linux/include/linux/sched.h
<<
>>
Prefs
   1#ifndef _LINUX_SCHED_H
   2#define _LINUX_SCHED_H
   3
   4#include <uapi/linux/sched.h>
   5
   6
   7struct sched_param {
   8        int sched_priority;
   9};
  10
  11#include <asm/param.h>  /* for HZ */
  12
  13#include <linux/capability.h>
  14#include <linux/threads.h>
  15#include <linux/kernel.h>
  16#include <linux/types.h>
  17#include <linux/timex.h>
  18#include <linux/jiffies.h>
  19#include <linux/rbtree.h>
  20#include <linux/thread_info.h>
  21#include <linux/cpumask.h>
  22#include <linux/errno.h>
  23#include <linux/nodemask.h>
  24#include <linux/mm_types.h>
  25
  26#include <asm/page.h>
  27#include <asm/ptrace.h>
  28#include <asm/cputime.h>
  29
  30#include <linux/smp.h>
  31#include <linux/sem.h>
  32#include <linux/signal.h>
  33#include <linux/compiler.h>
  34#include <linux/completion.h>
  35#include <linux/pid.h>
  36#include <linux/percpu.h>
  37#include <linux/topology.h>
  38#include <linux/proportions.h>
  39#include <linux/seccomp.h>
  40#include <linux/rcupdate.h>
  41#include <linux/rculist.h>
  42#include <linux/rtmutex.h>
  43
  44#include <linux/time.h>
  45#include <linux/param.h>
  46#include <linux/resource.h>
  47#include <linux/timer.h>
  48#include <linux/hrtimer.h>
  49#include <linux/task_io_accounting.h>
  50#include <linux/latencytop.h>
  51#include <linux/cred.h>
  52#include <linux/llist.h>
  53#include <linux/uidgid.h>
  54
  55#include <asm/processor.h>
  56
  57struct exec_domain;
  58struct futex_pi_state;
  59struct robust_list_head;
  60struct bio_list;
  61struct fs_struct;
  62struct perf_event_context;
  63struct blk_plug;
  64
  65/*
  66 * List of flags we want to share for kernel threads,
  67 * if only because they are not used by them anyway.
  68 */
  69#define CLONE_KERNEL    (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
  70
  71/*
  72 * These are the constant used to fake the fixed-point load-average
  73 * counting. Some notes:
  74 *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
  75 *    a load-average precision of 10 bits integer + 11 bits fractional
  76 *  - if you want to count load-averages more often, you need more
  77 *    precision, or rounding will get you. With 2-second counting freq,
  78 *    the EXP_n values would be 1981, 2034 and 2043 if still using only
  79 *    11 bit fractions.
  80 */
  81extern unsigned long avenrun[];         /* Load averages */
  82extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
  83
  84#define FSHIFT          11              /* nr of bits of precision */
  85#define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
  86#define LOAD_FREQ       (5*HZ+1)        /* 5 sec intervals */
  87#define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
  88#define EXP_5           2014            /* 1/exp(5sec/5min) */
  89#define EXP_15          2037            /* 1/exp(5sec/15min) */
  90
  91#define CALC_LOAD(load,exp,n) \
  92        load *= exp; \
  93        load += n*(FIXED_1-exp); \
  94        load >>= FSHIFT;
  95
  96extern unsigned long total_forks;
  97extern int nr_threads;
  98DECLARE_PER_CPU(unsigned long, process_counts);
  99extern int nr_processes(void);
 100extern unsigned long nr_running(void);
 101extern unsigned long nr_uninterruptible(void);
 102extern unsigned long nr_iowait(void);
 103extern unsigned long nr_iowait_cpu(int cpu);
 104extern unsigned long this_cpu_load(void);
 105
 106
 107extern void calc_global_load(unsigned long ticks);
 108extern void update_cpu_load_nohz(void);
 109
 110/* Notifier for when a task gets migrated to a new CPU */
 111struct task_migration_notifier {
 112        struct task_struct *task;
 113        int from_cpu;
 114        int to_cpu;
 115};
 116extern void register_task_migration_notifier(struct notifier_block *n);
 117
 118extern unsigned long get_parent_ip(unsigned long addr);
 119
 120extern void dump_cpu_task(int cpu);
 121
 122struct seq_file;
 123struct cfs_rq;
 124struct task_group;
 125#ifdef CONFIG_SCHED_DEBUG
 126extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
 127extern void proc_sched_set_task(struct task_struct *p);
 128extern void
 129print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
 130#else
 131static inline void
 132proc_sched_show_task(struct task_struct *p, struct seq_file *m)
 133{
 134}
 135static inline void proc_sched_set_task(struct task_struct *p)
 136{
 137}
 138static inline void
 139print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
 140{
 141}
 142#endif
 143
 144/*
 145 * Task state bitmask. NOTE! These bits are also
 146 * encoded in fs/proc/array.c: get_task_state().
 147 *
 148 * We have two separate sets of flags: task->state
 149 * is about runnability, while task->exit_state are
 150 * about the task exiting. Confusing, but this way
 151 * modifying one set can't modify the other one by
 152 * mistake.
 153 */
 154#define TASK_RUNNING            0
 155#define TASK_INTERRUPTIBLE      1
 156#define TASK_UNINTERRUPTIBLE    2
 157#define __TASK_STOPPED          4
 158#define __TASK_TRACED           8
 159/* in tsk->exit_state */
 160#define EXIT_ZOMBIE             16
 161#define EXIT_DEAD               32
 162/* in tsk->state again */
 163#define TASK_DEAD               64
 164#define TASK_WAKEKILL           128
 165#define TASK_WAKING             256
 166#define TASK_STATE_MAX          512
 167
 168#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
 169
 170extern char ___assert_task_state[1 - 2*!!(
 171                sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
 172
 173/* Convenience macros for the sake of set_task_state */
 174#define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
 175#define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
 176#define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
 177
 178/* Convenience macros for the sake of wake_up */
 179#define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 180#define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
 181
 182/* get_task_state() */
 183#define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 184                                 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 185                                 __TASK_TRACED)
 186
 187#define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
 188#define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
 189#define task_is_dead(task)      ((task)->exit_state != 0)
 190#define task_is_stopped_or_traced(task) \
 191                        ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 192#define task_contributes_to_load(task)  \
 193                                ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
 194                                 (task->flags & PF_FROZEN) == 0)
 195
 196#define __set_task_state(tsk, state_value)              \
 197        do { (tsk)->state = (state_value); } while (0)
 198#define set_task_state(tsk, state_value)                \
 199        set_mb((tsk)->state, (state_value))
 200
 201/*
 202 * set_current_state() includes a barrier so that the write of current->state
 203 * is correctly serialised wrt the caller's subsequent test of whether to
 204 * actually sleep:
 205 *
 206 *      set_current_state(TASK_UNINTERRUPTIBLE);
 207 *      if (do_i_need_to_sleep())
 208 *              schedule();
 209 *
 210 * If the caller does not need such serialisation then use __set_current_state()
 211 */
 212#define __set_current_state(state_value)                        \
 213        do { current->state = (state_value); } while (0)
 214#define set_current_state(state_value)          \
 215        set_mb(current->state, (state_value))
 216
 217/* Task command name length */
 218#define TASK_COMM_LEN 16
 219
 220#include <linux/spinlock.h>
 221
 222/*
 223 * This serializes "schedule()" and also protects
 224 * the run-queue from deletions/modifications (but
 225 * _adding_ to the beginning of the run-queue has
 226 * a separate lock).
 227 */
 228extern rwlock_t tasklist_lock;
 229extern spinlock_t mmlist_lock;
 230
 231struct task_struct;
 232
 233#ifdef CONFIG_PROVE_RCU
 234extern int lockdep_tasklist_lock_is_held(void);
 235#endif /* #ifdef CONFIG_PROVE_RCU */
 236
 237extern void sched_init(void);
 238extern void sched_init_smp(void);
 239extern asmlinkage void schedule_tail(struct task_struct *prev);
 240extern void init_idle(struct task_struct *idle, int cpu);
 241extern void init_idle_bootup_task(struct task_struct *idle);
 242
 243extern int runqueue_is_locked(int cpu);
 244
 245#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
 246extern void nohz_balance_enter_idle(int cpu);
 247extern void set_cpu_sd_state_idle(void);
 248extern int get_nohz_timer_target(void);
 249#else
 250static inline void nohz_balance_enter_idle(int cpu) { }
 251static inline void set_cpu_sd_state_idle(void) { }
 252#endif
 253
 254/*
 255 * Only dump TASK_* tasks. (0 for all tasks)
 256 */
 257extern void show_state_filter(unsigned long state_filter);
 258
 259static inline void show_state(void)
 260{
 261        show_state_filter(0);
 262}
 263
 264extern void show_regs(struct pt_regs *);
 265
 266/*
 267 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 268 * task), SP is the stack pointer of the first frame that should be shown in the back
 269 * trace (or NULL if the entire call-chain of the task should be shown).
 270 */
 271extern void show_stack(struct task_struct *task, unsigned long *sp);
 272
 273void io_schedule(void);
 274long io_schedule_timeout(long timeout);
 275
 276extern void cpu_init (void);
 277extern void trap_init(void);
 278extern void update_process_times(int user);
 279extern void scheduler_tick(void);
 280
 281extern void sched_show_task(struct task_struct *p);
 282
 283#ifdef CONFIG_LOCKUP_DETECTOR
 284extern void touch_softlockup_watchdog(void);
 285extern void touch_softlockup_watchdog_sync(void);
 286extern void touch_all_softlockup_watchdogs(void);
 287extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
 288                                  void __user *buffer,
 289                                  size_t *lenp, loff_t *ppos);
 290extern unsigned int  softlockup_panic;
 291void lockup_detector_init(void);
 292#else
 293static inline void touch_softlockup_watchdog(void)
 294{
 295}
 296static inline void touch_softlockup_watchdog_sync(void)
 297{
 298}
 299static inline void touch_all_softlockup_watchdogs(void)
 300{
 301}
 302static inline void lockup_detector_init(void)
 303{
 304}
 305#endif
 306
 307#ifdef CONFIG_DETECT_HUNG_TASK
 308extern unsigned int  sysctl_hung_task_panic;
 309extern unsigned long sysctl_hung_task_check_count;
 310extern unsigned long sysctl_hung_task_timeout_secs;
 311extern unsigned long sysctl_hung_task_warnings;
 312extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
 313                                         void __user *buffer,
 314                                         size_t *lenp, loff_t *ppos);
 315#else
 316/* Avoid need for ifdefs elsewhere in the code */
 317enum { sysctl_hung_task_timeout_secs = 0 };
 318#endif
 319
 320/* Attach to any functions which should be ignored in wchan output. */
 321#define __sched         __attribute__((__section__(".sched.text")))
 322
 323/* Linker adds these: start and end of __sched functions */
 324extern char __sched_text_start[], __sched_text_end[];
 325
 326/* Is this address in the __sched functions? */
 327extern int in_sched_functions(unsigned long addr);
 328
 329#define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 330extern signed long schedule_timeout(signed long timeout);
 331extern signed long schedule_timeout_interruptible(signed long timeout);
 332extern signed long schedule_timeout_killable(signed long timeout);
 333extern signed long schedule_timeout_uninterruptible(signed long timeout);
 334asmlinkage void schedule(void);
 335extern void schedule_preempt_disabled(void);
 336extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
 337
 338struct nsproxy;
 339struct user_namespace;
 340
 341/*
 342 * Default maximum number of active map areas, this limits the number of vmas
 343 * per mm struct. Users can overwrite this number by sysctl but there is a
 344 * problem.
 345 *
 346 * When a program's coredump is generated as ELF format, a section is created
 347 * per a vma. In ELF, the number of sections is represented in unsigned short.
 348 * This means the number of sections should be smaller than 65535 at coredump.
 349 * Because the kernel adds some informative sections to a image of program at
 350 * generating coredump, we need some margin. The number of extra sections is
 351 * 1-3 now and depends on arch. We use "5" as safe margin, here.
