */
 909         unsigned int sbe_count;
 910         unsigned int sbe_balanced;
 911         unsigned int sbe_pushed;
 912
 913         /* SD_BALANCE_FORK stats */
 914         unsigned int sbf_count;
 915         unsigned int sbf_balanced;
 916         unsigned int sbf_pushed;
 917
 918         /* try_to_wake_up() stats */
 919         unsigned int ttwu_wake_remote;
 920         unsigned int ttwu_move_affine;
 921         unsigned int ttwu_move_balance;
 922 #endif
 923 #ifdef CONFIG_SCHED_DEBUG
 924         char *name;
 925 #endif
 926
 927         /* span of all CPUs in this domain */
 928         unsigned long span[];
 929 };
 930
 931 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
 932 {
 933         return to_cpumask(sd->span);
 934 }
 935
 936 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
 937                                     struct sched_domain_attr *dattr_new);
 938
 939 /* Test a flag in parent sched domain */
 940 static inline int test_sd_parent(struct sched_domain *sd, int flag)
 941 {
 942         if (sd->parent && (sd->parent->flags & flag))
 943                 return 1;
 944
 945         return 0;
 946 }
 947
 948 #else /* CONFIG_SMP */
 949
 950 struct sched_domain_attr;
 951
 952 static inline void
 953 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
 954                         struct sched_domain_attr *dattr_new)
 955 {
 956 }
 957 #endif  /* !CONFIG_SMP */
 958
 959 struct io_context;                      /* See blkdev.h */
 960
 961
 962 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
 963 extern void prefetch_stack(struct task_struct *t);
 964 #else
 965 static inline void prefetch_stack(struct task_struct *t) { }
 966 #endif
 967
 968 struct audit_context;           /* See audit.c */
 969 struct mempolicy;
 970 struct pipe_inode_info;
 971 struct uts_namespace;
 972
 973 struct rq;
 974 struct sched_domain;
 975
 976 struct sched_class {
 977         const struct sched_class *next;
 978
 979         void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
 980         void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
 981         void (*yield_task) (struct rq *rq);
 982
 983         void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
 984
 985         struct task_struct * (*pick_next_task) (struct rq *rq);
 986         void (*put_prev_task) (struct rq *rq, struct task_struct *p);
 987
 988 #ifdef CONFIG_SMP
 989         int  (*select_task_rq)(struct task_struct *p, int sync);
 990
 991         unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
 992                         struct rq *busiest, unsigned long max_load_move,
 993                         struct sched_domain *sd, enum cpu_idle_type idle,
 994                         int *all_pinned, int *this_best_prio);
 995
 996         int (*move_one_task) (struct rq *this_rq, int this_cpu,
 997                               struct rq *busiest, struct sched_domain *sd,
 998                               enum cpu_idle_type idle);
 999         void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1000         int (*needs_post_schedule) (struct rq *this_rq);
1001         void (*post_schedule) (struct rq *this_rq);
1002         void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1003
1004         void (*set_cpus_allowed)(struct task_struct *p,
1005                                  const struct cpumask *newmask);
1006
1007         void (*rq_online)(struct rq *rq);
1008         void (*rq_offline)(struct rq *rq);
1009 #endif
1010
1011         void (*set_curr_task) (struct rq *rq);
1012         void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1013         void (*task_new) (struct rq *rq, struct task_struct *p);
1014
1015         void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1016                                int running);
1017         void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1018                              int running);
1019         void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1020                              int oldprio, int running);
1021
1022 #ifdef CONFIG_FAIR_GROUP_SCHED
1023         void (*moved_group) (struct task_struct *p);
1024 #endif
1025 };
1026
1027 struct load_weight {
1028         unsigned long weight, inv_weight;
1029 };
1030
1031 /*
1032  * CFS stats for a schedulable entity (task, task-group etc)
1033  *
1034  * Current field usage histogram:
1035  *
1036  *     4 se->block_start
1037  *     4 se->run_node
1038  *     4 se->sleep_start
1039  *     6 se->load.weight
1040  */
1041 struct sched_entity {
1042         struct load_weight      load;           /* for load-balancing */
1043         struct rb_node          run_node;
1044         struct list_head        group_node;
1045         unsigned int            on_rq;
1046
1047         u64                     exec_start;
1048         u64                     sum_exec_runtime;
1049         u64                     vruntime;
1050         u64                     prev_sum_exec_runtime;
1051
1052         u64                     last_wakeup;
1053         u64                     avg_overlap;
1054
1055         u64                     start_runtime;
1056         u64                     avg_wakeup;
1057         u64                     nr_migrations;
1058
1059 #ifdef CONFIG_SCHEDSTATS
1060         u64                     wait_start;
1061         u64                     wait_max;
1062         u64                     wait_count;
1063         u64                     wait_sum;
1064
1065         u64                     sleep_start;
1066         u64                     sleep_max;
1067         s64                     sum_sleep_runtime;
1068
1069         u64                     block_start;
1070         u64                     block_max;
1071         u64                     exec_max;
1072         u64                     slice_max;
1073
1074         u64                     nr_migrations_cold;
1075         u64                     nr_failed_migrations_affine;
1076         u64                     nr_failed_migrations_running;
1077         u64                     nr_failed_migrations_hot;
1078         u64                     nr_forced_migrations;
1079         u64                     nr_forced2_migrations;
1080
1081         u64                     nr_wakeups;
1082         u64                     nr_wakeups_sync;
1083         u64                     nr_wakeups_migrate;
1084         u64                     nr_wakeups_local;
