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/plist.h>
  20#include <linux/rbtree.h>
  21#include <linux/thread_info.h>
  22#include <linux/cpumask.h>
  23#include <linux/errno.h>
  24#include <linux/nodemask.h>
  25#include <linux/mm_types.h>
  26#include <linux/preempt_mask.h>
  27
  28#include <asm/page.h>
  29#include <asm/ptrace.h>
  30#include <asm/cputime.h>
  31
  32#include <linux/smp.h>
  33#include <linux/sem.h>
  34#include <linux/signal.h>
  35#include <linux/compiler.h>
  36#include <linux/completion.h>
  37#include <linux/pid.h>
  38#include <linux/percpu.h>
  39#include <linux/topology.h>
  40#include <linux/proportions.h>
  41#include <linux/seccomp.h>
  42#include <linux/rcupdate.h>
  43#include <linux/rculist.h>
  44#include <linux/rtmutex.h>
  45
  46#include <linux/time.h>
  47#include <linux/param.h>
  48#include <linux/resource.h>
  49#include <linux/timer.h>
  50#include <linux/hrtimer.h>
  51#include <linux/task_io_accounting.h>
  52#include <linux/latencytop.h>
  53#include <linux/cred.h>
  54#include <linux/llist.h>
  55#include <linux/uidgid.h>
  56#include <linux/gfp.h>
  57
  58#include <asm/processor.h>
  59
  60#define SCHED_ATTR_SIZE_VER0    48      /* sizeof first published struct */
  61
  62/*
  63 * Extended scheduling parameters data structure.
  64 *
  65 * This is needed because the original struct sched_param can not be
  66 * altered without introducing ABI issues with legacy applications
  67 * (e.g., in sched_getparam()).
  68 *
  69 * However, the possibility of specifying more than just a priority for
  70 * the tasks may be useful for a wide variety of application fields, e.g.,
  71 * multimedia, streaming, automation and control, and many others.
  72 *
  73 * This variant (sched_attr) is meant at describing a so-called
  74 * sporadic time-constrained task. In such model a task is specified by:
  75 *  - the activation period or minimum instance inter-arrival time;
  76 *  - the maximum (or average, depending on the actual scheduling
  77 *    discipline) computation time of all instances, a.k.a. runtime;
  78 *  - the deadline (relative to the actual activation time) of each
  79 *    instance.
  80 * Very briefly, a periodic (sporadic) task asks for the execution of
  81 * some specific computation --which is typically called an instance--
  82 * (at most) every period. Moreover, each instance typically lasts no more
  83 * than the runtime and must be completed by time instant t equal to
  84 * the instance activation time + the deadline.
  85 *
  86 * This is reflected by the actual fields of the sched_attr structure:
  87 *
  88 *  @size               size of the structure, for fwd/bwd compat.
  89 *
  90 *  @sched_policy       task's scheduling policy
  91 *  @sched_flags        for customizing the scheduler behaviour
  92 *  @sched_nice         task's nice value      (SCHED_NORMAL/BATCH)
  93 *  @sched_priority     task's static priority (SCHED_FIFO/RR)
  94 *  @sched_deadline     representative of the task's deadline
  95 *  @sched_runtime      representative of the task's runtime
  96 *  @sched_period       representative of the task's period
  97 *
  98 * Given this task model, there are a multiplicity of scheduling algorithms
  99 * and policies, that can be used to ensure all the tasks will make their
 100 * timing constraints.
 101 *
 102 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
 103 * only user of this new interface. More information about the algorithm
 104 * available in the scheduling class file or in Documentation/.
 105 */
 106struct sched_attr {
 107        u32 size;
 108
 109        u32 sched_policy;
 110        u64 sched_flags;
 111
 112        /* SCHED_NORMAL, SCHED_BATCH */
 113        s32 sched_nice;
 114
 115        /* SCHED_FIFO, SCHED_RR */
 116        u32 sched_priority;
 117
 118        /* SCHED_DEADLINE */
 119        u64 sched_runtime;
 120        u64 sched_deadline;
 121        u64 sched_period;
 122};
 123
 124struct exec_domain;
 125struct futex_pi_state;
 126struct robust_list_head;
 127struct bio_list;
 128struct fs_struct;
 129struct perf_event_context;
 130struct blk_plug;
 131struct filename;
 132
 133/*
 134 * List of flags we want to share for kernel threads,
 135 * if only because they are not used by them anyway.
 136 */
 137#define CLONE_KERNEL    (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
 138
 139/*
 140 * These are the constant used to fake the fixed-point load-average
 141 * counting. Some notes:
 142 *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
 143 *    a load-average precision of 10 bits integer + 11 bits fractional
 144 *  - if you want to count load-averages more often, you need more
 145 *    precision, or rounding will get you. With 2-second counting freq,
 146 *    the EXP_n values would be 1981, 2034 and 2043 if still using only
 147 *    11 bit fractions.
 148 */
 149extern unsigned long avenrun[];         /* Load averages */
 150extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
 151
 152#define FSHIFT          11              /* nr of bits of precision */
 153#define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
 154#define LOAD_FREQ       (5*HZ+1)        /* 5 sec intervals */
 155#define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
 156#define EXP_5           2014            /* 1/exp(5sec/5min) */
 157#define EXP_15          2037            /* 1/exp(5sec/15min) */
 158
 159#define CALC_LOAD(load,exp,n) \
 160        load *= exp; \
 161        load += n*(FIXED_1-exp); \
 162        load >>= FSHIFT;
 163
 164extern unsigned long total_forks;
 165extern int nr_threads;
 166DECLARE_PER_CPU(unsigned long, process_counts);
 167extern int nr_processes(void);
 168extern unsigned long nr_running(void);
 169extern unsigned long nr_iowait(void);
 170extern unsigned long nr_iowait_cpu(int cpu);
 171extern unsigned long this_cpu_load(void);
 172
 173
 174extern void calc_global_load(unsigned long ticks);
 175extern void update_cpu_load_nohz(void);
 176
 177extern unsigned long get_parent_ip(unsigned long addr);
 178
 179extern void dump_cpu_task(int cpu);
 180
 181struct seq_file;
 182struct cfs_rq;
 183struct task_group;
 184#ifdef CONFIG_SCHED_DEBUG
 185extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
 186extern void proc_sched_set_task(struct task_struct *p);
 187extern void
 188print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
 189#endif
 190
 191/*
 192 * Task state bitmask. NOTE! These bits are also
 193 * encoded in fs/proc/array.c: get_task_state().
 194 *
 195 * We have two separate sets of flags: task->state
 196 * is about runnability, while task->exit_state are
 197 * about the task exiting. Confusing, but this way
 198 * modifying one set can't modify the other one by
 199 * mistake.
 200 */
 201#define TASK_RUNNING            0
 202#define TASK_INTERRUPTIBLE      1
 203#define TASK_UNINTERRUPTIBLE    2
 204#define __TASK_STOPPED          4
 205#define __TASK_TRACED           8
 206/* in tsk->exit_state */
 207#define EXIT_ZOMBIE             16
 208#define EXIT_DEAD               32
 209/* in tsk->state again */
 210#define TASK_DEAD               64
 211#define TASK_WAKEKILL           128
 212#define TASK_WAKING             256
 213#define TASK_PARKED             512
 214#define TASK_STATE_MAX          1024
 215
 216#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
 217
 218extern char ___assert_task_state[1 - 2*!!(
 219                sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
 220
 221/* Convenience macros for the sake of set_task_state */
 222#define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
 223#define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
 224#define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
 225
 226/* Convenience macros for the sake of wake_up */
 227#define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 228#define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
 229
 230/* get_task_state() */
 231#define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 232                                 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 233                                 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
 234
 235#define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
 236#define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
 237#define task_is_stopped_or_traced(task) \
 238                        ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 239#define task_contributes_to_load(task)  \
 240                                ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
 241                                 (task->flags & PF_FROZEN) == 0)
 242
 243#define __set_task_state(tsk, state_value)              \
 244        do { (tsk)->state = (state_value); } while (0)
 245#define set_task_state(tsk, state_value)                \
 246        set_mb((tsk)->state, (state_value))
 247
 248/*
 249 * set_current_state() includes a barrier so that the write of current->state
 250 * is correctly serialised wrt the caller's subsequent test of whether to
 251 * actually sleep:
 252 *
 253 *      set_current_state(TASK_UNINTERRUPTIBLE);
 254 *      if (do_i_need_to_sleep())
 255 *              schedule();
 256 *
 257 * If the caller does not need such serialisation then use __set_current_state()
 258 */
 259#define __set_current_state(state_value)                        \
 260        do { current->state = (state_value); } while (0)
 261#define set_current_state(state_value)          \
 262        set_mb(current->state, (state_value))
 263
 264/* Task command name length */
 265#define TASK_COMM_LEN 16
 266
 267#include <linux/spinlock.h>
 268
 269/*
 270 * This serializes "schedule()" and also protects
 271 * the run-queue from deletions/modifications (but
 272 * _adding_ to the beginning of the run-queue has
 273 * a separate lock).
 274 */
 275extern rwlock_t tasklist_lock;
 276extern spinlock_t mmlist_lock;
 277
 278struct task_struct;
 279
 280#ifdef CONFIG_PROVE_RCU
 281extern int lockdep_tasklist_lock_is_held(void);
 282#endif /* #ifdef CONFIG_PROVE_RCU */
 283
 284extern void sched_init(void);
 285extern void sched_init_smp(void);
 286extern asmlinkage void schedule_tail(struct task_struct *prev);
 287extern void init_idle(struct task_struct *idle, int cpu);
 288extern void init_idle_bootup_task(struct task_struct *idle);
 289
 290extern int runqueue_is_locked(int cpu);
 291
 292#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
 293extern void nohz_balance_enter_idle(int cpu);
 294extern void set_cpu_sd_state_idle(void);
 295extern int get_nohz_timer_target(void);
 296#else
 297static inline void nohz_balance_enter_idle(int cpu) { }
 298static inline void set_cpu_sd_state_idle(void) { }
 299#endif
 300
 301/*
 302 * Only dump TASK_* tasks. (0 for all tasks)
 303 */
 304extern void show_state_filter(unsigned long state_filter);
 305
 306static inline void show_state(void)
 307{
 308        show_state_filter(0);
 309}
 310
 311extern void show_regs(struct pt_regs *);
 312
 313/*
 314 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 315 * task), SP is the stack pointer of the first frame that should be shown in the back
 316 * trace (or NULL if the entire call-chain of the task should be shown).
 317 */
 318extern void show_stack(struct task_struct *task, unsigned long *sp);
 319
 320void io_schedule(void);
 321long io_schedule_timeout(long timeout);
 322
 323extern void cpu_init (void);
 324extern void trap_init(void);
 325extern void update_process_times(int user);
 326extern void scheduler_tick(void);
 327
 328extern void sched_show_task(struct task_struct *p);
 329
 330#ifdef CONFIG_LOCKUP_DETECTOR
 331extern void touch_softlockup_watchdog(void);
 332extern void touch_softlockup_watchdog_sync(void);
 333extern void touch_all_softlockup_watchdogs(void);
 334extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
 335                                  void __user *buffer,
 336                                  size_t *lenp, loff_t *ppos);
 337extern unsigned int  softlockup_panic;
 338void lockup_detector_init(void);
 339#else
 340static inline void touch_softlockup_watchdog(void)
 341{
 342}
 343static inline void touch_softlockup_watchdog_sync(void)
 344{
 345}
 346static inline void touch_all_softlockup_watchdogs(void)
 347{
 348}
 349static inline void lockup_detector_init(void)
 350{
 351}
 352#endif
 353
 354#ifdef CONFIG_DETECT_HUNG_TASK
 355void reset_hung_task_detector(void);
 356#else
 357static inline void reset_hung_task_detector(void)
 358{
 359}
 360#endif
 361
 362/* Attach to any functions which should be ignored in wchan output. */
 363#define __sched         __attribute__((__section__(".sched.text")))
 364
 365/* Linker adds these: start and end of __sched functions */
 366extern char __sched_text_start[], __sched_text_end[];
 367
 368/* Is this address in the __sched functions? */
 369extern int in_sched_functions(unsigned long addr);
 370
 371#define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 372extern signed long schedule_timeout(signed long timeout);
 373extern signed long schedule_timeout_interruptible(signed long timeout);
 374extern signed long schedule_timeout_killable(signed long timeout);
 375extern signed long schedule_timeout_uninterruptible(signed long timeout);
 376asmlinkage void schedule(void);
 377extern void schedule_preempt_disabled(void);
 378
 379struct nsproxy;
 380struct user_namespace;
 381
 382#ifdef CONFIG_MMU
 383extern void arch_pick_mmap_layout(struct mm_struct *mm);
 384extern unsigned long
 385arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 386                       unsigned long, unsigned long);
 387extern unsigned long
 388arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 389                          unsigned long len, unsigned long pgoff,
 390                          unsigned long flags);
 391#else
 392static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
 393#endif
 394
 395#define SUID_DUMP_DISABLE       0       /* No setuid dumping */
 396#define SUID_DUMP_USER          1       /* Dump as user of process */
 397#define SUID_DUMP_ROOT          2       /* Dump as root */
 398
 399/* mm flags */
 400
 401/* for SUID_DUMP_* above */
 402#define MMF_DUMPABLE_BITS 2
 403#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
 404
 405extern void set_dumpable(struct mm_struct *mm, int value);
 406/*
 407 * This returns the actual value of the suid_dumpable flag. For things
 408 * that are using this for checking for privilege transitions, it must
 409 * test against SUID_DUMP_USER rather than treating it as a boolean
 410 * value.
