linux/include/linux/sched.h
<<
>>
Prefs
   1#ifndef _LINUX_SCHED_H
   2#define _LINUX_SCHED_H
   3
   4/*
   5 * cloning flags:
   6 */
   7#define CSIGNAL         0x000000ff      /* signal mask to be sent at exit */
   8#define CLONE_VM        0x00000100      /* set if VM shared between processes */
   9#define CLONE_FS        0x00000200      /* set if fs info shared between processes */
  10#define CLONE_FILES     0x00000400      /* set if open files shared between processes */
  11#define CLONE_SIGHAND   0x00000800      /* set if signal handlers and blocked signals shared */
  12#define CLONE_PTRACE    0x00002000      /* set if we want to let tracing continue on the child too */
  13#define CLONE_VFORK     0x00004000      /* set if the parent wants the child to wake it up on mm_release */
  14#define CLONE_PARENT    0x00008000      /* set if we want to have the same parent as the cloner */
  15#define CLONE_THREAD    0x00010000      /* Same thread group? */
  16#define CLONE_NEWNS     0x00020000      /* New namespace group? */
  17#define CLONE_SYSVSEM   0x00040000      /* share system V SEM_UNDO semantics */
  18#define CLONE_SETTLS    0x00080000      /* create a new TLS for the child */
  19#define CLONE_PARENT_SETTID     0x00100000      /* set the TID in the parent */
  20#define CLONE_CHILD_CLEARTID    0x00200000      /* clear the TID in the child */
  21#define CLONE_DETACHED          0x00400000      /* Unused, ignored */
  22#define CLONE_UNTRACED          0x00800000      /* set if the tracing process can't force CLONE_PTRACE on this clone */
  23#define CLONE_CHILD_SETTID      0x01000000      /* set the TID in the child */
  24/* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
  25   and is now available for re-use. */
  26#define CLONE_NEWUTS            0x04000000      /* New utsname group? */
  27#define CLONE_NEWIPC            0x08000000      /* New ipcs */
  28#define CLONE_NEWUSER           0x10000000      /* New user namespace */
  29#define CLONE_NEWPID            0x20000000      /* New pid namespace */
  30#define CLONE_NEWNET            0x40000000      /* New network namespace */
  31#define CLONE_IO                0x80000000      /* Clone io context */
  32
  33/*
  34 * Scheduling policies
  35 */
  36#define SCHED_NORMAL            0
  37#define SCHED_FIFO              1
  38#define SCHED_RR                2
  39#define SCHED_BATCH             3
  40/* SCHED_ISO: reserved but not implemented yet */
  41#define SCHED_IDLE              5
  42/* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
  43#define SCHED_RESET_ON_FORK     0x40000000
  44
  45#ifdef __KERNEL__
  46
  47struct sched_param {
  48        int sched_priority;
  49};
  50
  51#include <asm/param.h>  /* for HZ */
  52
  53#include <linux/capability.h>
  54#include <linux/threads.h>
  55#include <linux/kernel.h>
  56#include <linux/types.h>
  57#include <linux/timex.h>
  58#include <linux/jiffies.h>
  59#include <linux/rbtree.h>
  60#include <linux/thread_info.h>
  61#include <linux/cpumask.h>
  62#include <linux/errno.h>
  63#include <linux/nodemask.h>
  64#include <linux/mm_types.h>
  65
  66#include <asm/page.h>
  67#include <asm/ptrace.h>
  68#include <asm/cputime.h>
  69
  70#include <linux/smp.h>
  71#include <linux/sem.h>
  72#include <linux/signal.h>
  73#include <linux/compiler.h>
  74#include <linux/completion.h>
  75#include <linux/pid.h>
  76#include <linux/percpu.h>
  77#include <linux/topology.h>
  78#include <linux/proportions.h>
  79#include <linux/seccomp.h>
  80#include <linux/rcupdate.h>
  81#include <linux/rculist.h>
  82#include <linux/rtmutex.h>
  83
  84#include <linux/time.h>
  85#include <linux/param.h>
  86#include <linux/resource.h>
  87#include <linux/timer.h>
  88#include <linux/hrtimer.h>
  89#include <linux/task_io_accounting.h>
  90#include <linux/latencytop.h>
  91#include <linux/cred.h>
  92#include <linux/llist.h>
  93#include <linux/uidgid.h>
  94
  95#include <asm/processor.h>
  96
  97struct exec_domain;
  98struct futex_pi_state;
  99struct robust_list_head;
 100struct bio_list;
 101struct fs_struct;
 102struct perf_event_context;
 103struct blk_plug;
 104
 105/*
 106 * List of flags we want to share for kernel threads,
 107 * if only because they are not used by them anyway.