1085         u64                     nr_wakeups_remote;
1086         u64                     nr_wakeups_affine;
1087         u64                     nr_wakeups_affine_attempts;
1088         u64                     nr_wakeups_passive;
1089         u64                     nr_wakeups_idle;
1090 #endif
1091
1092 #ifdef CONFIG_FAIR_GROUP_SCHED
1093         struct sched_entity     *parent;
1094         /* rq on which this entity is (to be) queued: */
1095         struct cfs_rq           *cfs_rq;
1096         /* rq "owned" by this entity/group: */
1097         struct cfs_rq           *my_q;
1098 #endif
1099 };
1100
1101 struct sched_rt_entity {
1102         struct list_head run_list;
1103         unsigned long timeout;
1104         unsigned int time_slice;
1105         int nr_cpus_allowed;
1106
1107         struct sched_rt_entity *back;
1108 #ifdef CONFIG_RT_GROUP_SCHED
1109         struct sched_rt_entity  *parent;
1110         /* rq on which this entity is (to be) queued: */
1111         struct rt_rq            *rt_rq;
1112         /* rq "owned" by this entity/group: */
1113         struct rt_rq            *my_q;
1114 #endif
1115 };
1116
1117 struct task_struct {
1118         volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1119         void *stack;
1120         atomic_t usage;
1121         unsigned int flags;     /* per process flags, defined below */
1122         unsigned int ptrace;
1123
1124         int lock_depth;         /* BKL lock depth */
1125
1126 #ifdef CONFIG_SMP
1127 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1128         int oncpu;
1129 #endif
1130 #endif
1131
1132         int prio, static_prio, normal_prio;
1133         unsigned int rt_priority;
1134         const struct sched_class *sched_class;
1135         struct sched_entity se;
1136         struct sched_rt_entity rt;
1137
1138 #ifdef CONFIG_PREEMPT_NOTIFIERS
1139         /* list of struct preempt_notifier: */
1140         struct hlist_head preempt_notifiers;
1141 #endif
1142
1143         /*
1144          * fpu_counter contains the number of consecutive context switches
1145          * that the FPU is used. If this is over a threshold, the lazy fpu
1146          * saving becomes unlazy to save the trap. This is an unsigned char
1147          * so that after 256 times the counter wraps and the behavior turns
1148          * lazy again; this to deal with bursty apps that only use FPU for
1149          * a short time
1150          */
1151         unsigned char fpu_counter;
1152         s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1153 #ifdef CONFIG_BLK_DEV_IO_TRACE
1154         unsigned int btrace_seq;
1155 #endif
1156
1157         unsigned int policy;
1158         cpumask_t cpus_allowed;
1159
1160 #ifdef CONFIG_PREEMPT_RCU
1161         int rcu_read_lock_nesting;
1162         int rcu_flipctr_idx;
1163 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1164
1165 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1166         struct sched_info sched_info;
1167 #endif
1168
1169         struct list_head tasks;
1170         struct plist_node pushable_tasks;
1171
1172         struct mm_struct *mm, *active_mm;
1173
1174 /* task state */
1175         struct linux_binfmt *binfmt;
1176         int exit_state;
1177         int exit_code, exit_signal;
1178         int pdeath_signal;  /*  The signal sent when the parent dies  */
1179         /* ??? */
1180         unsigned int personality;
1181         unsigned did_exec:1;
1182         unsigned in_execve:1;   /* Tell the LSMs that the process is doing an
1183                                  * execve */
1184         pid_t pid;
1185         pid_t tgid;
1186
1187         /* Canary value for the -fstack-protector gcc feature */
1188         unsigned long stack_canary;
1189
1190         /*
1191          * pointers to (original) parent process, youngest child, younger sibling,
1192          * older sibling, respectively.  (p->father can be replaced with
1193          * p->real_parent->pid)
1194          */
1195         struct task_struct *real_parent; /* real parent process */
1196         struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1197         /*
1198          * children/sibling forms the list of my natural children
1199          */
1200         struct list_head children;      /* list of my children */
1201         struct list_head sibling;       /* linkage in my parent's children list */
1202         struct task_struct *group_leader;       /* threadgroup leader */
1203
1204         /*
1205          * ptraced is the list of tasks this task is using ptrace on.
1206          * This includes both natural children and PTRACE_ATTACH targets.
1207          * p->ptrace_entry is p's link on the p->parent->ptraced list.
1208          */
1209         struct list_head ptraced;
1210         struct list_head ptrace_entry;
1211
1212 #ifdef CONFIG_X86_PTRACE_BTS
1213         /*
1214          * This is the tracer handle for the ptrace BTS extension.
1215          * This field actually belongs to the ptracer task.
1216          */
1217         struct bts_tracer *bts;
1218         /*
1219          * The buffer to hold the BTS data.
1220          */
1221         void *bts_buffer;
1222         size_t bts_size;
1223 #endif /* CONFIG_X86_PTRACE_BTS */
1224
1225         /* PID/PID hash table linkage. */
1226         struct pid_link pids[PIDTYPE_MAX];
1227         struct list_head thread_group;
1228
1229         struct completion *vfork_done;          /* for vfork() */
1230         int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1231         int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1232
1233         cputime_t utime, stime, utimescaled, stimescaled;
1234         cputime_t gtime;
1235         cputime_t prev_utime, prev_stime;
1236         unsigned long nvcsw, nivcsw; /* context switch counts */
1237         struct timespec start_time;             /* monotonic time */
1238         struct timespec real_start_time;        /* boot based time */
1239 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1240         unsigned long min_flt, maj_flt;
1241
1242         struct task_cputime cputime_expires;
1243         struct list_head cpu_timers[3];
1244
1245 /* process credentials */
1246         const struct cred *real_cred;   /* objective and real subjective task
1247                                          * credentials (COW) */
1248         const struct cred *cred;        /* effective (overridable) subjective task
