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#include <linux/sched/prio.h>
   7
   8
   9struct sched_param {
  10        int sched_priority;
  11};
  12
  13#include <asm/param.h>  /* for HZ */
  14
  15#include <linux/capability.h>
  16#include <linux/threads.h>
  17#include <linux/kernel.h>
  18#include <linux/types.h>
  19#include <linux/timex.h>
  20#include <linux/jiffies.h>
  21#include <linux/plist.h>
  22#include <linux/rbtree.h>
  23#include <linux/thread_info.h>
  24#include <linux/cpumask.h>
  25#include <linux/errno.h>
  26#include <linux/nodemask.h>
  27#include <linux/mm_types.h>
  28#include <linux/preempt.h>
  29
  30#include <asm/page.h>
  31#include <asm/ptrace.h>
  32#include <linux/cputime.h>
  33
  34#include <linux/smp.h>
  35#include <linux/sem.h>
  36#include <linux/shm.h>
  37#include <linux/signal.h>
  38#include <linux/compiler.h>
  39#include <linux/completion.h>
  40#include <linux/pid.h>
  41#include <linux/percpu.h>
  42#include <linux/topology.h>
  43#include <linux/proportions.h>
  44#include <linux/seccomp.h>
  45#include <linux/rcupdate.h>
  46#include <linux/rculist.h>
  47#include <linux/rtmutex.h>
  48
  49#include <linux/time.h>
  50#include <linux/param.h>
  51#include <linux/resource.h>
  52#include <linux/timer.h>
  53#include <linux/hrtimer.h>
  54#include <linux/kcov.h>
  55#include <linux/task_io_accounting.h>
  56#include <linux/latencytop.h>
  57#include <linux/cred.h>
  58#include <linux/llist.h>
  59#include <linux/uidgid.h>
  60#include <linux/gfp.h>
  61#include <linux/magic.h>
  62#include <linux/cgroup-defs.h>
  63
  64#include <asm/processor.h>
  65
  66#define SCHED_ATTR_SIZE_VER0    48      /* sizeof first published struct */
  67
  68/*
  69 * Extended scheduling parameters data structure.
  70 *
  71 * This is needed because the original struct sched_param can not be
  72 * altered without introducing ABI issues with legacy applications
  73 * (e.g., in sched_getparam()).
  74 *
  75 * However, the possibility of specifying more than just a priority for
  76 * the tasks may be useful for a wide variety of application fields, e.g.,
  77 * multimedia, streaming, automation and control, and many others.
  78 *
  79 * This variant (sched_attr) is meant at describing a so-called
  80 * sporadic time-constrained task. In such model a task is specified by:
  81 *  - the activation period or minimum instance inter-arrival time;
  82 *  - the maximum (or average, depending on the actual scheduling
  83 *    discipline) computation time of all instances, a.k.a. runtime;
  84 *  - the deadline (relative to the actual activation time) of each
  85 *    instance.
  86 * Very briefly, a periodic (sporadic) task asks for the execution of
  87 * some specific computation --which is typically called an instance--
  88 * (at most) every period. Moreover, each instance typically lasts no more
  89 * than the runtime and must be completed by time instant t equal to
  90 * the instance activation time + the deadline.
  91 *
  92 * This is reflected by the actual fields of the sched_attr structure:
  93 *
  94 *  @size               size of the structure, for fwd/bwd compat.
  95 *
  96 *  @sched_policy       task's scheduling policy
  97 *  @sched_flags        for customizing the scheduler behaviour
  98 *  @sched_nice         task's nice value      (SCHED_NORMAL/BATCH)
  99 *  @sched_priority     task's static priority (SCHED_FIFO/RR)
 100 *  @sched_deadline     representative of the task's deadline
 101 *  @sched_runtime      representative of the task's runtime
 102 *  @sched_period       representative of the task's period
 103 *
 104 * Given this task model, there are a multiplicity of scheduling algorithms
 105 * and policies, that can be used to ensure all the tasks will make their
 106 * timing constraints.
 107 *
 108 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
 109 * only user of this new interface. More information about the algorithm
 110 * available in the scheduling class file or in Documentation/.
 111 */
 112struct sched_attr {
 113        u32 size;
 114
 115        u32 sched_policy;
 116        u64 sched_flags;
 117
 118        /* SCHED_NORMAL, SCHED_BATCH */
 119        s32 sched_nice;
 120
 121        /* SCHED_FIFO, SCHED_RR */
 122        u32 sched_priority;
 123
 124        /* SCHED_DEADLINE */
 125        u64 sched_runtime;
 126        u64 sched_deadline;
 127        u64 sched_period;
 128};
 129
 130struct futex_pi_state;
 131struct robust_list_head;
 132struct bio_list;
 133struct fs_struct;
 134struct perf_event_context;
 135struct blk_plug;
 136struct filename;
 137struct nameidata;
 138
 139#define VMACACHE_BITS 2
 140#define VMACACHE_SIZE (1U << VMACACHE_BITS)
 141#define VMACACHE_MASK (VMACACHE_SIZE - 1)
 142
 143/*
 144 * These are the constant used to fake the fixed-point load-average
 145 * counting. Some notes:
 146 *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
 147 *    a load-average precision of 10 bits integer + 11 bits fractional
 148 *  - if you want to count load-averages more often, you need more
 149 *    precision, or rounding will get you. With 2-second counting freq,
 150 *    the EXP_n values would be 1981, 2034 and 2043 if still using only
 151 *    11 bit fractions.
 152 */
 153extern unsigned long avenrun[];         /* Load averages */
 154extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
 155
 156#define FSHIFT          11              /* nr of bits of precision */
 157#define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
 158#define LOAD_FREQ       (5*HZ+1)        /* 5 sec intervals */
 159#define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
 160#define EXP_5           2014            /* 1/exp(5sec/5min) */
 161#define EXP_15          2037            /* 1/exp(5sec/15min) */
 162
 163#define CALC_LOAD(load,exp,n) \
 164        load *= exp; \
 165        load += n*(FIXED_1-exp); \
 166        load >>= FSHIFT;
 167
 168extern unsigned long total_forks;
 169extern int nr_threads;
 170DECLARE_PER_CPU(unsigned long, process_counts);
 171extern int nr_processes(void);
 172extern unsigned long nr_running(void);
 173extern bool single_task_running(void);
 174extern unsigned long nr_iowait(void);
 175extern unsigned long nr_iowait_cpu(int cpu);
 176extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
 177
 178extern void calc_global_load(unsigned long ticks);
 179
 180#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
 181extern void update_cpu_load_nohz(int active);
 182#else
 183static inline void update_cpu_load_nohz(int active) { }
 184#endif
 185
 186extern void dump_cpu_task(int cpu);
 187
 188struct seq_file;
 189struct cfs_rq;
 190struct task_group;
 191#ifdef CONFIG_SCHED_DEBUG
 192extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
 193extern void proc_sched_set_task(struct task_struct *p);
 194#endif
 195
 196/*
 197 * Task state bitmask. NOTE! These bits are also
 198 * encoded in fs/proc/array.c: get_task_state().
 199 *
 200 * We have two separate sets of flags: task->state
 201 * is about runnability, while task->exit_state are
 202 * about the task exiting. Confusing, but this way
 203 * modifying one set can't modify the other one by
 204 * mistake.
 205 */
 206#define TASK_RUNNING            0
 207#define TASK_INTERRUPTIBLE      1
 208#define TASK_UNINTERRUPTIBLE    2
 209#define __TASK_STOPPED          4
 210#define __TASK_TRACED           8
 211/* in tsk->exit_state */
 212#define EXIT_DEAD               16
 213#define EXIT_ZOMBIE             32
 214#define EXIT_TRACE              (EXIT_ZOMBIE | EXIT_DEAD)
 215/* in tsk->state again */
 216#define TASK_DEAD               64
 217#define TASK_WAKEKILL           128
 218#define TASK_WAKING             256
 219#define TASK_PARKED             512
 220#define TASK_NOLOAD             1024
 221#define TASK_STATE_MAX          2048
 222
 223#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
 224
 225extern char ___assert_task_state[1 - 2*!!(
 226                sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
 227
 228/* Convenience macros for the sake of set_task_state */
 229#define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
 230#define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
 231#define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
 232
 233#define TASK_IDLE               (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
 234
 235/* Convenience macros for the sake of wake_up */
 236#define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 237#define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
 238
 239/* get_task_state() */
 240#define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 241                                 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 242                                 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
 243
 244#define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
 245#define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
 246#define task_is_stopped_or_traced(task) \
 247                        ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 248#define task_contributes_to_load(task)  \
 249                                ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
 250                                 (task->flags & PF_FROZEN) == 0 && \
 251                                 (task->state & TASK_NOLOAD) == 0)
 252
 253#ifdef CONFIG_DEBUG_ATOMIC_SLEEP
 254
 255#define __set_task_state(tsk, state_value)                      \
 256        do {                                                    \
 257                (tsk)->task_state_change = _THIS_IP_;           \
 258                (tsk)->state = (state_value);                   \
 259        } while (0)
 260#define set_task_state(tsk, state_value)                        \
 261        do {                                                    \
 262                (tsk)->task_state_change = _THIS_IP_;           \
 263                smp_store_mb((tsk)->state, (state_value));              \
 264        } while (0)
 265
 266/*
 267 * set_current_state() includes a barrier so that the write of current->state
 268 * is correctly serialised wrt the caller's subsequent test of whether to
 269 * actually sleep:
 270 *
 271 *      set_current_state(TASK_UNINTERRUPTIBLE);
 272 *      if (do_i_need_to_sleep())
 273 *              schedule();
 274 *
 275 * If the caller does not need such serialisation then use __set_current_state()
 276 */
 277#define __set_current_state(state_value)                        \
 278        do {                                                    \
 279                current->task_state_change = _THIS_IP_;         \
 280                current->state = (state_value);                 \
 281        } while (0)
 282#define set_current_state(state_value)                          \
 283        do {                                                    \
 284                current->task_state_change = _THIS_IP_;         \
 285                smp_store_mb(current->state, (state_value));            \
 286        } while (0)
 287
 288#else
 289
 290#define __set_task_state(tsk, state_value)              \
 291        do { (tsk)->state = (state_value); } while (0)
 292#define set_task_state(tsk, state_value)                \
 293        smp_store_mb((tsk)->state, (state_value))
 294
 295/*
 296 * set_current_state() includes a barrier so that the write of current->state
 297 * is correctly serialised wrt the caller's subsequent test of whether to
 298 * actually sleep:
 299 *
 300 *      set_current_state(TASK_UNINTERRUPTIBLE);
 301 *      if (do_i_need_to_sleep())
 302 *              schedule();
 303 *
 304 * If the caller does not need such serialisation then use __set_current_state()
 305 */
 306#define __set_current_state(state_value)                \
 307        do { current->state = (state_value); } while (0)
 308#define set_current_state(state_value)                  \
 309        smp_store_mb(current->state, (state_value))
 310
 311#endif
 312
 313/* Task command name length */
 314#define TASK_COMM_LEN 16
 315
 316#include <linux/spinlock.h>
 317
 318/*
 319 * This serializes "schedule()" and also protects
 320 * the run-queue from deletions/modifications (but
 321 * _adding_ to the beginning of the run-queue has
 322 * a separate lock).
 323 */
 324extern rwlock_t tasklist_lock;
 325extern spinlock_t mmlist_lock;
 326
 327struct task_struct;
 328
 329#ifdef CONFIG_PROVE_RCU
 330extern int lockdep_tasklist_lock_is_held(void);
 331#endif /* #ifdef CONFIG_PROVE_RCU */
 332
 333extern void sched_init(void);
 334extern void sched_init_smp(void);
 335extern asmlinkage void schedule_tail(struct task_struct *prev);
 336extern void init_idle(struct task_struct *idle, int cpu);
 337extern void init_idle_bootup_task(struct task_struct *idle);
 338
 339extern cpumask_var_t cpu_isolated_map;
 340
 341extern int runqueue_is_locked(int cpu);
 342
 343#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
 344extern void nohz_balance_enter_idle(int cpu);
 345extern void set_cpu_sd_state_idle(void);
 346extern int get_nohz_timer_target(void);
 347#else
 348static inline void nohz_balance_enter_idle(int cpu) { }
 349static inline void set_cpu_sd_state_idle(void) { }
 350#endif
 351
 352/*
 353 * Only dump TASK_* tasks. (0 for all tasks)
 354 */
 355extern void show_state_filter(unsigned long state_filter);
 356
 357static inline void show_state(void)
 358{
 359        show_state_filter(0);
 360}
 361
 362extern void show_regs(struct pt_regs *);
 363
 364/*
 365 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 366 * task), SP is the stack pointer of the first frame that should be shown in the back
 367 * trace (or NULL if the entire call-chain of the task should be shown).