 352 */
 353#define MAPCOUNT_ELF_CORE_MARGIN        (5)
 354#define DEFAULT_MAX_MAP_COUNT   (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
 355
 356extern int sysctl_max_map_count;
 357
 358#include <linux/aio.h>
 359
 360#ifdef CONFIG_MMU
 361extern void arch_pick_mmap_layout(struct mm_struct *mm);
 362extern unsigned long
 363arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 364                       unsigned long, unsigned long);
 365extern unsigned long
 366arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 367                          unsigned long len, unsigned long pgoff,
 368                          unsigned long flags);
 369extern void arch_unmap_area(struct mm_struct *, unsigned long);
 370extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
 371#else
 372static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
 373#endif
 374
 375
 376extern void set_dumpable(struct mm_struct *mm, int value);
 377extern int get_dumpable(struct mm_struct *mm);
 378
 379/* get/set_dumpable() values */
 380#define SUID_DUMPABLE_DISABLED  0
 381#define SUID_DUMPABLE_ENABLED   1
 382#define SUID_DUMPABLE_SAFE      2
 383
 384/* mm flags */
 385/* dumpable bits */
 386#define MMF_DUMPABLE      0  /* core dump is permitted */
 387#define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
 388
 389#define MMF_DUMPABLE_BITS 2
 390#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
 391
 392/* coredump filter bits */
 393#define MMF_DUMP_ANON_PRIVATE   2
 394#define MMF_DUMP_ANON_SHARED    3
 395#define MMF_DUMP_MAPPED_PRIVATE 4
 396#define MMF_DUMP_MAPPED_SHARED  5
 397#define MMF_DUMP_ELF_HEADERS    6
 398#define MMF_DUMP_HUGETLB_PRIVATE 7
 399#define MMF_DUMP_HUGETLB_SHARED  8
 400
 401#define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
 402#define MMF_DUMP_FILTER_BITS    7
 403#define MMF_DUMP_FILTER_MASK \
 404        (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
 405#define MMF_DUMP_FILTER_DEFAULT \
 406        ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
 407         (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
 408
 409#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
 410# define MMF_DUMP_MASK_DEFAULT_ELF      (1 << MMF_DUMP_ELF_HEADERS)
 411#else
 412# define MMF_DUMP_MASK_DEFAULT_ELF      0
 413#endif
 414                                        /* leave room for more dump flags */
 415#define MMF_VM_MERGEABLE        16      /* KSM may merge identical pages */
 416#define MMF_VM_HUGEPAGE         17      /* set when VM_HUGEPAGE is set on vma */
 417#define MMF_EXE_FILE_CHANGED    18      /* see prctl_set_mm_exe_file() */
 418
 419#define MMF_HAS_UPROBES         19      /* has uprobes */
 420#define MMF_RECALC_UPROBES      20      /* MMF_HAS_UPROBES can be wrong */
 421
 422#define MMF_INIT_MASK           (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
 423
 424struct sighand_struct {
 425        atomic_t                count;
 426        struct k_sigaction      action[_NSIG];
 427        spinlock_t              siglock;
 428        wait_queue_head_t       signalfd_wqh;
 429};
 430
 431struct pacct_struct {
 432        int                     ac_flag;
 433        long                    ac_exitcode;
 434        unsigned long           ac_mem;
 435        cputime_t               ac_utime, ac_stime;
 436        unsigned long           ac_minflt, ac_majflt;
 437};
 438
 439struct cpu_itimer {
 440        cputime_t expires;
 441        cputime_t incr;
 442        u32 error;
 443        u32 incr_error;
 444};
 445
 446/**
 447 * struct cputime - snaphsot of system and user cputime
 448 * @utime: time spent in user mode
 449 * @stime: time spent in system mode
 450 *
 451 * Gathers a generic snapshot of user and system time.
 452 */
 453struct cputime {
 454        cputime_t utime;
 455        cputime_t stime;
 456};
 457
 458/**
 459 * struct task_cputime - collected CPU time counts
 460 * @utime:              time spent in user mode, in &cputime_t units
 461 * @stime:              time spent in kernel mode, in &cputime_t units
 462 * @sum_exec_runtime:   total time spent on the CPU, in nanoseconds
 463 *
 464 * This is an extension of struct cputime that includes the total runtime
 465 * spent by the task from the scheduler point of view.
 466 *
 467 * As a result, this structure groups together three kinds of CPU time
 468 * that are tracked for threads and thread groups.  Most things considering
 469 * CPU time want to group these counts together and treat all three
 470 * of them in parallel.
 471 */
 472struct task_cputime {
 473        cputime_t utime;
 474        cputime_t stime;
 475        unsigned long long sum_exec_runtime;
 476};
 477/* Alternate field names when used to cache expirations. */
 478#define prof_exp        stime
 479#define virt_exp        utime
 480#define sched_exp       sum_exec_runtime
 481
 482#define INIT_CPUTIME    \
 483        (struct task_cputime) {                                 \
 484                .utime = 0,                                     \
 485                .stime = 0,                                     \
 486                .sum_exec_runtime = 0,                          \
 487        }
 488
 489/*
 490 * Disable preemption until the scheduler is running.
 491 * Reset by start_kernel()->sched_init()->init_idle().
 492 *
 493 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
 494 * before the scheduler is active -- see should_resched().
 495 */
 496#define INIT_PREEMPT_COUNT      (1 + PREEMPT_ACTIVE)
 497
 498/**
 499 * struct thread_group_cputimer - thread group interval timer counts
 500 * @cputime:            thread group interval timers.
 501 * @running:            non-zero when there are timers running and
 502 *                      @cputime receives updates.
 503 * @lock:               lock for fields in this struct.
 504 *
 505 * This structure contains the version of task_cputime, above, that is
 506 * used for thread group CPU timer calculations.
 507 */
 508struct thread_group_cputimer {
 509        struct task_cputime cputime;
 510        int running;
 511        raw_spinlock_t lock;
 512};
 513
 514#include <linux/rwsem.h>
 515struct autogroup;
 516
 517/*
 518 * NOTE! "signal_struct" does not have its own
 519 * locking, because a shared signal_struct always
 520 * implies a shared sighand_struct, so locking
 521 * sighand_struct is always a proper superset of
 522 * the locking of signal_struct.
 523 */
 524struct signal_struct {
 525        atomic_t                sigcnt;
 526        atomic_t                live;
 527        int                     nr_threads;
 528
 529        wait_queue_head_t       wait_chldexit;  /* for wait4() */
 530
 531        /* current thread group signal load-balancing target: */
 532        struct task_struct      *curr_target;
 533
 534        /* shared signal handling: */
 535        struct sigpending       shared_pending;
 536
 537        /* thread group exit support */
 538        int                     group_exit_code;
 539        /* overloaded:
 540         * - notify group_exit_task when ->count is equal to notify_count
 541         * - everyone except group_exit_task is stopped during signal delivery
 542         *   of fatal signals, group_exit_task processes the signal.
 543         */
 544        int                     notify_count;
 545        struct task_struct      *group_exit_task;
 546
 547        /* thread group stop support, overloads group_exit_code too */
 548        int                     group_stop_count;
 549        unsigned int            flags; /* see SIGNAL_* flags below */
 550
 551        /*
 552         * PR_SET_CHILD_SUBREAPER marks a process, like a service
 553         * manager, to re-parent orphan (double-forking) child processes
 554         * to this process instead of 'init'. The service manager is
 555         * able to receive SIGCHLD signals and is able to investigate
 556         * the process until it calls wait(). All children of this
 557         * process will inherit a flag if they should look for a
 558         * child_subreaper process at exit.
 559         */
 560        unsigned int            is_child_subreaper:1;
 561        unsigned int            has_child_subreaper:1;
 562
 563        /* POSIX.1b Interval Timers */
 564        struct list_head posix_timers;
 565
 566        /* ITIMER_REAL timer for the process */
 567        struct hrtimer real_timer;
 568        struct pid *leader_pid;
 569        ktime_t it_real_incr;
 570
 571        /*
 572         * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
 573         * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
 574         * values are defined to 0 and 1 respectively
 575         */
 576        struct cpu_itimer it[2];
 577
 578        /*
 579         * Thread group totals for process CPU timers.
 580         * See thread_group_cputimer(), et al, for details.
 581         */
 582        struct thread_group_cputimer cputimer;
 583
 584        /* Earliest-expiration cache. */
 585        struct task_cputime cputime_expires;
 586
 587        struct list_head cpu_timers[3];
 588
 589        struct pid *tty_old_pgrp;
 590
 591        /* boolean value for session group leader */
 592        int leader;
 593
 594        struct tty_struct *tty; /* NULL if no tty */
 595
 596#ifdef CONFIG_SCHED_AUTOGROUP
 597        struct autogroup *autogroup;
 598#endif
 599        /*
 600         * Cumulative resource counters for dead threads in the group,
 601         * and for reaped dead child processes forked by this group.
 602         * Live threads maintain their own counters and add to these
 603         * in __exit_signal, except for the group leader.
 604         */
 605        cputime_t utime, stime, cutime, cstime;
 606        cputime_t gtime;
 607        cputime_t cgtime;
 608#ifndef CONFIG_VIRT_CPU_ACCOUNTING
 609        struct cputime prev_cputime;
 610#endif
 611        unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 612        unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 613        unsigned long inblock, oublock, cinblock, coublock;
 614        unsigned long maxrss, cmaxrss;
 615        struct task_io_accounting ioac;
 616
 617        /*
 618         * Cumulative ns of schedule CPU time fo dead threads in the
 619         * group, not including a zombie group leader, (This only differs
 620         * from jiffies_to_ns(utime + stime) if sched_clock uses something
 621         * other than jiffies.)
 622         */
 623        unsigned long long sum_sched_runtime;
 624
 625        /*
 626         * We don't bother to synchronize most readers of this at all,
 627         * because there is no reader checking a limit that actually needs
 628         * to get both rlim_cur and rlim_max atomically, and either one
 629         * alone is a single word that can safely be read normally.
 630         * getrlimit/setrlimit use task_lock(current->group_leader) to
 631         * protect this instead of the siglock, because they really
 632         * have no need to disable irqs.
 633         */
 634        struct rlimit rlim[RLIM_NLIMITS];
 635
 636#ifdef CONFIG_BSD_PROCESS_ACCT
 637        struct pacct_struct pacct;      /* per-process accounting information */
 638#endif
 639#ifdef CONFIG_TASKSTATS
 640        struct taskstats *stats;
 641#endif
 642#ifdef CONFIG_AUDIT
 643        unsigned audit_tty;
 644        struct tty_audit_buf *tty_audit_buf;
 645#endif
 646#ifdef CONFIG_CGROUPS
 647        /*
 648         * group_rwsem prevents new tasks from entering the threadgroup and
 649         * member tasks from exiting,a more specifically, setting of
 650         * PF_EXITING.  fork and exit paths are protected with this rwsem
 651         * using threadgroup_change_begin/end().  Users which require
 652         * threadgroup to remain stable should use threadgroup_[un]lock()
 653         * which also takes care of exec path.  Currently, cgroup is the
 654         * only user.
 655         */
 656        struct rw_semaphore group_rwsem;
 657#endif
 658
 659        oom_flags_t oom_flags;
 660        short oom_score_adj;            /* OOM kill score adjustment */
 661        short oom_score_adj_min;        /* OOM kill score adjustment min value.
 662                                         * Only settable by CAP_SYS_RESOURCE. */
 663
 664        struct mutex cred_guard_mutex;  /* guard against foreign influences on
 665                                         * credential calculations
 666                                         * (notably. ptrace) */
 667};
 668
 669/*
 670 * Bits in flags field of signal_struct.
 671 */
 672#define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 673#define SIGNAL_STOP_CONTINUED   0x00000002 /* SIGCONT since WCONTINUED reap */
 674#define SIGNAL_GROUP_EXIT       0x00000004 /* group exit in progress */
 675/*
 676 * Pending notifications to parent.