1249                                          * credentials (COW) */
1250         struct mutex cred_exec_mutex;   /* execve vs ptrace cred calculation mutex */
1251
1252         char comm[TASK_COMM_LEN]; /* executable name excluding path
1253                                      - access with [gs]et_task_comm (which lock
1254                                        it with task_lock())
1255                                      - initialized normally by flush_old_exec */
1256 /* file system info */
1257         int link_count, total_link_count;
1258 #ifdef CONFIG_SYSVIPC
1259 /* ipc stuff */
1260         struct sysv_sem sysvsem;
1261 #endif
1262 #ifdef CONFIG_DETECT_HUNG_TASK
1263 /* hung task detection */
1264         unsigned long last_switch_count;
1265 #endif
1266 /* CPU-specific state of this task */
1267         struct thread_struct thread;
1268 /* filesystem information */
1269         struct fs_struct *fs;
1270 /* open file information */
1271         struct files_struct *files;
1272 /* namespaces */
1273         struct nsproxy *nsproxy;
1274 /* signal handlers */
1275         struct signal_struct *signal;
1276         struct sighand_struct *sighand;
1277
1278         sigset_t blocked, real_blocked;
1279         sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1280         struct sigpending pending;
1281
1282         unsigned long sas_ss_sp;
1283         size_t sas_ss_size;
1284         int (*notifier)(void *priv);
1285         void *notifier_data;
1286         sigset_t *notifier_mask;
1287         struct audit_context *audit_context;
1288 #ifdef CONFIG_AUDITSYSCALL
1289         uid_t loginuid;
1290         unsigned int sessionid;
1291 #endif
1292         seccomp_t seccomp;
1293
1294 /* Thread group tracking */
1295         u32 parent_exec_id;
1296         u32 self_exec_id;
1297 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1298         spinlock_t alloc_lock;
1299
1300 #ifdef CONFIG_GENERIC_HARDIRQS
1301         /* IRQ handler threads */
1302         struct irqaction *irqaction;
1303 #endif
1304
1305         /* Protection of the PI data structures: */
1306         spinlock_t pi_lock;
1307
1308 #ifdef CONFIG_RT_MUTEXES
1309         /* PI waiters blocked on a rt_mutex held by this task */
1310         struct plist_head pi_waiters;
1311         /* Deadlock detection and priority inheritance handling */
1312         struct rt_mutex_waiter *pi_blocked_on;
1313 #endif
1314
1315 #ifdef CONFIG_DEBUG_MUTEXES
1316         /* mutex deadlock detection */
1317         struct mutex_waiter *blocked_on;
1318 #endif
1319 #ifdef CONFIG_TRACE_IRQFLAGS
1320         unsigned int irq_events;
1321         int hardirqs_enabled;
1322         unsigned long hardirq_enable_ip;
1323         unsigned int hardirq_enable_event;
1324         unsigned long hardirq_disable_ip;
1325         unsigned int hardirq_disable_event;
1326         int softirqs_enabled;
1327         unsigned long softirq_disable_ip;
1328         unsigned int softirq_disable_event;
1329         unsigned long softirq_enable_ip;
1330         unsigned int softirq_enable_event;
1331         int hardirq_context;
1332         int softirq_context;
1333 #endif
1334 #ifdef CONFIG_LOCKDEP
1335 # define MAX_LOCK_DEPTH 48UL
1336         u64 curr_chain_key;
1337         int lockdep_depth;
1338         unsigned int lockdep_recursion;
1339         struct held_lock held_locks[MAX_LOCK_DEPTH];
1340         gfp_t lockdep_reclaim_gfp;
1341 #endif
1342
1343 /* journalling filesystem info */
1344         void *journal_info;
1345
1346 /* stacked block device info */
1347         struct bio *bio_list, **bio_tail;
1348
1349 /* VM state */
1350         struct reclaim_state *reclaim_state;
1351
1352         struct backing_dev_info *backing_dev_info;
1353
1354         struct io_context *io_context;
1355
1356         unsigned long ptrace_message;
1357         siginfo_t *last_siginfo; /* For ptrace use.  */
1358         struct task_io_accounting ioac;
1359 #if defined(CONFIG_TASK_XACCT)
1360         u64 acct_rss_mem1;      /* accumulated rss usage */
1361         u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1362         cputime_t acct_timexpd; /* stime + utime since last update */
1363 #endif
1364 #ifdef CONFIG_CPUSETS
1365         nodemask_t mems_allowed;
1366         int cpuset_mems_generation;
1367         int cpuset_mem_spread_rotor;
1368 #endif
1369 #ifdef CONFIG_CGROUPS
1370         /* Control Group info protected by css_set_lock */
1371         struct css_set *cgroups;
1372         /* cg_list protected by css_set_lock and tsk->alloc_lock */
1373         struct list_head cg_list;
1374 #endif
1375 #ifdef CONFIG_FUTEX
1376         struct robust_list_head __user *robust_list;
1377 #ifdef CONFIG_COMPAT
1378         struct compat_robust_list_head __user *compat_robust_list;
1379 #endif
1380         struct list_head pi_state_list;
1381         struct futex_pi_state *pi_state_cache;
1382 #endif
1383 #ifdef CONFIG_NUMA
1384         struct mempolicy *mempolicy;
1385         short il_next;
1386 #endif
1387         atomic_t fs_excl;       /* holding fs exclusive resources */
1388         struct rcu_head rcu;
1389
1390         /*
1391          * cache last used pipe for splice
1392          */
1393         struct pipe_inode_info *splice_pipe;
1394 #ifdef  CONFIG_TASK_DELAY_ACCT
1395         struct task_delay_info *delays;
1396 #endif
1397 #ifdef CONFIG_FAULT_INJECTION
1398         int make_it_fail;
1399 #endif
1400         struct prop_local_single dirties;
1401 #ifdef CONFIG_LATENCYTOP
1402         int latency_record_count;
1403         struct latency_record latency_record[LT_SAVECOUNT];
1404 #endif
1405         /*
1406          * time slack values; these are used to round up poll() and
1407          * select() etc timeout values. These are in nanoseconds.
1408          */
1409         unsigned long timer_slack_ns;
1410         unsigned long default_timer_slack_ns;
1411
1412         struct list_head        *scm_work_list;
1413 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1414         /* Index of current stored adress in ret_stack */
1415         int curr_ret_stack;
1416         /* Stack of return addresses for return function tracing */
1417         struct ftrace_ret_stack *ret_stack;
1418         /* time stamp for last schedule */
1419         unsigned long long ftrace_timestamp;
1420         /*
1421          * Number of functions that haven't been traced
1422          * because of depth overrun.