 368 */
 369extern void show_stack(struct task_struct *task, unsigned long *sp);
 370
 371extern void cpu_init (void);
 372extern void trap_init(void);
 373extern void update_process_times(int user);
 374extern void scheduler_tick(void);
 375
 376extern void sched_show_task(struct task_struct *p);
 377
 378#ifdef CONFIG_LOCKUP_DETECTOR
 379extern void touch_softlockup_watchdog_sched(void);
 380extern void touch_softlockup_watchdog(void);
 381extern void touch_softlockup_watchdog_sync(void);
 382extern void touch_all_softlockup_watchdogs(void);
 383extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
 384                                  void __user *buffer,
 385                                  size_t *lenp, loff_t *ppos);
 386extern unsigned int  softlockup_panic;
 387extern unsigned int  hardlockup_panic;
 388void lockup_detector_init(void);
 389#else
 390static inline void touch_softlockup_watchdog_sched(void)
 391{
 392}
 393static inline void touch_softlockup_watchdog(void)
 394{
 395}
 396static inline void touch_softlockup_watchdog_sync(void)
 397{
 398}
 399static inline void touch_all_softlockup_watchdogs(void)
 400{
 401}
 402static inline void lockup_detector_init(void)
 403{
 404}
 405#endif
 406
 407#ifdef CONFIG_DETECT_HUNG_TASK
 408void reset_hung_task_detector(void);
 409#else
 410static inline void reset_hung_task_detector(void)
 411{
 412}
 413#endif
 414
 415/* Attach to any functions which should be ignored in wchan output. */
 416#define __sched         __attribute__((__section__(".sched.text")))
 417
 418/* Linker adds these: start and end of __sched functions */
 419extern char __sched_text_start[], __sched_text_end[];
 420
 421/* Is this address in the __sched functions? */
 422extern int in_sched_functions(unsigned long addr);
 423
 424#define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 425extern signed long schedule_timeout(signed long timeout);
 426extern signed long schedule_timeout_interruptible(signed long timeout);
 427extern signed long schedule_timeout_killable(signed long timeout);
 428extern signed long schedule_timeout_uninterruptible(signed long timeout);
 429extern signed long schedule_timeout_idle(signed long timeout);
 430asmlinkage void schedule(void);
 431extern void schedule_preempt_disabled(void);
 432
 433extern long io_schedule_timeout(long timeout);
 434
 435static inline void io_schedule(void)
 436{
 437        io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
 438}
 439
 440struct nsproxy;
 441struct user_namespace;
 442
 443#ifdef CONFIG_MMU
 444extern void arch_pick_mmap_layout(struct mm_struct *mm);
 445extern unsigned long
 446arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 447                       unsigned long, unsigned long);
 448extern unsigned long
 449arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 450                          unsigned long len, unsigned long pgoff,
 451                          unsigned long flags);
 452#else
 453static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
 454#endif
 455
 456#define SUID_DUMP_DISABLE       0       /* No setuid dumping */
 457#define SUID_DUMP_USER          1       /* Dump as user of process */
 458#define SUID_DUMP_ROOT          2       /* Dump as root */
 459
 460/* mm flags */
 461
 462/* for SUID_DUMP_* above */
 463#define MMF_DUMPABLE_BITS 2
 464#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
 465
 466extern void set_dumpable(struct mm_struct *mm, int value);
 467/*
 468 * This returns the actual value of the suid_dumpable flag. For things
 469 * that are using this for checking for privilege transitions, it must
 470 * test against SUID_DUMP_USER rather than treating it as a boolean
 471 * value.
 472 */
 473static inline int __get_dumpable(unsigned long mm_flags)
 474{
 475        return mm_flags & MMF_DUMPABLE_MASK;
 476}
 477
 478static inline int get_dumpable(struct mm_struct *mm)
 479{
 480        return __get_dumpable(mm->flags);
 481}
 482
 483/* coredump filter bits */
 484#define MMF_DUMP_ANON_PRIVATE   2
 485#define MMF_DUMP_ANON_SHARED    3
 486#define MMF_DUMP_MAPPED_PRIVATE 4
 487#define MMF_DUMP_MAPPED_SHARED  5
 488#define MMF_DUMP_ELF_HEADERS    6
 489#define MMF_DUMP_HUGETLB_PRIVATE 7
 490#define MMF_DUMP_HUGETLB_SHARED  8
 491#define MMF_DUMP_DAX_PRIVATE    9
 492#define MMF_DUMP_DAX_SHARED     10
 493
 494#define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
 495#define MMF_DUMP_FILTER_BITS    9
 496#define MMF_DUMP_FILTER_MASK \
 497        (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
 498#define MMF_DUMP_FILTER_DEFAULT \
 499        ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
 500         (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
 501
 502#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
 503# define MMF_DUMP_MASK_DEFAULT_ELF      (1 << MMF_DUMP_ELF_HEADERS)
 504#else
 505# define MMF_DUMP_MASK_DEFAULT_ELF      0
 506#endif
 507                                        /* leave room for more dump flags */
 508#define MMF_VM_MERGEABLE        16      /* KSM may merge identical pages */
 509#define MMF_VM_HUGEPAGE         17      /* set when VM_HUGEPAGE is set on vma */
 510#define MMF_EXE_FILE_CHANGED    18      /* see prctl_set_mm_exe_file() */
 511
 512#define MMF_HAS_UPROBES         19      /* has uprobes */
 513#define MMF_RECALC_UPROBES      20      /* MMF_HAS_UPROBES can be wrong */
 514
 515#define MMF_INIT_MASK           (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
 516
 517struct sighand_struct {
 518        atomic_t                count;
 519        struct k_sigaction      action[_NSIG];
 520        spinlock_t              siglock;
 521        wait_queue_head_t       signalfd_wqh;
 522};
 523
 524struct pacct_struct {
 525        int                     ac_flag;
 526        long                    ac_exitcode;
 527        unsigned long           ac_mem;
 528        cputime_t               ac_utime, ac_stime;
 529        unsigned long           ac_minflt, ac_majflt;
 530};
 531
 532struct cpu_itimer {
 533        cputime_t expires;
 534        cputime_t incr;
 535        u32 error;
 536        u32 incr_error;
 537};
 538
 539/**
 540 * struct prev_cputime - snaphsot of system and user cputime
 541 * @utime: time spent in user mode
 542 * @stime: time spent in system mode
 543 * @lock: protects the above two fields
 544 *
 545 * Stores previous user/system time values such that we can guarantee
 546 * monotonicity.
 547 */
 548struct prev_cputime {
 549#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
 550        cputime_t utime;
 551        cputime_t stime;
 552        raw_spinlock_t lock;
 553#endif
 554};
 555
 556static inline void prev_cputime_init(struct prev_cputime *prev)
 557{
 558#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
 559        prev->utime = prev->stime = 0;
 560        raw_spin_lock_init(&prev->lock);
 561#endif
 562}
 563
 564/**
 565 * struct task_cputime - collected CPU time counts
 566 * @utime:              time spent in user mode, in &cputime_t units
 567 * @stime:              time spent in kernel mode, in &cputime_t units
 568 * @sum_exec_runtime:   total time spent on the CPU, in nanoseconds
 569 *
 570 * This structure groups together three kinds of CPU time that are tracked for
 571 * threads and thread groups.  Most things considering CPU time want to group
 572 * these counts together and treat all three of them in parallel.
 573 */
 574struct task_cputime {
 575        cputime_t utime;
 576        cputime_t stime;
 577        unsigned long long sum_exec_runtime;
 578};
 579
 580/* Alternate field names when used to cache expirations. */
 581#define virt_exp        utime
 582#define prof_exp        stime
 583#define sched_exp       sum_exec_runtime
 584
 585#define INIT_CPUTIME    \
 586        (struct task_cputime) {                                 \
 587                .utime = 0,                                     \
 588                .stime = 0,                                     \
 589                .sum_exec_runtime = 0,                          \
 590        }
 591
 592/*
 593 * This is the atomic variant of task_cputime, which can be used for
 594 * storing and updating task_cputime statistics without locking.
 595 */
 596struct task_cputime_atomic {
 597        atomic64_t utime;
 598        atomic64_t stime;
 599        atomic64_t sum_exec_runtime;
 600};
 601
 602#define INIT_CPUTIME_ATOMIC \
 603        (struct task_cputime_atomic) {                          \
 604                .utime = ATOMIC64_INIT(0),                      \
 605                .stime = ATOMIC64_INIT(0),                      \
 606                .sum_exec_runtime = ATOMIC64_INIT(0),           \
 607        }
 608
 609#define PREEMPT_DISABLED        (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
 610
 611/*
 612 * Disable preemption until the scheduler is running -- use an unconditional
 613 * value so that it also works on !PREEMPT_COUNT kernels.
 614 *
 615 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
 616 */
 617#define INIT_PREEMPT_COUNT      PREEMPT_OFFSET
 618
 619/*
 620 * Initial preempt_count value; reflects the preempt_count schedule invariant
 621 * which states that during context switches:
 622 *
 623 *    preempt_count() == 2*PREEMPT_DISABLE_OFFSET
 624 *
 625 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
 626 * Note: See finish_task_switch().
 627 */
 628#define FORK_PREEMPT_COUNT      (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
 629
 630/**
 631 * struct thread_group_cputimer - thread group interval timer counts
 632 * @cputime_atomic:     atomic thread group interval timers.
 633 * @running:            true when there are timers running and
 634 *                      @cputime_atomic receives updates.
 635 * @checking_timer:     true when a thread in the group is in the
 636 *                      process of checking for thread group timers.
 637 *
 638 * This structure contains the version of task_cputime, above, that is
 639 * used for thread group CPU timer calculations.
 640 */
 641struct thread_group_cputimer {
 642        struct task_cputime_atomic cputime_atomic;
 643        bool running;
 644        bool checking_timer;
 645};
 646
 647#include <linux/rwsem.h>
 648struct autogroup;
 649
 650/*
 651 * NOTE! "signal_struct" does not have its own
 652 * locking, because a shared signal_struct always
 653 * implies a shared sighand_struct, so locking
 654 * sighand_struct is always a proper superset of
 655 * the locking of signal_struct.
 656 */
 657struct signal_struct {
 658        atomic_t                sigcnt;
 659        atomic_t                live;
 660        int                     nr_threads;
 661        struct list_head        thread_head;
 662
 663        wait_queue_head_t       wait_chldexit;  /* for wait4() */
 664
 665        /* current thread group signal load-balancing target: */
 666        struct task_struct      *curr_target;
 667
 668        /* shared signal handling: */
 669        struct sigpending       shared_pending;
 670
 671        /* thread group exit support */
 672        int                     group_exit_code;
 673        /* overloaded:
 674         * - notify group_exit_task when ->count is equal to notify_count
 675         * - everyone except group_exit_task is stopped during signal delivery
 676         *   of fatal signals, group_exit_task processes the signal.
 677         */
 678        int                     notify_count;
 679        struct task_struct      *group_exit_task;
 680
 681        /* thread group stop support, overloads group_exit_code too */
 682        int                     group_stop_count;
 683        unsigned int            flags; /* see SIGNAL_* flags below */
 684
 685        /*
 686         * PR_SET_CHILD_SUBREAPER marks a process, like a service
 687         * manager, to re-parent orphan (double-forking) child processes
 688         * to this process instead of 'init'. The service manager is
 689         * able to receive SIGCHLD signals and is able to investigate
 690         * the process until it calls wait(). All children of this
 691         * process will inherit a flag if they should look for a
 692         * child_subreaper process at exit.
 693         */
 694        unsigned int            is_child_subreaper:1;
 695        unsigned int            has_child_subreaper:1;
 696
 697        /* POSIX.1b Interval Timers */
 698        int                     posix_timer_id;
 699        struct list_head        posix_timers;
 700
 701        /* ITIMER_REAL timer for the process */
 702        struct hrtimer real_timer;
 703        struct pid *leader_pid;
 704        ktime_t it_real_incr;
 705
 706        /*
 707         * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
 708         * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
 709         * values are defined to 0 and 1 respectively
 710         */
 711        struct cpu_itimer it[2];
 712
 713        /*
 714         * Thread group totals for process CPU timers.
 715         * See thread_group_cputimer(), et al, for details.
 716         */
 717        struct thread_group_cputimer cputimer;
 718
 719        /* Earliest-expiration cache. */
 720        struct task_cputime cputime_expires;
 721
 722#ifdef CONFIG_NO_HZ_FULL
 723        atomic_t tick_dep_mask;
 724#endif
 725
 726        struct list_head cpu_timers[3];
 727
 728        struct pid *tty_old_pgrp;
 729
 730        /* boolean value for session group leader */
 731        int leader;
 732
 733        struct tty_struct *tty; /* NULL if no tty */
 734
 735#ifdef CONFIG_SCHED_AUTOGROUP
 736        struct autogroup *autogroup;
 737#endif
 738        /*
 739         * Cumulative resource counters for dead threads in the group,
 740         * and for reaped dead child processes forked by this group.
 741         * Live threads maintain their own counters and add to these
 742         * in __exit_signal, except for the group leader.
 743         */
 744        seqlock_t stats_lock;
 745        cputime_t utime, stime, cutime, cstime;
 746        cputime_t gtime;
 747        cputime_t cgtime;
 748        struct prev_cputime prev_cputime;
 749        unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 750        unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 751        unsigned long inblock, oublock, cinblock, coublock;
 752        unsigned long maxrss, cmaxrss;
 753        struct task_io_accounting ioac;
 754
 755        /*
 756         * Cumulative ns of schedule CPU time fo dead threads in the
 757         * group, not including a zombie group leader, (This only differs
 758         * from jiffies_to_ns(utime + stime) if sched_clock uses something
 759         * other than jiffies.)
 760         */
 761        unsigned long long sum_sched_runtime;
 762
 763        /*
 764         * We don't bother to synchronize most readers of this at all,
 765         * because there is no reader checking a limit that actually needs
 766         * to get both rlim_cur and rlim_max atomically, and either one
 767         * alone is a single word that can safely be read normally.
 768         * getrlimit/setrlimit use task_lock(current->group_leader) to
 769         * protect this instead of the siglock, because they really
 770         * have no need to disable irqs.