 411 */
 412static inline int __get_dumpable(unsigned long mm_flags)
 413{
 414        return mm_flags & MMF_DUMPABLE_MASK;
 415}
 416
 417static inline int get_dumpable(struct mm_struct *mm)
 418{
 419        return __get_dumpable(mm->flags);
 420}
 421
 422/* coredump filter bits */
 423#define MMF_DUMP_ANON_PRIVATE   2
 424#define MMF_DUMP_ANON_SHARED    3
 425#define MMF_DUMP_MAPPED_PRIVATE 4
 426#define MMF_DUMP_MAPPED_SHARED  5
 427#define MMF_DUMP_ELF_HEADERS    6
 428#define MMF_DUMP_HUGETLB_PRIVATE 7
 429#define MMF_DUMP_HUGETLB_SHARED  8
 430
 431#define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
 432#define MMF_DUMP_FILTER_BITS    7
 433#define MMF_DUMP_FILTER_MASK \
 434        (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
 435#define MMF_DUMP_FILTER_DEFAULT \
 436        ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
 437         (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
 438
 439#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
 440# define MMF_DUMP_MASK_DEFAULT_ELF      (1 << MMF_DUMP_ELF_HEADERS)
 441#else
 442# define MMF_DUMP_MASK_DEFAULT_ELF      0
 443#endif
 444                                        /* leave room for more dump flags */
 445#define MMF_VM_MERGEABLE        16      /* KSM may merge identical pages */
 446#define MMF_VM_HUGEPAGE         17      /* set when VM_HUGEPAGE is set on vma */
 447#define MMF_EXE_FILE_CHANGED    18      /* see prctl_set_mm_exe_file() */
 448
 449#define MMF_HAS_UPROBES         19      /* has uprobes */
 450#define MMF_RECALC_UPROBES      20      /* MMF_HAS_UPROBES can be wrong */
 451
 452#define MMF_INIT_MASK           (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
 453
 454struct sighand_struct {
 455        atomic_t                count;
 456        struct k_sigaction      action[_NSIG];
 457        spinlock_t              siglock;
 458        wait_queue_head_t       signalfd_wqh;
 459};
 460
 461struct pacct_struct {
 462        int                     ac_flag;
 463        long                    ac_exitcode;
 464        unsigned long           ac_mem;
 465        cputime_t               ac_utime, ac_stime;
 466        unsigned long           ac_minflt, ac_majflt;
 467};
 468
 469struct cpu_itimer {
 470        cputime_t expires;
 471        cputime_t incr;
 472        u32 error;
 473        u32 incr_error;
 474};
 475
 476/**
 477 * struct cputime - snaphsot of system and user cputime
 478 * @utime: time spent in user mode
 479 * @stime: time spent in system mode
 480 *
 481 * Gathers a generic snapshot of user and system time.
 482 */
 483struct cputime {
 484        cputime_t utime;
 485        cputime_t stime;
 486};
 487
 488/**
 489 * struct task_cputime - collected CPU time counts
 490 * @utime:              time spent in user mode, in &cputime_t units
 491 * @stime:              time spent in kernel mode, in &cputime_t units
 492 * @sum_exec_runtime:   total time spent on the CPU, in nanoseconds
 493 *
 494 * This is an extension of struct cputime that includes the total runtime
 495 * spent by the task from the scheduler point of view.
 496 *
 497 * As a result, this structure groups together three kinds of CPU time
 498 * that are tracked for threads and thread groups.  Most things considering
 499 * CPU time want to group these counts together and treat all three
 500 * of them in parallel.
 501 */
 502struct task_cputime {
 503        cputime_t utime;
 504        cputime_t stime;
 505        unsigned long long sum_exec_runtime;
 506};
 507/* Alternate field names when used to cache expirations. */
 508#define prof_exp        stime
 509#define virt_exp        utime
 510#define sched_exp       sum_exec_runtime
 511
 512#define INIT_CPUTIME    \
 513        (struct task_cputime) {                                 \
 514                .utime = 0,                                     \
 515                .stime = 0,                                     \
 516                .sum_exec_runtime = 0,                          \
 517        }
 518
 519#ifdef CONFIG_PREEMPT_COUNT
 520#define PREEMPT_DISABLED        (1 + PREEMPT_ENABLED)
 521#else
 522#define PREEMPT_DISABLED        PREEMPT_ENABLED
 523#endif
 524
 525/*
 526 * Disable preemption until the scheduler is running.
 527 * Reset by start_kernel()->sched_init()->init_idle().
 528 *
 529 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
 530 * before the scheduler is active -- see should_resched().
 531 */
 532#define INIT_PREEMPT_COUNT      (PREEMPT_DISABLED + PREEMPT_ACTIVE)
 533
 534/**
 535 * struct thread_group_cputimer - thread group interval timer counts
 536 * @cputime:            thread group interval timers.
 537 * @running:            non-zero when there are timers running and
 538 *                      @cputime receives updates.
 539 * @lock:               lock for fields in this struct.
 540 *
 541 * This structure contains the version of task_cputime, above, that is
 542 * used for thread group CPU timer calculations.
 543 */
 544struct thread_group_cputimer {
 545        struct task_cputime cputime;
 546        int running;
 547        raw_spinlock_t lock;
 548};
 549
 550#include <linux/rwsem.h>
 551struct autogroup;
 552
 553/*
 554 * NOTE! "signal_struct" does not have its own
 555 * locking, because a shared signal_struct always
 556 * implies a shared sighand_struct, so locking
 557 * sighand_struct is always a proper superset of
 558 * the locking of signal_struct.
 559 */
 560struct signal_struct {
 561        atomic_t                sigcnt;
 562        atomic_t                live;
 563        int                     nr_threads;
 564        struct list_head        thread_head;
 565
 566        wait_queue_head_t       wait_chldexit;  /* for wait4() */
 567
 568        /* current thread group signal load-balancing target: */
 569        struct task_struct      *curr_target;
 570
 571        /* shared signal handling: */
 572        struct sigpending       shared_pending;
 573
 574        /* thread group exit support */
 575        int                     group_exit_code;
 576        /* overloaded:
 577         * - notify group_exit_task when ->count is equal to notify_count
 578         * - everyone except group_exit_task is stopped during signal delivery
 579         *   of fatal signals, group_exit_task processes the signal.
 580         */
 581        int                     notify_count;
 582        struct task_struct      *group_exit_task;
 583
 584        /* thread group stop support, overloads group_exit_code too */
 585        int                     group_stop_count;
 586        unsigned int            flags; /* see SIGNAL_* flags below */
 587
 588        /*
 589         * PR_SET_CHILD_SUBREAPER marks a process, like a service
 590         * manager, to re-parent orphan (double-forking) child processes
 591         * to this process instead of 'init'. The service manager is
 592         * able to receive SIGCHLD signals and is able to investigate
 593         * the process until it calls wait(). All children of this
 594         * process will inherit a flag if they should look for a
 595         * child_subreaper process at exit.
 596         */
 597        unsigned int            is_child_subreaper:1;
 598        unsigned int            has_child_subreaper:1;
 599
 600        /* POSIX.1b Interval Timers */
 601        int                     posix_timer_id;
 602        struct list_head        posix_timers;
 603
 604        /* ITIMER_REAL timer for the process */
 605        struct hrtimer real_timer;
 606        struct pid *leader_pid;
 607        ktime_t it_real_incr;
 608
 609        /*
 610         * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
 611         * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
 612         * values are defined to 0 and 1 respectively
 613         */
 614        struct cpu_itimer it[2];
 615
 616        /*
 617         * Thread group totals for process CPU timers.
 618         * See thread_group_cputimer(), et al, for details.
 619         */
 620        struct thread_group_cputimer cputimer;
 621
 622        /* Earliest-expiration cache. */
 623        struct task_cputime cputime_expires;
 624
 625        struct list_head cpu_timers[3];
 626
 627        struct pid *tty_old_pgrp;
 628
 629        /* boolean value for session group leader */
 630        int leader;
 631
 632        struct tty_struct *tty; /* NULL if no tty */
 633
 634#ifdef CONFIG_SCHED_AUTOGROUP
 635        struct autogroup *autogroup;
 636#endif
 637        /*
 638         * Cumulative resource counters for dead threads in the group,
 639         * and for reaped dead child processes forked by this group.
 640         * Live threads maintain their own counters and add to these
 641         * in __exit_signal, except for the group leader.
 642         */
 643        cputime_t utime, stime, cutime, cstime;
 644        cputime_t gtime;
 645        cputime_t cgtime;
 646#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
 647        struct cputime prev_cputime;
 648#endif
 649        unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 650        unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 651        unsigned long inblock, oublock, cinblock, coublock;
 652        unsigned long maxrss, cmaxrss;
 653        struct task_io_accounting ioac;
 654
 655        /*
 656         * Cumulative ns of schedule CPU time fo dead threads in the
 657         * group, not including a zombie group leader, (This only differs
 658         * from jiffies_to_ns(utime + stime) if sched_clock uses something
 659         * other than jiffies.)
 660         */
 661        unsigned long long sum_sched_runtime;
 662
 663        /*
 664         * We don't bother to synchronize most readers of this at all,
 665         * because there is no reader checking a limit that actually needs
 666         * to get both rlim_cur and rlim_max atomically, and either one
 667         * alone is a single word that can safely be read normally.
 668         * getrlimit/setrlimit use task_lock(current->group_leader) to
 669         * protect this instead of the siglock, because they really
 670         * have no need to disable irqs.
 671         */
 672        struct rlimit rlim[RLIM_NLIMITS];
 673
 674#ifdef CONFIG_BSD_PROCESS_ACCT
 675        struct pacct_struct pacct;      /* per-process accounting information */
 676#endif
 677#ifdef CONFIG_TASKSTATS
 678        struct taskstats *stats;
 679#endif
 680#ifdef CONFIG_AUDIT
 681        unsigned audit_tty;
 682        unsigned audit_tty_log_passwd;
 683        struct tty_audit_buf *tty_audit_buf;
 684#endif
 685#ifdef CONFIG_CGROUPS
 686        /*
 687         * group_rwsem prevents new tasks from entering the threadgroup and
 688         * member tasks from exiting,a more specifically, setting of
 689         * PF_EXITING.  fork and exit paths are protected with this rwsem
 690         * using threadgroup_change_begin/end().  Users which require
 691         * threadgroup to remain stable should use threadgroup_[un]lock()
 692         * which also takes care of exec path.  Currently, cgroup is the
 693         * only user.
 694         */
 695        struct rw_semaphore group_rwsem;
 696#endif
 697
 698        oom_flags_t oom_flags;
 699        short oom_score_adj;            /* OOM kill score adjustment */
 700        short oom_score_adj_min;        /* OOM kill score adjustment min value.