1423          */
1424         atomic_t trace_overrun;
1425         /* Pause for the tracing */
1426         atomic_t tracing_graph_pause;
1427 #endif
1428 #ifdef CONFIG_TRACING
1429         /* state flags for use by tracers */
1430         unsigned long trace;
1431 #endif
1432 };
1433
1434 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1435 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1436
1437 /*
1438  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1439  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1440  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1441  * values are inverted: lower p->prio value means higher priority.
1442  *
1443  * The MAX_USER_RT_PRIO value allows the actual maximum
1444  * RT priority to be separate from the value exported to
1445  * user-space.  This allows kernel threads to set their
1446  * priority to a value higher than any user task. Note:
1447  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1448  */
1449
1450 #define MAX_USER_RT_PRIO        100
1451 #define MAX_RT_PRIO             MAX_USER_RT_PRIO
1452
1453 #define MAX_PRIO                (MAX_RT_PRIO + 40)
1454 #define DEFAULT_PRIO            (MAX_RT_PRIO + 20)
1455
1456 static inline int rt_prio(int prio)
1457 {
1458         if (unlikely(prio < MAX_RT_PRIO))
1459                 return 1;
1460         return 0;
1461 }
1462
1463 static inline int rt_task(struct task_struct *p)
1464 {
1465         return rt_prio(p->prio);
1466 }
1467
1468 static inline struct pid *task_pid(struct task_struct *task)
1469 {
1470         return task->pids[PIDTYPE_PID].pid;
1471 }
1472
1473 static inline struct pid *task_tgid(struct task_struct *task)
1474 {
1475         return task->group_leader->pids[PIDTYPE_PID].pid;
1476 }
1477
1478 /*
1479  * Without tasklist or rcu lock it is not safe to dereference
1480  * the result of task_pgrp/task_session even if task == current,
1481  * we can race with another thread doing sys_setsid/sys_setpgid.
1482  */
1483 static inline struct pid *task_pgrp(struct task_struct *task)
1484 {
1485         return task->group_leader->pids[PIDTYPE_PGID].pid;
1486 }
1487
1488 static inline struct pid *task_session(struct task_struct *task)
1489 {
1490         return task->group_leader->pids[PIDTYPE_SID].pid;
1491 }
1492
1493 struct pid_namespace;
1494
1495 /*
1496  * the helpers to get the task's different pids as they are seen
1497  * from various namespaces
1498  *
1499  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1500  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1501  *                     current.
1502  * task_xid_nr_ns()  : id seen from the ns specified;
1503  *
1504  * set_task_vxid()   : assigns a virtual id to a task;
1505  *
1506  * see also pid_nr() etc in include/linux/pid.h
1507  */
1508 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1509                         struct pid_namespace *ns);
1510
1511 static inline pid_t task_pid_nr(struct task_struct *tsk)
1512 {
1513         return tsk->pid;
1514 }
1515
1516 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1517                                         struct pid_namespace *ns)
1518 {
1519         return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1520 }
1521
1522 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1523 {
1524         return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1525 }
1526
1527
1528 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1529 {
1530         return tsk->tgid;
1531 }
1532
1533 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1534
1535 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1536 {
1537         return pid_vnr(task_tgid(tsk));
1538 }
1539
1540
1541 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1542                                         struct pid_namespace *ns)
1543 {
1544         return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1545 }
1546
1547 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1548 {
1549         return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1550 }
1551
1552
1553 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1554                                         struct pid_namespace *ns)
1555 {
1556         return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1557 }
1558
1559 static inline pid_t task_session_vnr(struct task_struct *tsk)
1560 {
1561         return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1562 }
1563
1564 /* obsolete, do not use */
1565 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1566 {
1567         return task_pgrp_nr_ns(tsk, &init_pid_ns);
1568 }
1569
1570 /**
1571  * pid_alive - check that a task structure is not stale
1572  * @p: Task structure to be checked.
1573  *
1574  * Test if a process is not yet dead (at most zombie state)
1575  * If pid_alive fails, then pointers within the task structure
1576  * can be stale and must not be dereferenced.
1577  */
1578 static inline int pid_alive(struct task_struct *p)
1579 {
1580         return p->pids[PIDTYPE_PID].pid != NULL;
1581 }
1582
1583 /**
1584  * is_global_init - check if a task structure is init
1585  * @tsk: Task structure to be checked.
1586  *
1587  * Check if a task structure is the first user space task the kernel created.
1588  */
1589 static inline int is_global_init(struct task_struct *tsk)
1590 {
1591         return tsk->pid == 1;
1592 }
1593
1594 /*
1595  * is_container_init:
1596  * check whether in the task is init in its own pid namespace.
1597  */
1598 extern int is_container_init(struct task_struct *tsk);
1599
1600 extern struct pid *cad_pid;
1601
1602 extern void free_task(struct task_struct *tsk);
1603 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1604
1605 extern void __put_task_struct(struct task_struct *t);
1606
1607 static inline void put_task_struct(struct task_struct *t)
1608 {
1609         if (atomic_dec_and_test(&t->usage))
1610                 __put_task_struct(t);
1611 }
1612
1613 extern cputime_t task_utime(struct task_struct *p);
1614 extern cputime_t task_stime(struct task_struct *p);
1615 extern cputime_t task_gtime(struct task_struct *p);
1616
1617 /*
1618  * Per process flags
1619  */
1620 #define PF_ALIGNWARN    0x00000001      /* Print alignment warning msgs */
1621                                         /* Not implemented yet, only for 486*/
1622 #define PF_STARTING     0x00000002      /* being created */
1623 #define PF_EXITING      0x00000004      /* getting shut down */
1624 #define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
1625 #define PF_VCPU         0x00000010      /* I'm a virtual CPU */
1626 #define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
1627 #define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
1628 #define PF_DUMPCORE     0x00000200      /* dumped core */
1629 #define PF_SIGNALED     0x00000400      /* killed by a signal */
1630 #define PF_MEMALLOC     0x00000800      /* Allocating memory */
1631 #define PF_FLUSHER      0x00001000      /* responsible for disk writeback */
1632 #define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
1633 #define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
1634 #define PF_FROZEN       0x00010000      /* frozen for system suspend */
1635 #define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
1636 #define PF_KSWAPD       0x00040000      /* I am kswapd */
1637 #define PF_SWAPOFF      0x00080000      /* I am in swapoff */
1638 #define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
1639 #define PF_KTHREAD      0x00200000      /* I am a kernel thread */
1640 #define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
1641 #define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
1642 #define PF_SPREAD_PAGE  0x01000000      /* Spread page cache over cpuset */
1643 #define PF_SPREAD_SLAB  0x02000000      /* Spread some slab caches over cpuset */
1644 #define PF_THREAD_BOUND 0x04000000      /* Thread bound to specific cpu */
1645 #define PF_MEMPOLICY    0x10000000      /* Non-default NUMA mempolicy */
1646 #define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
1647 #define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezeable */
1648 #define PF_FREEZER_NOSIG 0x80000000     /* Freezer won't send signals to it */
1649
1650 /*
1651  * Only the _current_ task can read/write to tsk->flags, but other
1652  * tasks can access tsk->flags in readonly mode for example
1653  * with tsk_used_math (like during threaded core dumping).