 108 */
 109#define CLONE_KERNEL    (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
 110
 111/*
 112 * These are the constant used to fake the fixed-point load-average
 113 * counting. Some notes:
 114 *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
 115 *    a load-average precision of 10 bits integer + 11 bits fractional
 116 *  - if you want to count load-averages more often, you need more
 117 *    precision, or rounding will get you. With 2-second counting freq,
 118 *    the EXP_n values would be 1981, 2034 and 2043 if still using only
 119 *    11 bit fractions.
 120 */
 121extern unsigned long avenrun[];         /* Load averages */
 122extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
 123
 124#define FSHIFT          11              /* nr of bits of precision */
 125#define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
 126#define LOAD_FREQ       (5*HZ+1)        /* 5 sec intervals */
 127#define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
 128#define EXP_5           2014            /* 1/exp(5sec/5min) */
 129#define EXP_15          2037            /* 1/exp(5sec/15min) */
 130
 131#define CALC_LOAD(load,exp,n) \
 132        load *= exp; \
 133        load += n*(FIXED_1-exp); \
 134        load >>= FSHIFT;
 135
 136extern unsigned long total_forks;
 137extern int nr_threads;
 138DECLARE_PER_CPU(unsigned long, process_counts);
 139extern int nr_processes(void);
 140extern unsigned long nr_running(void);
 141extern unsigned long nr_uninterruptible(void);
 142extern unsigned long nr_iowait(void);
 143extern unsigned long nr_iowait_cpu(int cpu);
 144extern unsigned long this_cpu_load(void);
 145
 146
 147extern void calc_global_load(unsigned long ticks);
 148extern void update_cpu_load_nohz(void);
 149
 150extern unsigned long get_parent_ip(unsigned long addr);
 151
 152struct seq_file;
 153struct cfs_rq;
 154struct task_group;
 155#ifdef CONFIG_SCHED_DEBUG
 156extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
 157extern void proc_sched_set_task(struct task_struct *p);
 158extern void
 159print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
 160#else
 161static inline void
 162proc_sched_show_task(struct task_struct *p, struct seq_file *m)
 163{
 164}
 165static inline void proc_sched_set_task(struct task_struct *p)
 166{
 167}
 168static inline void
 169print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
 170{
 171}
 172#endif
 173
 174/*
 175 * Task state bitmask. NOTE! These bits are also
 176 * encoded in fs/proc/array.c: get_task_state().
 177 *
 178 * We have two separate sets of flags: task->state
 179 * is about runnability, while task->exit_state are
 180 * about the task exiting. Confusing, but this way
 181 * modifying one set can't modify the other one by
 182 * mistake.