 701                                         * Only settable by CAP_SYS_RESOURCE. */
 702
 703        struct mutex cred_guard_mutex;  /* guard against foreign influences on
 704                                         * credential calculations
 705                                         * (notably. ptrace) */
 706};
 707
 708/*
 709 * Bits in flags field of signal_struct.
 710 */
 711#define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 712#define SIGNAL_STOP_CONTINUED   0x00000002 /* SIGCONT since WCONTINUED reap */
 713#define SIGNAL_GROUP_EXIT       0x00000004 /* group exit in progress */
 714#define SIGNAL_GROUP_COREDUMP   0x00000008 /* coredump in progress */
 715/*
 716 * Pending notifications to parent.
 717 */
 718#define SIGNAL_CLD_STOPPED      0x00000010
 719#define SIGNAL_CLD_CONTINUED    0x00000020
 720#define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
 721
 722#define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
 723
 724/* If true, all threads except ->group_exit_task have pending SIGKILL */
 725static inline int signal_group_exit(const struct signal_struct *sig)
 726{
 727        return  (sig->flags & SIGNAL_GROUP_EXIT) ||
 728                (sig->group_exit_task != NULL);
 729}
 730
 731/*
 732 * Some day this will be a full-fledged user tracking system..
 733 */
 734struct user_struct {
 735        atomic_t __count;       /* reference count */
 736        atomic_t processes;     /* How many processes does this user have? */
 737        atomic_t files;         /* How many open files does this user have? */
 738        atomic_t sigpending;    /* How many pending signals does this user have? */
 739#ifdef CONFIG_INOTIFY_USER
 740        atomic_t inotify_watches; /* How many inotify watches does this user have? */
 741        atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 742#endif
 743#ifdef CONFIG_FANOTIFY
 744        atomic_t fanotify_listeners;
 745#endif
 746#ifdef CONFIG_EPOLL
 747        atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
 748#endif
 749#ifdef CONFIG_POSIX_MQUEUE
 750        /* protected by mq_lock */
 751        unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 752#endif
 753        unsigned long locked_shm; /* How many pages of mlocked shm ? */
 754
 755#ifdef CONFIG_KEYS
 756        struct key *uid_keyring;        /* UID specific keyring */
 757        struct key *session_keyring;    /* UID's default session keyring */
 758#endif
 759
 760        /* Hash table maintenance information */
 761        struct hlist_node uidhash_node;
 762        kuid_t uid;
 763
 764#ifdef CONFIG_PERF_EVENTS
 765        atomic_long_t locked_vm;
 766#endif
 767};
 768
 769extern int uids_sysfs_init(void);
 770
 771extern struct user_struct *find_user(kuid_t);
 772
 773extern struct user_struct root_user;
 774#define INIT_USER (&root_user)
 775
 776
 777struct backing_dev_info;
 778struct reclaim_state;
 779
 780#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 781struct sched_info {
 782        /* cumulative counters */
 783        unsigned long pcount;         /* # of times run on this cpu */
 784        unsigned long long run_delay; /* time spent waiting on a runqueue */
 785
 786        /* timestamps */
 787        unsigned long long last_arrival,/* when we last ran on a cpu */
 788                           last_queued; /* when we were last queued to run */
 789};
 790#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
 791
 792#ifdef CONFIG_TASK_DELAY_ACCT
 793struct task_delay_info {
 794        spinlock_t      lock;
 795        unsigned int    flags;  /* Private per-task flags */
 796
 797        /* For each stat XXX, add following, aligned appropriately
 798         *
 799         * struct timespec XXX_start, XXX_end;
 800         * u64 XXX_delay;
 801         * u32 XXX_count;
 802         *
 803         * Atomicity of updates to XXX_delay, XXX_count protected by
 804         * single lock above (split into XXX_lock if contention is an issue).
 805         */
 806
 807        /*
 808         * XXX_count is incremented on every XXX operation, the delay
 809         * associated with the operation is added to XXX_delay.
 810         * XXX_delay contains the accumulated delay time in nanoseconds.
 811         */
 812        struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
 813        u64 blkio_delay;        /* wait for sync block io completion */
 814        u64 swapin_delay;       /* wait for swapin block io completion */
 815        u32 blkio_count;        /* total count of the number of sync block */
 816                                /* io operations performed */
 817        u32 swapin_count;       /* total count of the number of swapin block */
 818                                /* io operations performed */
 819
 820        struct timespec freepages_start, freepages_end;
 821        u64 freepages_delay;    /* wait for memory reclaim */
 822        u32 freepages_count;    /* total count of memory reclaim */
 823};
 824#endif  /* CONFIG_TASK_DELAY_ACCT */
 825
 826static inline int sched_info_on(void)
 827{
 828#ifdef CONFIG_SCHEDSTATS
 829        return 1;
 830#elif defined(CONFIG_TASK_DELAY_ACCT)
 831        extern int delayacct_on;
 832        return delayacct_on;
 833#else
 834        return 0;
 835#endif
 836}
 837
 838enum cpu_idle_type {
 839        CPU_IDLE,
 840        CPU_NOT_IDLE,
 841        CPU_NEWLY_IDLE,
 842        CPU_MAX_IDLE_TYPES
 843};
 844
 845/*
 846 * Increase resolution of cpu_power calculations
 847 */
 848#define SCHED_POWER_SHIFT       10
 849#define SCHED_POWER_SCALE       (1L << SCHED_POWER_SHIFT)
 850
 851/*
 852 * sched-domains (multiprocessor balancing) declarations:
 853 */
 854#ifdef CONFIG_SMP
 855#define SD_LOAD_BALANCE         0x0001  /* Do load balancing on this domain. */
 856#define SD_BALANCE_NEWIDLE      0x0002  /* Balance when about to become idle */
 857#define SD_BALANCE_EXEC         0x0004  /* Balance on exec */
 858#define SD_BALANCE_FORK         0x0008  /* Balance on fork, clone */
 859#define SD_BALANCE_WAKE         0x0010  /* Balance on wakeup */
 860#define SD_WAKE_AFFINE          0x0020  /* Wake task to waking CPU */
 861#define SD_SHARE_CPUPOWER       0x0080  /* Domain members share cpu power */
 862#define SD_SHARE_PKG_RESOURCES  0x0200  /* Domain members share cpu pkg resources */
 863#define SD_SERIALIZE            0x0400  /* Only a single load balancing instance */
 864#define SD_ASYM_PACKING         0x0800  /* Place busy groups earlier in the domain */
 865#define SD_PREFER_SIBLING       0x1000  /* Prefer to place tasks in a sibling domain */
 866#define SD_OVERLAP              0x2000  /* sched_domains of this level overlap */
 867#define SD_NUMA                 0x4000  /* cross-node balancing */
 868
 869extern int __weak arch_sd_sibiling_asym_packing(void);
 870
 871struct sched_domain_attr {
 872        int relax_domain_level;
 873};
 874
 875#define SD_ATTR_INIT    (struct sched_domain_attr) {    \
 876        .relax_domain_level = -1,                       \
 877}
 878
 879extern int sched_domain_level_max;
 880
 881struct sched_group;
 882
 883struct sched_domain {
 884        /* These fields must be setup */
 885        struct sched_domain *parent;    /* top domain must be null terminated */
 886        struct sched_domain *child;     /* bottom domain must be null terminated */
 887        struct sched_group *groups;     /* the balancing groups of the domain */
 888        unsigned long min_interval;     /* Minimum balance interval ms */
 889        unsigned long max_interval;     /* Maximum balance interval ms */
 890        unsigned int busy_factor;       /* less balancing by factor if busy */
 891        unsigned int imbalance_pct;     /* No balance until over watermark */
 892        unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
 893        unsigned int busy_idx;
 894        unsigned int idle_idx;
 895        unsigned int newidle_idx;
 896        unsigned int wake_idx;
 897        unsigned int forkexec_idx;
 898        unsigned int smt_gain;
 899
 900        int nohz_idle;                  /* NOHZ IDLE status */
 901        int flags;                      /* See SD_* */
 902        int level;
 903
 904        /* Runtime fields. */
 905        unsigned long last_balance;     /* init to jiffies. units in jiffies */
 906        unsigned int balance_interval;  /* initialise to 1. units in ms. */
 907        unsigned int nr_balance_failed; /* initialise to 0 */
 908
 909        /* idle_balance() stats */
 910        u64 max_newidle_lb_cost;
 911        unsigned long next_decay_max_lb_cost;
 912
 913#ifdef CONFIG_SCHEDSTATS
 914        /* load_balance() stats */
 915        unsigned int lb_count[CPU_MAX_IDLE_TYPES];
 916        unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
 917        unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
 918        unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
 919        unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
 920        unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
 921        unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
 922        unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
 923
 924        /* Active load balancing */
 925        unsigned int alb_count;
 926        unsigned int alb_failed;
 927        unsigned int alb_pushed;
 928
 929        /* SD_BALANCE_EXEC stats */
 930        unsigned int sbe_count;
 931        unsigned int sbe_balanced;
 932        unsigned int sbe_pushed;
 933
 934        /* SD_BALANCE_FORK stats */
 935        unsigned int sbf_count;
 936        unsigned int sbf_balanced;
 937        unsigned int sbf_pushed;
 938
 939        /* try_to_wake_up() stats */
 940        unsigned int ttwu_wake_remote;
 941        unsigned int ttwu_move_affine;
 942        unsigned int ttwu_move_balance;
 943#endif
 944#ifdef CONFIG_SCHED_DEBUG
 945        char *name;
 946#endif
 947        union {
 948                void *private;          /* used during construction */
 949                struct rcu_head rcu;    /* used during destruction */
 950        };
 951
 952        unsigned int span_weight;
 953        /*
 954         * Span of all CPUs in this domain.
 955         *
 956         * NOTE: this field is variable length. (Allocated dynamically
 957         * by attaching extra space to the end of the structure,
 958         * depending on how many CPUs the kernel has booted up with)
 959         */
 960        unsigned long span[0];
 961};
 962
 963static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
 964{
 965        return to_cpumask(sd->span);
 966}
 967
 968extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
 969                                    struct sched_domain_attr *dattr_new);
 970
 971/* Allocate an array of sched domains, for partition_sched_domains(). */
 972cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
 973void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
 974
 975bool cpus_share_cache(int this_cpu, int that_cpu);
 976
 977#else /* CONFIG_SMP */
 978
 979struct sched_domain_attr;
 980
 981static inline void
 982partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
 983                        struct sched_domain_attr *dattr_new)
 984{
 985}
 986
 987static inline bool cpus_share_cache(int this_cpu, int that_cpu)
 988{
 989        return true;
 990}
 991
 992#endif  /* !CONFIG_SMP */
 993
 994
 995struct io_context;                      /* See blkdev.h */
 996
 997
 998#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
 999extern void prefetch_stack(struct task_struct *t);
1000#else
1001static inline void prefetch_stack(struct task_struct *t) { }
1002#endif
1003
1004struct audit_context;           /* See audit.c */
1005struct mempolicy;
1006struct pipe_inode_info;
1007struct uts_namespace;
1008
1009struct load_weight {
1010        unsigned long weight;
1011        u32 inv_weight;
1012};
1013
1014struct sched_avg {
1015        /*
1016         * These sums represent an infinite geometric series and so are bound
1017         * above by 1024/(1-y).  Thus we only need a u32 to store them for all
1018         * choices of y < 1-2^(-32)*1024.