1654  * There is however an exception to this rule during ptrace
1655  * or during fork: the ptracer task is allowed to write to the
1656  * child->flags of its traced child (same goes for fork, the parent
1657  * can write to the child->flags), because we're guaranteed the
1658  * child is not running and in turn not changing child->flags
1659  * at the same time the parent does it.
1660  */
1661 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1662 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1663 #define clear_used_math() clear_stopped_child_used_math(current)
1664 #define set_used_math() set_stopped_child_used_math(current)
1665 #define conditional_stopped_child_used_math(condition, child) \
1666         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1667 #define conditional_used_math(condition) \
1668         conditional_stopped_child_used_math(condition, current)
1669 #define copy_to_stopped_child_used_math(child) \
1670         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1671 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1672 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1673 #define used_math() tsk_used_math(current)
1674
1675 #ifdef CONFIG_SMP
1676 extern int set_cpus_allowed_ptr(struct task_struct *p,
1677                                 const struct cpumask *new_mask);
1678 #else
1679 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1680                                        const struct cpumask *new_mask)
1681 {
1682         if (!cpumask_test_cpu(0, new_mask))
1683                 return -EINVAL;
1684         return 0;
1685 }
1686 #endif
1687 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1688 {
1689         return set_cpus_allowed_ptr(p, &new_mask);
1690 }
1691
1692 /*
1693  * Architectures can set this to 1 if they have specified
1694  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1695  * but then during bootup it turns out that sched_clock()
1696  * is reliable after all:
1697  */
1698 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1699 extern int sched_clock_stable;
1700 #endif
1701
1702 extern unsigned long long sched_clock(void);
1703
1704 extern void sched_clock_init(void);
1705 extern u64 sched_clock_cpu(int cpu);
1706
1707 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1708 static inline void sched_clock_tick(void)
1709 {
1710 }
1711
1712 static inline void sched_clock_idle_sleep_event(void)
1713 {
1714 }
1715
1716 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1717 {
1718 }
1719 #else
1720 extern void sched_clock_tick(void);
1721 extern void sched_clock_idle_sleep_event(void);
1722 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1723 #endif
1724
1725 /*
1726  * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1727  * clock constructed from sched_clock():
1728  */
1729 extern unsigned long long cpu_clock(int cpu);
1730
1731 extern unsigned long long
1732 task_sched_runtime(struct task_struct *task);
1733 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1734
1735 /* sched_exec is called by processes performing an exec */
1736 #ifdef CONFIG_SMP
1737 extern void sched_exec(void);
1738 #else
1739 #define sched_exec()   {}
1740 #endif
1741
1742 extern void sched_clock_idle_sleep_event(void);
1743 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1744
1745 #ifdef CONFIG_HOTPLUG_CPU
1746 extern void idle_task_exit(void);
1747 #else
1748 static inline void idle_task_exit(void) {}
1749 #endif
1750
1751 extern void sched_idle_next(void);
1752
1753 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1754 extern void wake_up_idle_cpu(int cpu);
1755 #else
1756 static inline void wake_up_idle_cpu(int cpu) { }
1757 #endif
1758
1759 extern unsigned int sysctl_sched_latency;
1760 extern unsigned int sysctl_sched_min_granularity;
1761 extern unsigned int sysctl_sched_wakeup_granularity;
1762 extern unsigned int sysctl_sched_shares_ratelimit;
1763 extern unsigned int sysctl_sched_shares_thresh;
1764 #ifdef CONFIG_SCHED_DEBUG
1765 extern unsigned int sysctl_sched_child_runs_first;
1766 extern unsigned int sysctl_sched_features;
1767 extern unsigned int sysctl_sched_migration_cost;
1768 extern unsigned int sysctl_sched_nr_migrate;
1769
1770 int sched_nr_latency_handler(struct ctl_table *table, int write,
1771                 struct file *file, void __user *buffer, size_t *length,
1772                 loff_t *ppos);
1773 #endif
1774 extern unsigned int sysctl_sched_rt_period;
1775 extern int sysctl_sched_rt_runtime;
1776
1777 int sched_rt_handler(struct ctl_table *table, int write,
1778                 struct file *filp, void __user *buffer, size_t *lenp,
1779                 loff_t *ppos);
1780
1781 extern unsigned int sysctl_sched_compat_yield;
1782
1783 #ifdef CONFIG_RT_MUTEXES
1784 extern int rt_mutex_getprio(struct task_struct *p);
1785 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1786 extern void rt_mutex_adjust_pi(struct task_struct *p);
1787 #else
1788 static inline int rt_mutex_getprio(struct task_struct *p)
1789 {
1790         return p->normal_prio;
1791 }
1792 # define rt_mutex_adjust_pi(p)          do { } while (0)
1793 #endif
1794
1795 extern void set_user_nice(struct task_struct *p, long nice);
1796 extern int task_prio(const struct task_struct *p);
1797 extern int task_nice(const struct task_struct *p);
1798 extern int can_nice(const struct task_struct *p, const int nice);
1799 extern int task_curr(const struct task_struct *p);
1800 extern int idle_cpu(int cpu);
1801 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1802 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1803                                       struct sched_param *);
1804 extern struct task_struct *idle_task(int cpu);
1805 extern struct task_struct *curr_task(int cpu);
1806 extern void set_curr_task(int cpu, struct task_struct *p);
1807
1808 void yield(void);
1809
1810 /*
1811  * The default (Linux) execution domain.