 771         */
 772        struct rlimit rlim[RLIM_NLIMITS];
 773
 774#ifdef CONFIG_BSD_PROCESS_ACCT
 775        struct pacct_struct pacct;      /* per-process accounting information */
 776#endif
 777#ifdef CONFIG_TASKSTATS
 778        struct taskstats *stats;
 779#endif
 780#ifdef CONFIG_AUDIT
 781        unsigned audit_tty;
 782        struct tty_audit_buf *tty_audit_buf;
 783#endif
 784
 785        oom_flags_t oom_flags;
 786        short oom_score_adj;            /* OOM kill score adjustment */
 787        short oom_score_adj_min;        /* OOM kill score adjustment min value.
 788                                         * Only settable by CAP_SYS_RESOURCE. */
 789
 790        struct mutex cred_guard_mutex;  /* guard against foreign influences on
 791                                         * credential calculations
 792                                         * (notably. ptrace) */
 793};
 794
 795/*
 796 * Bits in flags field of signal_struct.
 797 */
 798#define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 799#define SIGNAL_STOP_CONTINUED   0x00000002 /* SIGCONT since WCONTINUED reap */
 800#define SIGNAL_GROUP_EXIT       0x00000004 /* group exit in progress */
 801#define SIGNAL_GROUP_COREDUMP   0x00000008 /* coredump in progress */
 802/*
 803 * Pending notifications to parent.
 804 */
 805#define SIGNAL_CLD_STOPPED      0x00000010
 806#define SIGNAL_CLD_CONTINUED    0x00000020
 807#define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
 808
 809#define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
 810
 811/* If true, all threads except ->group_exit_task have pending SIGKILL */
 812static inline int signal_group_exit(const struct signal_struct *sig)
 813{
 814        return  (sig->flags & SIGNAL_GROUP_EXIT) ||
 815                (sig->group_exit_task != NULL);
 816}
 817
 818/*
 819 * Some day this will be a full-fledged user tracking system..
 820 */
 821struct user_struct {
 822        atomic_t __count;       /* reference count */
 823        atomic_t processes;     /* How many processes does this user have? */
 824        atomic_t sigpending;    /* How many pending signals does this user have? */
 825#ifdef CONFIG_INOTIFY_USER
 826        atomic_t inotify_watches; /* How many inotify watches does this user have? */
 827        atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 828#endif
 829#ifdef CONFIG_FANOTIFY
 830        atomic_t fanotify_listeners;
 831#endif
 832#ifdef CONFIG_EPOLL
 833        atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
 834#endif
 835#ifdef CONFIG_POSIX_MQUEUE
 836        /* protected by mq_lock */
 837        unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 838#endif
 839        unsigned long locked_shm; /* How many pages of mlocked shm ? */
 840        unsigned long unix_inflight;    /* How many files in flight in unix sockets */
 841        atomic_long_t pipe_bufs;  /* how many pages are allocated in pipe buffers */
 842
 843#ifdef CONFIG_KEYS
 844        struct key *uid_keyring;        /* UID specific keyring */
 845        struct key *session_keyring;    /* UID's default session keyring */
 846#endif
 847
 848        /* Hash table maintenance information */
 849        struct hlist_node uidhash_node;
 850        kuid_t uid;
 851
 852#if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
 853        atomic_long_t locked_vm;
 854#endif
 855};
 856
 857extern int uids_sysfs_init(void);
 858
 859extern struct user_struct *find_user(kuid_t);
 860
 861extern struct user_struct root_user;
 862#define INIT_USER (&root_user)
 863
 864
 865struct backing_dev_info;
 866struct reclaim_state;
 867
 868#ifdef CONFIG_SCHED_INFO
 869struct sched_info {
 870        /* cumulative counters */
 871        unsigned long pcount;         /* # of times run on this cpu */
 872        unsigned long long run_delay; /* time spent waiting on a runqueue */
 873
 874        /* timestamps */
 875        unsigned long long last_arrival,/* when we last ran on a cpu */
 876                           last_queued; /* when we were last queued to run */
 877};
 878#endif /* CONFIG_SCHED_INFO */
 879
 880#ifdef CONFIG_TASK_DELAY_ACCT
 881struct task_delay_info {
 882        spinlock_t      lock;
 883        unsigned int    flags;  /* Private per-task flags */
 884
 885        /* For each stat XXX, add following, aligned appropriately
 886         *
 887         * struct timespec XXX_start, XXX_end;
 888         * u64 XXX_delay;
 889         * u32 XXX_count;
 890         *
 891         * Atomicity of updates to XXX_delay, XXX_count protected by
 892         * single lock above (split into XXX_lock if contention is an issue).
 893         */
 894
 895        /*
 896         * XXX_count is incremented on every XXX operation, the delay
 897         * associated with the operation is added to XXX_delay.
 898         * XXX_delay contains the accumulated delay time in nanoseconds.
 899         */
 900        u64 blkio_start;        /* Shared by blkio, swapin */
 901        u64 blkio_delay;        /* wait for sync block io completion */
 902        u64 swapin_delay;       /* wait for swapin block io completion */
 903        u32 blkio_count;        /* total count of the number of sync block */
 904                                /* io operations performed */
 905        u32 swapin_count;       /* total count of the number of swapin block */
 906                                /* io operations performed */
 907
 908        u64 freepages_start;
 909        u64 freepages_delay;    /* wait for memory reclaim */
 910        u32 freepages_count;    /* total count of memory reclaim */
 911};
 912#endif  /* CONFIG_TASK_DELAY_ACCT */
 913
 914static inline int sched_info_on(void)
 915{
 916#ifdef CONFIG_SCHEDSTATS
 917        return 1;
 918#elif defined(CONFIG_TASK_DELAY_ACCT)
 919        extern int delayacct_on;
 920        return delayacct_on;
 921#else
 922        return 0;
 923#endif
 924}
 925
 926#ifdef CONFIG_SCHEDSTATS
 927void force_schedstat_enabled(void);
 928#endif
 929
 930enum cpu_idle_type {
 931        CPU_IDLE,
 932        CPU_NOT_IDLE,
 933        CPU_NEWLY_IDLE,
 934        CPU_MAX_IDLE_TYPES
 935};
 936
 937/*
 938 * Increase resolution of cpu_capacity calculations
 939 */
 940#define SCHED_CAPACITY_SHIFT    10
 941#define SCHED_CAPACITY_SCALE    (1L << SCHED_CAPACITY_SHIFT)
 942
 943/*
 944 * Wake-queues are lists of tasks with a pending wakeup, whose
 945 * callers have already marked the task as woken internally,
 946 * and can thus carry on. A common use case is being able to
 947 * do the wakeups once the corresponding user lock as been
 948 * released.
 949 *
 950 * We hold reference to each task in the list across the wakeup,
 951 * thus guaranteeing that the memory is still valid by the time
 952 * the actual wakeups are performed in wake_up_q().
 953 *
 954 * One per task suffices, because there's never a need for a task to be
 955 * in two wake queues simultaneously; it is forbidden to abandon a task
 956 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
 957 * already in a wake queue, the wakeup will happen soon and the second
 958 * waker can just skip it.
 959 *
 960 * The WAKE_Q macro declares and initializes the list head.
 961 * wake_up_q() does NOT reinitialize the list; it's expected to be
 962 * called near the end of a function, where the fact that the queue is
 963 * not used again will be easy to see by inspection.
 964 *
 965 * Note that this can cause spurious wakeups. schedule() callers
 966 * must ensure the call is done inside a loop, confirming that the
 967 * wakeup condition has in fact occurred.
 968 */
 969struct wake_q_node {
 970        struct wake_q_node *next;
 971};
 972
 973struct wake_q_head {
 974        struct wake_q_node *first;
 975        struct wake_q_node **lastp;
 976};
 977
 978#define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
 979
 980#define WAKE_Q(name)                                    \
 981        struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
 982
 983extern void wake_q_add(struct wake_q_head *head,
 984                       struct task_struct *task);
 985extern void wake_up_q(struct wake_q_head *head);
 986
 987/*
 988 * sched-domains (multiprocessor balancing) declarations:
 989 */
 990#ifdef CONFIG_SMP
 991#define SD_LOAD_BALANCE         0x0001  /* Do load balancing on this domain. */
 992#define SD_BALANCE_NEWIDLE      0x0002  /* Balance when about to become idle */
 993#define SD_BALANCE_EXEC         0x0004  /* Balance on exec */
 994#define SD_BALANCE_FORK         0x0008  /* Balance on fork, clone */
 995#define SD_BALANCE_WAKE         0x0010  /* Balance on wakeup */
 996#define SD_WAKE_AFFINE          0x0020  /* Wake task to waking CPU */
 997#define SD_SHARE_CPUCAPACITY    0x0080  /* Domain members share cpu power */
 998#define SD_SHARE_POWERDOMAIN    0x0100  /* Domain members share power domain */
 999#define SD_SHARE_PKG_RESOURCES  0x0200  /* Domain members share cpu pkg resources */
1000#define SD_SERIALIZE            0x0400  /* Only a single load balancing instance */
1001#define SD_ASYM_PACKING         0x0800  /* Place busy groups earlier in the domain */
1002#define SD_PREFER_SIBLING       0x1000  /* Prefer to place tasks in a sibling domain */
1003#define SD_OVERLAP              0x2000  /* sched_domains of this level overlap */
1004#define SD_NUMA                 0x4000  /* cross-node balancing */
1005
1006#ifdef CONFIG_SCHED_SMT
1007static inline int cpu_smt_flags(void)
1008{
1009        return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1010}
1011#endif
1012
1013#ifdef CONFIG_SCHED_MC
1014static inline int cpu_core_flags(void)
1015{
1016        return SD_SHARE_PKG_RESOURCES;
1017}
1018#endif
1019
1020#ifdef CONFIG_NUMA
1021static inline int cpu_numa_flags(void)
1022{
1023        return SD_NUMA;
1024}
1025#endif
1026
1027struct sched_domain_attr {
1028        int relax_domain_level;
1029};
1030
1031#define SD_ATTR_INIT    (struct sched_domain_attr) {    \
1032        .relax_domain_level = -1,                       \
1033}
1034
1035extern int sched_domain_level_max;
1036
1037struct sched_group;
1038
1039struct sched_domain {
1040        /* These fields must be setup */
1041        struct sched_domain *parent;    /* top domain must be null terminated */
1042        struct sched_domain *child;     /* bottom domain must be null terminated */
1043        struct sched_group *groups;     /* the balancing groups of the domain */
1044        unsigned long min_interval;     /* Minimum balance interval ms */
1045        unsigned long max_interval;     /* Maximum balance interval ms */
1046        unsigned int busy_factor;       /* less balancing by factor if busy */
1047        unsigned int imbalance_pct;     /* No balance until over watermark */
1048        unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
1049        unsigned int busy_idx;
1050        unsigned int idle_idx;
1051        unsigned int newidle_idx;
1052        unsigned int wake_idx;
1053        unsigned int forkexec_idx;
1054        unsigned int smt_gain;
1055
1056        int nohz_idle;                  /* NOHZ IDLE status */
1057        int flags;                      /* See SD_* */
1058        int level;
1059
1060        /* Runtime fields. */
1061        unsigned long last_balance;     /* init to jiffies. units in jiffies */
1062        unsigned int balance_interval;  /* initialise to 1. units in ms. */
1063        unsigned int nr_balance_failed; /* initialise to 0 */
1064
1065        /* idle_balance() stats */
1066        u64 max_newidle_lb_cost;
1067        unsigned long next_decay_max_lb_cost;
1068
1069#ifdef CONFIG_SCHEDSTATS
1070        /* load_balance() stats */
1071        unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1072        unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1073        unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1074        unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1075        unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1076        unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1077        unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1078        unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1079
1080        /* Active load balancing */
1081        unsigned int alb_count;
1082        unsigned int alb_failed;
1083        unsigned int alb_pushed;
1084
1085        /* SD_BALANCE_EXEC stats */
1086        unsigned int sbe_count;
1087        unsigned int sbe_balanced;
1088        unsigned int sbe_pushed;
1089
1090        /* SD_BALANCE_FORK stats */
1091        unsigned int sbf_count;
1092        unsigned int sbf_balanced;
1093        unsigned int sbf_pushed;
1094
1095        /* try_to_wake_up() stats */
1096        unsigned int ttwu_wake_remote;
1097        unsigned int ttwu_move_affine;
1098        unsigned int ttwu_move_balance;
1099#endif
1100#ifdef CONFIG_SCHED_DEBUG
1101        char *name;
1102#endif
1103        union {
1104                void *private;          /* used during construction */
1105                struct rcu_head rcu;    /* used during destruction */
1106        };
1107
1108        unsigned int span_weight;
1109        /*
1110         * Span of all CPUs in this domain.
