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