 183 */
 184#define TASK_RUNNING            0
 185#define TASK_INTERRUPTIBLE      1
 186#define TASK_UNINTERRUPTIBLE    2
 187#define __TASK_STOPPED          4
 188#define __TASK_TRACED           8
 189/* in tsk->exit_state */
 190#define EXIT_ZOMBIE             16
 191#define EXIT_DEAD               32
 192/* in tsk->state again */
 193#define TASK_DEAD               64
 194#define TASK_WAKEKILL           128
 195#define TASK_WAKING             256
 196#define TASK_STATE_MAX          512
 197
 198#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
 199
 200extern char ___assert_task_state[1 - 2*!!(
 201                sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
 202
 203/* Convenience macros for the sake of set_task_state */
 204#define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
 205#define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
 206#define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
 207
 208/* Convenience macros for the sake of wake_up */
 209#define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 210#define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
 211
 212/* get_task_state() */
 213#define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 214                                 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 215                                 __TASK_TRACED)
 216
 217#define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
 218#define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
 219#define task_is_dead(task)      ((task)->exit_state != 0)
 220#define task_is_stopped_or_traced(task) \
 221                        ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 222#define task_contributes_to_load(task)  \
 223                                ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
 224                                 (task->flags & PF_FROZEN) == 0)
 225
 226#define __set_task_state(tsk, state_value)              \
 227        do { (tsk)->state = (state_value); } while (0)
 228#define set_task_state(tsk, state_value)                \
 229        set_mb((tsk)->state, (state_value))
 230
 231/*
 232 * set_current_state() includes a barrier so that the write of current->state
 233 * is correctly serialised wrt the caller's subsequent test of whether to
 234 * actually sleep:
 235 *
 236 *      set_current_state(TASK_UNINTERRUPTIBLE);
 237 *      if (do_i_need_to_sleep())
 238 *              schedule();
 239 *
 240 * If the caller does not need such serialisation then use __set_current_state()
 241 */
 242#define __set_current_state(state_value)                        \
 243        do { current->state = (state_value); } while (0)
 244#define set_current_state(state_value)          \
 245        set_mb(current->state, (state_value))
 246
 247/* Task command name length */
 248#define TASK_COMM_LEN 16
 249
 250#include <linux/spinlock.h>
 251
 252/*
 253 * This serializes "schedule()" and also protects
 254 * the run-queue from deletions/modifications (but
 255 * _adding_ to the beginning of the run-queue has
 256 * a separate lock).
 257 */
 258extern rwlock_t tasklist_lock;
 259extern spinlock_t mmlist_lock;
 260
 261struct task_struct;
 262
 263#ifdef CONFIG_PROVE_RCU
 264extern int lockdep_tasklist_lock_is_held(void);
 265#endif /* #ifdef CONFIG_PROVE_RCU */
 266
 267extern void sched_init(void);
 268extern void sched_init_smp(void);
 269extern asmlinkage void schedule_tail(struct task_struct *prev);
 270extern void init_idle(struct task_struct *idle, int cpu);
 271extern void init_idle_bootup_task(struct task_struct *idle);
 272
 273extern int runqueue_is_locked(int cpu);
 274
 275#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
 276extern void select_nohz_load_balancer(int stop_tick);
 277extern void set_cpu_sd_state_idle(void);
 278extern int get_nohz_timer_target(void);
 279#else
 280static inline void select_nohz_load_balancer(int stop_tick) { }
 281static inline void set_cpu_sd_state_idle(void) { }
 282#endif
 283
 284/*
 285 * Only dump TASK_* tasks. (0 for all tasks)
 286 */
 287extern void show_state_filter(unsigned long state_filter);
 288
 289static inline void show_state(void)
 290{
 291        show_state_filter(0);
 292}
 293
 294extern void show_regs(struct pt_regs *);
 295
 296/*
 297 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 298 * task), SP is the stack pointer of the first frame that should be shown in the back
 299 * trace (or NULL if the entire call-chain of the task should be shown).
 300 */
 301extern void show_stack(struct task_struct *task, unsigned long *sp);
 302
 303void io_schedule(void);
 304long io_schedule_timeout(long timeout);
 305
 306extern void cpu_init (void);
 307extern void trap_init(void);
 308extern void update_process_times(int user);
 309extern void scheduler_tick(void);
 310
 311extern void sched_show_task(struct task_struct *p);
 312
 313#ifdef CONFIG_LOCKUP_DETECTOR
 314extern void touch_softlockup_watchdog(void);
 315extern void touch_softlockup_watchdog_sync(void);
 316extern void touch_all_softlockup_watchdogs(void);
 317extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
 318                                  void __user *buffer,
 319                                  size_t *lenp, loff_t *ppos);
 320extern unsigned int  softlockup_panic;
 321void lockup_detector_init(void);
 322#else
 323static inline void touch_softlockup_watchdog(void)
 324{
 325}
 326static inline void touch_softlockup_watchdog_sync(void)
 327{
 328}
 329static inline void touch_all_softlockup_watchdogs(void)
 330{
 331}
 332static inline void lockup_detector_init(void)
 333{
 334}
 335#endif
 336
 337#ifdef CONFIG_DETECT_HUNG_TASK
 338extern unsigned int  sysctl_hung_task_panic;
 339extern unsigned long sysctl_hung_task_check_count;
 340extern unsigned long sysctl_hung_task_timeout_secs;
 341extern unsigned long sysctl_hung_task_warnings;
 342extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
 343                                         void __user *buffer,
 344                                         size_t *lenp, loff_t *ppos);
 345#else
 346/* Avoid need for ifdefs elsewhere in the code */
 347enum { sysctl_hung_task_timeout_secs = 0 };
 348#endif
 349
 350/* Attach to any functions which should be ignored in wchan output. */
 351#define __sched         __attribute__((__section__(".sched.text")))
 352
 353/* Linker adds these: start and end of __sched functions */
 354extern char __sched_text_start[], __sched_text_end[];
 355
 356/* Is this address in the __sched functions? */
 357extern int in_sched_functions(unsigned long addr);
 358
 359#define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 360extern signed long schedule_timeout(signed long timeout);
 361extern signed long schedule_timeout_interruptible(signed long timeout);
 362extern signed long schedule_timeout_killable(signed long timeout);
 363extern signed long schedule_timeout_uninterruptible(signed long timeout);
 364asmlinkage void schedule(void);
 365extern void schedule_preempt_disabled(void);
 366extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
 367
 368struct nsproxy;
 369struct user_namespace;
 370
 371/*
 372 * Default maximum number of active map areas, this limits the number of vmas
 373 * per mm struct. Users can overwrite this number by sysctl but there is a
 374 * problem.