 677 */
 678#define SIGNAL_CLD_STOPPED      0x00000010
 679#define SIGNAL_CLD_CONTINUED    0x00000020
 680#define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
 681
 682#define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
 683
 684/* If true, all threads except ->group_exit_task have pending SIGKILL */
 685static inline int signal_group_exit(const struct signal_struct *sig)
 686{
 687        return  (sig->flags & SIGNAL_GROUP_EXIT) ||
 688                (sig->group_exit_task != NULL);
 689}
 690
 691/*
 692 * Some day this will be a full-fledged user tracking system..
 693 */
 694struct user_struct {
 695        atomic_t __count;       /* reference count */
 696        atomic_t processes;     /* How many processes does this user have? */
 697        atomic_t files;         /* How many open files does this user have? */
 698        atomic_t sigpending;    /* How many pending signals does this user have? */
 699#ifdef CONFIG_INOTIFY_USER
 700        atomic_t inotify_watches; /* How many inotify watches does this user have? */
 701        atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 702#endif
 703#ifdef CONFIG_FANOTIFY
 704        atomic_t fanotify_listeners;
 705#endif
 706#ifdef CONFIG_EPOLL
 707        atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
 708#endif
 709#ifdef CONFIG_POSIX_MQUEUE
 710        /* protected by mq_lock */
 711        unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 712#endif
 713        unsigned long locked_shm; /* How many pages of mlocked shm ? */
 714
 715#ifdef CONFIG_KEYS
 716        struct key *uid_keyring;        /* UID specific keyring */
 717        struct key *session_keyring;    /* UID's default session keyring */
 718#endif
 719
 720        /* Hash table maintenance information */
 721        struct hlist_node uidhash_node;
 722        kuid_t uid;
 723
 724#ifdef CONFIG_PERF_EVENTS
 725        atomic_long_t locked_vm;
 726#endif
 727};
 728
 729extern int uids_sysfs_init(void);
 730
 731extern struct user_struct *find_user(kuid_t);
 732
 733extern struct user_struct root_user;
 734#define INIT_USER (&root_user)
 735
 736
 737struct backing_dev_info;
 738struct reclaim_state;
 739
 740#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 741struct sched_info {
 742        /* cumulative counters */
 743        unsigned long pcount;         /* # of times run on this cpu */
 744        unsigned long long run_delay; /* time spent waiting on a runqueue */
 745
 746        /* timestamps */
 747        unsigned long long last_arrival,/* when we last ran on a cpu */
 748                           last_queued; /* when we were last queued to run */
 749};
 750#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
 751
 752#ifdef CONFIG_TASK_DELAY_ACCT
 753struct task_delay_info {
 754        spinlock_t      lock;
 755        unsigned int    flags;  /* Private per-task flags */
 756
 757        /* For each stat XXX, add following, aligned appropriately
 758         *
 759         * struct timespec XXX_start, XXX_end;
 760         * u64 XXX_delay;
 761         * u32 XXX_count;
 762         *
 763         * Atomicity of updates to XXX_delay, XXX_count protected by
 764         * single lock above (split into XXX_lock if contention is an issue).
 765         */
 766
 767        /*
 768         * XXX_count is incremented on every XXX operation, the delay
 769         * associated with the operation is added to XXX_delay.
 770         * XXX_delay contains the accumulated delay time in nanoseconds.
 771         */
 772        struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
 773        u64 blkio_delay;        /* wait for sync block io completion */
 774        u64 swapin_delay;       /* wait for swapin block io completion */
 775        u32 blkio_count;        /* total count of the number of sync block */
 776                                /* io operations performed */
 777        u32 swapin_count;       /* total count of the number of swapin block */
 778                                /* io operations performed */
 779
 780        struct timespec freepages_start, freepages_end;
 781        u64 freepages_delay;    /* wait for memory reclaim */
 782        u32 freepages_count;    /* total count of memory reclaim */
 783};
 784#endif  /* CONFIG_TASK_DELAY_ACCT */
 785
 786static inline int sched_info_on(void)
 787{
 788#ifdef CONFIG_SCHEDSTATS
 789        return 1;
 790#elif defined(CONFIG_TASK_DELAY_ACCT)
 791        extern int delayacct_on;
 792        return delayacct_on;
 793#else
 794        return 0;
 795#endif
 796}
 797
 798enum cpu_idle_type {
 799        CPU_IDLE,
 800        CPU_NOT_IDLE,
 801        CPU_NEWLY_IDLE,
 802        CPU_MAX_IDLE_TYPES
 803};
 804
 805/*
 806 * Increase resolution of nice-level calculations for 64-bit architectures.
 807 * The extra resolution improves shares distribution and load balancing of
 808 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
 809 * hierarchies, especially on larger systems. This is not a user-visible change
 810 * and does not change the user-interface for setting shares/weights.
 811 *
 812 * We increase resolution only if we have enough bits to allow this increased
 813 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
 814 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
 815 * increased costs.
 816 */
 817#if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load  */
 818# define SCHED_LOAD_RESOLUTION  10
 819# define scale_load(w)          ((w) << SCHED_LOAD_RESOLUTION)
 820# define scale_load_down(w)     ((w) >> SCHED_LOAD_RESOLUTION)
 821#else
 822# define SCHED_LOAD_RESOLUTION  0
 823# define scale_load(w)          (w)
 824# define scale_load_down(w)     (w)
 825#endif
 826
 827#define SCHED_LOAD_SHIFT        (10 + SCHED_LOAD_RESOLUTION)
 828#define SCHED_LOAD_SCALE        (1L << SCHED_LOAD_SHIFT)
 829
 830/*
 831 * Increase resolution of cpu_power calculations
 832 */
 833#define SCHED_POWER_SHIFT       10
 834#define SCHED_POWER_SCALE       (1L << SCHED_POWER_SHIFT)
 835
 836/*
 837 * sched-domains (multiprocessor balancing) declarations:
 838 */
 839#ifdef CONFIG_SMP
 840#define SD_LOAD_BALANCE         0x0001  /* Do load balancing on this domain. */
 841#define SD_BALANCE_NEWIDLE      0x0002  /* Balance when about to become idle */
 842#define SD_BALANCE_EXEC         0x0004  /* Balance on exec */
 843#define SD_BALANCE_FORK         0x0008  /* Balance on fork, clone */
 844#define SD_BALANCE_WAKE         0x0010  /* Balance on wakeup */
 845#define SD_WAKE_AFFINE          0x0020  /* Wake task to waking CPU */
 846#define SD_SHARE_CPUPOWER       0x0080  /* Domain members share cpu power */
 847#define SD_SHARE_PKG_RESOURCES  0x0200  /* Domain members share cpu pkg resources */
 848#define SD_SERIALIZE            0x0400  /* Only a single load balancing instance */
 849#define SD_ASYM_PACKING         0x0800  /* Place busy groups earlier in the domain */
 850#define SD_PREFER_SIBLING       0x1000  /* Prefer to place tasks in a sibling domain */
 851#define SD_OVERLAP              0x2000  /* sched_domains of this level overlap */
 852
 853extern int __weak arch_sd_sibiling_asym_packing(void);
 854
 855struct sched_group_power {
 856        atomic_t ref;
 857        /*
 858         * CPU power of this group, SCHED_LOAD_SCALE being max power for a
 859         * single CPU.
 860         */
 861        unsigned int power, power_orig;
 862        unsigned long next_update;
 863        /*
 864         * Number of busy cpus in this group.
 865         */
 866        atomic_t nr_busy_cpus;
 867
 868        unsigned long cpumask[0]; /* iteration mask */
 869};
 870
 871struct sched_group {
 872        struct sched_group *next;       /* Must be a circular list */
 873        atomic_t ref;
 874
 875        unsigned int group_weight;
 876        struct sched_group_power *sgp;
 877
 878        /*
 879         * The CPUs this group covers.
 880         *
 881         * NOTE: this field is variable length. (Allocated dynamically
 882         * by attaching extra space to the end of the structure,
 883         * depending on how many CPUs the kernel has booted up with)
 884         */
 885        unsigned long cpumask[0];
 886};
 887
 888static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
 889{
 890        return to_cpumask(sg->cpumask);
 891}
 892
 893/*
 894 * cpumask masking which cpus in the group are allowed to iterate up the domain
 895 * tree.
 896 */
 897static inline struct cpumask *sched_group_mask(struct sched_group *sg)
 898{
 899        return to_cpumask(sg->sgp->cpumask);
 900}
 901
 902/**
 903 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
 904 * @group: The group whose first cpu is to be returned.
 905 */
 906static inline unsigned int group_first_cpu(struct sched_group *group)
 907{
 908        return cpumask_first(sched_group_cpus(group));
 909}
 910
 911struct sched_domain_attr {
 912        int relax_domain_level;
 913};
 914
 915#define SD_ATTR_INIT    (struct sched_domain_attr) {    \
 916        .relax_domain_level = -1,                       \
 917}
 918
 919extern int sched_domain_level_max;
 920
 921struct sched_domain {
 922        /* These fields must be setup */
 923        struct sched_domain *parent;    /* top domain must be null terminated */
 924        struct sched_domain *child;     /* bottom domain must be null terminated */
 925        struct sched_group *groups;     /* the balancing groups of the domain */
 926        unsigned long min_interval;     /* Minimum balance interval ms */
 927        unsigned long max_interval;     /* Maximum balance interval ms */
 928        unsigned int busy_factor;       /* less balancing by factor if busy */
 929        unsigned int imbalance_pct;     /* No balance until over watermark */
 930        unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
 931        unsigned int busy_idx;
 932        unsigned int idle_idx;
 933        unsigned int newidle_idx;
 934        unsigned int wake_idx;
 935        unsigned int forkexec_idx;
 936        unsigned int smt_gain;
 937        int flags;                      /* See SD_* */
 938        int level;
 939
 940        /* Runtime fields. */
 941        unsigned long last_balance;     /* init to jiffies. units in jiffies */
 942        unsigned int balance_interval;  /* initialise to 1. units in ms. */
 943        unsigned int nr_balance_failed; /* initialise to 0 */
 944
 945        u64 last_update;
 946
 947#ifdef CONFIG_SCHEDSTATS
 948        /* load_balance() stats */
 949        unsigned int lb_count[CPU_MAX_IDLE_TYPES];
 950        unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
 951        unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
 952        unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
 953        unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
 954        unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
 955        unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
 956        unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
 957
 958        /* Active load balancing */
 959        unsigned int alb_count;
 960        unsigned int alb_failed;
 961        unsigned int alb_pushed;
 962
 963        /* SD_BALANCE_EXEC stats */
 964        unsigned int sbe_count;
 965        unsigned int sbe_balanced;
 966        unsigned int sbe_pushed;
 967
 968        /* SD_BALANCE_FORK stats */
 969        unsigned int sbf_count;
 970        unsigned int sbf_balanced;
 971        unsigned int sbf_pushed;
 972
 973        /* try_to_wake_up() stats */
 974        unsigned int ttwu_wake_remote;
 975        unsigned int ttwu_move_affine;
 976        unsigned int ttwu_move_balance;
 977#endif
 978#ifdef CONFIG_SCHED_DEBUG
 979        char *name;
 980#endif
 981        union {
 982                void *private;          /* used during construction */
 983                struct rcu_head rcu;    /* used during destruction */
 984        };
 985
 986        unsigned int span_weight;
 987        /*
 988         * Span of all CPUs in this domain.