1111         *
1112         * NOTE: this field is variable length. (Allocated dynamically
1113         * by attaching extra space to the end of the structure,
1114         * depending on how many CPUs the kernel has booted up with)
1115         */
1116        unsigned long span[0];
1117};
1118
1119static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1120{
1121        return to_cpumask(sd->span);
1122}
1123
1124extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1125                                    struct sched_domain_attr *dattr_new);
1126
1127/* Allocate an array of sched domains, for partition_sched_domains(). */
1128cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1129void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1130
1131bool cpus_share_cache(int this_cpu, int that_cpu);
1132
1133typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1134typedef int (*sched_domain_flags_f)(void);
1135
1136#define SDTL_OVERLAP    0x01
1137
1138struct sd_data {
1139        struct sched_domain **__percpu sd;
1140        struct sched_group **__percpu sg;
1141        struct sched_group_capacity **__percpu sgc;
1142};
1143
1144struct sched_domain_topology_level {
1145        sched_domain_mask_f mask;
1146        sched_domain_flags_f sd_flags;
1147        int                 flags;
1148        int                 numa_level;
1149        struct sd_data      data;
1150#ifdef CONFIG_SCHED_DEBUG
1151        char                *name;
1152#endif
1153};
1154
1155extern void set_sched_topology(struct sched_domain_topology_level *tl);
1156extern void wake_up_if_idle(int cpu);
1157
1158#ifdef CONFIG_SCHED_DEBUG
1159# define SD_INIT_NAME(type)             .name = #type
1160#else
1161# define SD_INIT_NAME(type)
1162#endif
1163
1164#else /* CONFIG_SMP */
1165
1166struct sched_domain_attr;
1167
1168static inline void
1169partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1170                        struct sched_domain_attr *dattr_new)
1171{
1172}
1173
1174static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1175{
1176        return true;
1177}
1178
1179#endif  /* !CONFIG_SMP */
1180
1181
1182struct io_context;                      /* See blkdev.h */
1183
1184
1185#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1186extern void prefetch_stack(struct task_struct *t);
1187#else
1188static inline void prefetch_stack(struct task_struct *t) { }
1189#endif
1190
1191struct audit_context;           /* See audit.c */
1192struct mempolicy;
1193struct pipe_inode_info;
1194struct uts_namespace;
1195
1196struct load_weight {
1197        unsigned long weight;
1198        u32 inv_weight;
1199};
1200
1201/*
1202 * The load_avg/util_avg accumulates an infinite geometric series.
1203 * 1) load_avg factors frequency scaling into the amount of time that a
1204 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1205 * aggregated such weights of all runnable and blocked sched_entities.
1206 * 2) util_avg factors frequency and cpu scaling into the amount of time
1207 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1208 * For cfs_rq, it is the aggregated such times of all runnable and
1209 * blocked sched_entities.
1210 * The 64 bit load_sum can:
1211 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1212 * the highest weight (=88761) always runnable, we should not overflow
1213 * 2) for entity, support any load.weight always runnable
1214 */
1215struct sched_avg {
1216        u64 last_update_time, load_sum;
1217        u32 util_sum, period_contrib;
1218        unsigned long load_avg, util_avg;
1219};
1220
1221#ifdef CONFIG_SCHEDSTATS
1222struct sched_statistics {
1223        u64                     wait_start;
1224        u64                     wait_max;
1225        u64                     wait_count;
1226        u64                     wait_sum;
1227        u64                     iowait_count;
1228        u64                     iowait_sum;
1229
1230        u64                     sleep_start;
1231        u64                     sleep_max;
1232        s64                     sum_sleep_runtime;
1233
1234        u64                     block_start;
1235        u64                     block_max;
1236        u64                     exec_max;
1237        u64                     slice_max;
1238
1239        u64                     nr_migrations_cold;
1240        u64                     nr_failed_migrations_affine;
1241        u64                     nr_failed_migrations_running;
1242        u64                     nr_failed_migrations_hot;
1243        u64                     nr_forced_migrations;
1244
1245        u64                     nr_wakeups;
1246        u64                     nr_wakeups_sync;
1247        u64                     nr_wakeups_migrate;
1248        u64                     nr_wakeups_local;
1249        u64                     nr_wakeups_remote;
1250        u64                     nr_wakeups_affine;
1251        u64                     nr_wakeups_affine_attempts;
1252        u64                     nr_wakeups_passive;
1253        u64                     nr_wakeups_idle;
1254};
1255#endif
1256
1257struct sched_entity {
1258        struct load_weight      load;           /* for load-balancing */
1259        struct rb_node          run_node;
1260        struct list_head        group_node;
1261        unsigned int            on_rq;
1262
1263        u64                     exec_start;
1264        u64                     sum_exec_runtime;
1265        u64                     vruntime;
1266        u64                     prev_sum_exec_runtime;
1267
1268        u64                     nr_migrations;
1269
1270#ifdef CONFIG_SCHEDSTATS
1271        struct sched_statistics statistics;
1272#endif
1273
1274#ifdef CONFIG_FAIR_GROUP_SCHED
1275        int                     depth;
1276        struct sched_entity     *parent;
1277        /* rq on which this entity is (to be) queued: */
1278        struct cfs_rq           *cfs_rq;
1279        /* rq "owned" by this entity/group: */
1280        struct cfs_rq           *my_q;
1281#endif
1282
1283#ifdef CONFIG_SMP
1284        /*
1285         * Per entity load average tracking.
1286         *
1287         * Put into separate cache line so it does not
1288         * collide with read-mostly values above.
1289         */
1290        struct sched_avg        avg ____cacheline_aligned_in_smp;
1291#endif
1292};
1293
1294struct sched_rt_entity {
1295        struct list_head run_list;
1296        unsigned long timeout;
1297        unsigned long watchdog_stamp;
1298        unsigned int time_slice;
1299        unsigned short on_rq;
1300        unsigned short on_list;
1301
1302        struct sched_rt_entity *back;
1303#ifdef CONFIG_RT_GROUP_SCHED
1304        struct sched_rt_entity  *parent;
1305        /* rq on which this entity is (to be) queued: */
1306        struct rt_rq            *rt_rq;
1307        /* rq "owned" by this entity/group: */
1308        struct rt_rq            *my_q;
1309#endif
1310};
1311
1312struct sched_dl_entity {
1313        struct rb_node  rb_node;
1314
1315        /*
1316         * Original scheduling parameters. Copied here from sched_attr
1317         * during sched_setattr(), they will remain the same until
1318         * the next sched_setattr().
1319         */
1320        u64 dl_runtime;         /* maximum runtime for each instance    */
1321        u64 dl_deadline;        /* relative deadline of each instance   */
1322        u64 dl_period;          /* separation of two instances (period) */
1323        u64 dl_bw;              /* dl_runtime / dl_deadline             */
1324
1325        /*
1326         * Actual scheduling parameters. Initialized with the values above,
1327         * they are continously updated during task execution. Note that
1328         * the remaining runtime could be < 0 in case we are in overrun.
1329         */
1330        s64 runtime;            /* remaining runtime for this instance  */
1331        u64 deadline;           /* absolute deadline for this instance  */
1332        unsigned int flags;     /* specifying the scheduler behaviour   */
1333
1334        /*
1335         * Some bool flags:
1336         *
1337         * @dl_throttled tells if we exhausted the runtime. If so, the
1338         * task has to wait for a replenishment to be performed at the
1339         * next firing of dl_timer.
1340         *
1341         * @dl_boosted tells if we are boosted due to DI. If so we are
1342         * outside bandwidth enforcement mechanism (but only until we
1343         * exit the critical section);
1344         *
1345         * @dl_yielded tells if task gave up the cpu before consuming
1346         * all its available runtime during the last job.
1347         */
1348        int dl_throttled, dl_boosted, dl_yielded;
1349
1350        /*
1351         * Bandwidth enforcement timer. Each -deadline task has its
1352         * own bandwidth to be enforced, thus we need one timer per task.
1353         */
1354        struct hrtimer dl_timer;
1355};
1356
1357union rcu_special {
1358        struct {
1359                u8 blocked;
1360                u8 need_qs;
1361                u8 exp_need_qs;
1362                u8 pad; /* Otherwise the compiler can store garbage here. */
1363        } b; /* Bits. */
1364        u32 s; /* Set of bits. */
1365};
1366struct rcu_node;
1367
1368enum perf_event_task_context {
1369        perf_invalid_context = -1,
1370        perf_hw_context = 0,
1371        perf_sw_context,
1372        perf_nr_task_contexts,
1373};
1374
1375/* Track pages that require TLB flushes */
1376struct tlbflush_unmap_batch {
1377        /*
1378         * Each bit set is a CPU that potentially has a TLB entry for one of
1379         * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1380         */
1381        struct cpumask cpumask;
1382
1383        /* True if any bit in cpumask is set */
1384        bool flush_required;
1385
1386        /*
1387         * If true then the PTE was dirty when unmapped. The entry must be
1388         * flushed before IO is initiated or a stale TLB entry potentially
1389         * allows an update without redirtying the page.
1390         */
1391        bool writable;
1392};
1393
1394struct task_struct {
1395        volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1396        void *stack;
1397        atomic_t usage;
1398        unsigned int flags;     /* per process flags, defined below */
1399        unsigned int ptrace;
1400
1401#ifdef CONFIG_SMP
1402        struct llist_node wake_entry;
1403        int on_cpu;
1404        unsigned int wakee_flips;
1405        unsigned long wakee_flip_decay_ts;
1406        struct task_struct *last_wakee;
1407
1408        int wake_cpu;
1409#endif
1410        int on_rq;
1411
1412        int prio, static_prio, normal_prio;
1413        unsigned int rt_priority;
1414        const struct sched_class *sched_class;
1415        struct sched_entity se;
1416        struct sched_rt_entity rt;
1417#ifdef CONFIG_CGROUP_SCHED
1418        struct task_group *sched_task_group;
1419#endif
1420        struct sched_dl_entity dl;
1421
1422#ifdef CONFIG_PREEMPT_NOTIFIERS
1423        /* list of struct preempt_notifier: */
1424        struct hlist_head preempt_notifiers;
1425#endif
1426
1427#ifdef CONFIG_BLK_DEV_IO_TRACE
1428        unsigned int btrace_seq;
1429#endif
1430
1431        unsigned int policy;
1432        int nr_cpus_allowed;
1433        cpumask_t cpus_allowed;
1434
1435#ifdef CONFIG_PREEMPT_RCU
1436        int rcu_read_lock_nesting;
1437        union rcu_special rcu_read_unlock_special;
1438        struct list_head rcu_node_entry;
1439        struct rcu_node *rcu_blocked_node;
1440#endif /* #ifdef CONFIG_PREEMPT_RCU */
1441#ifdef CONFIG_TASKS_RCU
1442        unsigned long rcu_tasks_nvcsw;
1443        bool rcu_tasks_holdout;
1444        struct list_head rcu_tasks_holdout_list;
1445        int rcu_tasks_idle_cpu;
1446#endif /* #ifdef CONFIG_TASKS_RCU */
1447
1448#ifdef CONFIG_SCHED_INFO
1449        struct sched_info sched_info;
1450#endif
1451
1452        struct list_head tasks;
1453#ifdef CONFIG_SMP
1454        struct plist_node pushable_tasks;
1455        struct rb_node pushable_dl_tasks;
1456#endif
1457
1458        struct mm_struct *mm, *active_mm;
1459        /* per-thread vma caching */
1460        u32 vmacache_seqnum;
1461        struct vm_area_struct *vmacache[VMACACHE_SIZE];
1462#if defined(SPLIT_RSS_COUNTING)
1463        struct task_rss_stat    rss_stat;
1464#endif
1465/* task state */
1466        int exit_state;
1467        int exit_code, exit_signal;
1468        int pdeath_signal;  /*  The signal sent when the parent dies  */
1469        unsigned long jobctl;   /* JOBCTL_*, siglock protected */
1470
1471        /* Used for emulating ABI behavior of previous Linux versions */
1472        unsigned int personality;
1473
1474        /* scheduler bits, serialized by scheduler locks */
1475        unsigned sched_reset_on_fork:1;
1476        unsigned sched_contributes_to_load:1;
1477        unsigned sched_migrated:1;
1478        unsigned :0; /* force alignment to the next boundary */
1479
1480        /* unserialized, strictly 'current' */
1481        unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1482        unsigned in_iowait:1;
1483#ifdef CONFIG_MEMCG
1484        unsigned memcg_may_oom:1;
1485#ifndef CONFIG_SLOB
1486        unsigned memcg_kmem_skip_account:1;
1487#endif
1488#endif
1489#ifdef CONFIG_COMPAT_BRK
1490        unsigned brk_randomized:1;
1491#endif
1492
1493        unsigned long atomic_flags; /* Flags needing atomic access. */
1494
1495        struct restart_block restart_block;
1496
1497        pid_t pid;
1498        pid_t tgid;
1499
1500#ifdef CONFIG_CC_STACKPROTECTOR
1501        /* Canary value for the -fstack-protector gcc feature */
1502        unsigned long stack_canary;
1503#endif
1504        /*
1505         * pointers to (original) parent process, youngest child, younger sibling,
1506         * older sibling, respectively.  (p->father can be replaced with
1507         * p->real_parent->pid)
1508         */
1509        struct task_struct __rcu *real_parent; /* real parent process */
1510        struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1511        /*
1512         * children/sibling forms the list of my natural children
1513         */
1514        struct list_head children;      /* list of my children */
1515        struct list_head sibling;       /* linkage in my parent's children list */
1516        struct task_struct *group_leader;       /* threadgroup leader */
1517
1518        /*
1519         * ptraced is the list of tasks this task is using ptrace on.
1520         * This includes both natural children and PTRACE_ATTACH targets.
1521         * p->ptrace_entry is p's link on the p->parent->ptraced list.