1812  */
1813 extern struct exec_domain       default_exec_domain;
1814
1815 union thread_union {
1816         struct thread_info thread_info;
1817         unsigned long stack[THREAD_SIZE/sizeof(long)];
1818 };
1819
1820 #ifndef __HAVE_ARCH_KSTACK_END
1821 static inline int kstack_end(void *addr)
1822 {
1823         /* Reliable end of stack detection:
1824          * Some APM bios versions misalign the stack
1825          */
1826         return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1827 }
1828 #endif
1829
1830 extern union thread_union init_thread_union;
1831 extern struct task_struct init_task;
1832
1833 extern struct   mm_struct init_mm;
1834
1835 extern struct pid_namespace init_pid_ns;
1836
1837 /*
1838  * find a task by one of its numerical ids
1839  *
1840  * find_task_by_pid_type_ns():
1841  *      it is the most generic call - it finds a task by all id,
1842  *      type and namespace specified
1843  * find_task_by_pid_ns():
1844  *      finds a task by its pid in the specified namespace
1845  * find_task_by_vpid():
1846  *      finds a task by its virtual pid
1847  *
1848  * see also find_vpid() etc in include/linux/pid.h
1849  */
1850
1851 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1852                 struct pid_namespace *ns);
1853
1854 extern struct task_struct *find_task_by_vpid(pid_t nr);
1855 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1856                 struct pid_namespace *ns);
1857
1858 extern void __set_special_pids(struct pid *pid);
1859
1860 /* per-UID process charging. */
1861 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1862 static inline struct user_struct *get_uid(struct user_struct *u)
1863 {
1864         atomic_inc(&u->__count);
1865         return u;
1866 }
1867 extern void free_uid(struct user_struct *);
1868 extern void release_uids(struct user_namespace *ns);
1869
1870 #include <asm/current.h>
1871
1872 extern void do_timer(unsigned long ticks);
1873
1874 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1875 extern int wake_up_process(struct task_struct *tsk);
1876 extern void wake_up_new_task(struct task_struct *tsk,
1877                                 unsigned long clone_flags);
1878 #ifdef CONFIG_SMP
1879  extern void kick_process(struct task_struct *tsk);
1880 #else
1881  static inline void kick_process(struct task_struct *tsk) { }
1882 #endif
1883 extern void sched_fork(struct task_struct *p, int clone_flags);
1884 extern void sched_dead(struct task_struct *p);
1885
1886 extern void proc_caches_init(void);
1887 extern void flush_signals(struct task_struct *);
1888 extern void ignore_signals(struct task_struct *);
1889 extern void flush_signal_handlers(struct task_struct *, int force_default);
1890 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1891
1892 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1893 {
1894         unsigned long flags;
1895         int ret;
1896
1897         spin_lock_irqsave(&tsk->sighand->siglock, flags);
1898         ret = dequeue_signal(tsk, mask, info);
1899         spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1900
1901         return ret;
1902 }
1903
1904 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1905                               sigset_t *mask);
1906 extern void unblock_all_signals(void);
1907 extern void release_task(struct task_struct * p);
1908 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1909 extern int force_sigsegv(int, struct task_struct *);
1910 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1911 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1912 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1913 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1914 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1915 extern int kill_pid(struct pid *pid, int sig, int priv);
1916 extern int kill_proc_info(int, struct siginfo *, pid_t);
1917 extern int do_notify_parent(struct task_struct *, int);
1918 extern void force_sig(int, struct task_struct *);
1919 extern void force_sig_specific(int, struct task_struct *);
1920 extern int send_sig(int, struct task_struct *, int);
1921 extern void zap_other_threads(struct task_struct *p);
1922 extern struct sigqueue *sigqueue_alloc(void);
1923 extern void sigqueue_free(struct sigqueue *);
1924 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
1925 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1926 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1927
1928 static inline int kill_cad_pid(int sig, int priv)
1929 {
1930         return kill_pid(cad_pid, sig, priv);
1931 }
1932
1933 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1934 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1935 #define SEND_SIG_PRIV   ((struct siginfo *) 1)
1936 #define SEND_SIG_FORCED ((struct siginfo *) 2)
1937
1938 static inline int is_si_special(const struct siginfo *info)
1939 {
1940         return info <= SEND_SIG_FORCED;
1941 }
1942
1943 /* True if we are on the alternate signal stack.  */
1944
1945 static inline int on_sig_stack(unsigned long sp)
1946 {
1947         return (sp - current->sas_ss_sp < current->sas_ss_size);
1948 }
1949
1950 static inline int sas_ss_flags(unsigned long sp)
1951 {
1952         return (current->sas_ss_size == 0 ? SS_DISABLE
1953                 : on_sig_stack(sp) ? SS_ONSTACK : 0);
1954 }
1955
1956 /*
1957  * Routines for handling mm_structs
1958  */
1959 extern struct mm_struct * mm_alloc(void);
1960
1961 /* mmdrop drops the mm and the page tables */
1962 extern void __mmdrop(struct mm_struct *);
1963 static inline void mmdrop(struct mm_struct * mm)
1964 {
1965         if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1966                 __mmdrop(mm);
1967 }
1968
1969 /* mmput gets rid of the mappings and all user-space */
1970 extern void mmput(struct mm_struct *);
1971 /* Grab a reference to a task's mm, if it is not already going away */
1972 extern struct mm_struct *get_task_mm(struct task_struct *task);
1973 /* Remove the current tasks stale references to the old mm_struct */
1974 extern void mm_release(struct task_struct *, struct mm_struct *);
1975 /* Allocate a new mm structure and copy contents from tsk->mm */
1976 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1977
1978 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1979                         struct task_struct *, struct pt_regs *);
1980 extern void flush_thread(void);
1981 extern void exit_thread(void);
1982
1983 extern void exit_files(struct task_struct *);
1984 extern void __cleanup_signal(struct signal_struct *);
1985 extern void __cleanup_sighand(struct sighand_struct *);
1986
1987 extern void exit_itimers(struct signal_struct *);
1988 extern void flush_itimer_signals(void);
1989
1990 extern NORET_TYPE void do_group_exit(int);
1991
1992 extern void daemonize(const char *, ...);
1993 extern int allow_signal(int);
1994 extern int disallow_signal(int);
1995
1996 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1997 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1998 struct task_struct *fork_idle(int);
1999
2000 extern void set_task_comm(struct task_struct *tsk, char *from);
2001 extern char *get_task_comm(char *to, struct task_struct *tsk);
2002
2003 #ifdef CONFIG_SMP
2004 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2005 #else
2006 static inline unsigned long wait_task_inactive(struct task_struct *p,
2007                                                long match_state)
2008 {
2009         return 1;
2010 }
2011 #endif
2012
2013 #define next_task(p)    list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2014
2015 #define for_each_process(p) \
2016         for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2017
2018 extern bool is_single_threaded(struct task_struct *);
2019
2020 /*
2021  * Careful: do_each_thread/while_each_thread is a double loop so
2022  *          'break' will not work as expected - use goto instead.