 989         *
 990         * NOTE: this field is variable length. (Allocated dynamically
 991         * by attaching extra space to the end of the structure,
 992         * depending on how many CPUs the kernel has booted up with)
 993         */
 994        unsigned long span[0];
 995};
 996
 997static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
 998{
 999        return to_cpumask(sd->span);
1000}
1001
1002extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1003                                    struct sched_domain_attr *dattr_new);
1004
1005/* Allocate an array of sched domains, for partition_sched_domains(). */
1006cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1007void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1008
1009/* Test a flag in parent sched domain */
1010static inline int test_sd_parent(struct sched_domain *sd, int flag)
1011{
1012        if (sd->parent && (sd->parent->flags & flag))
1013                return 1;
1014
1015        return 0;
1016}
1017
1018unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1019unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1020
1021bool cpus_share_cache(int this_cpu, int that_cpu);
1022
1023#else /* CONFIG_SMP */
1024
1025struct sched_domain_attr;
1026
1027static inline void
1028partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1029                        struct sched_domain_attr *dattr_new)
1030{
1031}
1032
1033static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1034{
1035        return true;
1036}
1037
1038#endif  /* !CONFIG_SMP */
1039
1040
1041struct io_context;                      /* See blkdev.h */
1042
1043
1044#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1045extern void prefetch_stack(struct task_struct *t);
1046#else
1047static inline void prefetch_stack(struct task_struct *t) { }
1048#endif
1049
1050struct audit_context;           /* See audit.c */
1051struct mempolicy;
1052struct pipe_inode_info;
1053struct uts_namespace;
1054
1055struct rq;
1056struct sched_domain;
1057
1058/*
1059 * wake flags
1060 */
1061#define WF_SYNC         0x01            /* waker goes to sleep after wakup */
1062#define WF_FORK         0x02            /* child wakeup after fork */
1063#define WF_MIGRATED     0x04            /* internal use, task got migrated */
1064
1065#define ENQUEUE_WAKEUP          1
1066#define ENQUEUE_HEAD            2
1067#ifdef CONFIG_SMP
1068#define ENQUEUE_WAKING          4       /* sched_class::task_waking was called */
1069#else
1070#define ENQUEUE_WAKING          0
1071#endif
1072
1073#define DEQUEUE_SLEEP           1
1074
1075struct sched_class {
1076        const struct sched_class *next;
1077
1078        void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1079        void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1080        void (*yield_task) (struct rq *rq);
1081        bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1082
1083        void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1084
1085        struct task_struct * (*pick_next_task) (struct rq *rq);
1086        void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1087
1088#ifdef CONFIG_SMP
1089        int  (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1090        void (*migrate_task_rq)(struct task_struct *p, int next_cpu);
1091
1092        void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1093        void (*post_schedule) (struct rq *this_rq);
1094        void (*task_waking) (struct task_struct *task);
1095        void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1096
1097        void (*set_cpus_allowed)(struct task_struct *p,
1098                                 const struct cpumask *newmask);
1099
1100        void (*rq_online)(struct rq *rq);
1101        void (*rq_offline)(struct rq *rq);
1102#endif
1103
1104        void (*set_curr_task) (struct rq *rq);
1105        void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1106        void (*task_fork) (struct task_struct *p);
1107
1108        void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1109        void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1110        void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1111                             int oldprio);
1112
1113        unsigned int (*get_rr_interval) (struct rq *rq,
1114                                         struct task_struct *task);
1115
1116#ifdef CONFIG_FAIR_GROUP_SCHED
1117        void (*task_move_group) (struct task_struct *p, int on_rq);
1118#endif
1119};
1120
1121struct load_weight {
1122        unsigned long weight, inv_weight;
1123};
1124
1125struct sched_avg {
1126        /*
1127         * These sums represent an infinite geometric series and so are bound
1128         * above by 1024/(1-y).  Thus we only need a u32 to store them for for all
1129         * choices of y < 1-2^(-32)*1024.
1130         */
1131        u32 runnable_avg_sum, runnable_avg_period;
1132        u64 last_runnable_update;
1133        s64 decay_count;
1134        unsigned long load_avg_contrib;
1135};
1136
1137#ifdef CONFIG_SCHEDSTATS
1138struct sched_statistics {
1139        u64                     wait_start;
1140        u64                     wait_max;
1141        u64                     wait_count;
1142        u64                     wait_sum;
1143        u64                     iowait_count;
1144        u64                     iowait_sum;
1145
1146        u64                     sleep_start;
1147        u64                     sleep_max;
1148        s64                     sum_sleep_runtime;
1149
1150        u64                     block_start;
1151        u64                     block_max;
1152        u64                     exec_max;
1153        u64                     slice_max;
1154
1155        u64                     nr_migrations_cold;
1156        u64                     nr_failed_migrations_affine;
1157        u64                     nr_failed_migrations_running;
1158        u64                     nr_failed_migrations_hot;
1159        u64                     nr_forced_migrations;
1160
1161        u64                     nr_wakeups;
1162        u64                     nr_wakeups_sync;
1163        u64                     nr_wakeups_migrate;
1164        u64                     nr_wakeups_local;
1165        u64                     nr_wakeups_remote;
1166        u64                     nr_wakeups_affine;
1167        u64                     nr_wakeups_affine_attempts;
1168        u64                     nr_wakeups_passive;
1169        u64                     nr_wakeups_idle;
1170};
1171#endif
1172
1173struct sched_entity {
1174        struct load_weight      load;           /* for load-balancing */
1175        struct rb_node          run_node;
1176        struct list_head        group_node;
1177        unsigned int            on_rq;
1178
1179        u64                     exec_start;
1180        u64                     sum_exec_runtime;
1181        u64                     vruntime;
1182        u64                     prev_sum_exec_runtime;
1183
1184        u64                     nr_migrations;
1185
1186#ifdef CONFIG_SCHEDSTATS
1187        struct sched_statistics statistics;
1188#endif
1189
1190#ifdef CONFIG_FAIR_GROUP_SCHED
1191        struct sched_entity     *parent;
1192        /* rq on which this entity is (to be) queued: */
1193        struct cfs_rq           *cfs_rq;
1194        /* rq "owned" by this entity/group: */
1195        struct cfs_rq           *my_q;
1196#endif
1197/*
1198 * Load-tracking only depends on SMP, FAIR_GROUP_SCHED dependency below may be
1199 * removed when useful for applications beyond shares distribution (e.g.
1200 * load-balance).
1201 */
1202#if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)
1203        /* Per-entity load-tracking */
1204        struct sched_avg        avg;
1205#endif
1206};
1207
1208struct sched_rt_entity {
1209        struct list_head run_list;
1210        unsigned long timeout;
1211        unsigned int time_slice;
1212
1213        struct sched_rt_entity *back;
1214#ifdef CONFIG_RT_GROUP_SCHED
1215        struct sched_rt_entity  *parent;
1216        /* rq on which this entity is (to be) queued: */
1217        struct rt_rq            *rt_rq;
1218        /* rq "owned" by this entity/group: */
1219        struct rt_rq            *my_q;
1220#endif
1221};
1222
1223/*
1224 * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1225 * Timeslices get refilled after they expire.
1226 */
1227#define RR_TIMESLICE            (100 * HZ / 1000)
1228
1229struct rcu_node;
1230
1231enum perf_event_task_context {
1232        perf_invalid_context = -1,
1233        perf_hw_context = 0,
1234        perf_sw_context,
1235        perf_nr_task_contexts,
1236};
1237
1238struct task_struct {
1239        volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1240        void *stack;
1241        atomic_t usage;
1242        unsigned int flags;     /* per process flags, defined below */
1243        unsigned int ptrace;
1244
1245#ifdef CONFIG_SMP
1246        struct llist_node wake_entry;
1247        int on_cpu;
1248#endif
1249        int on_rq;
1250
1251        int prio, static_prio, normal_prio;
1252        unsigned int rt_priority;
1253        const struct sched_class *sched_class;
1254        struct sched_entity se;
1255        struct sched_rt_entity rt;
1256#ifdef CONFIG_CGROUP_SCHED
1257        struct task_group *sched_task_group;
1258#endif
1259
1260#ifdef CONFIG_PREEMPT_NOTIFIERS
1261        /* list of struct preempt_notifier: */
1262        struct hlist_head preempt_notifiers;
1263#endif
1264
1265        /*
1266         * fpu_counter contains the number of consecutive context switches
1267         * that the FPU is used. If this is over a threshold, the lazy fpu
1268         * saving becomes unlazy to save the trap. This is an unsigned char
1269         * so that after 256 times the counter wraps and the behavior turns
1270         * lazy again; this to deal with bursty apps that only use FPU for
1271         * a short time
1272         */
1273        unsigned char fpu_counter;
1274#ifdef CONFIG_BLK_DEV_IO_TRACE
1275        unsigned int btrace_seq;
1276#endif
1277
1278        unsigned int policy;
1279        int nr_cpus_allowed;
1280        cpumask_t cpus_allowed;
1281
1282#ifdef CONFIG_PREEMPT_RCU
1283        int rcu_read_lock_nesting;
1284        char rcu_read_unlock_special;
1285        struct list_head rcu_node_entry;
1286#endif /* #ifdef CONFIG_PREEMPT_RCU */
1287#ifdef CONFIG_TREE_PREEMPT_RCU
1288        struct rcu_node *rcu_blocked_node;
1289#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1290#ifdef CONFIG_RCU_BOOST
1291        struct rt_mutex *rcu_boost_mutex;
1292#endif /* #ifdef CONFIG_RCU_BOOST */
1293
1294#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1295        struct sched_info sched_info;
1296#endif
1297
1298        struct list_head tasks;
1299#ifdef CONFIG_SMP
1300        struct plist_node pushable_tasks;
1301#endif
1302
1303        struct mm_struct *mm, *active_mm;
1304#ifdef CONFIG_COMPAT_BRK
1305        unsigned brk_randomized:1;
1306#endif
1307#if defined(SPLIT_RSS_COUNTING)
1308        struct task_rss_stat    rss_stat;
1309#endif
1310/* task state */
1311        int exit_state;
1312        int exit_code, exit_signal;
1313        int pdeath_signal;  /*  The signal sent when the parent dies  */
1314        unsigned int jobctl;    /* JOBCTL_*, siglock protected */
1315        /* ??? */
1316        unsigned int personality;
1317        unsigned did_exec:1;
1318        unsigned in_execve:1;   /* Tell the LSMs that the process is doing an
1319                                 * execve */
1320        unsigned in_iowait:1;
1321
1322        /* task may not gain privileges */
1323        unsigned no_new_privs:1;
1324
1325        /* Revert to default priority/policy when forking */
1326        unsigned sched_reset_on_fork:1;
1327        unsigned sched_contributes_to_load:1;
1328
1329        pid_t pid;
1330        pid_t tgid;
1331
1332#ifdef CONFIG_CC_STACKPROTECTOR
1333        /* Canary value for the -fstack-protector gcc feature */
1334        unsigned long stack_canary;
1335#endif
1336        /*
1337         * pointers to (original) parent process, youngest child, younger sibling,
1338         * older sibling, respectively.  (p->father can be replaced with
1339         * p->real_parent->pid)
1340         */
1341        struct task_struct __rcu *real_parent; /* real parent process */
1342        struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1343        /*
1344         * children/sibling forms the list of my natural children
1345         */
1346        struct list_head children;      /* list of my children */
1347        struct list_head sibling;       /* linkage in my parent's children list */
1348        struct task_struct *group_leader;       /* threadgroup leader */
1349
1350        /*
1351         * ptraced is the list of tasks this task is using ptrace on.
1352         * This includes both natural children and PTRACE_ATTACH targets.
1353         * p->ptrace_entry is p's link on the p->parent->ptraced list.