1522         */
1523        struct list_head ptraced;
1524        struct list_head ptrace_entry;
1525
1526        /* PID/PID hash table linkage. */
1527        struct pid_link pids[PIDTYPE_MAX];
1528        struct list_head thread_group;
1529        struct list_head thread_node;
1530
1531        struct completion *vfork_done;          /* for vfork() */
1532        int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1533        int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1534
1535        cputime_t utime, stime, utimescaled, stimescaled;
1536        cputime_t gtime;
1537        struct prev_cputime prev_cputime;
1538#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1539        seqcount_t vtime_seqcount;
1540        unsigned long long vtime_snap;
1541        enum {
1542                /* Task is sleeping or running in a CPU with VTIME inactive */
1543                VTIME_INACTIVE = 0,
1544                /* Task runs in userspace in a CPU with VTIME active */
1545                VTIME_USER,
1546                /* Task runs in kernelspace in a CPU with VTIME active */
1547                VTIME_SYS,
1548        } vtime_snap_whence;
1549#endif
1550
1551#ifdef CONFIG_NO_HZ_FULL
1552        atomic_t tick_dep_mask;
1553#endif
1554        unsigned long nvcsw, nivcsw; /* context switch counts */
1555        u64 start_time;         /* monotonic time in nsec */
1556        u64 real_start_time;    /* boot based time in nsec */
1557/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1558        unsigned long min_flt, maj_flt;
1559
1560        struct task_cputime cputime_expires;
1561        struct list_head cpu_timers[3];
1562
1563/* process credentials */
1564        const struct cred __rcu *real_cred; /* objective and real subjective task
1565                                         * credentials (COW) */
1566        const struct cred __rcu *cred;  /* effective (overridable) subjective task
1567                                         * credentials (COW) */
1568        char comm[TASK_COMM_LEN]; /* executable name excluding path
1569                                     - access with [gs]et_task_comm (which lock
1570                                       it with task_lock())
1571                                     - initialized normally by setup_new_exec */
1572/* file system info */
1573        struct nameidata *nameidata;
1574#ifdef CONFIG_SYSVIPC
1575/* ipc stuff */
1576        struct sysv_sem sysvsem;
1577        struct sysv_shm sysvshm;
1578#endif
1579#ifdef CONFIG_DETECT_HUNG_TASK
1580/* hung task detection */
1581        unsigned long last_switch_count;
1582#endif
1583/* filesystem information */
1584        struct fs_struct *fs;
1585/* open file information */
1586        struct files_struct *files;
1587/* namespaces */
1588        struct nsproxy *nsproxy;
1589/* signal handlers */
1590        struct signal_struct *signal;
1591        struct sighand_struct *sighand;
1592
1593        sigset_t blocked, real_blocked;
1594        sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1595        struct sigpending pending;
1596
1597        unsigned long sas_ss_sp;
1598        size_t sas_ss_size;
1599
1600        struct callback_head *task_works;
1601
1602        struct audit_context *audit_context;
1603#ifdef CONFIG_AUDITSYSCALL
1604        kuid_t loginuid;
1605        unsigned int sessionid;
1606#endif
1607        struct seccomp seccomp;
1608
1609/* Thread group tracking */
1610        u32 parent_exec_id;
1611        u32 self_exec_id;
1612/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1613 * mempolicy */
1614        spinlock_t alloc_lock;
1615
1616        /* Protection of the PI data structures: */
1617        raw_spinlock_t pi_lock;
1618
1619        struct wake_q_node wake_q;
1620
1621#ifdef CONFIG_RT_MUTEXES
1622        /* PI waiters blocked on a rt_mutex held by this task */
1623        struct rb_root pi_waiters;
1624        struct rb_node *pi_waiters_leftmost;
1625        /* Deadlock detection and priority inheritance handling */
1626        struct rt_mutex_waiter *pi_blocked_on;
1627#endif
1628
1629#ifdef CONFIG_DEBUG_MUTEXES
1630        /* mutex deadlock detection */
1631        struct mutex_waiter *blocked_on;
1632#endif
1633#ifdef CONFIG_TRACE_IRQFLAGS
1634        unsigned int irq_events;
1635        unsigned long hardirq_enable_ip;
1636        unsigned long hardirq_disable_ip;
1637        unsigned int hardirq_enable_event;
1638        unsigned int hardirq_disable_event;
1639        int hardirqs_enabled;
1640        int hardirq_context;
1641        unsigned long softirq_disable_ip;
1642        unsigned long softirq_enable_ip;
1643        unsigned int softirq_disable_event;
1644        unsigned int softirq_enable_event;
1645        int softirqs_enabled;
1646        int softirq_context;
1647#endif
1648#ifdef CONFIG_LOCKDEP
1649# define MAX_LOCK_DEPTH 48UL
1650        u64 curr_chain_key;
1651        int lockdep_depth;
1652        unsigned int lockdep_recursion;
1653        struct held_lock held_locks[MAX_LOCK_DEPTH];
1654        gfp_t lockdep_reclaim_gfp;
1655#endif
1656#ifdef CONFIG_UBSAN
1657        unsigned int in_ubsan;
1658#endif
1659
1660/* journalling filesystem info */
1661        void *journal_info;
1662
1663/* stacked block device info */
1664        struct bio_list *bio_list;
1665
1666#ifdef CONFIG_BLOCK
1667/* stack plugging */
1668        struct blk_plug *plug;
1669#endif
1670
1671/* VM state */
1672        struct reclaim_state *reclaim_state;
1673
1674        struct backing_dev_info *backing_dev_info;
1675
1676        struct io_context *io_context;
1677
1678        unsigned long ptrace_message;
1679        siginfo_t *last_siginfo; /* For ptrace use.  */
1680        struct task_io_accounting ioac;
1681#if defined(CONFIG_TASK_XACCT)
1682        u64 acct_rss_mem1;      /* accumulated rss usage */
1683        u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1684        cputime_t acct_timexpd; /* stime + utime since last update */
1685#endif
1686#ifdef CONFIG_CPUSETS
1687        nodemask_t mems_allowed;        /* Protected by alloc_lock */
1688        seqcount_t mems_allowed_seq;    /* Seqence no to catch updates */
1689        int cpuset_mem_spread_rotor;
1690        int cpuset_slab_spread_rotor;
1691#endif
1692#ifdef CONFIG_CGROUPS
1693        /* Control Group info protected by css_set_lock */
1694        struct css_set __rcu *cgroups;
1695        /* cg_list protected by css_set_lock and tsk->alloc_lock */
1696        struct list_head cg_list;
1697#endif
1698#ifdef CONFIG_FUTEX
1699        struct robust_list_head __user *robust_list;
1700#ifdef CONFIG_COMPAT
1701        struct compat_robust_list_head __user *compat_robust_list;
1702#endif
1703        struct list_head pi_state_list;
1704        struct futex_pi_state *pi_state_cache;
1705#endif
1706#ifdef CONFIG_PERF_EVENTS
1707        struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1708        struct mutex perf_event_mutex;
1709        struct list_head perf_event_list;
1710#endif
1711#ifdef CONFIG_DEBUG_PREEMPT
1712        unsigned long preempt_disable_ip;
1713#endif
1714#ifdef CONFIG_NUMA
1715        struct mempolicy *mempolicy;    /* Protected by alloc_lock */
1716        short il_next;
1717        short pref_node_fork;
1718#endif
1719#ifdef CONFIG_NUMA_BALANCING
1720        int numa_scan_seq;
1721        unsigned int numa_scan_period;
1722        unsigned int numa_scan_period_max;
1723        int numa_preferred_nid;
1724        unsigned long numa_migrate_retry;
1725        u64 node_stamp;                 /* migration stamp  */
1726        u64 last_task_numa_placement;
1727        u64 last_sum_exec_runtime;
1728        struct callback_head numa_work;
1729
1730        struct list_head numa_entry;
1731        struct numa_group *numa_group;
1732
1733        /*
1734         * numa_faults is an array split into four regions:
1735         * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1736         * in this precise order.
1737         *
1738         * faults_memory: Exponential decaying average of faults on a per-node
1739         * basis. Scheduling placement decisions are made based on these
1740         * counts. The values remain static for the duration of a PTE scan.
1741         * faults_cpu: Track the nodes the process was running on when a NUMA
1742         * hinting fault was incurred.
1743         * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1744         * during the current scan window. When the scan completes, the counts
1745         * in faults_memory and faults_cpu decay and these values are copied.
1746         */
1747        unsigned long *numa_faults;
1748        unsigned long total_numa_faults;
1749
1750        /*
1751         * numa_faults_locality tracks if faults recorded during the last
1752         * scan window were remote/local or failed to migrate. The task scan
1753         * period is adapted based on the locality of the faults with different
1754         * weights depending on whether they were shared or private faults
1755         */
1756        unsigned long numa_faults_locality[3];
1757
1758        unsigned long numa_pages_migrated;
1759#endif /* CONFIG_NUMA_BALANCING */
1760
1761#ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1762        struct tlbflush_unmap_batch tlb_ubc;
1763#endif
1764
1765        struct rcu_head rcu;
1766
1767        /*
1768         * cache last used pipe for splice
1769         */
1770        struct pipe_inode_info *splice_pipe;
1771
1772        struct page_frag task_frag;
1773
1774#ifdef  CONFIG_TASK_DELAY_ACCT
1775        struct task_delay_info *delays;
1776#endif
1777#ifdef CONFIG_FAULT_INJECTION
1778        int make_it_fail;
1779#endif
1780        /*
1781         * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1782         * balance_dirty_pages() for some dirty throttling pause
1783         */
1784        int nr_dirtied;
1785        int nr_dirtied_pause;
1786        unsigned long dirty_paused_when; /* start of a write-and-pause period */
1787
1788#ifdef CONFIG_LATENCYTOP
1789        int latency_record_count;
1790        struct latency_record latency_record[LT_SAVECOUNT];
1791#endif
1792        /*
1793         * time slack values; these are used to round up poll() and
1794         * select() etc timeout values. These are in nanoseconds.
1795         */
1796        u64 timer_slack_ns;
1797        u64 default_timer_slack_ns;
1798
1799#ifdef CONFIG_KASAN
1800        unsigned int kasan_depth;
1801#endif
1802#ifdef CONFIG_FUNCTION_GRAPH_TRACER
1803        /* Index of current stored address in ret_stack */
1804        int curr_ret_stack;
1805        /* Stack of return addresses for return function tracing */
1806        struct ftrace_ret_stack *ret_stack;
1807        /* time stamp for last schedule */
1808        unsigned long long ftrace_timestamp;
1809        /*
1810         * Number of functions that haven't been traced
1811         * because of depth overrun.
1812         */
1813        atomic_t trace_overrun;
1814        /* Pause for the tracing */
1815        atomic_t tracing_graph_pause;
1816#endif
1817#ifdef CONFIG_TRACING
1818        /* state flags for use by tracers */
1819        unsigned long trace;
1820        /* bitmask and counter of trace recursion */
1821        unsigned long trace_recursion;
1822#endif /* CONFIG_TRACING */
1823#ifdef CONFIG_KCOV
1824        /* Coverage collection mode enabled for this task (0 if disabled). */
1825        enum kcov_mode kcov_mode;
1826        /* Size of the kcov_area. */
1827        unsigned        kcov_size;
1828        /* Buffer for coverage collection. */
1829        void            *kcov_area;
1830        /* kcov desciptor wired with this task or NULL. */
1831        struct kcov     *kcov;
1832#endif
1833#ifdef CONFIG_MEMCG
1834        struct mem_cgroup *memcg_in_oom;
1835        gfp_t memcg_oom_gfp_mask;
1836        int memcg_oom_order;
1837
1838        /* number of pages to reclaim on returning to userland */
1839        unsigned int memcg_nr_pages_over_high;
1840#endif
1841#ifdef CONFIG_UPROBES
1842        struct uprobe_task *utask;
1843#endif
1844#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1845        unsigned int    sequential_io;
1846        unsigned int    sequential_io_avg;
1847#endif
1848#ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1849        unsigned long   task_state_change;
1850#endif
1851        int pagefault_disabled;
1852#ifdef CONFIG_MMU
1853        struct task_struct *oom_reaper_list;
1854#endif
1855/* CPU-specific state of this task */
1856        struct thread_struct thread;
1857/*
1858 * WARNING: on x86, 'thread_struct' contains a variable-sized
1859 * structure.  It *MUST* be at the end of 'task_struct'.
1860 *
1861 * Do not put anything below here!
1862 */
1863};
1864
1865#ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1866extern int arch_task_struct_size __read_mostly;
1867#else
1868# define arch_task_struct_size (sizeof(struct task_struct))
1869#endif
1870
1871/* Future-safe accessor for struct task_struct's cpus_allowed. */
1872#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1873
1874#define TNF_MIGRATED    0x01
1875#define TNF_NO_GROUP    0x02
1876#define TNF_SHARED      0x04
1877#define TNF_FAULT_LOCAL 0x08
1878#define TNF_MIGRATE_FAIL 0x10
1879
1880#ifdef CONFIG_NUMA_BALANCING
1881extern void task_numa_fault(int last_node, int node, int pages, int flags);
1882extern pid_t task_numa_group_id(struct task_struct *p);
1883extern void set_numabalancing_state(bool enabled);
1884extern void task_numa_free(struct task_struct *p);
1885extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1886                                        int src_nid, int dst_cpu);
1887#else
1888static inline void task_numa_fault(int last_node, int node, int pages,
1889                                   int flags)
1890{
1891}
1892static inline pid_t task_numa_group_id(struct task_struct *p)
1893{
1894        return 0;
1895}
1896static inline void set_numabalancing_state(bool enabled)
1897{
1898}
1899static inline void task_numa_free(struct task_struct *p)
1900{
1901}
1902static inline bool should_numa_migrate_memory(struct task_struct *p,
1903                                struct page *page, int src_nid, int dst_cpu)
1904{
1905        return true;
1906}
1907#endif
1908
1909static inline struct pid *task_pid(struct task_struct *task)
1910{
1911        return task->pids[PIDTYPE_PID].pid;
1912}
1913
1914static inline struct pid *task_tgid(struct task_struct *task)
1915{
1916        return task->group_leader->pids[PIDTYPE_PID].pid;
1917}
1918
1919/*
1920 * Without tasklist or rcu lock it is not safe to dereference
1921 * the result of task_pgrp/task_session even if task == current,
1922 * we can race with another thread doing sys_setsid/sys_setpgid.