2023  */
2024 #define do_each_thread(g, t) \
2025         for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2026
2027 #define while_each_thread(g, t) \
2028         while ((t = next_thread(t)) != g)
2029
2030 /* de_thread depends on thread_group_leader not being a pid based check */
2031 #define thread_group_leader(p)  (p == p->group_leader)
2032
2033 /* Do to the insanities of de_thread it is possible for a process
2034  * to have the pid of the thread group leader without actually being
2035  * the thread group leader.  For iteration through the pids in proc
2036  * all we care about is that we have a task with the appropriate
2037  * pid, we don't actually care if we have the right task.
2038  */
2039 static inline int has_group_leader_pid(struct task_struct *p)
2040 {
2041         return p->pid == p->tgid;
2042 }
2043
2044 static inline
2045 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2046 {
2047         return p1->tgid == p2->tgid;
2048 }
2049
2050 static inline struct task_struct *next_thread(const struct task_struct *p)
2051 {
2052         return list_entry(rcu_dereference(p->thread_group.next),
2053                           struct task_struct, thread_group);
2054 }
2055
2056 static inline int thread_group_empty(struct task_struct *p)
2057 {
2058         return list_empty(&p->thread_group);
2059 }
2060
2061 #define delay_group_leader(p) \
2062                 (thread_group_leader(p) && !thread_group_empty(p))
2063
2064 static inline int task_detached(struct task_struct *p)
2065 {
2066         return p->exit_signal == -1;
2067 }
2068
2069 /*
2070  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2071  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2072  * pins the final release of task.io_context.  Also protects ->cpuset and
2073  * ->cgroup.subsys[].
2074  *
2075  * Nests both inside and outside of read_lock(&tasklist_lock).
2076  * It must not be nested with write_lock_irq(&tasklist_lock),
2077  * neither inside nor outside.
2078  */
2079 static inline void task_lock(struct task_struct *p)
2080 {
2081         spin_lock(&p->alloc_lock);
2082 }
2083
2084 static inline void task_unlock(struct task_struct *p)
2085 {
2086         spin_unlock(&p->alloc_lock);
2087 }
2088
2089 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2090                                                         unsigned long *flags);
2091
2092 static inline void unlock_task_sighand(struct task_struct *tsk,
2093                                                 unsigned long *flags)
2094 {
2095         spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2096 }
2097
2098 #ifndef __HAVE_THREAD_FUNCTIONS
2099
2100 #define task_thread_info(task)  ((struct thread_info *)(task)->stack)
2101 #define task_stack_page(task)   ((task)->stack)
2102
2103 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2104 {
2105         *task_thread_info(p) = *task_thread_info(org);
2106         task_thread_info(p)->task = p;
2107 }
2108
2109 static inline unsigned long *end_of_stack(struct task_struct *p)
2110 {
2111         return (unsigned long *)(task_thread_info(p) + 1);
2112 }
2113
2114 #endif
2115
2116 static inline int object_is_on_stack(void *obj)
2117 {
2118         void *stack = task_stack_page(current);
2119
2120         return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2121 }
2122
2123 extern void thread_info_cache_init(void);
2124
2125 #ifdef CONFIG_DEBUG_STACK_USAGE
2126 static inline unsigned long stack_not_used(struct task_struct *p)
2127 {
2128         unsigned long *n = end_of_stack(p);
2129
2130         do {    /* Skip over canary */
2131                 n++;
2132         } while (!*n);
2133
2134         return (unsigned long)n - (unsigned long)end_of_stack(p);
2135 }
2136 #endif
2137
2138 /* set thread flags in other task's structures
2139  * - see asm/thread_info.h for TIF_xxxx flags available
2140  */
2141 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2142 {
2143         set_ti_thread_flag(task_thread_info(tsk), flag);
2144 }
2145
2146 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2147 {
2148         clear_ti_thread_flag(task_thread_info(tsk), flag);
2149 }
2150
2151 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2152 {
2153         return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2154 }
2155
2156 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2157 {
2158         return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2159 }
2160
2161 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2162 {
2163         return test_ti_thread_flag(task_thread_info(tsk), flag);
2164 }
2165
2166 static inline void set_tsk_need_resched(struct task_struct *tsk)
2167 {
2168         set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2169 }
2170
2171 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2172 {
2173         clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2174 }
2175
2176 static inline int test_tsk_need_resched(struct task_struct *tsk)
2177 {
2178         return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2179 }
2180
2181 static inline int signal_pending(struct task_struct *p)
2182 {
2183         return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2184 }
2185
2186 extern int __fatal_signal_pending(struct task_struct *p);
2187
2188 static inline int fatal_signal_pending(struct task_struct *p)
2189 {
2190         return signal_pending(p) && __fatal_signal_pending(p);
2191 }
2192
2193 static inline int signal_pending_state(long state, struct task_struct *p)
2194 {
2195         if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2196                 return 0;
2197         if (!signal_pending(p))
2198                 return 0;
2199
2200         return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2201 }
2202
2203 static inline int need_resched(void)
2204 {
2205         return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2206 }
2207
2208 /*
2209  * cond_resched() and cond_resched_lock(): latency reduction via
2210  * explicit rescheduling in places that are safe. The return
2211  * value indicates whether a reschedule was done in fact.