1019         */
1020        u32 runnable_avg_sum, runnable_avg_period;
1021        u64 last_runnable_update;
1022        s64 decay_count;
1023        unsigned long load_avg_contrib;
1024};
1025
1026#ifdef CONFIG_SCHEDSTATS
1027struct sched_statistics {
1028        u64                     wait_start;
1029        u64                     wait_max;
1030        u64                     wait_count;
1031        u64                     wait_sum;
1032        u64                     iowait_count;
1033        u64                     iowait_sum;
1034
1035        u64                     sleep_start;
1036        u64                     sleep_max;
1037        s64                     sum_sleep_runtime;
1038
1039        u64                     block_start;
1040        u64                     block_max;
1041        u64                     exec_max;
1042        u64                     slice_max;
1043
1044        u64                     nr_migrations_cold;
1045        u64                     nr_failed_migrations_affine;
1046        u64                     nr_failed_migrations_running;
1047        u64                     nr_failed_migrations_hot;
1048        u64                     nr_forced_migrations;
1049
1050        u64                     nr_wakeups;
1051        u64                     nr_wakeups_sync;
1052        u64                     nr_wakeups_migrate;
1053        u64                     nr_wakeups_local;
1054        u64                     nr_wakeups_remote;
1055        u64                     nr_wakeups_affine;
1056        u64                     nr_wakeups_affine_attempts;
1057        u64                     nr_wakeups_passive;
1058        u64                     nr_wakeups_idle;
1059};
1060#endif
1061
1062struct sched_entity {
1063        struct load_weight      load;           /* for load-balancing */
1064        struct rb_node          run_node;
1065        struct list_head        group_node;
1066        unsigned int            on_rq;
1067
1068        u64                     exec_start;
1069        u64                     sum_exec_runtime;
1070        u64                     vruntime;
1071        u64                     prev_sum_exec_runtime;
1072
1073        u64                     nr_migrations;
1074
1075#ifdef CONFIG_SCHEDSTATS
1076        struct sched_statistics statistics;
1077#endif
1078
1079#ifdef CONFIG_FAIR_GROUP_SCHED
1080        struct sched_entity     *parent;
1081        /* rq on which this entity is (to be) queued: */
1082        struct cfs_rq           *cfs_rq;
1083        /* rq "owned" by this entity/group: */
1084        struct cfs_rq           *my_q;
1085#endif
1086
1087#ifdef CONFIG_SMP
1088        /* Per-entity load-tracking */
1089        struct sched_avg        avg;
1090#endif
1091};
1092
1093struct sched_rt_entity {
1094        struct list_head run_list;
1095        unsigned long timeout;
1096        unsigned long watchdog_stamp;
1097        unsigned int time_slice;
1098
1099        struct sched_rt_entity *back;
1100#ifdef CONFIG_RT_GROUP_SCHED
1101        struct sched_rt_entity  *parent;
1102        /* rq on which this entity is (to be) queued: */
1103        struct rt_rq            *rt_rq;
1104        /* rq "owned" by this entity/group: */
1105        struct rt_rq            *my_q;
1106#endif
1107};
1108
1109struct sched_dl_entity {
1110        struct rb_node  rb_node;
1111
1112        /*
1113         * Original scheduling parameters. Copied here from sched_attr
1114         * during sched_setscheduler2(), they will remain the same until
1115         * the next sched_setscheduler2().
1116         */
1117        u64 dl_runtime;         /* maximum runtime for each instance    */
1118        u64 dl_deadline;        /* relative deadline of each instance   */
1119        u64 dl_period;          /* separation of two instances (period) */
1120        u64 dl_bw;              /* dl_runtime / dl_deadline             */
1121
1122        /*
1123         * Actual scheduling parameters. Initialized with the values above,
1124         * they are continously updated during task execution. Note that
1125         * the remaining runtime could be < 0 in case we are in overrun.
1126         */
1127        s64 runtime;            /* remaining runtime for this instance  */
1128        u64 deadline;           /* absolute deadline for this instance  */
1129        unsigned int flags;     /* specifying the scheduler behaviour   */
1130
1131        /*
1132         * Some bool flags:
1133         *
1134         * @dl_throttled tells if we exhausted the runtime. If so, the
1135         * task has to wait for a replenishment to be performed at the
1136         * next firing of dl_timer.
1137         *
1138         * @dl_new tells if a new instance arrived. If so we must
1139         * start executing it with full runtime and reset its absolute
1140         * deadline;
1141         *
1142         * @dl_boosted tells if we are boosted due to DI. If so we are
1143         * outside bandwidth enforcement mechanism (but only until we
1144         * exit the critical section).
1145         */
1146        int dl_throttled, dl_new, dl_boosted;
1147
1148        /*
1149         * Bandwidth enforcement timer. Each -deadline task has its
1150         * own bandwidth to be enforced, thus we need one timer per task.
1151         */
1152        struct hrtimer dl_timer;
1153};
1154
1155struct rcu_node;
1156
1157enum perf_event_task_context {
1158        perf_invalid_context = -1,
1159        perf_hw_context = 0,
1160        perf_sw_context,
1161        perf_nr_task_contexts,
1162};
1163
1164struct task_struct {
1165        volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1166        void *stack;
1167        atomic_t usage;
1168        unsigned int flags;     /* per process flags, defined below */
1169        unsigned int ptrace;
1170
1171#ifdef CONFIG_SMP
1172        struct llist_node wake_entry;
1173        int on_cpu;
1174        struct task_struct *last_wakee;
1175        unsigned long wakee_flips;
1176        unsigned long wakee_flip_decay_ts;
1177
1178        int wake_cpu;
1179#endif
1180        int on_rq;
1181
1182        int prio, static_prio, normal_prio;
1183        unsigned int rt_priority;
1184        const struct sched_class *sched_class;
1185        struct sched_entity se;
1186        struct sched_rt_entity rt;
1187#ifdef CONFIG_CGROUP_SCHED
1188        struct task_group *sched_task_group;
1189#endif
1190        struct sched_dl_entity dl;
1191
1192#ifdef CONFIG_PREEMPT_NOTIFIERS
1193        /* list of struct preempt_notifier: */
1194        struct hlist_head preempt_notifiers;
1195#endif
1196
1197#ifdef CONFIG_BLK_DEV_IO_TRACE
1198        unsigned int btrace_seq;
1199#endif
1200
1201        unsigned int policy;
1202        int nr_cpus_allowed;
1203        cpumask_t cpus_allowed;
1204
1205#ifdef CONFIG_PREEMPT_RCU
1206        int rcu_read_lock_nesting;
1207        char rcu_read_unlock_special;
1208        struct list_head rcu_node_entry;
1209#endif /* #ifdef CONFIG_PREEMPT_RCU */
1210#ifdef CONFIG_TREE_PREEMPT_RCU
1211        struct rcu_node *rcu_blocked_node;
1212#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1213#ifdef CONFIG_RCU_BOOST
1214        struct rt_mutex *rcu_boost_mutex;
1215#endif /* #ifdef CONFIG_RCU_BOOST */
1216
1217#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1218        struct sched_info sched_info;
1219#endif
1220
1221        struct list_head tasks;
1222#ifdef CONFIG_SMP
1223        struct plist_node pushable_tasks;
1224        struct rb_node pushable_dl_tasks;
1225#endif
1226
1227        struct mm_struct *mm, *active_mm;
1228#ifdef CONFIG_COMPAT_BRK
1229        unsigned brk_randomized:1;
1230#endif
1231#if defined(SPLIT_RSS_COUNTING)
1232        struct task_rss_stat    rss_stat;
1233#endif
1234/* task state */
1235        int exit_state;
1236        int exit_code, exit_signal;
1237        int pdeath_signal;  /*  The signal sent when the parent dies  */
1238        unsigned int jobctl;    /* JOBCTL_*, siglock protected */
1239
1240        /* Used for emulating ABI behavior of previous Linux versions */
1241        unsigned int personality;
1242
1243        unsigned in_execve:1;   /* Tell the LSMs that the process is doing an
1244                                 * execve */
1245        unsigned in_iowait:1;
1246
1247        /* task may not gain privileges */
1248        unsigned no_new_privs:1;
1249
1250        /* Revert to default priority/policy when forking */
1251        unsigned sched_reset_on_fork:1;
1252        unsigned sched_contributes_to_load:1;
1253
1254        pid_t pid;
1255        pid_t tgid;
1256
1257#ifdef CONFIG_CC_STACKPROTECTOR
1258        /* Canary value for the -fstack-protector gcc feature */
1259        unsigned long stack_canary;
1260#endif
1261        /*
1262         * pointers to (original) parent process, youngest child, younger sibling,
1263         * older sibling, respectively.  (p->father can be replaced with
1264         * p->real_parent->pid)
1265         */
1266        struct task_struct __rcu *real_parent; /* real parent process */
1267        struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1268        /*
1269         * children/sibling forms the list of my natural children
1270         */
1271        struct list_head children;      /* list of my children */
1272        struct list_head sibling;       /* linkage in my parent's children list */
1273        struct task_struct *group_leader;       /* threadgroup leader */
1274
1275        /*
1276         * ptraced is the list of tasks this task is using ptrace on.
1277         * This includes both natural children and PTRACE_ATTACH targets.
1278         * p->ptrace_entry is p's link on the p->parent->ptraced list.