1354         */
1355        struct list_head ptraced;
1356        struct list_head ptrace_entry;
1357
1358        /* PID/PID hash table linkage. */
1359        struct pid_link pids[PIDTYPE_MAX];
1360        struct list_head thread_group;
1361
1362        struct completion *vfork_done;          /* for vfork() */
1363        int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1364        int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1365
1366        cputime_t utime, stime, utimescaled, stimescaled;
1367        cputime_t gtime;
1368#ifndef CONFIG_VIRT_CPU_ACCOUNTING
1369        struct cputime prev_cputime;
1370#endif
1371        unsigned long nvcsw, nivcsw; /* context switch counts */
1372        struct timespec start_time;             /* monotonic time */
1373        struct timespec real_start_time;        /* boot based time */
1374/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1375        unsigned long min_flt, maj_flt;
1376
1377        struct task_cputime cputime_expires;
1378        struct list_head cpu_timers[3];
1379
1380/* process credentials */
1381        const struct cred __rcu *real_cred; /* objective and real subjective task
1382                                         * credentials (COW) */
1383        const struct cred __rcu *cred;  /* effective (overridable) subjective task
1384                                         * credentials (COW) */
1385        char comm[TASK_COMM_LEN]; /* executable name excluding path
1386                                     - access with [gs]et_task_comm (which lock
1387                                       it with task_lock())
1388                                     - initialized normally by setup_new_exec */
1389/* file system info */
1390        int link_count, total_link_count;
1391#ifdef CONFIG_SYSVIPC
1392/* ipc stuff */
1393        struct sysv_sem sysvsem;
1394#endif
1395#ifdef CONFIG_DETECT_HUNG_TASK
1396/* hung task detection */
1397        unsigned long last_switch_count;
1398#endif
1399/* CPU-specific state of this task */
1400        struct thread_struct thread;
1401/* filesystem information */
1402        struct fs_struct *fs;
1403/* open file information */
1404        struct files_struct *files;
1405/* namespaces */
1406        struct nsproxy *nsproxy;
1407/* signal handlers */
1408        struct signal_struct *signal;
1409        struct sighand_struct *sighand;
1410
1411        sigset_t blocked, real_blocked;
1412        sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1413        struct sigpending pending;
1414
1415        unsigned long sas_ss_sp;
1416        size_t sas_ss_size;
1417        int (*notifier)(void *priv);
1418        void *notifier_data;
1419        sigset_t *notifier_mask;
1420        struct callback_head *task_works;
1421
1422        struct audit_context *audit_context;
1423#ifdef CONFIG_AUDITSYSCALL
1424        kuid_t loginuid;
1425        unsigned int sessionid;
1426#endif
1427        struct seccomp seccomp;
1428
1429/* Thread group tracking */
1430        u32 parent_exec_id;
1431        u32 self_exec_id;
1432/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1433 * mempolicy */
1434        spinlock_t alloc_lock;
1435
1436        /* Protection of the PI data structures: */
1437        raw_spinlock_t pi_lock;
1438
1439#ifdef CONFIG_RT_MUTEXES
1440        /* PI waiters blocked on a rt_mutex held by this task */
1441        struct plist_head pi_waiters;
1442        /* Deadlock detection and priority inheritance handling */
1443        struct rt_mutex_waiter *pi_blocked_on;
1444#endif
1445
1446#ifdef CONFIG_DEBUG_MUTEXES
1447        /* mutex deadlock detection */
1448        struct mutex_waiter *blocked_on;
1449#endif
1450#ifdef CONFIG_TRACE_IRQFLAGS
1451        unsigned int irq_events;
1452        unsigned long hardirq_enable_ip;
1453        unsigned long hardirq_disable_ip;
1454        unsigned int hardirq_enable_event;
1455        unsigned int hardirq_disable_event;
1456        int hardirqs_enabled;
1457        int hardirq_context;
1458        unsigned long softirq_disable_ip;
1459        unsigned long softirq_enable_ip;
1460        unsigned int softirq_disable_event;
1461        unsigned int softirq_enable_event;
1462        int softirqs_enabled;
1463        int softirq_context;
1464#endif
1465#ifdef CONFIG_LOCKDEP
1466# define MAX_LOCK_DEPTH 48UL
1467        u64 curr_chain_key;
1468        int lockdep_depth;
1469        unsigned int lockdep_recursion;
1470        struct held_lock held_locks[MAX_LOCK_DEPTH];
1471        gfp_t lockdep_reclaim_gfp;
1472#endif
1473
1474/* journalling filesystem info */
1475        void *journal_info;
1476
1477/* stacked block device info */
1478        struct bio_list *bio_list;
1479
1480#ifdef CONFIG_BLOCK
1481/* stack plugging */
1482        struct blk_plug *plug;
1483#endif
1484
1485/* VM state */
1486        struct reclaim_state *reclaim_state;
1487
1488        struct backing_dev_info *backing_dev_info;
1489
1490        struct io_context *io_context;
1491
1492        unsigned long ptrace_message;
1493        siginfo_t *last_siginfo; /* For ptrace use.  */
1494        struct task_io_accounting ioac;
1495#if defined(CONFIG_TASK_XACCT)
1496        u64 acct_rss_mem1;      /* accumulated rss usage */
1497        u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1498        cputime_t acct_timexpd; /* stime + utime since last update */
1499#endif
1500#ifdef CONFIG_CPUSETS
1501        nodemask_t mems_allowed;        /* Protected by alloc_lock */
1502        seqcount_t mems_allowed_seq;    /* Seqence no to catch updates */
1503        int cpuset_mem_spread_rotor;
1504        int cpuset_slab_spread_rotor;
1505#endif
1506#ifdef CONFIG_CGROUPS
1507        /* Control Group info protected by css_set_lock */
1508        struct css_set __rcu *cgroups;
1509        /* cg_list protected by css_set_lock and tsk->alloc_lock */
1510        struct list_head cg_list;
1511#endif
1512#ifdef CONFIG_FUTEX
1513        struct robust_list_head __user *robust_list;
1514#ifdef CONFIG_COMPAT
1515        struct compat_robust_list_head __user *compat_robust_list;
1516#endif
1517        struct list_head pi_state_list;
1518        struct futex_pi_state *pi_state_cache;
1519#endif
1520#ifdef CONFIG_PERF_EVENTS
1521        struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1522        struct mutex perf_event_mutex;
1523        struct list_head perf_event_list;
1524#endif
1525#ifdef CONFIG_NUMA
1526        struct mempolicy *mempolicy;    /* Protected by alloc_lock */
1527        short il_next;
1528        short pref_node_fork;
1529#endif
1530#ifdef CONFIG_NUMA_BALANCING
1531        int numa_scan_seq;
1532        int numa_migrate_seq;
1533        unsigned int numa_scan_period;
1534        u64 node_stamp;                 /* migration stamp  */
1535        struct callback_head numa_work;
1536#endif /* CONFIG_NUMA_BALANCING */
1537
1538        struct rcu_head rcu;
1539
1540        /*
1541         * cache last used pipe for splice
1542         */
1543        struct pipe_inode_info *splice_pipe;
1544
1545        struct page_frag task_frag;
1546
1547#ifdef  CONFIG_TASK_DELAY_ACCT
1548        struct task_delay_info *delays;
1549#endif
1550#ifdef CONFIG_FAULT_INJECTION
1551        int make_it_fail;
1552#endif
1553        /*
1554         * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1555         * balance_dirty_pages() for some dirty throttling pause
1556         */
1557        int nr_dirtied;
1558        int nr_dirtied_pause;
1559        unsigned long dirty_paused_when; /* start of a write-and-pause period */
1560
1561#ifdef CONFIG_LATENCYTOP
1562        int latency_record_count;
1563        struct latency_record latency_record[LT_SAVECOUNT];
1564#endif
1565        /*
1566         * time slack values; these are used to round up poll() and
1567         * select() etc timeout values. These are in nanoseconds.
1568         */
1569        unsigned long timer_slack_ns;
1570        unsigned long default_timer_slack_ns;
1571
1572#ifdef CONFIG_FUNCTION_GRAPH_TRACER
1573        /* Index of current stored address in ret_stack */
1574        int curr_ret_stack;
1575        /* Stack of return addresses for return function tracing */
1576        struct ftrace_ret_stack *ret_stack;
1577        /* time stamp for last schedule */
1578        unsigned long long ftrace_timestamp;
1579        /*
1580         * Number of functions that haven't been traced
1581         * because of depth overrun.
1582         */
1583        atomic_t trace_overrun;
1584        /* Pause for the tracing */
1585        atomic_t tracing_graph_pause;
1586#endif
1587#ifdef CONFIG_TRACING
1588        /* state flags for use by tracers */
1589        unsigned long trace;
1590        /* bitmask and counter of trace recursion */
1591        unsigned long trace_recursion;
1592#endif /* CONFIG_TRACING */
1593#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1594        struct memcg_batch_info {
1595                int do_batch;   /* incremented when batch uncharge started */
1596                struct mem_cgroup *memcg; /* target memcg of uncharge */
1597                unsigned long nr_pages; /* uncharged usage */
1598                unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1599        } memcg_batch;
1600        unsigned int memcg_kmem_skip_account;
1601#endif
1602#ifdef CONFIG_HAVE_HW_BREAKPOINT
1603        atomic_t ptrace_bp_refcnt;
1604#endif
1605#ifdef CONFIG_UPROBES
1606        struct uprobe_task *utask;
1607#endif
1608};
1609
1610/* Future-safe accessor for struct task_struct's cpus_allowed. */
1611#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1612
1613#ifdef CONFIG_NUMA_BALANCING
1614extern void task_numa_fault(int node, int pages, bool migrated);
1615extern void set_numabalancing_state(bool enabled);
1616#else
1617static inline void task_numa_fault(int node, int pages, bool migrated)
1618{
1619}
1620static inline void set_numabalancing_state(bool enabled)
1621{
1622}
1623#endif
1624
1625/*
1626 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1627 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1628 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1629 * values are inverted: lower p->prio value means higher priority.
1630 *
1631 * The MAX_USER_RT_PRIO value allows the actual maximum
1632 * RT priority to be separate from the value exported to
1633 * user-space.  This allows kernel threads to set their
1634 * priority to a value higher than any user task. Note:
1635 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1636 */
1637
1638#define MAX_USER_RT_PRIO        100
1639#define MAX_RT_PRIO             MAX_USER_RT_PRIO
1640
1641#define MAX_PRIO                (MAX_RT_PRIO + 40)
1642#define DEFAULT_PRIO            (MAX_RT_PRIO + 20)
1643
1644static inline int rt_prio(int prio)
1645{
1646        if (unlikely(prio < MAX_RT_PRIO))
1647                return 1;
1648        return 0;
1649}
1650
1651static inline int rt_task(struct task_struct *p)
1652{
1653        return rt_prio(p->prio);
1654}
1655
1656static inline struct pid *task_pid(struct task_struct *task)
1657{
1658        return task->pids[PIDTYPE_PID].pid;
1659}
1660
1661static inline struct pid *task_tgid(struct task_struct *task)
1662{
1663        return task->group_leader->pids[PIDTYPE_PID].pid;
1664}
1665
1666/*
1667 * Without tasklist or rcu lock it is not safe to dereference
1668 * the result of task_pgrp/task_session even if task == current,
1669 * we can race with another thread doing sys_setsid/sys_setpgid.
1670 */
1671static inline struct pid *task_pgrp(struct task_struct *task)
1672{
1673        return task->group_leader->pids[PIDTYPE_PGID].pid;
1674}
1675
1676static inline struct pid *task_session(struct task_struct *task)
1677{
1678        return task->group_leader->pids[PIDTYPE_SID].pid;
1679}
1680
1681struct pid_namespace;
1682
1683/*
1684 * the helpers to get the task's different pids as they are seen
1685 * from various namespaces
1686 *
1687 * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1688 * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1689 *                     current.
1690 * task_xid_nr_ns()  : id seen from the ns specified;
1691 *
1692 * set_task_vxid()   : assigns a virtual id to a task;
1693 *
1694 * see also pid_nr() etc in include/linux/pid.h
1695 */
1696pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1697                        struct pid_namespace *ns);
1698
1699static inline pid_t task_pid_nr(struct task_struct *tsk)
1700{
1701        return tsk->pid;
1702}
1703
1704static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1705                                        struct pid_namespace *ns)
1706{
1707        return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1708}
1709
1710static inline pid_t task_pid_vnr(struct task_struct *tsk)
1711{
1712        return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1713}
1714
1715
1716static inline pid_t task_tgid_nr(struct task_struct *tsk)
1717{
1718        return tsk->tgid;
1719}
1720
1721pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1722
1723static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1724{
1725        return pid_vnr(task_tgid(tsk));
1726}
1727
1728
1729static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1730                                        struct pid_namespace *ns)
1731{
1732        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1733}
1734
1735static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1736{
1737        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1738}
1739
1740
1741static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1742                                        struct pid_namespace *ns)
1743{
1744        return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1745}
1746
1747static inline pid_t task_session_vnr(struct task_struct *tsk)
1748{
1749        return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1750}
1751
1752/* obsolete, do not use */
1753static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1754{
1755        return task_pgrp_nr_ns(tsk, &init_pid_ns);
1756}
1757
1758/**
1759 * pid_alive - check that a task structure is not stale
1760 * @p: Task structure to be checked.