1923 */
1924static inline struct pid *task_pgrp(struct task_struct *task)
1925{
1926        return task->group_leader->pids[PIDTYPE_PGID].pid;
1927}
1928
1929static inline struct pid *task_session(struct task_struct *task)
1930{
1931        return task->group_leader->pids[PIDTYPE_SID].pid;
1932}
1933
1934struct pid_namespace;
1935
1936/*
1937 * the helpers to get the task's different pids as they are seen
1938 * from various namespaces
1939 *
1940 * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1941 * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1942 *                     current.
1943 * task_xid_nr_ns()  : id seen from the ns specified;
1944 *
1945 * set_task_vxid()   : assigns a virtual id to a task;
1946 *
1947 * see also pid_nr() etc in include/linux/pid.h
1948 */
1949pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1950                        struct pid_namespace *ns);
1951
1952static inline pid_t task_pid_nr(struct task_struct *tsk)
1953{
1954        return tsk->pid;
1955}
1956
1957static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1958                                        struct pid_namespace *ns)
1959{
1960        return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1961}
1962
1963static inline pid_t task_pid_vnr(struct task_struct *tsk)
1964{
1965        return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1966}
1967
1968
1969static inline pid_t task_tgid_nr(struct task_struct *tsk)
1970{
1971        return tsk->tgid;
1972}
1973
1974pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1975
1976static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1977{
1978        return pid_vnr(task_tgid(tsk));
1979}
1980
1981
1982static inline int pid_alive(const struct task_struct *p);
1983static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1984{
1985        pid_t pid = 0;
1986
1987        rcu_read_lock();
1988        if (pid_alive(tsk))
1989                pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1990        rcu_read_unlock();
1991
1992        return pid;
1993}
1994
1995static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1996{
1997        return task_ppid_nr_ns(tsk, &init_pid_ns);
1998}
1999
2000static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
2001                                        struct pid_namespace *ns)
2002{
2003        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
2004}
2005
2006static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
2007{
2008        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
2009}
2010
2011
2012static inline pid_t task_session_nr_ns(struct task_struct *tsk,
2013                                        struct pid_namespace *ns)
2014{
2015        return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
2016}
2017
2018static inline pid_t task_session_vnr(struct task_struct *tsk)
2019{
2020        return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
2021}
2022
2023/* obsolete, do not use */
2024static inline pid_t task_pgrp_nr(struct task_struct *tsk)
2025{
2026        return task_pgrp_nr_ns(tsk, &init_pid_ns);
2027}
2028
2029/**
2030 * pid_alive - check that a task structure is not stale
2031 * @p: Task structure to be checked.
2032 *
2033 * Test if a process is not yet dead (at most zombie state)
2034 * If pid_alive fails, then pointers within the task structure
2035 * can be stale and must not be dereferenced.
2036 *
2037 * Return: 1 if the process is alive. 0 otherwise.
2038 */
2039static inline int pid_alive(const struct task_struct *p)
2040{
2041        return p->pids[PIDTYPE_PID].pid != NULL;
2042}
2043
2044/**
2045 * is_global_init - check if a task structure is init. Since init
2046 * is free to have sub-threads we need to check tgid.
2047 * @tsk: Task structure to be checked.
2048 *
2049 * Check if a task structure is the first user space task the kernel created.
2050 *
2051 * Return: 1 if the task structure is init. 0 otherwise.
2052 */
2053static inline int is_global_init(struct task_struct *tsk)
2054{
2055        return task_tgid_nr(tsk) == 1;
2056}
2057
2058extern struct pid *cad_pid;
2059
2060extern void free_task(struct task_struct *tsk);
2061#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2062
2063extern void __put_task_struct(struct task_struct *t);
2064
2065static inline void put_task_struct(struct task_struct *t)
2066{
2067        if (atomic_dec_and_test(&t->usage))
2068                __put_task_struct(t);
2069}
2070
2071#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2072extern void task_cputime(struct task_struct *t,
2073                         cputime_t *utime, cputime_t *stime);
2074extern void task_cputime_scaled(struct task_struct *t,
2075                                cputime_t *utimescaled, cputime_t *stimescaled);
2076extern cputime_t task_gtime(struct task_struct *t);
2077#else
2078static inline void task_cputime(struct task_struct *t,
2079                                cputime_t *utime, cputime_t *stime)
2080{
2081        if (utime)
2082                *utime = t->utime;
2083        if (stime)
2084                *stime = t->stime;
2085}
2086
2087static inline void task_cputime_scaled(struct task_struct *t,
2088                                       cputime_t *utimescaled,
2089                                       cputime_t *stimescaled)
2090{
2091        if (utimescaled)
2092                *utimescaled = t->utimescaled;
2093        if (stimescaled)
2094                *stimescaled = t->stimescaled;
2095}
2096
2097static inline cputime_t task_gtime(struct task_struct *t)
2098{
2099        return t->gtime;
2100}
2101#endif
2102extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2103extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2104
2105/*
2106 * Per process flags
2107 */
2108#define PF_EXITING      0x00000004      /* getting shut down */
2109#define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
2110#define PF_VCPU         0x00000010      /* I'm a virtual CPU */
2111#define PF_WQ_WORKER    0x00000020      /* I'm a workqueue worker */
2112#define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
2113#define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
2114#define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
2115#define PF_DUMPCORE     0x00000200      /* dumped core */
2116#define PF_SIGNALED     0x00000400      /* killed by a signal */
2117#define PF_MEMALLOC     0x00000800      /* Allocating memory */
2118#define PF_NPROC_EXCEEDED 0x00001000    /* set_user noticed that RLIMIT_NPROC was exceeded */
2119#define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
2120#define PF_USED_ASYNC   0x00004000      /* used async_schedule*(), used by module init */
2121#define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
2122#define PF_FROZEN       0x00010000      /* frozen for system suspend */
2123#define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
2124#define PF_KSWAPD       0x00040000      /* I am kswapd */
2125#define PF_MEMALLOC_NOIO 0x00080000     /* Allocating memory without IO involved */
2126#define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
2127#define PF_KTHREAD      0x00200000      /* I am a kernel thread */
2128#define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
2129#define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
2130#define PF_NO_SETAFFINITY 0x04000000    /* Userland is not allowed to meddle with cpus_allowed */
2131#define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
2132#define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
2133#define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezable */
2134#define PF_SUSPEND_TASK 0x80000000      /* this thread called freeze_processes and should not be frozen */
2135
2136/*
2137 * Only the _current_ task can read/write to tsk->flags, but other
2138 * tasks can access tsk->flags in readonly mode for example
2139 * with tsk_used_math (like during threaded core dumping).
2140 * There is however an exception to this rule during ptrace
2141 * or during fork: the ptracer task is allowed to write to the
2142 * child->flags of its traced child (same goes for fork, the parent
2143 * can write to the child->flags), because we're guaranteed the
2144 * child is not running and in turn not changing child->flags
2145 * at the same time the parent does it.
2146 */
2147#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2148#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2149#define clear_used_math() clear_stopped_child_used_math(current)
2150#define set_used_math() set_stopped_child_used_math(current)
2151#define conditional_stopped_child_used_math(condition, child) \
2152        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2153#define conditional_used_math(condition) \
2154        conditional_stopped_child_used_math(condition, current)
2155#define copy_to_stopped_child_used_math(child) \
2156        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2157/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2158#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2159#define used_math() tsk_used_math(current)
2160
2161/* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2162 * __GFP_FS is also cleared as it implies __GFP_IO.
2163 */
2164static inline gfp_t memalloc_noio_flags(gfp_t flags)
2165{
2166        if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2167                flags &= ~(__GFP_IO | __GFP_FS);
2168        return flags;
2169}
2170
2171static inline unsigned int memalloc_noio_save(void)
2172{
2173        unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2174        current->flags |= PF_MEMALLOC_NOIO;
2175        return flags;
2176}
2177
2178static inline void memalloc_noio_restore(unsigned int flags)
2179{
2180        current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2181}
2182
2183/* Per-process atomic flags. */
2184#define PFA_NO_NEW_PRIVS 0      /* May not gain new privileges. */
2185#define PFA_SPREAD_PAGE  1      /* Spread page cache over cpuset */
2186#define PFA_SPREAD_SLAB  2      /* Spread some slab caches over cpuset */
2187
2188
2189#define TASK_PFA_TEST(name, func)                                       \
2190        static inline bool task_##func(struct task_struct *p)           \
2191        { return test_bit(PFA_##name, &p->atomic_flags); }
2192#define TASK_PFA_SET(name, func)                                        \
2193        static inline void task_set_##func(struct task_struct *p)       \
2194        { set_bit(PFA_##name, &p->atomic_flags); }
2195#define TASK_PFA_CLEAR(name, func)                                      \
2196        static inline void task_clear_##func(struct task_struct *p)     \
2197        { clear_bit(PFA_##name, &p->atomic_flags); }
2198
2199TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2200TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2201
2202TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2203TASK_PFA_SET(SPREAD_PAGE, spread_page)
2204TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2205
2206TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2207TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2208TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2209
2210/*
2211 * task->jobctl flags
2212 */
2213#define JOBCTL_STOP_SIGMASK     0xffff  /* signr of the last group stop */
2214
2215#define JOBCTL_STOP_DEQUEUED_BIT 16     /* stop signal dequeued */
2216#define JOBCTL_STOP_PENDING_BIT 17      /* task should stop for group stop */
2217#define JOBCTL_STOP_CONSUME_BIT 18      /* consume group stop count */
2218#define JOBCTL_TRAP_STOP_BIT    19      /* trap for STOP */
2219#define JOBCTL_TRAP_NOTIFY_BIT  20      /* trap for NOTIFY */
2220#define JOBCTL_TRAPPING_BIT     21      /* switching to TRACED */
2221#define JOBCTL_LISTENING_BIT    22      /* ptracer is listening for events */
2222
2223#define JOBCTL_STOP_DEQUEUED    (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2224#define JOBCTL_STOP_PENDING     (1UL << JOBCTL_STOP_PENDING_BIT)
2225#define JOBCTL_STOP_CONSUME     (1UL << JOBCTL_STOP_CONSUME_BIT)
2226#define JOBCTL_TRAP_STOP        (1UL << JOBCTL_TRAP_STOP_BIT)
2227#define JOBCTL_TRAP_NOTIFY      (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2228#define JOBCTL_TRAPPING         (1UL << JOBCTL_TRAPPING_BIT)
2229#define JOBCTL_LISTENING        (1UL << JOBCTL_LISTENING_BIT)
2230
2231#define JOBCTL_TRAP_MASK        (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2232#define JOBCTL_PENDING_MASK     (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2233
2234extern bool task_set_jobctl_pending(struct task_struct *task,
2235                                    unsigned long mask);
2236extern void task_clear_jobctl_trapping(struct task_struct *task);
2237extern void task_clear_jobctl_pending(struct task_struct *task,
2238                                      unsigned long mask);
2239
2240static inline void rcu_copy_process(struct task_struct *p)
2241{
2242#ifdef CONFIG_PREEMPT_RCU
2243        p->rcu_read_lock_nesting = 0;
2244        p->rcu_read_unlock_special.s = 0;
2245        p->rcu_blocked_node = NULL;
2246        INIT_LIST_HEAD(&p->rcu_node_entry);
2247#endif /* #ifdef CONFIG_PREEMPT_RCU */
2248#ifdef CONFIG_TASKS_RCU
2249        p->rcu_tasks_holdout = false;
2250        INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2251        p->rcu_tasks_idle_cpu = -1;
2252#endif /* #ifdef CONFIG_TASKS_RCU */
2253}
2254
2255static inline void tsk_restore_flags(struct task_struct *task,
2256                                unsigned long orig_flags, unsigned long flags)
2257{
2258        task->flags &= ~flags;
2259        task->flags |= orig_flags & flags;
2260}
2261
2262extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2263                                     const struct cpumask *trial);
2264extern int task_can_attach(struct task_struct *p,
2265                           const struct cpumask *cs_cpus_allowed);
2266#ifdef CONFIG_SMP
2267extern void do_set_cpus_allowed(struct task_struct *p,
2268                               const struct cpumask *new_mask);
2269
2270extern int set_cpus_allowed_ptr(struct task_struct *p,
2271                                const struct cpumask *new_mask);
2272#else
2273static inline void do_set_cpus_allowed(struct task_struct *p,
2274                                      const struct cpumask *new_mask)
2275{
2276}
2277static inline int set_cpus_allowed_ptr(struct task_struct *p,
2278                                       const struct cpumask *new_mask)
2279{
2280        if (!cpumask_test_cpu(0, new_mask))
2281                return -EINVAL;
2282        return 0;
2283}
2284#endif
2285
2286#ifdef CONFIG_NO_HZ_COMMON
2287void calc_load_enter_idle(void);
2288void calc_load_exit_idle(void);
2289#else
2290static inline void calc_load_enter_idle(void) { }
2291static inline void calc_load_exit_idle(void) { }
2292#endif /* CONFIG_NO_HZ_COMMON */
2293
2294/*
2295 * Do not use outside of architecture code which knows its limitations.
2296 *
2297 * sched_clock() has no promise of monotonicity or bounded drift between
2298 * CPUs, use (which you should not) requires disabling IRQs.
2299 *
2300 * Please use one of the three interfaces below.