1279         */
1280        struct list_head ptraced;
1281        struct list_head ptrace_entry;
1282
1283        /* PID/PID hash table linkage. */
1284        struct pid_link pids[PIDTYPE_MAX];
1285        struct list_head thread_group;
1286        struct list_head thread_node;
1287
1288        struct completion *vfork_done;          /* for vfork() */
1289        int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1290        int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1291
1292        cputime_t utime, stime, utimescaled, stimescaled;
1293        cputime_t gtime;
1294#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1295        struct cputime prev_cputime;
1296#endif
1297#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1298        seqlock_t vtime_seqlock;
1299        unsigned long long vtime_snap;
1300        enum {
1301                VTIME_SLEEPING = 0,
1302                VTIME_USER,
1303                VTIME_SYS,
1304        } vtime_snap_whence;
1305#endif
1306        unsigned long nvcsw, nivcsw; /* context switch counts */
1307        struct timespec start_time;             /* monotonic time */
1308        struct timespec real_start_time;        /* boot based time */
1309/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1310        unsigned long min_flt, maj_flt;
1311
1312        struct task_cputime cputime_expires;
1313        struct list_head cpu_timers[3];
1314
1315/* process credentials */
1316        const struct cred __rcu *real_cred; /* objective and real subjective task
1317                                         * credentials (COW) */
1318        const struct cred __rcu *cred;  /* effective (overridable) subjective task
1319                                         * credentials (COW) */
1320        char comm[TASK_COMM_LEN]; /* executable name excluding path
1321                                     - access with [gs]et_task_comm (which lock
1322                                       it with task_lock())
1323                                     - initialized normally by setup_new_exec */
1324/* file system info */
1325        int link_count, total_link_count;
1326#ifdef CONFIG_SYSVIPC
1327/* ipc stuff */
1328        struct sysv_sem sysvsem;
1329#endif
1330#ifdef CONFIG_DETECT_HUNG_TASK
1331/* hung task detection */
1332        unsigned long last_switch_count;
1333#endif
1334/* CPU-specific state of this task */
1335        struct thread_struct thread;
1336/* filesystem information */
1337        struct fs_struct *fs;
1338/* open file information */
1339        struct files_struct *files;
1340/* namespaces */
1341        struct nsproxy *nsproxy;
1342/* signal handlers */
1343        struct signal_struct *signal;
1344        struct sighand_struct *sighand;
1345
1346        sigset_t blocked, real_blocked;
1347        sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1348        struct sigpending pending;
1349
1350        unsigned long sas_ss_sp;
1351        size_t sas_ss_size;
1352        int (*notifier)(void *priv);
1353        void *notifier_data;
1354        sigset_t *notifier_mask;
1355        struct callback_head *task_works;
1356
1357        struct audit_context *audit_context;
1358#ifdef CONFIG_AUDITSYSCALL
1359        kuid_t loginuid;
1360        unsigned int sessionid;
1361#endif
1362        struct seccomp seccomp;
1363
1364/* Thread group tracking */
1365        u32 parent_exec_id;
1366        u32 self_exec_id;
1367/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1368 * mempolicy */
1369        spinlock_t alloc_lock;
1370
1371        /* Protection of the PI data structures: */
1372        raw_spinlock_t pi_lock;
1373
1374#ifdef CONFIG_RT_MUTEXES
1375        /* PI waiters blocked on a rt_mutex held by this task */
1376        struct rb_root pi_waiters;
1377        struct rb_node *pi_waiters_leftmost;
1378        /* Deadlock detection and priority inheritance handling */
1379        struct rt_mutex_waiter *pi_blocked_on;
1380        /* Top pi_waiters task */
1381        struct task_struct *pi_top_task;
1382#endif
1383
1384#ifdef CONFIG_DEBUG_MUTEXES
1385        /* mutex deadlock detection */
1386        struct mutex_waiter *blocked_on;
1387#endif
1388#ifdef CONFIG_TRACE_IRQFLAGS
1389        unsigned int irq_events;
1390        unsigned long hardirq_enable_ip;
1391        unsigned long hardirq_disable_ip;
1392        unsigned int hardirq_enable_event;
1393        unsigned int hardirq_disable_event;
1394        int hardirqs_enabled;
1395        int hardirq_context;
1396        unsigned long softirq_disable_ip;
1397        unsigned long softirq_enable_ip;
1398        unsigned int softirq_disable_event;
1399        unsigned int softirq_enable_event;
1400        int softirqs_enabled;
1401        int softirq_context;
1402#endif
1403#ifdef CONFIG_LOCKDEP
1404# define MAX_LOCK_DEPTH 48UL
1405        u64 curr_chain_key;
1406        int lockdep_depth;
1407        unsigned int lockdep_recursion;
1408        struct held_lock held_locks[MAX_LOCK_DEPTH];
1409        gfp_t lockdep_reclaim_gfp;
1410#endif
1411
1412/* journalling filesystem info */
1413        void *journal_info;
1414
1415/* stacked block device info */
1416        struct bio_list *bio_list;
1417
1418#ifdef CONFIG_BLOCK
1419/* stack plugging */
1420        struct blk_plug *plug;
1421#endif
1422
1423/* VM state */
1424        struct reclaim_state *reclaim_state;
1425
1426        struct backing_dev_info *backing_dev_info;
1427
1428        struct io_context *io_context;
1429
1430        unsigned long ptrace_message;
1431        siginfo_t *last_siginfo; /* For ptrace use.  */
1432        struct task_io_accounting ioac;
1433#if defined(CONFIG_TASK_XACCT)
1434        u64 acct_rss_mem1;      /* accumulated rss usage */
1435        u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1436        cputime_t acct_timexpd; /* stime + utime since last update */
1437#endif
1438#ifdef CONFIG_CPUSETS
1439        nodemask_t mems_allowed;        /* Protected by alloc_lock */
1440        seqcount_t mems_allowed_seq;    /* Seqence no to catch updates */
1441        int cpuset_mem_spread_rotor;
1442        int cpuset_slab_spread_rotor;
1443#endif
1444#ifdef CONFIG_CGROUPS
1445        /* Control Group info protected by css_set_lock */
1446        struct css_set __rcu *cgroups;
1447        /* cg_list protected by css_set_lock and tsk->alloc_lock */
1448        struct list_head cg_list;
1449#endif
1450#ifdef CONFIG_FUTEX
1451        struct robust_list_head __user *robust_list;
1452#ifdef CONFIG_COMPAT
1453        struct compat_robust_list_head __user *compat_robust_list;
1454#endif
1455        struct list_head pi_state_list;
1456        struct futex_pi_state *pi_state_cache;
1457#endif
1458#ifdef CONFIG_PERF_EVENTS
1459        struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1460        struct mutex perf_event_mutex;
1461        struct list_head perf_event_list;
1462#endif
1463#ifdef CONFIG_NUMA
1464        struct mempolicy *mempolicy;    /* Protected by alloc_lock */
1465        short il_next;
1466        short pref_node_fork;
1467#endif
1468#ifdef CONFIG_NUMA_BALANCING
1469        int numa_scan_seq;
1470        unsigned int numa_scan_period;
1471        unsigned int numa_scan_period_max;
1472        int numa_preferred_nid;
1473        int numa_migrate_deferred;
1474        unsigned long numa_migrate_retry;
1475        u64 node_stamp;                 /* migration stamp  */
1476        struct callback_head numa_work;
1477
1478        struct list_head numa_entry;
1479        struct numa_group *numa_group;
1480
1481        /*
1482         * Exponential decaying average of faults on a per-node basis.
1483         * Scheduling placement decisions are made based on the these counts.
1484         * The values remain static for the duration of a PTE scan
1485         */
1486        unsigned long *numa_faults;
1487        unsigned long total_numa_faults;
1488
1489        /*
1490         * numa_faults_buffer records faults per node during the current
1491         * scan window. When the scan completes, the counts in numa_faults
1492         * decay and these values are copied.
1493         */
1494        unsigned long *numa_faults_buffer;
1495
1496        /*
1497         * numa_faults_locality tracks if faults recorded during the last
1498         * scan window were remote/local. The task scan period is adapted
1499         * based on the locality of the faults with different weights
1500         * depending on whether they were shared or private faults
1501         */
1502        unsigned long numa_faults_locality[2];
1503
1504        unsigned long numa_pages_migrated;
1505#endif /* CONFIG_NUMA_BALANCING */
1506
1507        struct rcu_head rcu;
1508
1509        /*
1510         * cache last used pipe for splice
1511         */
1512        struct pipe_inode_info *splice_pipe;
1513
1514        struct page_frag task_frag;
1515
1516#ifdef  CONFIG_TASK_DELAY_ACCT
1517        struct task_delay_info *delays;
1518#endif
1519#ifdef CONFIG_FAULT_INJECTION
1520        int make_it_fail;
1521#endif
1522        /*
1523         * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1524         * balance_dirty_pages() for some dirty throttling pause
1525         */
1526        int nr_dirtied;
1527        int nr_dirtied_pause;
1528        unsigned long dirty_paused_when; /* start of a write-and-pause period */
1529
1530#ifdef CONFIG_LATENCYTOP
1531        int latency_record_count;
1532        struct latency_record latency_record[LT_SAVECOUNT];
1533#endif
1534        /*
1535         * time slack values; these are used to round up poll() and
1536         * select() etc timeout values. These are in nanoseconds.
1537         */
1538        unsigned long timer_slack_ns;
1539        unsigned long default_timer_slack_ns;
1540
1541#ifdef CONFIG_FUNCTION_GRAPH_TRACER
1542        /* Index of current stored address in ret_stack */
1543        int curr_ret_stack;
1544        /* Stack of return addresses for return function tracing */
1545        struct ftrace_ret_stack *ret_stack;
1546        /* time stamp for last schedule */
1547        unsigned long long ftrace_timestamp;
1548        /*
1549         * Number of functions that haven't been traced
1550         * because of depth overrun.
1551         */
1552        atomic_t trace_overrun;
1553        /* Pause for the tracing */
1554        atomic_t tracing_graph_pause;
1555#endif
1556#ifdef CONFIG_TRACING
1557        /* state flags for use by tracers */
1558        unsigned long trace;
1559        /* bitmask and counter of trace recursion */
1560        unsigned long trace_recursion;
1561#endif /* CONFIG_TRACING */
1562#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1563        struct memcg_batch_info {
1564                int do_batch;   /* incremented when batch uncharge started */
1565                struct mem_cgroup *memcg; /* target memcg of uncharge */
1566                unsigned long nr_pages; /* uncharged usage */
1567                unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1568        } memcg_batch;
1569        unsigned int memcg_kmem_skip_account;
1570        struct memcg_oom_info {
1571                struct mem_cgroup *memcg;
1572                gfp_t gfp_mask;
1573                int order;
1574                unsigned int may_oom:1;
1575        } memcg_oom;
1576#endif
1577#ifdef CONFIG_UPROBES
1578        struct uprobe_task *utask;
1579#endif
1580#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1581        unsigned int    sequential_io;
1582        unsigned int    sequential_io_avg;
1583#endif
1584};
1585
1586/* Future-safe accessor for struct task_struct's cpus_allowed. */
1587#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1588
1589#define TNF_MIGRATED    0x01
1590#define TNF_NO_GROUP    0x02
1591#define TNF_SHARED      0x04
1592#define TNF_FAULT_LOCAL 0x08
1593
1594#ifdef CONFIG_NUMA_BALANCING
1595extern void task_numa_fault(int last_node, int node, int pages, int flags);
1596extern pid_t task_numa_group_id(struct task_struct *p);
1597extern void set_numabalancing_state(bool enabled);
1598extern void task_numa_free(struct task_struct *p);
1599
1600extern unsigned int sysctl_numa_balancing_migrate_deferred;
1601#else
1602static inline void task_numa_fault(int last_node, int node, int pages,
1603                                   int flags)
1604{
1605}
1606static inline pid_t task_numa_group_id(struct task_struct *p)
1607{
1608        return 0;
1609}
1610static inline void set_numabalancing_state(bool enabled)
1611{
1612}
1613static inline void task_numa_free(struct task_struct *p)
1614{
1615}
1616#endif
1617
1618static inline struct pid *task_pid(struct task_struct *task)
1619{
1620        return task->pids[PIDTYPE_PID].pid;
1621}
1622
1623static inline struct pid *task_tgid(struct task_struct *task)
1624{
1625        return task->group_leader->pids[PIDTYPE_PID].pid;
1626}
1627
1628/*
1629 * Without tasklist or rcu lock it is not safe to dereference
1630 * the result of task_pgrp/task_session even if task == current,
1631 * we can race with another thread doing sys_setsid/sys_setpgid.
1632 */
1633static inline struct pid *task_pgrp(struct task_struct *task)
1634{
1635        return task->group_leader->pids[PIDTYPE_PGID].pid;
1636}
1637
1638static inline struct pid *task_session(struct task_struct *task)
1639{
1640        return task->group_leader->pids[PIDTYPE_SID].pid;
1641}
1642
1643struct pid_namespace;
1644
1645/*
1646 * the helpers to get the task's different pids as they are seen
1647 * from various namespaces
1648 *
1649 * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1650 * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1651 *                     current.
1652 * task_xid_nr_ns()  : id seen from the ns specified;
1653 *
1654 * set_task_vxid()   : assigns a virtual id to a task;
1655 *
1656 * see also pid_nr() etc in include/linux/pid.h
1657 */
1658pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1659                        struct pid_namespace *ns);
1660
1661static inline pid_t task_pid_nr(struct task_struct *tsk)
1662{
1663        return tsk->pid;
1664}
1665
1666static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1667                                        struct pid_namespace *ns)
1668{
1669        return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1670}
1671
1672static inline pid_t task_pid_vnr(struct task_struct *tsk)
1673{
1674        return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1675}
1676
1677
1678static inline pid_t task_tgid_nr(struct task_struct *tsk)
1679{
1680        return tsk->tgid;
1681}
1682
1683pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1684
1685static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1686{
1687        return pid_vnr(task_tgid(tsk));
1688}
1689
1690
1691static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1692                                        struct pid_namespace *ns)
1693{
1694        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1695}
1696
1697static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1698{
1699        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1700}
1701
1702
1703static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1704                                        struct pid_namespace *ns)
1705{
1706        return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1707}
1708
1709static inline pid_t task_session_vnr(struct task_struct *tsk)
1710{
1711        return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1712}
1713
1714/* obsolete, do not use */
1715static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1716{
1717        return task_pgrp_nr_ns(tsk, &init_pid_ns);
1718}
1719
1720/**
1721 * pid_alive - check that a task structure is not stale
1722 * @p: Task structure to be checked.
1723 *
1724 * Test if a process is not yet dead (at most zombie state)
1725 * If pid_alive fails, then pointers within the task structure
1726 * can be stale and must not be dereferenced.
1727 *
1728 * Return: 1 if the process is alive. 0 otherwise.
1729 */
1730static inline int pid_alive(struct task_struct *p)
1731{
1732        return p->pids[PIDTYPE_PID].pid != NULL;
1733}
1734
1735/**
1736 * is_global_init - check if a task structure is init
1737 * @tsk: Task structure to be checked.
1738 *
1739 * Check if a task structure is the first user space task the kernel created.
1740 *
1741 * Return: 1 if the task structure is init. 0 otherwise.