1761 *
1762 * Test if a process is not yet dead (at most zombie state)
1763 * If pid_alive fails, then pointers within the task structure
1764 * can be stale and must not be dereferenced.
1765 */
1766static inline int pid_alive(struct task_struct *p)
1767{
1768        return p->pids[PIDTYPE_PID].pid != NULL;
1769}
1770
1771/**
1772 * is_global_init - check if a task structure is init
1773 * @tsk: Task structure to be checked.
1774 *
1775 * Check if a task structure is the first user space task the kernel created.
1776 */
1777static inline int is_global_init(struct task_struct *tsk)
1778{
1779        return tsk->pid == 1;
1780}
1781
1782extern struct pid *cad_pid;
1783
1784extern void free_task(struct task_struct *tsk);
1785#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1786
1787extern void __put_task_struct(struct task_struct *t);
1788
1789static inline void put_task_struct(struct task_struct *t)
1790{
1791        if (atomic_dec_and_test(&t->usage))
1792                __put_task_struct(t);
1793}
1794
1795extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1796extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1797
1798/*
1799 * Per process flags
1800 */
1801#define PF_EXITING      0x00000004      /* getting shut down */
1802#define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
1803#define PF_VCPU         0x00000010      /* I'm a virtual CPU */
1804#define PF_WQ_WORKER    0x00000020      /* I'm a workqueue worker */
1805#define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
1806#define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1807#define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
1808#define PF_DUMPCORE     0x00000200      /* dumped core */
1809#define PF_SIGNALED     0x00000400      /* killed by a signal */
1810#define PF_MEMALLOC     0x00000800      /* Allocating memory */
1811#define PF_NPROC_EXCEEDED 0x00001000    /* set_user noticed that RLIMIT_NPROC was exceeded */
1812#define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
1813#define PF_USED_ASYNC   0x00004000      /* used async_schedule*(), used by module init */
1814#define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
1815#define PF_FROZEN       0x00010000      /* frozen for system suspend */
1816#define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
1817#define PF_KSWAPD       0x00040000      /* I am kswapd */
1818#define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
1819#define PF_KTHREAD      0x00200000      /* I am a kernel thread */
1820#define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
1821#define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
1822#define PF_SPREAD_PAGE  0x01000000      /* Spread page cache over cpuset */
1823#define PF_SPREAD_SLAB  0x02000000      /* Spread some slab caches over cpuset */
1824#define PF_THREAD_BOUND 0x04000000      /* Thread bound to specific cpu */
1825#define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1826#define PF_MEMPOLICY    0x10000000      /* Non-default NUMA mempolicy */
1827#define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
1828#define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezable */
1829
1830/*
1831 * Only the _current_ task can read/write to tsk->flags, but other
1832 * tasks can access tsk->flags in readonly mode for example
1833 * with tsk_used_math (like during threaded core dumping).
1834 * There is however an exception to this rule during ptrace
1835 * or during fork: the ptracer task is allowed to write to the
1836 * child->flags of its traced child (same goes for fork, the parent
1837 * can write to the child->flags), because we're guaranteed the
1838 * child is not running and in turn not changing child->flags
1839 * at the same time the parent does it.
1840 */
1841#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1842#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1843#define clear_used_math() clear_stopped_child_used_math(current)
1844#define set_used_math() set_stopped_child_used_math(current)
1845#define conditional_stopped_child_used_math(condition, child) \
1846        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1847#define conditional_used_math(condition) \
1848        conditional_stopped_child_used_math(condition, current)
1849#define copy_to_stopped_child_used_math(child) \
1850        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1851/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1852#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1853#define used_math() tsk_used_math(current)
1854
1855/*
1856 * task->jobctl flags
1857 */
1858#define JOBCTL_STOP_SIGMASK     0xffff  /* signr of the last group stop */
1859
1860#define JOBCTL_STOP_DEQUEUED_BIT 16     /* stop signal dequeued */
1861#define JOBCTL_STOP_PENDING_BIT 17      /* task should stop for group stop */
1862#define JOBCTL_STOP_CONSUME_BIT 18      /* consume group stop count */
1863#define JOBCTL_TRAP_STOP_BIT    19      /* trap for STOP */
1864#define JOBCTL_TRAP_NOTIFY_BIT  20      /* trap for NOTIFY */
1865#define JOBCTL_TRAPPING_BIT     21      /* switching to TRACED */
1866#define JOBCTL_LISTENING_BIT    22      /* ptracer is listening for events */
1867
1868#define JOBCTL_STOP_DEQUEUED    (1 << JOBCTL_STOP_DEQUEUED_BIT)
1869#define JOBCTL_STOP_PENDING     (1 << JOBCTL_STOP_PENDING_BIT)
1870#define JOBCTL_STOP_CONSUME     (1 << JOBCTL_STOP_CONSUME_BIT)
1871#define JOBCTL_TRAP_STOP        (1 << JOBCTL_TRAP_STOP_BIT)
1872#define JOBCTL_TRAP_NOTIFY      (1 << JOBCTL_TRAP_NOTIFY_BIT)
1873#define JOBCTL_TRAPPING         (1 << JOBCTL_TRAPPING_BIT)
1874#define JOBCTL_LISTENING        (1 << JOBCTL_LISTENING_BIT)
1875
1876#define JOBCTL_TRAP_MASK        (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1877#define JOBCTL_PENDING_MASK     (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1878
1879extern bool task_set_jobctl_pending(struct task_struct *task,
1880                                    unsigned int mask);
1881extern void task_clear_jobctl_trapping(struct task_struct *task);
1882extern void task_clear_jobctl_pending(struct task_struct *task,
1883                                      unsigned int mask);
1884
1885#ifdef CONFIG_PREEMPT_RCU
1886
1887#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1888#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1889
1890static inline void rcu_copy_process(struct task_struct *p)
1891{
1892        p->rcu_read_lock_nesting = 0;
1893        p->rcu_read_unlock_special = 0;
1894#ifdef CONFIG_TREE_PREEMPT_RCU
1895        p->rcu_blocked_node = NULL;
1896#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1897#ifdef CONFIG_RCU_BOOST
1898        p->rcu_boost_mutex = NULL;
1899#endif /* #ifdef CONFIG_RCU_BOOST */
1900        INIT_LIST_HEAD(&p->rcu_node_entry);
1901}
1902
1903#else
1904
1905static inline void rcu_copy_process(struct task_struct *p)
1906{
1907}
1908
1909#endif
1910
1911static inline void tsk_restore_flags(struct task_struct *task,
1912                                unsigned long orig_flags, unsigned long flags)
1913{
1914        task->flags &= ~flags;
1915        task->flags |= orig_flags & flags;
1916}
1917
1918#ifdef CONFIG_SMP
1919extern void do_set_cpus_allowed(struct task_struct *p,
1920                               const struct cpumask *new_mask);
1921
1922extern int set_cpus_allowed_ptr(struct task_struct *p,
1923                                const struct cpumask *new_mask);
1924#else
1925static inline void do_set_cpus_allowed(struct task_struct *p,
1926                                      const struct cpumask *new_mask)
1927{
1928}
1929static inline int set_cpus_allowed_ptr(struct task_struct *p,
1930                                       const struct cpumask *new_mask)
1931{
1932        if (!cpumask_test_cpu(0, new_mask))
1933                return -EINVAL;
1934        return 0;
1935}
1936#endif
1937
1938#ifdef CONFIG_NO_HZ
1939void calc_load_enter_idle(void);
1940void calc_load_exit_idle(void);
1941#else
1942static inline void calc_load_enter_idle(void) { }
1943static inline void calc_load_exit_idle(void) { }
1944#endif /* CONFIG_NO_HZ */
1945
1946#ifndef CONFIG_CPUMASK_OFFSTACK
1947static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1948{
1949        return set_cpus_allowed_ptr(p, &new_mask);
1950}
1951#endif
1952
1953/*
1954 * Do not use outside of architecture code which knows its limitations.
1955 *
1956 * sched_clock() has no promise of monotonicity or bounded drift between
1957 * CPUs, use (which you should not) requires disabling IRQs.
1958 *
1959 * Please use one of the three interfaces below.
1960 */
1961extern unsigned long long notrace sched_clock(void);
1962/*
1963 * See the comment in kernel/sched/clock.c
1964 */
1965extern u64 cpu_clock(int cpu);
1966extern u64 local_clock(void);
1967extern u64 sched_clock_cpu(int cpu);
1968
1969
1970extern void sched_clock_init(void);
1971
1972#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1973static inline void sched_clock_tick(void)
1974{
1975}
1976
1977static inline void sched_clock_idle_sleep_event(void)
1978{
1979}
1980
1981static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1982{
1983}
1984#else
1985/*
1986 * Architectures can set this to 1 if they have specified
1987 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1988 * but then during bootup it turns out that sched_clock()
1989 * is reliable after all:
1990 */
1991extern int sched_clock_stable;
1992
1993extern void sched_clock_tick(void);
1994extern void sched_clock_idle_sleep_event(void);
1995extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1996#endif
1997
1998#ifdef CONFIG_IRQ_TIME_ACCOUNTING
1999/*
2000 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2001 * The reason for this explicit opt-in is not to have perf penalty with
2002 * slow sched_clocks.