2212  * cond_resched_lock() will drop the spinlock before scheduling,
2213  * cond_resched_softirq() will enable bhs before scheduling.
2214  */
2215 extern int _cond_resched(void);
2216 #ifdef CONFIG_PREEMPT_BKL
2217 static inline int cond_resched(void)
2218 {
2219         return 0;
2220 }
2221 #else
2222 static inline int cond_resched(void)
2223 {
2224         return _cond_resched();
2225 }
2226 #endif
2227 extern int cond_resched_lock(spinlock_t * lock);
2228 extern int cond_resched_softirq(void);
2229 static inline int cond_resched_bkl(void)
2230 {
2231         return _cond_resched();
2232 }
2233
2234 /*
2235  * Does a critical section need to be broken due to another
2236  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2237  * but a general need for low latency)
2238  */
2239 static inline int spin_needbreak(spinlock_t *lock)
2240 {
2241 #ifdef CONFIG_PREEMPT
2242         return spin_is_contended(lock);
2243 #else
2244         return 0;
2245 #endif
2246 }
2247
2248 /*
2249  * Thread group CPU time accounting.
2250  */
2251 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2252 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2253
2254 static inline void thread_group_cputime_init(struct signal_struct *sig)
2255 {
2256         sig->cputimer.cputime = INIT_CPUTIME;
2257         spin_lock_init(&sig->cputimer.lock);
2258         sig->cputimer.running = 0;
2259 }
2260
2261 static inline void thread_group_cputime_free(struct signal_struct *sig)
2262 {
2263 }
2264
2265 /*
2266  * Reevaluate whether the task has signals pending delivery.
2267  * Wake the task if so.
2268  * This is required every time the blocked sigset_t changes.
2269  * callers must hold sighand->siglock.
2270  */
2271 extern void recalc_sigpending_and_wake(struct task_struct *t);
2272 extern void recalc_sigpending(void);
2273
2274 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2275
2276 /*
2277  * Wrappers for p->thread_info->cpu access. No-op on UP.
2278  */
2279 #ifdef CONFIG_SMP
2280
2281 static inline unsigned int task_cpu(const struct task_struct *p)
2282 {
2283         return task_thread_info(p)->cpu;
2284 }
2285
2286 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2287
2288 #else
2289
2290 static inline unsigned int task_cpu(const struct task_struct *p)
2291 {
2292         return 0;
2293 }
2294
2295 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2296 {
2297 }
2298
2299 #endif /* CONFIG_SMP */
2300
2301 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2302
2303 #ifdef CONFIG_TRACING
2304 extern void
2305 __trace_special(void *__tr, void *__data,
2306                 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2307 #else
2308 static inline void
2309 __trace_special(void *__tr, void *__data,
2310                 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2311 {
2312 }
2313 #endif
2314
2315 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2316 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2317
2318 extern void normalize_rt_tasks(void);
2319
2320 #ifdef CONFIG_GROUP_SCHED
2321
2322 extern struct task_group init_task_group;
2323 #ifdef CONFIG_USER_SCHED
2324 extern struct task_group root_task_group;
2325 extern void set_tg_uid(struct user_struct *user);
2326 #endif
2327
2328 extern struct task_group *sched_create_group(struct task_group *parent);
2329 extern void sched_destroy_group(struct task_group *tg);
2330 extern void sched_move_task(struct task_struct *tsk);
2331 #ifdef CONFIG_FAIR_GROUP_SCHED
2332 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2333 extern unsigned long sched_group_shares(struct task_group *tg);
2334 #endif
2335 #ifdef CONFIG_RT_GROUP_SCHED
2336 extern int sched_group_set_rt_runtime(struct task_group *tg,
2337                                       long rt_runtime_us);
2338 extern long sched_group_rt_runtime(struct task_group *tg);
2339 extern int sched_group_set_rt_period(struct task_group *tg,
2340                                       long rt_period_us);
2341 extern long sched_group_rt_period(struct task_group *tg);
2342 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2343 #endif
2344 #endif
2345
2346 extern int task_can_switch_user(struct user_struct *up,
2347                                         struct task_struct *tsk);
2348
2349 #ifdef CONFIG_TASK_XACCT
2350 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2351 {
2352         tsk->ioac.rchar += amt;
2353 }
2354
2355 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2356 {
2357         tsk->ioac.wchar += amt;
2358 }
2359
2360 static inline void inc_syscr(struct task_struct *tsk)
2361 {
2362         tsk->ioac.syscr++;
2363 }
2364
2365 static inline void inc_syscw(struct task_struct *tsk)
2366 {
2367         tsk->ioac.syscw++;
2368 }
2369 #else
2370 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2371 {
2372 }
2373
2374 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2375 {
2376 }
2377
2378 static inline void inc_syscr(struct task_struct *tsk)
2379 {
2380 }
2381
2382 static inline void inc_syscw(struct task_struct *tsk)
2383 {
2384 }
2385 #endif
2386
2387 #ifndef TASK_SIZE_OF
2388 #define TASK_SIZE_OF(tsk)       TASK_SIZE
2389 #endif
2390
2391 #ifdef CONFIG_MM_OWNER
2392 extern void mm_update_next_owner(struct mm_struct *mm);
2393 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2394 #else
2395 static inline void mm_update_next_owner(struct mm_struct *mm)
2396 {
2397 }
2398
2399 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2400 {
2401 }
2402 #endif /* CONFIG_MM_OWNER */
2403
2404 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2405
2406 #endif /* __KERNEL__ */
2407
2408 #endif