1742 */
1743static inline int is_global_init(struct task_struct *tsk)
1744{
1745        return tsk->pid == 1;
1746}
1747
1748extern struct pid *cad_pid;
1749
1750extern void free_task(struct task_struct *tsk);
1751#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1752
1753extern void __put_task_struct(struct task_struct *t);
1754
1755static inline void put_task_struct(struct task_struct *t)
1756{
1757        if (atomic_dec_and_test(&t->usage))
1758                __put_task_struct(t);
1759}
1760
1761#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1762extern void task_cputime(struct task_struct *t,
1763                         cputime_t *utime, cputime_t *stime);
1764extern void task_cputime_scaled(struct task_struct *t,
1765                                cputime_t *utimescaled, cputime_t *stimescaled);
1766extern cputime_t task_gtime(struct task_struct *t);
1767#else
1768static inline void task_cputime(struct task_struct *t,
1769                                cputime_t *utime, cputime_t *stime)
1770{
1771        if (utime)
1772                *utime = t->utime;
1773        if (stime)
1774                *stime = t->stime;
1775}
1776
1777static inline void task_cputime_scaled(struct task_struct *t,
1778                                       cputime_t *utimescaled,
1779                                       cputime_t *stimescaled)
1780{
1781        if (utimescaled)
1782                *utimescaled = t->utimescaled;
1783        if (stimescaled)
1784                *stimescaled = t->stimescaled;
1785}
1786
1787static inline cputime_t task_gtime(struct task_struct *t)
1788{
1789        return t->gtime;
1790}
1791#endif
1792extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1793extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1794
1795/*
1796 * Per process flags
1797 */
1798#define PF_EXITING      0x00000004      /* getting shut down */
1799#define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
1800#define PF_VCPU         0x00000010      /* I'm a virtual CPU */
1801#define PF_WQ_WORKER    0x00000020      /* I'm a workqueue worker */
1802#define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
1803#define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1804#define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
1805#define PF_DUMPCORE     0x00000200      /* dumped core */
1806#define PF_SIGNALED     0x00000400      /* killed by a signal */
1807#define PF_MEMALLOC     0x00000800      /* Allocating memory */
1808#define PF_NPROC_EXCEEDED 0x00001000    /* set_user noticed that RLIMIT_NPROC was exceeded */
1809#define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
1810#define PF_USED_ASYNC   0x00004000      /* used async_schedule*(), used by module init */
1811#define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
1812#define PF_FROZEN       0x00010000      /* frozen for system suspend */
1813#define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
1814#define PF_KSWAPD       0x00040000      /* I am kswapd */
1815#define PF_MEMALLOC_NOIO 0x00080000     /* Allocating memory without IO involved */
1816#define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
1817#define PF_KTHREAD      0x00200000      /* I am a kernel thread */
1818#define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
1819#define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
1820#define PF_SPREAD_PAGE  0x01000000      /* Spread page cache over cpuset */
1821#define PF_SPREAD_SLAB  0x02000000      /* Spread some slab caches over cpuset */
1822#define PF_NO_SETAFFINITY 0x04000000    /* Userland is not allowed to meddle with cpus_allowed */
1823#define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1824#define PF_MEMPOLICY    0x10000000      /* Non-default NUMA mempolicy */
1825#define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
1826#define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezable */
1827#define PF_SUSPEND_TASK 0x80000000      /* this thread called freeze_processes and should not be frozen */
1828
1829/*
1830 * Only the _current_ task can read/write to tsk->flags, but other
1831 * tasks can access tsk->flags in readonly mode for example
1832 * with tsk_used_math (like during threaded core dumping).
1833 * There is however an exception to this rule during ptrace
1834 * or during fork: the ptracer task is allowed to write to the
1835 * child->flags of its traced child (same goes for fork, the parent
1836 * can write to the child->flags), because we're guaranteed the
1837 * child is not running and in turn not changing child->flags
1838 * at the same time the parent does it.
1839 */
1840#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1841#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1842#define clear_used_math() clear_stopped_child_used_math(current)
1843#define set_used_math() set_stopped_child_used_math(current)
1844#define conditional_stopped_child_used_math(condition, child) \
1845        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1846#define conditional_used_math(condition) \
1847        conditional_stopped_child_used_math(condition, current)
1848#define copy_to_stopped_child_used_math(child) \
1849        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1850/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1851#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1852#define used_math() tsk_used_math(current)
1853
1854/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1855static inline gfp_t memalloc_noio_flags(gfp_t flags)
1856{
1857        if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1858                flags &= ~__GFP_IO;
1859        return flags;
1860}
1861
1862static inline unsigned int memalloc_noio_save(void)
1863{
1864        unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1865        current->flags |= PF_MEMALLOC_NOIO;
1866        return flags;
1867}
1868
1869static inline void memalloc_noio_restore(unsigned int flags)
1870{
1871        current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1872}
1873
1874/*
1875 * task->jobctl flags
1876 */
1877#define JOBCTL_STOP_SIGMASK     0xffff  /* signr of the last group stop */
1878
1879#define JOBCTL_STOP_DEQUEUED_BIT 16     /* stop signal dequeued */
1880#define JOBCTL_STOP_PENDING_BIT 17      /* task should stop for group stop */
1881#define JOBCTL_STOP_CONSUME_BIT 18      /* consume group stop count */
1882#define JOBCTL_TRAP_STOP_BIT    19      /* trap for STOP */
1883#define JOBCTL_TRAP_NOTIFY_BIT  20      /* trap for NOTIFY */
1884#define JOBCTL_TRAPPING_BIT     21      /* switching to TRACED */
1885#define JOBCTL_LISTENING_BIT    22      /* ptracer is listening for events */
1886
1887#define JOBCTL_STOP_DEQUEUED    (1 << JOBCTL_STOP_DEQUEUED_BIT)
1888#define JOBCTL_STOP_PENDING     (1 << JOBCTL_STOP_PENDING_BIT)
1889#define JOBCTL_STOP_CONSUME     (1 << JOBCTL_STOP_CONSUME_BIT)
1890#define JOBCTL_TRAP_STOP        (1 << JOBCTL_TRAP_STOP_BIT)
1891#define JOBCTL_TRAP_NOTIFY      (1 << JOBCTL_TRAP_NOTIFY_BIT)
1892#define JOBCTL_TRAPPING         (1 << JOBCTL_TRAPPING_BIT)
1893#define JOBCTL_LISTENING        (1 << JOBCTL_LISTENING_BIT)
1894
1895#define JOBCTL_TRAP_MASK        (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1896#define JOBCTL_PENDING_MASK     (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1897
1898extern bool task_set_jobctl_pending(struct task_struct *task,
1899                                    unsigned int mask);
1900extern void task_clear_jobctl_trapping(struct task_struct *task);
1901extern void task_clear_jobctl_pending(struct task_struct *task,
1902                                      unsigned int mask);
1903
1904#ifdef CONFIG_PREEMPT_RCU
1905
1906#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1907#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1908
1909static inline void rcu_copy_process(struct task_struct *p)
1910{
1911        p->rcu_read_lock_nesting = 0;
1912        p->rcu_read_unlock_special = 0;
1913#ifdef CONFIG_TREE_PREEMPT_RCU
1914        p->rcu_blocked_node = NULL;
1915#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1916#ifdef CONFIG_RCU_BOOST
1917        p->rcu_boost_mutex = NULL;
1918#endif /* #ifdef CONFIG_RCU_BOOST */
1919        INIT_LIST_HEAD(&p->rcu_node_entry);
1920}
1921
1922#else
1923
1924static inline void rcu_copy_process(struct task_struct *p)
1925{
1926}
1927
1928#endif
1929
1930static inline void tsk_restore_flags(struct task_struct *task,
1931                                unsigned long orig_flags, unsigned long flags)
1932{
1933        task->flags &= ~flags;
1934        task->flags |= orig_flags & flags;
1935}
1936
1937#ifdef CONFIG_SMP
1938extern void do_set_cpus_allowed(struct task_struct *p,
1939                               const struct cpumask *new_mask);
1940
1941extern int set_cpus_allowed_ptr(struct task_struct *p,
1942                                const struct cpumask *new_mask);
1943#else
1944static inline void do_set_cpus_allowed(struct task_struct *p,
1945                                      const struct cpumask *new_mask)
1946{
1947}
1948static inline int set_cpus_allowed_ptr(struct task_struct *p,
1949                                       const struct cpumask *new_mask)
1950{
1951        if (!cpumask_test_cpu(0, new_mask))
1952                return -EINVAL;
1953        return 0;
1954}
1955#endif
1956
1957#ifdef CONFIG_NO_HZ_COMMON
1958void calc_load_enter_idle(void);
1959void calc_load_exit_idle(void);
1960#else
1961static inline void calc_load_enter_idle(void) { }
1962static inline void calc_load_exit_idle(void) { }
1963#endif /* CONFIG_NO_HZ_COMMON */
1964
1965#ifndef CONFIG_CPUMASK_OFFSTACK
1966static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1967{
1968        return set_cpus_allowed_ptr(p, &new_mask);
1969}
1970#endif
1971
1972/*
1973 * Do not use outside of architecture code which knows its limitations.
1974 *
1975 * sched_clock() has no promise of monotonicity or bounded drift between
1976 * CPUs, use (which you should not) requires disabling IRQs.
1977 *
1978 * Please use one of the three interfaces below.
1979 */
1980extern unsigned long long notrace sched_clock(void);
1981/*
1982 * See the comment in kernel/sched/clock.c
1983 */
1984extern u64 cpu_clock(int cpu);
1985extern u64 local_clock(void);
1986extern u64 sched_clock_cpu(int cpu);
1987
1988
1989extern void sched_clock_init(void);
1990
1991#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1992static inline void sched_clock_tick(void)
1993{
1994}
1995
1996static inline void sched_clock_idle_sleep_event(void)
1997{
1998}
1999
2000static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2001{
2002}
2003#else
2004/*
2005 * Architectures can set this to 1 if they have specified
2006 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2007 * but then during bootup it turns out that sched_clock()
2008 * is reliable after all:
2009 */
2010extern int sched_clock_stable(void);
2011extern void set_sched_clock_stable(void);
2012extern void clear_sched_clock_stable(void);
2013
2014extern void sched_clock_tick(void);
2015extern void sched_clock_idle_sleep_event(void);
2016extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2017#endif
2018
2019#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2020/*
2021 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2022 * The reason for this explicit opt-in is not to have perf penalty with
2023 * slow sched_clocks.
2024 */
2025extern void enable_sched_clock_irqtime(void);
2026extern void disable_sched_clock_irqtime(void);
2027#else
2028static inline void enable_sched_clock_irqtime(void) {}
2029static inline void disable_sched_clock_irqtime(void) {}
2030#endif
2031
2032extern unsigned long long
2033task_sched_runtime(struct task_struct *task);
2034
2035/* sched_exec is called by processes performing an exec */
2036#ifdef CONFIG_SMP
2037extern void sched_exec(void);
2038#else
2039#define sched_exec()   {}
2040#endif
2041
2042extern void sched_clock_idle_sleep_event(void);
2043extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2044
2045#ifdef CONFIG_HOTPLUG_CPU
2046extern void idle_task_exit(void);
2047#else
2048static inline void idle_task_exit(void) {}
2049#endif
2050
2051#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2052extern void wake_up_nohz_cpu(int cpu);
2053#else
2054static inline void wake_up_nohz_cpu(int cpu) { }
2055#endif
2056
2057#ifdef CONFIG_NO_HZ_FULL
2058extern bool sched_can_stop_tick(void);
2059extern u64 scheduler_tick_max_deferment(void);
2060#else
2061static inline bool sched_can_stop_tick(void) { return false; }
2062#endif
2063
2064#ifdef CONFIG_SCHED_AUTOGROUP
2065extern void sched_autogroup_create_attach(struct task_struct *p);
2066extern void sched_autogroup_detach(struct task_struct *p);
2067extern void sched_autogroup_fork(struct signal_struct *sig);
2068extern void sched_autogroup_exit(struct signal_struct *sig);
2069#ifdef CONFIG_PROC_FS
2070extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2071extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2072#endif
2073#else
2074static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2075static inline void sched_autogroup_detach(struct task_struct *p) { }
2076static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2077static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2078#endif
2079
2080extern bool yield_to(struct task_struct *p, bool preempt);
2081extern void set_user_nice(struct task_struct *p, long nice);
2082extern int task_prio(const struct task_struct *p);
2083extern int task_nice(const struct task_struct *p);
2084extern int can_nice(const struct task_struct *p, const int nice);
2085extern int task_curr(const struct task_struct *p);
2086extern int idle_cpu(int cpu);
2087extern int sched_setscheduler(struct task_struct *, int,
2088                              const struct sched_param *);
2089extern int sched_setscheduler_nocheck(struct task_struct *, int,
2090                                      const struct sched_param *);
2091extern int sched_setattr(struct task_struct *,
2092                         const struct sched_attr *);
2093extern struct task_struct *idle_task(int cpu);
2094/**
2095 * is_idle_task - is the specified task an idle task?