2003 */
2004extern void enable_sched_clock_irqtime(void);
2005extern void disable_sched_clock_irqtime(void);
2006#else
2007static inline void enable_sched_clock_irqtime(void) {}
2008static inline void disable_sched_clock_irqtime(void) {}
2009#endif
2010
2011extern unsigned long long
2012task_sched_runtime(struct task_struct *task);
2013
2014/* sched_exec is called by processes performing an exec */
2015#ifdef CONFIG_SMP
2016extern void sched_exec(void);
2017#else
2018#define sched_exec()   {}
2019#endif
2020
2021extern void sched_clock_idle_sleep_event(void);
2022extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2023
2024#ifdef CONFIG_HOTPLUG_CPU
2025extern void idle_task_exit(void);
2026#else
2027static inline void idle_task_exit(void) {}
2028#endif
2029
2030#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2031extern void wake_up_idle_cpu(int cpu);
2032#else
2033static inline void wake_up_idle_cpu(int cpu) { }
2034#endif
2035
2036extern unsigned int sysctl_sched_latency;
2037extern unsigned int sysctl_sched_min_granularity;
2038extern unsigned int sysctl_sched_wakeup_granularity;
2039extern unsigned int sysctl_sched_child_runs_first;
2040
2041enum sched_tunable_scaling {
2042        SCHED_TUNABLESCALING_NONE,
2043        SCHED_TUNABLESCALING_LOG,
2044        SCHED_TUNABLESCALING_LINEAR,
2045        SCHED_TUNABLESCALING_END,
2046};
2047extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2048
2049extern unsigned int sysctl_numa_balancing_scan_delay;
2050extern unsigned int sysctl_numa_balancing_scan_period_min;
2051extern unsigned int sysctl_numa_balancing_scan_period_max;
2052extern unsigned int sysctl_numa_balancing_scan_period_reset;
2053extern unsigned int sysctl_numa_balancing_scan_size;
2054extern unsigned int sysctl_numa_balancing_settle_count;
2055
2056#ifdef CONFIG_SCHED_DEBUG
2057extern unsigned int sysctl_sched_migration_cost;
2058extern unsigned int sysctl_sched_nr_migrate;
2059extern unsigned int sysctl_sched_time_avg;
2060extern unsigned int sysctl_timer_migration;
2061extern unsigned int sysctl_sched_shares_window;
2062
2063int sched_proc_update_handler(struct ctl_table *table, int write,
2064                void __user *buffer, size_t *length,
2065                loff_t *ppos);
2066#endif
2067#ifdef CONFIG_SCHED_DEBUG
2068static inline unsigned int get_sysctl_timer_migration(void)
2069{
2070        return sysctl_timer_migration;
2071}
2072#else
2073static inline unsigned int get_sysctl_timer_migration(void)
2074{
2075        return 1;
2076}
2077#endif
2078extern unsigned int sysctl_sched_rt_period;
2079extern int sysctl_sched_rt_runtime;
2080
2081int sched_rt_handler(struct ctl_table *table, int write,
2082                void __user *buffer, size_t *lenp,
2083                loff_t *ppos);
2084
2085#ifdef CONFIG_SCHED_AUTOGROUP
2086extern unsigned int sysctl_sched_autogroup_enabled;
2087
2088extern void sched_autogroup_create_attach(struct task_struct *p);
2089extern void sched_autogroup_detach(struct task_struct *p);
2090extern void sched_autogroup_fork(struct signal_struct *sig);
2091extern void sched_autogroup_exit(struct signal_struct *sig);
2092#ifdef CONFIG_PROC_FS
2093extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2094extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2095#endif
2096#else
2097static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2098static inline void sched_autogroup_detach(struct task_struct *p) { }
2099static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2100static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2101#endif
2102
2103#ifdef CONFIG_CFS_BANDWIDTH
2104extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2105#endif
2106
2107#ifdef CONFIG_RT_MUTEXES
2108extern int rt_mutex_getprio(struct task_struct *p);
2109extern void rt_mutex_setprio(struct task_struct *p, int prio);
2110extern void rt_mutex_adjust_pi(struct task_struct *p);
2111static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2112{
2113        return tsk->pi_blocked_on != NULL;
2114}
2115#else
2116static inline int rt_mutex_getprio(struct task_struct *p)
2117{
2118        return p->normal_prio;
2119}
2120# define rt_mutex_adjust_pi(p)          do { } while (0)
2121static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2122{
2123        return false;
2124}
2125#endif
2126
2127extern bool yield_to(struct task_struct *p, bool preempt);
2128extern void set_user_nice(struct task_struct *p, long nice);
2129extern int task_prio(const struct task_struct *p);
2130extern int task_nice(const struct task_struct *p);
2131extern int can_nice(const struct task_struct *p, const int nice);
2132extern int task_curr(const struct task_struct *p);
2133extern int idle_cpu(int cpu);
2134extern int sched_setscheduler(struct task_struct *, int,
2135                              const struct sched_param *);
2136extern int sched_setscheduler_nocheck(struct task_struct *, int,
2137                                      const struct sched_param *);
2138extern struct task_struct *idle_task(int cpu);
2139/**
2140 * is_idle_task - is the specified task an idle task?
2141 * @p: the task in question.
2142 */
2143static inline bool is_idle_task(const struct task_struct *p)
2144{
2145        return p->pid == 0;
2146}
2147extern struct task_struct *curr_task(int cpu);
2148extern void set_curr_task(int cpu, struct task_struct *p);
2149
2150void yield(void);
2151
2152/*
2153 * The default (Linux) execution domain.
2154 */
2155extern struct exec_domain       default_exec_domain;
2156
2157union thread_union {
2158        struct thread_info thread_info;
2159        unsigned long stack[THREAD_SIZE/sizeof(long)];
2160};
2161
2162#ifndef __HAVE_ARCH_KSTACK_END
2163static inline int kstack_end(void *addr)
2164{
2165        /* Reliable end of stack detection:
2166         * Some APM bios versions misalign the stack
2167         */
2168        return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2169}
2170#endif
2171
2172extern union thread_union init_thread_union;
2173extern struct task_struct init_task;
2174
2175extern struct   mm_struct init_mm;
2176
2177extern struct pid_namespace init_pid_ns;
2178
2179/*
2180 * find a task by one of its numerical ids
2181 *
2182 * find_task_by_pid_ns():
2183 *      finds a task by its pid in the specified namespace
2184 * find_task_by_vpid():
2185 *      finds a task by its virtual pid
2186 *
2187 * see also find_vpid() etc in include/linux/pid.h
2188 */
2189
2190extern struct task_struct *find_task_by_vpid(pid_t nr);
2191extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2192                struct pid_namespace *ns);
2193
2194extern void __set_special_pids(struct pid *pid);
2195
2196/* per-UID process charging. */
2197extern struct user_struct * alloc_uid(kuid_t);
2198static inline struct user_struct *get_uid(struct user_struct *u)
2199{
2200        atomic_inc(&u->__count);
2201        return u;
2202}
2203extern void free_uid(struct user_struct *);
2204
2205#include <asm/current.h>
2206
2207extern void xtime_update(unsigned long ticks);
2208
2209extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2210extern int wake_up_process(struct task_struct *tsk);
2211extern void wake_up_new_task(struct task_struct *tsk);
2212#ifdef CONFIG_SMP
2213 extern void kick_process(struct task_struct *tsk);
2214#else
2215 static inline void kick_process(struct task_struct *tsk) { }
2216#endif
2217extern void sched_fork(struct task_struct *p);
2218extern void sched_dead(struct task_struct *p);
2219
2220extern void proc_caches_init(void);
2221extern void flush_signals(struct task_struct *);
2222extern void __flush_signals(struct task_struct *);
2223extern void ignore_signals(struct task_struct *);
2224extern void flush_signal_handlers(struct task_struct *, int force_default);
2225extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2226
2227static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2228{
2229        unsigned long flags;
2230        int ret;
2231
2232        spin_lock_irqsave(&tsk->sighand->siglock, flags);
2233        ret = dequeue_signal(tsk, mask, info);
2234        spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2235
2236        return ret;
2237}
2238
2239extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2240                              sigset_t *mask);
2241extern void unblock_all_signals(void);
2242extern void release_task(struct task_struct * p);
2243extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2244extern int force_sigsegv(int, struct task_struct *);
2245extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2246extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2247extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2248extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2249                                const struct cred *, u32);
2250extern int kill_pgrp(struct pid *pid, int sig, int priv);
2251extern int kill_pid(struct pid *pid, int sig, int priv);
2252extern int kill_proc_info(int, struct siginfo *, pid_t);
2253extern __must_check bool do_notify_parent(struct task_struct *, int);
2254extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2255extern void force_sig(int, struct task_struct *);
2256extern int send_sig(int, struct task_struct *, int);
2257extern int zap_other_threads(struct task_struct *p);
2258extern struct sigqueue *sigqueue_alloc(void);
2259extern void sigqueue_free(struct sigqueue *);
2260extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2261extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2262extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2263
2264static inline void restore_saved_sigmask(void)
2265{
2266        if (test_and_clear_restore_sigmask())
2267                __set_current_blocked(&current->saved_sigmask);
2268}
2269
2270static inline sigset_t *sigmask_to_save(void)
2271{
2272        sigset_t *res = &current->blocked;
2273        if (unlikely(test_restore_sigmask()))
2274                res = &current->saved_sigmask;
2275        return res;
2276}
2277
2278static inline int kill_cad_pid(int sig, int priv)
2279{
2280        return kill_pid(cad_pid, sig, priv);
2281}
2282
2283/* These can be the second arg to send_sig_info/send_group_sig_info.  */
2284#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2285#define SEND_SIG_PRIV   ((struct siginfo *) 1)
2286#define SEND_SIG_FORCED ((struct siginfo *) 2)
2287
2288/*
2289 * True if we are on the alternate signal stack.
2290 */
2291static inline int on_sig_stack(unsigned long sp)
2292{
2293#ifdef CONFIG_STACK_GROWSUP
2294        return sp >= current->sas_ss_sp &&
2295                sp - current->sas_ss_sp < current->sas_ss_size;
2296#else
2297        return sp > current->sas_ss_sp &&
2298                sp - current->sas_ss_sp <= current->sas_ss_size;
2299#endif
2300}
2301
2302static inline int sas_ss_flags(unsigned long sp)
2303{
2304        return (current->sas_ss_size == 0 ? SS_DISABLE
2305                : on_sig_stack(sp) ? SS_ONSTACK : 0);
2306}
2307
2308/*
2309 * Routines for handling mm_structs
2310 */
2311extern struct mm_struct * mm_alloc(void);
2312
2313/* mmdrop drops the mm and the page tables */
2314extern void __mmdrop(struct mm_struct *);
2315static inline void mmdrop(struct mm_struct * mm)
2316{
2317        if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2318                __mmdrop(mm);
2319}
2320
2321/* mmput gets rid of the mappings and all user-space */
2322extern void mmput(struct mm_struct *);
2323/* Grab a reference to a task's mm, if it is not already going away */
2324extern struct mm_struct *get_task_mm(struct task_struct *task);
2325/*
2326 * Grab a reference to a task's mm, if it is not already going away
2327 * and ptrace_may_access with the mode parameter passed to it
2328 * succeeds.
2329 */
2330extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2331/* Remove the current tasks stale references to the old mm_struct */
2332extern void mm_release(struct task_struct *, struct mm_struct *);
2333/* Allocate a new mm structure and copy contents from tsk->mm */
2334extern struct mm_struct *dup_mm(struct task_struct *tsk);
2335
2336extern int copy_thread(unsigned long, unsigned long, unsigned long,
2337                        struct task_struct *);
2338extern void flush_thread(void);
2339extern void exit_thread(void);
2340
2341extern void exit_files(struct task_struct *);
2342extern void __cleanup_sighand(struct sighand_struct *);
2343
2344extern void exit_itimers(struct signal_struct *);
2345extern void flush_itimer_signals(void);
2346
2347extern void do_group_exit(int);
2348
2349extern int allow_signal(int);
2350extern int disallow_signal(int);
2351
2352extern int do_execve(const char *,
2353                     const char __user * const __user *,
2354                     const char __user * const __user *);
2355extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2356struct task_struct *fork_idle(int);
2357extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2358
2359extern void set_task_comm(struct task_struct *tsk, char *from);
2360extern char *get_task_comm(char *to, struct task_struct *tsk);
2361
2362#ifdef CONFIG_SMP
2363void scheduler_ipi(void);
2364extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2365#else
2366static inline void scheduler_ipi(void) { }
2367static inline unsigned long wait_task_inactive(struct task_struct *p,
2368                                               long match_state)
2369{
2370        return 1;
2371}
2372#endif
2373
2374#define next_task(p) \
2375        list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2376
2377#define for_each_process(p) \
2378        for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2379
2380extern bool current_is_single_threaded(void);
2381
2382/*
2383 * Careful: do_each_thread/while_each_thread is a double loop so
2384 *          'break' will not work as expected - use goto instead.
2385 */
2386#define do_each_thread(g, t) \
2387        for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2388
2389#define while_each_thread(g, t) \
2390        while ((t = next_thread(t)) != g)
2391
2392static inline int get_nr_threads(struct task_struct *tsk)
2393{
2394        return tsk->signal->nr_threads;
2395}
2396
2397static inline bool thread_group_leader(struct task_struct *p)
2398{
2399        return p->exit_signal >= 0;
2400}
2401
2402/* Do to the insanities of de_thread it is possible for a process
2403 * to have the pid of the thread group leader without actually being
2404 * the thread group leader.  For iteration through the pids in proc
2405 * all we care about is that we have a task with the appropriate
2406 * pid, we don't actually care if we have the right task.
2407 */
2408static inline int has_group_leader_pid(struct task_struct *p)
2409{
2410        return p->pid == p->tgid;
2411}
2412
2413static inline
2414int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2415{
2416        return p1->tgid == p2->tgid;
2417}
2418
2419static inline struct task_struct *next_thread(const struct task_struct *p)
2420{
2421        return list_entry_rcu(p->thread_group.next,
2422                              struct task_struct, thread_group);
2423}
2424
2425static inline int thread_group_empty(struct task_struct *p)
2426{
2427        return list_empty(&p->thread_group);
2428}
2429
2430#define delay_group_leader(p) \
2431                (thread_group_leader(p) && !thread_group_empty(p))
2432
2433/*
2434 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2435 * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2436 * pins the final release of task.io_context.  Also protects ->cpuset and
2437 * ->cgroup.subsys[]. And ->vfork_done.