2301 */
2302extern unsigned long long notrace sched_clock(void);
2303/*
2304 * See the comment in kernel/sched/clock.c
2305 */
2306extern u64 cpu_clock(int cpu);
2307extern u64 local_clock(void);
2308extern u64 running_clock(void);
2309extern u64 sched_clock_cpu(int cpu);
2310
2311
2312extern void sched_clock_init(void);
2313
2314#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2315static inline void sched_clock_tick(void)
2316{
2317}
2318
2319static inline void sched_clock_idle_sleep_event(void)
2320{
2321}
2322
2323static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2324{
2325}
2326#else
2327/*
2328 * Architectures can set this to 1 if they have specified
2329 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2330 * but then during bootup it turns out that sched_clock()
2331 * is reliable after all:
2332 */
2333extern int sched_clock_stable(void);
2334extern void set_sched_clock_stable(void);
2335extern void clear_sched_clock_stable(void);
2336
2337extern void sched_clock_tick(void);
2338extern void sched_clock_idle_sleep_event(void);
2339extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2340#endif
2341
2342#ifdef CONFIG_IRQ_TIME_ACCOUNTING
2343/*
2344 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2345 * The reason for this explicit opt-in is not to have perf penalty with
2346 * slow sched_clocks.
2347 */
2348extern void enable_sched_clock_irqtime(void);
2349extern void disable_sched_clock_irqtime(void);
2350#else
2351static inline void enable_sched_clock_irqtime(void) {}
2352static inline void disable_sched_clock_irqtime(void) {}
2353#endif
2354
2355extern unsigned long long
2356task_sched_runtime(struct task_struct *task);
2357
2358/* sched_exec is called by processes performing an exec */
2359#ifdef CONFIG_SMP
2360extern void sched_exec(void);
2361#else
2362#define sched_exec()   {}
2363#endif
2364
2365extern void sched_clock_idle_sleep_event(void);
2366extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2367
2368#ifdef CONFIG_HOTPLUG_CPU
2369extern void idle_task_exit(void);
2370#else
2371static inline void idle_task_exit(void) {}
2372#endif
2373
2374#if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2375extern void wake_up_nohz_cpu(int cpu);
2376#else
2377static inline void wake_up_nohz_cpu(int cpu) { }
2378#endif
2379
2380#ifdef CONFIG_NO_HZ_FULL
2381extern u64 scheduler_tick_max_deferment(void);
2382#endif
2383
2384#ifdef CONFIG_SCHED_AUTOGROUP
2385extern void sched_autogroup_create_attach(struct task_struct *p);
2386extern void sched_autogroup_detach(struct task_struct *p);
2387extern void sched_autogroup_fork(struct signal_struct *sig);
2388extern void sched_autogroup_exit(struct signal_struct *sig);
2389#ifdef CONFIG_PROC_FS
2390extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2391extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2392#endif
2393#else
2394static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2395static inline void sched_autogroup_detach(struct task_struct *p) { }
2396static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2397static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2398#endif
2399
2400extern int yield_to(struct task_struct *p, bool preempt);
2401extern void set_user_nice(struct task_struct *p, long nice);
2402extern int task_prio(const struct task_struct *p);
2403/**
2404 * task_nice - return the nice value of a given task.
2405 * @p: the task in question.
2406 *
2407 * Return: The nice value [ -20 ... 0 ... 19 ].
2408 */
2409static inline int task_nice(const struct task_struct *p)
2410{
2411        return PRIO_TO_NICE((p)->static_prio);
2412}
2413extern int can_nice(const struct task_struct *p, const int nice);
2414extern int task_curr(const struct task_struct *p);
2415extern int idle_cpu(int cpu);
2416extern int sched_setscheduler(struct task_struct *, int,
2417                              const struct sched_param *);
2418extern int sched_setscheduler_nocheck(struct task_struct *, int,
2419                                      const struct sched_param *);
2420extern int sched_setattr(struct task_struct *,
2421                         const struct sched_attr *);
2422extern struct task_struct *idle_task(int cpu);
2423/**
2424 * is_idle_task - is the specified task an idle task?
2425 * @p: the task in question.
2426 *
2427 * Return: 1 if @p is an idle task. 0 otherwise.
2428 */
2429static inline bool is_idle_task(const struct task_struct *p)
2430{
2431        return p->pid == 0;
2432}
2433extern struct task_struct *curr_task(int cpu);
2434extern void set_curr_task(int cpu, struct task_struct *p);
2435
2436void yield(void);
2437
2438union thread_union {
2439        struct thread_info thread_info;
2440        unsigned long stack[THREAD_SIZE/sizeof(long)];
2441};
2442
2443#ifndef __HAVE_ARCH_KSTACK_END
2444static inline int kstack_end(void *addr)
2445{
2446        /* Reliable end of stack detection:
2447         * Some APM bios versions misalign the stack
2448         */
2449        return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2450}
2451#endif
2452
2453extern union thread_union init_thread_union;
2454extern struct task_struct init_task;
2455
2456extern struct   mm_struct init_mm;
2457
2458extern struct pid_namespace init_pid_ns;
2459
2460/*
2461 * find a task by one of its numerical ids
2462 *
2463 * find_task_by_pid_ns():
2464 *      finds a task by its pid in the specified namespace
2465 * find_task_by_vpid():
2466 *      finds a task by its virtual pid
2467 *
2468 * see also find_vpid() etc in include/linux/pid.h
2469 */
2470
2471extern struct task_struct *find_task_by_vpid(pid_t nr);
2472extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2473                struct pid_namespace *ns);
2474
2475/* per-UID process charging. */
2476extern struct user_struct * alloc_uid(kuid_t);
2477static inline struct user_struct *get_uid(struct user_struct *u)
2478{
2479        atomic_inc(&u->__count);
2480        return u;
2481}
2482extern void free_uid(struct user_struct *);
2483
2484#include <asm/current.h>
2485
2486extern void xtime_update(unsigned long ticks);
2487
2488extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2489extern int wake_up_process(struct task_struct *tsk);
2490extern void wake_up_new_task(struct task_struct *tsk);
2491#ifdef CONFIG_SMP
2492 extern void kick_process(struct task_struct *tsk);
2493#else
2494 static inline void kick_process(struct task_struct *tsk) { }
2495#endif
2496extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2497extern void sched_dead(struct task_struct *p);
2498
2499extern void proc_caches_init(void);
2500extern void flush_signals(struct task_struct *);
2501extern void ignore_signals(struct task_struct *);
2502extern void flush_signal_handlers(struct task_struct *, int force_default);
2503extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2504
2505static inline int kernel_dequeue_signal(siginfo_t *info)
2506{
2507        struct task_struct *tsk = current;
2508        siginfo_t __info;
2509        int ret;
2510
2511        spin_lock_irq(&tsk->sighand->siglock);
2512        ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2513        spin_unlock_irq(&tsk->sighand->siglock);
2514
2515        return ret;
2516}
2517
2518static inline void kernel_signal_stop(void)
2519{
2520        spin_lock_irq(&current->sighand->siglock);
2521        if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2522                __set_current_state(TASK_STOPPED);
2523        spin_unlock_irq(&current->sighand->siglock);
2524
2525        schedule();
2526}
2527
2528extern void release_task(struct task_struct * p);
2529extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2530extern int force_sigsegv(int, struct task_struct *);
2531extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2532extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2533extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2534extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2535                                const struct cred *, u32);
2536extern int kill_pgrp(struct pid *pid, int sig, int priv);
2537extern int kill_pid(struct pid *pid, int sig, int priv);
2538extern int kill_proc_info(int, struct siginfo *, pid_t);
2539extern __must_check bool do_notify_parent(struct task_struct *, int);
2540extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2541extern void force_sig(int, struct task_struct *);
2542extern int send_sig(int, struct task_struct *, int);
2543extern int zap_other_threads(struct task_struct *p);
2544extern struct sigqueue *sigqueue_alloc(void);
2545extern void sigqueue_free(struct sigqueue *);
2546extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2547extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2548
2549static inline void restore_saved_sigmask(void)
2550{
2551        if (test_and_clear_restore_sigmask())
2552                __set_current_blocked(&current->saved_sigmask);
2553}
2554
2555static inline sigset_t *sigmask_to_save(void)
2556{
2557        sigset_t *res = &current->blocked;
2558        if (unlikely(test_restore_sigmask()))
2559                res = &current->saved_sigmask;
2560        return res;
2561}
2562
2563static inline int kill_cad_pid(int sig, int priv)
2564{
2565        return kill_pid(cad_pid, sig, priv);
2566}
2567
2568/* These can be the second arg to send_sig_info/send_group_sig_info.  */
2569#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2570#define SEND_SIG_PRIV   ((struct siginfo *) 1)
2571#define SEND_SIG_FORCED ((struct siginfo *) 2)
2572
2573/*
2574 * True if we are on the alternate signal stack.
2575 */
2576static inline int on_sig_stack(unsigned long sp)
2577{
2578#ifdef CONFIG_STACK_GROWSUP
2579        return sp >= current->sas_ss_sp &&
2580                sp - current->sas_ss_sp < current->sas_ss_size;
2581#else
2582        return sp > current->sas_ss_sp &&
2583                sp - current->sas_ss_sp <= current->sas_ss_size;
2584#endif
2585}
2586
2587static inline int sas_ss_flags(unsigned long sp)
2588{
2589        if (!current->sas_ss_size)
2590                return SS_DISABLE;
2591
2592        return on_sig_stack(sp) ? SS_ONSTACK : 0;
2593}
2594
2595static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2596{
2597        if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2598#ifdef CONFIG_STACK_GROWSUP
2599                return current->sas_ss_sp;
2600#else
2601                return current->sas_ss_sp + current->sas_ss_size;
2602#endif
2603        return sp;
2604}
2605
2606/*
2607 * Routines for handling mm_structs
2608 */
2609extern struct mm_struct * mm_alloc(void);
2610
2611/* mmdrop drops the mm and the page tables */
2612extern void __mmdrop(struct mm_struct *);
2613static inline void mmdrop(struct mm_struct * mm)
2614{
2615        if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2616                __mmdrop(mm);
2617}
2618
2619/* mmput gets rid of the mappings and all user-space */
2620extern void mmput(struct mm_struct *);
2621/* Grab a reference to a task's mm, if it is not already going away */
2622extern struct mm_struct *get_task_mm(struct task_struct *task);
2623/*
2624 * Grab a reference to a task's mm, if it is not already going away
2625 * and ptrace_may_access with the mode parameter passed to it
2626 * succeeds.
2627 */
2628extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2629/* Remove the current tasks stale references to the old mm_struct */
2630extern void mm_release(struct task_struct *, struct mm_struct *);
2631
2632#ifdef CONFIG_HAVE_COPY_THREAD_TLS
2633extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2634                        struct task_struct *, unsigned long);
2635#else
2636extern int copy_thread(unsigned long, unsigned long, unsigned long,
2637                        struct task_struct *);
2638
2639/* Architectures that haven't opted into copy_thread_tls get the tls argument
2640 * via pt_regs, so ignore the tls argument passed via C. */
2641static inline int copy_thread_tls(
2642                unsigned long clone_flags, unsigned long sp, unsigned long arg,
2643                struct task_struct *p, unsigned long tls)
2644{
2645        return copy_thread(clone_flags, sp, arg, p);
2646}
2647#endif
2648extern void flush_thread(void);
2649extern void exit_thread(void);
2650
2651extern void exit_files(struct task_struct *);
2652extern void __cleanup_sighand(struct sighand_struct *);
2653
2654extern void exit_itimers(struct signal_struct *);
2655extern void flush_itimer_signals(void);
2656
2657extern void do_group_exit(int);
2658
2659extern int do_execve(struct filename *,
2660                     const char __user * const __user *,
2661                     const char __user * const __user *);
2662extern int do_execveat(int, struct filename *,
2663                       const char __user * const __user *,
2664                       const char __user * const __user *,
2665                       int);
2666extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2667extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2668struct task_struct *fork_idle(int);
2669extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2670
2671extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2672static inline void set_task_comm(struct task_struct *tsk, const char *from)
2673{
2674        __set_task_comm(tsk, from, false);
2675}
2676extern char *get_task_comm(char *to, struct task_struct *tsk);
2677
2678#ifdef CONFIG_SMP
2679void scheduler_ipi(void);
2680extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2681#else
2682static inline void scheduler_ipi(void) { }
2683static inline unsigned long wait_task_inactive(struct task_struct *p,
2684                                               long match_state)
2685{
2686        return 1;
2687}
2688#endif
2689
2690#define tasklist_empty() \
2691        list_empty(&init_task.tasks)
2692
2693#define next_task(p) \
2694        list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2695
2696#define for_each_process(p) \
2697        for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2698
2699extern bool current_is_single_threaded(void);
2700
2701/*
2702 * Careful: do_each_thread/while_each_thread is a double loop so
2703 *          'break' will not work as expected - use goto instead.
2704 */
2705#define do_each_thread(g, t) \
2706        for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2707
2708#define while_each_thread(g, t) \
2709        while ((t = next_thread(t)) != g)
2710
2711#define __for_each_thread(signal, t)    \
2712        list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2713
2714#define for_each_thread(p, t)           \
2715        __for_each_thread((p)->signal, t)
2716
2717/* Careful: this is a double loop, 'break' won't work as expected. */
2718#define for_each_process_thread(p, t)   \
2719        for_each_process(p) for_each_thread(p, t)
2720
2721static inline int get_nr_threads(struct task_struct *tsk)
2722{
2723        return tsk->signal->nr_threads;
2724}
2725
2726static inline bool thread_group_leader(struct task_struct *p)
2727{
2728        return p->exit_signal >= 0;
2729}
2730
2731/* Do to the insanities of de_thread it is possible for a process
2732 * to have the pid of the thread group leader without actually being
2733 * the thread group leader.  For iteration through the pids in proc
2734 * all we care about is that we have a task with the appropriate
2735 * pid, we don't actually care if we have the right task.