2096 * @p: the task in question.
2097 *
2098 * Return: 1 if @p is an idle task. 0 otherwise.
2099 */
2100static inline bool is_idle_task(const struct task_struct *p)
2101{
2102        return p->pid == 0;
2103}
2104extern struct task_struct *curr_task(int cpu);
2105extern void set_curr_task(int cpu, struct task_struct *p);
2106
2107void yield(void);
2108
2109/*
2110 * The default (Linux) execution domain.
2111 */
2112extern struct exec_domain       default_exec_domain;
2113
2114union thread_union {
2115        struct thread_info thread_info;
2116        unsigned long stack[THREAD_SIZE/sizeof(long)];
2117};
2118
2119#ifndef __HAVE_ARCH_KSTACK_END
2120static inline int kstack_end(void *addr)
2121{
2122        /* Reliable end of stack detection:
2123         * Some APM bios versions misalign the stack
2124         */
2125        return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2126}
2127#endif
2128
2129extern union thread_union init_thread_union;
2130extern struct task_struct init_task;
2131
2132extern struct   mm_struct init_mm;
2133
2134extern struct pid_namespace init_pid_ns;
2135
2136/*
2137 * find a task by one of its numerical ids
2138 *
2139 * find_task_by_pid_ns():
2140 *      finds a task by its pid in the specified namespace
2141 * find_task_by_vpid():
2142 *      finds a task by its virtual pid
2143 *
2144 * see also find_vpid() etc in include/linux/pid.h
2145 */
2146
2147extern struct task_struct *find_task_by_vpid(pid_t nr);
2148extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2149                struct pid_namespace *ns);
2150
2151/* per-UID process charging. */
2152extern struct user_struct * alloc_uid(kuid_t);
2153static inline struct user_struct *get_uid(struct user_struct *u)
2154{
2155        atomic_inc(&u->__count);
2156        return u;
2157}
2158extern void free_uid(struct user_struct *);
2159
2160#include <asm/current.h>
2161
2162extern void xtime_update(unsigned long ticks);
2163
2164extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2165extern int wake_up_process(struct task_struct *tsk);
2166extern void wake_up_new_task(struct task_struct *tsk);
2167#ifdef CONFIG_SMP
2168 extern void kick_process(struct task_struct *tsk);
2169#else
2170 static inline void kick_process(struct task_struct *tsk) { }
2171#endif
2172extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2173extern void sched_dead(struct task_struct *p);
2174
2175extern void proc_caches_init(void);
2176extern void flush_signals(struct task_struct *);
2177extern void __flush_signals(struct task_struct *);
2178extern void ignore_signals(struct task_struct *);
2179extern void flush_signal_handlers(struct task_struct *, int force_default);
2180extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2181
2182static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2183{
2184        unsigned long flags;
2185        int ret;
2186
2187        spin_lock_irqsave(&tsk->sighand->siglock, flags);
2188        ret = dequeue_signal(tsk, mask, info);
2189        spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2190
2191        return ret;
2192}
2193
2194extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2195                              sigset_t *mask);
2196extern void unblock_all_signals(void);
2197extern void release_task(struct task_struct * p);
2198extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2199extern int force_sigsegv(int, struct task_struct *);
2200extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2201extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2202extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2203extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2204                                const struct cred *, u32);
2205extern int kill_pgrp(struct pid *pid, int sig, int priv);
2206extern int kill_pid(struct pid *pid, int sig, int priv);
2207extern int kill_proc_info(int, struct siginfo *, pid_t);
2208extern __must_check bool do_notify_parent(struct task_struct *, int);
2209extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2210extern void force_sig(int, struct task_struct *);
2211extern int send_sig(int, struct task_struct *, int);
2212extern int zap_other_threads(struct task_struct *p);
2213extern struct sigqueue *sigqueue_alloc(void);
2214extern void sigqueue_free(struct sigqueue *);
2215extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2216extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2217
2218static inline void restore_saved_sigmask(void)
2219{
2220        if (test_and_clear_restore_sigmask())
2221                __set_current_blocked(&current->saved_sigmask);
2222}
2223
2224static inline sigset_t *sigmask_to_save(void)
2225{
2226        sigset_t *res = &current->blocked;
2227        if (unlikely(test_restore_sigmask()))
2228                res = &current->saved_sigmask;
2229        return res;
2230}
2231
2232static inline int kill_cad_pid(int sig, int priv)
2233{
2234        return kill_pid(cad_pid, sig, priv);
2235}
2236
2237/* These can be the second arg to send_sig_info/send_group_sig_info.  */
2238#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2239#define SEND_SIG_PRIV   ((struct siginfo *) 1)
2240#define SEND_SIG_FORCED ((struct siginfo *) 2)
2241
2242/*
2243 * True if we are on the alternate signal stack.
2244 */
2245static inline int on_sig_stack(unsigned long sp)
2246{
2247#ifdef CONFIG_STACK_GROWSUP
2248        return sp >= current->sas_ss_sp &&
2249                sp - current->sas_ss_sp < current->sas_ss_size;
2250#else
2251        return sp > current->sas_ss_sp &&
2252                sp - current->sas_ss_sp <= current->sas_ss_size;
2253#endif
2254}
2255
2256static inline int sas_ss_flags(unsigned long sp)
2257{
2258        return (current->sas_ss_size == 0 ? SS_DISABLE
2259                : on_sig_stack(sp) ? SS_ONSTACK : 0);
2260}
2261
2262static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2263{
2264        if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2265#ifdef CONFIG_STACK_GROWSUP
2266                return current->sas_ss_sp;
2267#else
2268                return current->sas_ss_sp + current->sas_ss_size;
2269#endif
2270        return sp;
2271}
2272
2273/*
2274 * Routines for handling mm_structs
2275 */
2276extern struct mm_struct * mm_alloc(void);
2277
2278/* mmdrop drops the mm and the page tables */
2279extern void __mmdrop(struct mm_struct *);
2280static inline void mmdrop(struct mm_struct * mm)
2281{
2282        if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2283                __mmdrop(mm);
2284}
2285
2286/* mmput gets rid of the mappings and all user-space */
2287extern void mmput(struct mm_struct *);
2288/* Grab a reference to a task's mm, if it is not already going away */
2289extern struct mm_struct *get_task_mm(struct task_struct *task);
2290/*
2291 * Grab a reference to a task's mm, if it is not already going away
2292 * and ptrace_may_access with the mode parameter passed to it
2293 * succeeds.
2294 */
2295extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2296/* Remove the current tasks stale references to the old mm_struct */
2297extern void mm_release(struct task_struct *, struct mm_struct *);
2298
2299extern int copy_thread(unsigned long, unsigned long, unsigned long,
2300                        struct task_struct *);
2301extern void flush_thread(void);
2302extern void exit_thread(void);
2303
2304extern void exit_files(struct task_struct *);
2305extern void __cleanup_sighand(struct sighand_struct *);
2306
2307extern void exit_itimers(struct signal_struct *);
2308extern void flush_itimer_signals(void);
2309
2310extern void do_group_exit(int);
2311
2312extern int allow_signal(int);
2313extern int disallow_signal(int);
2314
2315extern int do_execve(struct filename *,
2316                     const char __user * const __user *,
2317                     const char __user * const __user *);
2318extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2319struct task_struct *fork_idle(int);
2320extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2321
2322extern void set_task_comm(struct task_struct *tsk, char *from);
2323extern char *get_task_comm(char *to, struct task_struct *tsk);
2324
2325#ifdef CONFIG_SMP
2326void scheduler_ipi(void);
2327extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2328#else
2329static inline void scheduler_ipi(void) { }
2330static inline unsigned long wait_task_inactive(struct task_struct *p,
2331                                               long match_state)
2332{
2333        return 1;
2334}
2335#endif
2336
2337#define next_task(p) \
2338        list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2339
2340#define for_each_process(p) \
2341        for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2342
2343extern bool current_is_single_threaded(void);
2344
2345/*
2346 * Careful: do_each_thread/while_each_thread is a double loop so
2347 *          'break' will not work as expected - use goto instead.
2348 */
2349#define do_each_thread(g, t) \
2350        for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2351
2352#define while_each_thread(g, t) \
2353        while ((t = next_thread(t)) != g)
2354
2355#define __for_each_thread(signal, t)    \
2356        list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2357
2358#define for_each_thread(p, t)           \
2359        __for_each_thread((p)->signal, t)
2360
2361/* Careful: this is a double loop, 'break' won't work as expected. */
2362#define for_each_process_thread(p, t)   \
2363        for_each_process(p) for_each_thread(p, t)
2364
2365static inline int get_nr_threads(struct task_struct *tsk)
2366{
2367        return tsk->signal->nr_threads;
2368}
2369
2370static inline bool thread_group_leader(struct task_struct *p)
2371{
2372        return p->exit_signal >= 0;
2373}
2374
2375/* Do to the insanities of de_thread it is possible for a process
2376 * to have the pid of the thread group leader without actually being
2377 * the thread group leader.  For iteration through the pids in proc
2378 * all we care about is that we have a task with the appropriate
2379 * pid, we don't actually care if we have the right task.
2380 */
2381static inline bool has_group_leader_pid(struct task_struct *p)
2382{
2383        return task_pid(p) == p->signal->leader_pid;
2384}
2385
2386static inline
2387bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2388{
2389        return p1->signal == p2->signal;
2390}
2391
2392static inline struct task_struct *next_thread(const struct task_struct *p)
2393{
2394        return list_entry_rcu(p->thread_group.next,
2395                              struct task_struct, thread_group);
2396}
2397
2398static inline int thread_group_empty(struct task_struct *p)
2399{
2400        return list_empty(&p->thread_group);
2401}
2402
2403#define delay_group_leader(p) \
2404                (thread_group_leader(p) && !thread_group_empty(p))
2405
2406/*
2407 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2408 * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2409 * pins the final release of task.io_context.  Also protects ->cpuset and
2410 * ->cgroup.subsys[]. And ->vfork_done.
2411 *
2412 * Nests both inside and outside of read_lock(&tasklist_lock).
2413 * It must not be nested with write_lock_irq(&tasklist_lock),
2414 * neither inside nor outside.
2415 */
2416static inline void task_lock(struct task_struct *p)
2417{
2418        spin_lock(&p->alloc_lock);
2419}
2420
2421static inline void task_unlock(struct task_struct *p)
2422{
2423        spin_unlock(&p->alloc_lock);
2424}
2425
2426extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2427                                                        unsigned long *flags);
2428
2429static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2430                                                       unsigned long *flags)
2431{
2432        struct sighand_struct *ret;
2433
2434        ret = __lock_task_sighand(tsk, flags);
2435        (void)__cond_lock(&tsk->sighand->siglock, ret);
2436        return ret;
2437}
2438
2439static inline void unlock_task_sighand(struct task_struct *tsk,
2440                                                unsigned long *flags)
2441{
2442        spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2443}
2444
2445#ifdef CONFIG_CGROUPS
2446static inline void threadgroup_change_begin(struct task_struct *tsk)
2447{
2448        down_read(&tsk->signal->group_rwsem);
2449}
2450static inline void threadgroup_change_end(struct task_struct *tsk)
2451{
2452        up_read(&tsk->signal->group_rwsem);
2453}
2454
2455/**
2456 * threadgroup_lock - lock threadgroup
2457 * @tsk: member task of the threadgroup to lock
2458 *
2459 * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2460 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2461 * change ->group_leader/pid.  This is useful for cases where the threadgroup
2462 * needs to stay stable across blockable operations.
2463 *
2464 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2465 * synchronization.  While held, no new task will be added to threadgroup
2466 * and no existing live task will have its PF_EXITING set.