2438 *
2439 * Nests both inside and outside of read_lock(&tasklist_lock).
2440 * It must not be nested with write_lock_irq(&tasklist_lock),
2441 * neither inside nor outside.
2442 */
2443static inline void task_lock(struct task_struct *p)
2444{
2445        spin_lock(&p->alloc_lock);
2446}
2447
2448static inline void task_unlock(struct task_struct *p)
2449{
2450        spin_unlock(&p->alloc_lock);
2451}
2452
2453extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2454                                                        unsigned long *flags);
2455
2456static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2457                                                       unsigned long *flags)
2458{
2459        struct sighand_struct *ret;
2460
2461        ret = __lock_task_sighand(tsk, flags);
2462        (void)__cond_lock(&tsk->sighand->siglock, ret);
2463        return ret;
2464}
2465
2466static inline void unlock_task_sighand(struct task_struct *tsk,
2467                                                unsigned long *flags)
2468{
2469        spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2470}
2471
2472#ifdef CONFIG_CGROUPS
2473static inline void threadgroup_change_begin(struct task_struct *tsk)
2474{
2475        down_read(&tsk->signal->group_rwsem);
2476}
2477static inline void threadgroup_change_end(struct task_struct *tsk)
2478{
2479        up_read(&tsk->signal->group_rwsem);
2480}
2481
2482/**
2483 * threadgroup_lock - lock threadgroup
2484 * @tsk: member task of the threadgroup to lock
2485 *
2486 * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2487 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2488 * perform exec.  This is useful for cases where the threadgroup needs to
2489 * stay stable across blockable operations.
2490 *
2491 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2492 * synchronization.  While held, no new task will be added to threadgroup
2493 * and no existing live task will have its PF_EXITING set.
2494 *
2495 * During exec, a task goes and puts its thread group through unusual
2496 * changes.  After de-threading, exclusive access is assumed to resources
2497 * which are usually shared by tasks in the same group - e.g. sighand may
2498 * be replaced with a new one.  Also, the exec'ing task takes over group
2499 * leader role including its pid.  Exclude these changes while locked by
2500 * grabbing cred_guard_mutex which is used to synchronize exec path.
2501 */
2502static inline void threadgroup_lock(struct task_struct *tsk)
2503{
2504        /*
2505         * exec uses exit for de-threading nesting group_rwsem inside
2506         * cred_guard_mutex. Grab cred_guard_mutex first.
2507         */
2508        mutex_lock(&tsk->signal->cred_guard_mutex);
2509        down_write(&tsk->signal->group_rwsem);
2510}
2511
2512/**
2513 * threadgroup_unlock - unlock threadgroup
2514 * @tsk: member task of the threadgroup to unlock
2515 *
2516 * Reverse threadgroup_lock().
2517 */
2518static inline void threadgroup_unlock(struct task_struct *tsk)
2519{
2520        up_write(&tsk->signal->group_rwsem);
2521        mutex_unlock(&tsk->signal->cred_guard_mutex);
2522}
2523#else
2524static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2525static inline void threadgroup_change_end(struct task_struct *tsk) {}
2526static inline void threadgroup_lock(struct task_struct *tsk) {}
2527static inline void threadgroup_unlock(struct task_struct *tsk) {}
2528#endif
2529
2530#ifndef __HAVE_THREAD_FUNCTIONS
2531
2532#define task_thread_info(task)  ((struct thread_info *)(task)->stack)
2533#define task_stack_page(task)   ((task)->stack)
2534
2535static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2536{
2537        *task_thread_info(p) = *task_thread_info(org);
2538        task_thread_info(p)->task = p;
2539}
2540
2541static inline unsigned long *end_of_stack(struct task_struct *p)
2542{
2543        return (unsigned long *)(task_thread_info(p) + 1);
2544}
2545
2546#endif
2547
2548static inline int object_is_on_stack(void *obj)
2549{
2550        void *stack = task_stack_page(current);
2551
2552        return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2553}
2554
2555extern void thread_info_cache_init(void);
2556
2557#ifdef CONFIG_DEBUG_STACK_USAGE
2558static inline unsigned long stack_not_used(struct task_struct *p)
2559{
2560        unsigned long *n = end_of_stack(p);
2561
2562        do {    /* Skip over canary */
2563                n++;
2564        } while (!*n);
2565
2566        return (unsigned long)n - (unsigned long)end_of_stack(p);
2567}
2568#endif
2569
2570/* set thread flags in other task's structures
2571 * - see asm/thread_info.h for TIF_xxxx flags available
2572 */
2573static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2574{
2575        set_ti_thread_flag(task_thread_info(tsk), flag);
2576}
2577
2578static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2579{
2580        clear_ti_thread_flag(task_thread_info(tsk), flag);
2581}
2582
2583static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2584{
2585        return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2586}
2587
2588static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2589{
2590        return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2591}
2592
2593static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2594{
2595        return test_ti_thread_flag(task_thread_info(tsk), flag);
2596}
2597
2598static inline void set_tsk_need_resched(struct task_struct *tsk)
2599{
2600        set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2601}
2602
2603static inline void clear_tsk_need_resched(struct task_struct *tsk)
2604{
2605        clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2606}
2607
2608static inline int test_tsk_need_resched(struct task_struct *tsk)
2609{
2610        return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2611}
2612
2613static inline int restart_syscall(void)
2614{
2615        set_tsk_thread_flag(current, TIF_SIGPENDING);
2616        return -ERESTARTNOINTR;
2617}
2618
2619static inline int signal_pending(struct task_struct *p)
2620{
2621        return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2622}
2623
2624static inline int __fatal_signal_pending(struct task_struct *p)
2625{
2626        return unlikely(sigismember(&p->pending.signal, SIGKILL));
2627}
2628
2629static inline int fatal_signal_pending(struct task_struct *p)
2630{
2631        return signal_pending(p) && __fatal_signal_pending(p);
2632}
2633
2634static inline int signal_pending_state(long state, struct task_struct *p)
2635{
2636        if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2637                return 0;
2638        if (!signal_pending(p))
2639                return 0;
2640
2641        return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2642}
2643
2644static inline int need_resched(void)
2645{
2646        return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2647}
2648
2649/*
2650 * cond_resched() and cond_resched_lock(): latency reduction via
2651 * explicit rescheduling in places that are safe. The return
2652 * value indicates whether a reschedule was done in fact.
2653 * cond_resched_lock() will drop the spinlock before scheduling,
2654 * cond_resched_softirq() will enable bhs before scheduling.
2655 */
2656extern int _cond_resched(void);
2657
2658#define cond_resched() ({                       \
2659        __might_sleep(__FILE__, __LINE__, 0);   \
2660        _cond_resched();                        \
2661})
2662
2663extern int __cond_resched_lock(spinlock_t *lock);
2664
2665#ifdef CONFIG_PREEMPT_COUNT
2666#define PREEMPT_LOCK_OFFSET     PREEMPT_OFFSET
2667#else
2668#define PREEMPT_LOCK_OFFSET     0
2669#endif
2670
2671#define cond_resched_lock(lock) ({                              \
2672        __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2673        __cond_resched_lock(lock);                              \
2674})
2675
2676extern int __cond_resched_softirq(void);
2677
2678#define cond_resched_softirq() ({                                       \
2679        __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);      \
2680        __cond_resched_softirq();                                       \
2681})
2682
2683/*
2684 * Does a critical section need to be broken due to another
2685 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2686 * but a general need for low latency)
2687 */
2688static inline int spin_needbreak(spinlock_t *lock)
2689{
2690#ifdef CONFIG_PREEMPT
2691        return spin_is_contended(lock);
2692#else
2693        return 0;
2694#endif
2695}
2696
2697/*
2698 * Thread group CPU time accounting.
2699 */
2700void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2701void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2702
2703static inline void thread_group_cputime_init(struct signal_struct *sig)
2704{
2705        raw_spin_lock_init(&sig->cputimer.lock);
2706}
2707
2708/*
2709 * Reevaluate whether the task has signals pending delivery.
2710 * Wake the task if so.
2711 * This is required every time the blocked sigset_t changes.
2712 * callers must hold sighand->siglock.
2713 */
2714extern void recalc_sigpending_and_wake(struct task_struct *t);
2715extern void recalc_sigpending(void);
2716
2717extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2718
2719static inline void signal_wake_up(struct task_struct *t, bool resume)
2720{
2721        signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2722}
2723static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2724{
2725        signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2726}
2727
2728/*
2729 * Wrappers for p->thread_info->cpu access. No-op on UP.
2730 */
2731#ifdef CONFIG_SMP
2732
2733static inline unsigned int task_cpu(const struct task_struct *p)
2734{
2735        return task_thread_info(p)->cpu;
2736}
2737
2738extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2739
2740#else
2741
2742static inline unsigned int task_cpu(const struct task_struct *p)
2743{
2744        return 0;
2745}
2746
2747static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2748{
2749}
2750
2751#endif /* CONFIG_SMP */
2752
2753extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2754extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2755
2756extern void normalize_rt_tasks(void);
2757
2758#ifdef CONFIG_CGROUP_SCHED
2759
2760extern struct task_group root_task_group;
2761
2762extern struct task_group *sched_create_group(struct task_group *parent);
2763extern void sched_destroy_group(struct task_group *tg);
2764extern void sched_move_task(struct task_struct *tsk);
2765#ifdef CONFIG_FAIR_GROUP_SCHED
2766extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2767extern unsigned long sched_group_shares(struct task_group *tg);
2768#endif
2769#ifdef CONFIG_RT_GROUP_SCHED
2770extern int sched_group_set_rt_runtime(struct task_group *tg,
2771                                      long rt_runtime_us);
2772extern long sched_group_rt_runtime(struct task_group *tg);
2773extern int sched_group_set_rt_period(struct task_group *tg,
2774                                      long rt_period_us);
2775extern long sched_group_rt_period(struct task_group *tg);
2776extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2777#endif
2778#endif /* CONFIG_CGROUP_SCHED */
2779
2780extern int task_can_switch_user(struct user_struct *up,
2781                                        struct task_struct *tsk);
2782
2783#ifdef CONFIG_TASK_XACCT
2784static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2785{
2786        tsk->ioac.rchar += amt;
2787}
2788
2789static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2790{
2791        tsk->ioac.wchar += amt;
2792}
2793
2794static inline void inc_syscr(struct task_struct *tsk)
2795{
2796        tsk->ioac.syscr++;
2797}
2798
2799static inline void inc_syscw(struct task_struct *tsk)
2800{
2801        tsk->ioac.syscw++;
2802}
2803#else
2804static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2805{
2806}
2807
2808static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2809{
2810}
2811
2812static inline void inc_syscr(struct task_struct *tsk)
2813{
2814}
2815
2816static inline void inc_syscw(struct task_struct *tsk)
2817{
2818}
2819#endif
2820
2821#ifndef TASK_SIZE_OF
2822#define TASK_SIZE_OF(tsk)       TASK_SIZE
2823#endif
2824
2825#ifdef CONFIG_MM_OWNER
2826extern void mm_update_next_owner(struct mm_struct *mm);
2827extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2828#else
2829static inline void mm_update_next_owner(struct mm_struct *mm)
2830{
2831}
2832
2833static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2834{
2835}
2836#endif /* CONFIG_MM_OWNER */
2837
2838static inline unsigned long task_rlimit(const struct task_struct *tsk,
2839                unsigned int limit)
2840{
2841        return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2842}
2843
2844static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2845                unsigned int limit)
2846{
2847        return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2848}
2849
2850static inline unsigned long rlimit(unsigned int limit)
2851{
2852        return task_rlimit(current, limit);
2853}
2854
2855static inline unsigned long rlimit_max(unsigned int limit)
2856{
2857        return task_rlimit_max(current, limit);
2858}
2859
2860#endif
2861
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.