2736 */
2737static inline bool has_group_leader_pid(struct task_struct *p)
2738{
2739        return task_pid(p) == p->signal->leader_pid;
2740}
2741
2742static inline
2743bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2744{
2745        return p1->signal == p2->signal;
2746}
2747
2748static inline struct task_struct *next_thread(const struct task_struct *p)
2749{
2750        return list_entry_rcu(p->thread_group.next,
2751                              struct task_struct, thread_group);
2752}
2753
2754static inline int thread_group_empty(struct task_struct *p)
2755{
2756        return list_empty(&p->thread_group);
2757}
2758
2759#define delay_group_leader(p) \
2760                (thread_group_leader(p) && !thread_group_empty(p))
2761
2762/*
2763 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2764 * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2765 * pins the final release of task.io_context.  Also protects ->cpuset and
2766 * ->cgroup.subsys[]. And ->vfork_done.
2767 *
2768 * Nests both inside and outside of read_lock(&tasklist_lock).
2769 * It must not be nested with write_lock_irq(&tasklist_lock),
2770 * neither inside nor outside.
2771 */
2772static inline void task_lock(struct task_struct *p)
2773{
2774        spin_lock(&p->alloc_lock);
2775}
2776
2777static inline void task_unlock(struct task_struct *p)
2778{
2779        spin_unlock(&p->alloc_lock);
2780}
2781
2782extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2783                                                        unsigned long *flags);
2784
2785static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2786                                                       unsigned long *flags)
2787{
2788        struct sighand_struct *ret;
2789
2790        ret = __lock_task_sighand(tsk, flags);
2791        (void)__cond_lock(&tsk->sighand->siglock, ret);
2792        return ret;
2793}
2794
2795static inline void unlock_task_sighand(struct task_struct *tsk,
2796                                                unsigned long *flags)
2797{
2798        spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2799}
2800
2801/**
2802 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2803 * @tsk: task causing the changes
2804 *
2805 * All operations which modify a threadgroup - a new thread joining the
2806 * group, death of a member thread (the assertion of PF_EXITING) and
2807 * exec(2) dethreading the process and replacing the leader - are wrapped
2808 * by threadgroup_change_{begin|end}().  This is to provide a place which
2809 * subsystems needing threadgroup stability can hook into for
2810 * synchronization.
2811 */
2812static inline void threadgroup_change_begin(struct task_struct *tsk)
2813{
2814        might_sleep();
2815        cgroup_threadgroup_change_begin(tsk);
2816}
2817
2818/**
2819 * threadgroup_change_end - mark the end of changes to a threadgroup
2820 * @tsk: task causing the changes
2821 *
2822 * See threadgroup_change_begin().
2823 */
2824static inline void threadgroup_change_end(struct task_struct *tsk)
2825{
2826        cgroup_threadgroup_change_end(tsk);
2827}
2828
2829#ifndef __HAVE_THREAD_FUNCTIONS
2830
2831#define task_thread_info(task)  ((struct thread_info *)(task)->stack)
2832#define task_stack_page(task)   ((task)->stack)
2833
2834static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2835{
2836        *task_thread_info(p) = *task_thread_info(org);
2837        task_thread_info(p)->task = p;
2838}
2839
2840/*
2841 * Return the address of the last usable long on the stack.
2842 *
2843 * When the stack grows down, this is just above the thread
2844 * info struct. Going any lower will corrupt the threadinfo.
2845 *
2846 * When the stack grows up, this is the highest address.
2847 * Beyond that position, we corrupt data on the next page.
2848 */
2849static inline unsigned long *end_of_stack(struct task_struct *p)
2850{
2851#ifdef CONFIG_STACK_GROWSUP
2852        return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2853#else
2854        return (unsigned long *)(task_thread_info(p) + 1);
2855#endif
2856}
2857
2858#endif
2859#define task_stack_end_corrupted(task) \
2860                (*(end_of_stack(task)) != STACK_END_MAGIC)
2861
2862static inline int object_is_on_stack(void *obj)
2863{
2864        void *stack = task_stack_page(current);
2865
2866        return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2867}
2868
2869extern void thread_info_cache_init(void);
2870
2871#ifdef CONFIG_DEBUG_STACK_USAGE
2872static inline unsigned long stack_not_used(struct task_struct *p)
2873{
2874        unsigned long *n = end_of_stack(p);
2875
2876        do {    /* Skip over canary */
2877# ifdef CONFIG_STACK_GROWSUP
2878                n--;
2879# else
2880                n++;
2881# endif
2882        } while (!*n);
2883
2884# ifdef CONFIG_STACK_GROWSUP
2885        return (unsigned long)end_of_stack(p) - (unsigned long)n;
2886# else
2887        return (unsigned long)n - (unsigned long)end_of_stack(p);
2888# endif
2889}
2890#endif
2891extern void set_task_stack_end_magic(struct task_struct *tsk);
2892
2893/* set thread flags in other task's structures
2894 * - see asm/thread_info.h for TIF_xxxx flags available
2895 */
2896static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2897{
2898        set_ti_thread_flag(task_thread_info(tsk), flag);
2899}
2900
2901static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2902{
2903        clear_ti_thread_flag(task_thread_info(tsk), flag);
2904}
2905
2906static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2907{
2908        return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2909}
2910
2911static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2912{
2913        return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2914}
2915
2916static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2917{
2918        return test_ti_thread_flag(task_thread_info(tsk), flag);
2919}
2920
2921static inline void set_tsk_need_resched(struct task_struct *tsk)
2922{
2923        set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2924}
2925
2926static inline void clear_tsk_need_resched(struct task_struct *tsk)
2927{
2928        clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2929}
2930
2931static inline int test_tsk_need_resched(struct task_struct *tsk)
2932{
2933        return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2934}
2935
2936static inline int restart_syscall(void)
2937{
2938        set_tsk_thread_flag(current, TIF_SIGPENDING);
2939        return -ERESTARTNOINTR;
2940}
2941
2942static inline int signal_pending(struct task_struct *p)
2943{
2944        return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2945}
2946
2947static inline int __fatal_signal_pending(struct task_struct *p)
2948{
2949        return unlikely(sigismember(&p->pending.signal, SIGKILL));
2950}
2951
2952static inline int fatal_signal_pending(struct task_struct *p)
2953{
2954        return signal_pending(p) && __fatal_signal_pending(p);
2955}
2956
2957static inline int signal_pending_state(long state, struct task_struct *p)
2958{
2959        if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2960                return 0;
2961        if (!signal_pending(p))
2962                return 0;
2963
2964        return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2965}
2966
2967/*
2968 * cond_resched() and cond_resched_lock(): latency reduction via
2969 * explicit rescheduling in places that are safe. The return
2970 * value indicates whether a reschedule was done in fact.
2971 * cond_resched_lock() will drop the spinlock before scheduling,
2972 * cond_resched_softirq() will enable bhs before scheduling.
2973 */
2974extern int _cond_resched(void);
2975
2976#define cond_resched() ({                       \
2977        ___might_sleep(__FILE__, __LINE__, 0);  \
2978        _cond_resched();                        \
2979})
2980
2981extern int __cond_resched_lock(spinlock_t *lock);
2982
2983#define cond_resched_lock(lock) ({                              \
2984        ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2985        __cond_resched_lock(lock);                              \
2986})
2987
2988extern int __cond_resched_softirq(void);
2989
2990#define cond_resched_softirq() ({                                       \
2991        ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);     \
2992        __cond_resched_softirq();                                       \
2993})
2994
2995static inline void cond_resched_rcu(void)
2996{
2997#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2998        rcu_read_unlock();
2999        cond_resched();
3000        rcu_read_lock();
3001#endif
3002}
3003
3004/*
3005 * Does a critical section need to be broken due to another
3006 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3007 * but a general need for low latency)
3008 */
3009static inline int spin_needbreak(spinlock_t *lock)
3010{
3011#ifdef CONFIG_PREEMPT
3012        return spin_is_contended(lock);
3013#else
3014        return 0;
3015#endif
3016}
3017
3018/*
3019 * Idle thread specific functions to determine the need_resched
3020 * polling state.
3021 */
3022#ifdef TIF_POLLING_NRFLAG
3023static inline int tsk_is_polling(struct task_struct *p)
3024{
3025        return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3026}
3027
3028static inline void __current_set_polling(void)
3029{
3030        set_thread_flag(TIF_POLLING_NRFLAG);
3031}
3032
3033static inline bool __must_check current_set_polling_and_test(void)
3034{
3035        __current_set_polling();
3036
3037        /*
3038         * Polling state must be visible before we test NEED_RESCHED,
3039         * paired by resched_curr()
3040         */
3041        smp_mb__after_atomic();
3042
3043        return unlikely(tif_need_resched());
3044}
3045
3046static inline void __current_clr_polling(void)
3047{
3048        clear_thread_flag(TIF_POLLING_NRFLAG);
3049}
3050
3051static inline bool __must_check current_clr_polling_and_test(void)
3052{
3053        __current_clr_polling();
3054
3055        /*
3056         * Polling state must be visible before we test NEED_RESCHED,
3057         * paired by resched_curr()
3058         */
3059        smp_mb__after_atomic();
3060
3061        return unlikely(tif_need_resched());
3062}
3063
3064#else
3065static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3066static inline void __current_set_polling(void) { }
3067static inline void __current_clr_polling(void) { }
3068
3069static inline bool __must_check current_set_polling_and_test(void)
3070{
3071        return unlikely(tif_need_resched());
3072}
3073static inline bool __must_check current_clr_polling_and_test(void)
3074{
3075        return unlikely(tif_need_resched());
3076}
3077#endif
3078
3079static inline void current_clr_polling(void)
3080{
3081        __current_clr_polling();
3082
3083        /*
3084         * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3085         * Once the bit is cleared, we'll get IPIs with every new
3086         * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3087         * fold.
3088         */
3089        smp_mb(); /* paired with resched_curr() */
3090
3091        preempt_fold_need_resched();
3092}
3093
3094static __always_inline bool need_resched(void)
3095{
3096        return unlikely(tif_need_resched());
3097}
3098
3099/*
3100 * Thread group CPU time accounting.
3101 */
3102void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3103void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3104
3105/*
3106 * Reevaluate whether the task has signals pending delivery.
3107 * Wake the task if so.
3108 * This is required every time the blocked sigset_t changes.
3109 * callers must hold sighand->siglock.
3110 */
3111extern void recalc_sigpending_and_wake(struct task_struct *t);
3112extern void recalc_sigpending(void);
3113
3114extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3115
3116static inline void signal_wake_up(struct task_struct *t, bool resume)
3117{
3118        signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3119}
3120static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3121{
3122        signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3123}
3124
3125/*
3126 * Wrappers for p->thread_info->cpu access. No-op on UP.
3127 */
3128#ifdef CONFIG_SMP
3129
3130static inline unsigned int task_cpu(const struct task_struct *p)
3131{
3132        return task_thread_info(p)->cpu;
3133}
3134
3135static inline int task_node(const struct task_struct *p)
3136{
3137        return cpu_to_node(task_cpu(p));
3138}
3139
3140extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3141
3142#else
3143
3144static inline unsigned int task_cpu(const struct task_struct *p)
3145{
3146        return 0;
3147}
3148
3149static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3150{
3151}
3152
3153#endif /* CONFIG_SMP */
3154
3155extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3156extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3157
3158#ifdef CONFIG_CGROUP_SCHED
3159extern struct task_group root_task_group;
3160#endif /* CONFIG_CGROUP_SCHED */
3161
3162extern int task_can_switch_user(struct user_struct *up,
3163                                        struct task_struct *tsk);
3164
3165#ifdef CONFIG_TASK_XACCT
3166static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3167{
3168        tsk->ioac.rchar += amt;
3169}
3170
3171static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3172{
3173        tsk->ioac.wchar += amt;
3174}
3175
3176static inline void inc_syscr(struct task_struct *tsk)
3177{
3178        tsk->ioac.syscr++;
3179}
3180
3181static inline void inc_syscw(struct task_struct *tsk)
3182{
3183        tsk->ioac.syscw++;
3184}
3185#else
3186static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3187{
3188}
3189
3190static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3191{
3192}
3193
3194static inline void inc_syscr(struct task_struct *tsk)
3195{
3196}
3197
3198static inline void inc_syscw(struct task_struct *tsk)
3199{
3200}
3201#endif
3202
3203#ifndef TASK_SIZE_OF
3204#define TASK_SIZE_OF(tsk)       TASK_SIZE
3205#endif
3206
3207#ifdef CONFIG_MEMCG
3208extern void mm_update_next_owner(struct mm_struct *mm);
3209#else
3210static inline void mm_update_next_owner(struct mm_struct *mm)
3211{
3212}
3213#endif /* CONFIG_MEMCG */
3214
3215static inline unsigned long task_rlimit(const struct task_struct *tsk,
3216                unsigned int limit)
3217{
3218        return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3219}
3220
3221static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3222                unsigned int limit)
3223{
3224        return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3225}
3226
3227static inline unsigned long rlimit(unsigned int limit)
3228{
3229        return task_rlimit(current, limit);
3230}
3231
3232static inline unsigned long rlimit_max(unsigned int limit)
3233{
3234        return task_rlimit_max(current, limit);
3235}
3236
3237#ifdef CONFIG_CPU_FREQ
3238struct update_util_data {
3239        void (*func)(struct update_util_data *data,
3240                     u64 time, unsigned long util, unsigned long max);
3241};
3242
3243void cpufreq_set_update_util_data(int cpu, struct update_util_data *data);
3244#endif /* CONFIG_CPU_FREQ */
3245
3246#endif
3247