2467 *
2468 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2469 * sub-thread becomes a new leader.
2470 */
2471static inline void threadgroup_lock(struct task_struct *tsk)
2472{
2473        down_write(&tsk->signal->group_rwsem);
2474}
2475
2476/**
2477 * threadgroup_unlock - unlock threadgroup
2478 * @tsk: member task of the threadgroup to unlock
2479 *
2480 * Reverse threadgroup_lock().
2481 */
2482static inline void threadgroup_unlock(struct task_struct *tsk)
2483{
2484        up_write(&tsk->signal->group_rwsem);
2485}
2486#else
2487static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2488static inline void threadgroup_change_end(struct task_struct *tsk) {}
2489static inline void threadgroup_lock(struct task_struct *tsk) {}
2490static inline void threadgroup_unlock(struct task_struct *tsk) {}
2491#endif
2492
2493#ifndef __HAVE_THREAD_FUNCTIONS
2494
2495#define task_thread_info(task)  ((struct thread_info *)(task)->stack)
2496#define task_stack_page(task)   ((task)->stack)
2497
2498static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2499{
2500        *task_thread_info(p) = *task_thread_info(org);
2501        task_thread_info(p)->task = p;
2502}
2503
2504static inline unsigned long *end_of_stack(struct task_struct *p)
2505{
2506        return (unsigned long *)(task_thread_info(p) + 1);
2507}
2508
2509#endif
2510
2511static inline int object_is_on_stack(void *obj)
2512{
2513        void *stack = task_stack_page(current);
2514
2515        return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2516}
2517
2518extern void thread_info_cache_init(void);
2519
2520#ifdef CONFIG_DEBUG_STACK_USAGE
2521static inline unsigned long stack_not_used(struct task_struct *p)
2522{
2523        unsigned long *n = end_of_stack(p);
2524
2525        do {    /* Skip over canary */
2526                n++;
2527        } while (!*n);
2528
2529        return (unsigned long)n - (unsigned long)end_of_stack(p);
2530}
2531#endif
2532
2533/* set thread flags in other task's structures
2534 * - see asm/thread_info.h for TIF_xxxx flags available
2535 */
2536static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2537{
2538        set_ti_thread_flag(task_thread_info(tsk), flag);
2539}
2540
2541static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2542{
2543        clear_ti_thread_flag(task_thread_info(tsk), flag);
2544}
2545
2546static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2547{
2548        return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2549}
2550
2551static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2552{
2553        return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2554}
2555
2556static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2557{
2558        return test_ti_thread_flag(task_thread_info(tsk), flag);
2559}
2560
2561static inline void set_tsk_need_resched(struct task_struct *tsk)
2562{
2563        set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2564}
2565
2566static inline void clear_tsk_need_resched(struct task_struct *tsk)
2567{
2568        clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2569}
2570
2571static inline int test_tsk_need_resched(struct task_struct *tsk)
2572{
2573        return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2574}
2575
2576static inline int restart_syscall(void)
2577{
2578        set_tsk_thread_flag(current, TIF_SIGPENDING);
2579        return -ERESTARTNOINTR;
2580}
2581
2582static inline int signal_pending(struct task_struct *p)
2583{
2584        return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2585}
2586
2587static inline int __fatal_signal_pending(struct task_struct *p)
2588{
2589        return unlikely(sigismember(&p->pending.signal, SIGKILL));
2590}
2591
2592static inline int fatal_signal_pending(struct task_struct *p)
2593{
2594        return signal_pending(p) && __fatal_signal_pending(p);
2595}
2596
2597static inline int signal_pending_state(long state, struct task_struct *p)
2598{
2599        if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2600                return 0;
2601        if (!signal_pending(p))
2602                return 0;
2603
2604        return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2605}
2606
2607/*
2608 * cond_resched() and cond_resched_lock(): latency reduction via
2609 * explicit rescheduling in places that are safe. The return
2610 * value indicates whether a reschedule was done in fact.
2611 * cond_resched_lock() will drop the spinlock before scheduling,
2612 * cond_resched_softirq() will enable bhs before scheduling.
2613 */
2614extern int _cond_resched(void);
2615
2616#define cond_resched() ({                       \
2617        __might_sleep(__FILE__, __LINE__, 0);   \
2618        _cond_resched();                        \
2619})
2620
2621extern int __cond_resched_lock(spinlock_t *lock);
2622
2623#ifdef CONFIG_PREEMPT_COUNT
2624#define PREEMPT_LOCK_OFFSET     PREEMPT_OFFSET
2625#else
2626#define PREEMPT_LOCK_OFFSET     0
2627#endif
2628
2629#define cond_resched_lock(lock) ({                              \
2630        __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2631        __cond_resched_lock(lock);                              \
2632})
2633
2634extern int __cond_resched_softirq(void);
2635
2636#define cond_resched_softirq() ({                                       \
2637        __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);      \
2638        __cond_resched_softirq();                                       \
2639})
2640
2641static inline void cond_resched_rcu(void)
2642{
2643#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2644        rcu_read_unlock();
2645        cond_resched();
2646        rcu_read_lock();
2647#endif
2648}
2649
2650/*
2651 * Does a critical section need to be broken due to another
2652 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2653 * but a general need for low latency)
2654 */
2655static inline int spin_needbreak(spinlock_t *lock)
2656{
2657#ifdef CONFIG_PREEMPT
2658        return spin_is_contended(lock);
2659#else
2660        return 0;
2661#endif
2662}
2663
2664/*
2665 * Idle thread specific functions to determine the need_resched
2666 * polling state. We have two versions, one based on TS_POLLING in
2667 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2668 * thread_info.flags
2669 */
2670#ifdef TS_POLLING
2671static inline int tsk_is_polling(struct task_struct *p)
2672{
2673        return task_thread_info(p)->status & TS_POLLING;
2674}
2675static inline void __current_set_polling(void)
2676{
2677        current_thread_info()->status |= TS_POLLING;
2678}
2679
2680static inline bool __must_check current_set_polling_and_test(void)
2681{
2682        __current_set_polling();
2683
2684        /*
2685         * Polling state must be visible before we test NEED_RESCHED,
2686         * paired by resched_task()
2687         */
2688        smp_mb();
2689
2690        return unlikely(tif_need_resched());
2691}
2692
2693static inline void __current_clr_polling(void)
2694{
2695        current_thread_info()->status &= ~TS_POLLING;
2696}
2697
2698static inline bool __must_check current_clr_polling_and_test(void)
2699{
2700        __current_clr_polling();
2701
2702        /*
2703         * Polling state must be visible before we test NEED_RESCHED,
2704         * paired by resched_task()
2705         */
2706        smp_mb();
2707
2708        return unlikely(tif_need_resched());
2709}
2710#elif defined(TIF_POLLING_NRFLAG)
2711static inline int tsk_is_polling(struct task_struct *p)
2712{
2713        return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2714}
2715
2716static inline void __current_set_polling(void)
2717{
2718        set_thread_flag(TIF_POLLING_NRFLAG);
2719}
2720
2721static inline bool __must_check current_set_polling_and_test(void)
2722{
2723        __current_set_polling();
2724
2725        /*
2726         * Polling state must be visible before we test NEED_RESCHED,
2727         * paired by resched_task()
2728         *
2729         * XXX: assumes set/clear bit are identical barrier wise.
2730         */
2731        smp_mb__after_clear_bit();
2732
2733        return unlikely(tif_need_resched());
2734}
2735
2736static inline void __current_clr_polling(void)
2737{
2738        clear_thread_flag(TIF_POLLING_NRFLAG);
2739}
2740
2741static inline bool __must_check current_clr_polling_and_test(void)
2742{
2743        __current_clr_polling();
2744
2745        /*
2746         * Polling state must be visible before we test NEED_RESCHED,
2747         * paired by resched_task()
2748         */
2749        smp_mb__after_clear_bit();
2750
2751        return unlikely(tif_need_resched());
2752}
2753
2754#else
2755static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2756static inline void __current_set_polling(void) { }
2757static inline void __current_clr_polling(void) { }
2758
2759static inline bool __must_check current_set_polling_and_test(void)
2760{
2761        return unlikely(tif_need_resched());
2762}
2763static inline bool __must_check current_clr_polling_and_test(void)
2764{
2765        return unlikely(tif_need_resched());
2766}
2767#endif
2768
2769static inline void current_clr_polling(void)
2770{
2771        __current_clr_polling();
2772
2773        /*
2774         * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2775         * Once the bit is cleared, we'll get IPIs with every new
2776         * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2777         * fold.
2778         */
2779        smp_mb(); /* paired with resched_task() */
2780
2781        preempt_fold_need_resched();
2782}
2783
2784static __always_inline bool need_resched(void)
2785{
2786        return unlikely(tif_need_resched());
2787}
2788
2789/*
2790 * Thread group CPU time accounting.
2791 */
2792void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2793void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2794
2795static inline void thread_group_cputime_init(struct signal_struct *sig)
2796{
2797        raw_spin_lock_init(&sig->cputimer.lock);
2798}
2799
2800/*
2801 * Reevaluate whether the task has signals pending delivery.
2802 * Wake the task if so.
2803 * This is required every time the blocked sigset_t changes.
2804 * callers must hold sighand->siglock.
2805 */
2806extern void recalc_sigpending_and_wake(struct task_struct *t);
2807extern void recalc_sigpending(void);
2808
2809extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2810
2811static inline void signal_wake_up(struct task_struct *t, bool resume)
2812{
2813        signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2814}
2815static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2816{
2817        signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2818}
2819
2820/*
2821 * Wrappers for p->thread_info->cpu access. No-op on UP.
2822 */
2823#ifdef CONFIG_SMP
2824
2825static inline unsigned int task_cpu(const struct task_struct *p)
2826{
2827        return task_thread_info(p)->cpu;
2828}
2829
2830static inline int task_node(const struct task_struct *p)
2831{
2832        return cpu_to_node(task_cpu(p));
2833}
2834
2835extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2836
2837#else
2838
2839static inline unsigned int task_cpu(const struct task_struct *p)
2840{
2841        return 0;
2842}
2843
2844static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2845{
2846}
2847
2848#endif /* CONFIG_SMP */
2849
2850extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2851extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2852
2853#ifdef CONFIG_CGROUP_SCHED
2854extern struct task_group root_task_group;
2855#endif /* CONFIG_CGROUP_SCHED */
2856
2857extern int task_can_switch_user(struct user_struct *up,
2858                                        struct task_struct *tsk);
2859
2860#ifdef CONFIG_TASK_XACCT
2861static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2862{
2863        tsk->ioac.rchar += amt;
2864}
2865
2866static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2867{
2868        tsk->ioac.wchar += amt;
2869}
2870
2871static inline void inc_syscr(struct task_struct *tsk)
2872{
2873        tsk->ioac.syscr++;
2874}
2875
2876static inline void inc_syscw(struct task_struct *tsk)
2877{
2878        tsk->ioac.syscw++;
2879}
2880#else
2881static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2882{
2883}
2884
2885static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2886{
2887}
2888
2889static inline void inc_syscr(struct task_struct *tsk)
2890{
2891}
2892
2893static inline void inc_syscw(struct task_struct *tsk)
2894{
2895}
2896#endif
2897
2898#ifndef TASK_SIZE_OF
2899#define TASK_SIZE_OF(tsk)       TASK_SIZE
2900#endif
2901
2902#ifdef CONFIG_MM_OWNER
2903extern void mm_update_next_owner(struct mm_struct *mm);
2904extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2905#else
2906static inline void mm_update_next_owner(struct mm_struct *mm)
2907{
2908}
2909
2910static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2911{
2912}
2913#endif /* CONFIG_MM_OWNER */
2914
2915static inline unsigned long task_rlimit(const struct task_struct *tsk,
2916                unsigned int limit)
2917{
2918        return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2919}
2920
2921static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2922                unsigned int limit)
2923{
2924        return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2925}
2926
2927static inline unsigned long rlimit(unsigned int limit)
2928{
2929        return task_rlimit(current, limit);
2930}
2931
2932static inline unsigned long rlimit_max(unsigned int limit)
2933{
2934        return task_rlimit_max(current, limit);
2935}
2936
2937#endif
2938
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.