 375 *
 376 * When a program's coredump is generated as ELF format, a section is created
 377 * per a vma. In ELF, the number of sections is represented in unsigned short.
 378 * This means the number of sections should be smaller than 65535 at coredump.
 379 * Because the kernel adds some informative sections to a image of program at
 380 * generating coredump, we need some margin. The number of extra sections is
 381 * 1-3 now and depends on arch. We use "5" as safe margin, here.
 382 */
 383#define MAPCOUNT_ELF_CORE_MARGIN        (5)
 384#define DEFAULT_MAX_MAP_COUNT   (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
 385
 386extern int sysctl_max_map_count;
 387
 388#include <linux/aio.h>
 389
 390#ifdef CONFIG_MMU
 391extern void arch_pick_mmap_layout(struct mm_struct *mm);
 392extern unsigned long
 393arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 394                       unsigned long, unsigned long);
 395extern unsigned long
 396arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 397                          unsigned long len, unsigned long pgoff,
 398                          unsigned long flags);
 399extern void arch_unmap_area(struct mm_struct *, unsigned long);
 400extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
 401#else
 402static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
 403#endif
 404
 405
 406extern void set_dumpable(struct mm_struct *mm, int value);
 407extern int get_dumpable(struct mm_struct *mm);
 408
 409/* get/set_dumpable() values */
 410#define SUID_DUMPABLE_DISABLED  0
 411#define SUID_DUMPABLE_ENABLED   1
 412#define SUID_DUMPABLE_SAFE      2
 413
 414/* mm flags */
 415/* dumpable bits */
 416#define MMF_DUMPABLE      0  /* core dump is permitted */
 417#define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
 418
 419#define MMF_DUMPABLE_BITS 2
 420#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
 421
 422/* coredump filter bits */
 423#define MMF_DUMP_ANON_PRIVATE   2
 424#define MMF_DUMP_ANON_SHARED    3
 425#define MMF_DUMP_MAPPED_PRIVATE 4
 426#define MMF_DUMP_MAPPED_SHARED  5
 427#define MMF_DUMP_ELF_HEADERS    6
 428#define MMF_DUMP_HUGETLB_PRIVATE 7
 429#define MMF_DUMP_HUGETLB_SHARED  8
 430
 431#define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
 432#define MMF_DUMP_FILTER_BITS    7
 433#define MMF_DUMP_FILTER_MASK \
 434        (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
 435#define MMF_DUMP_FILTER_DEFAULT \
 436        ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
 437         (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
 438
 439#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
 440# define MMF_DUMP_MASK_DEFAULT_ELF      (1 << MMF_DUMP_ELF_HEADERS)
 441#else
 442# define MMF_DUMP_MASK_DEFAULT_ELF      0
 443#endif
 444                                        /* leave room for more dump flags */
 445#define MMF_VM_MERGEABLE        16      /* KSM may merge identical pages */
 446#define MMF_VM_HUGEPAGE         17      /* set when VM_HUGEPAGE is set on vma */
 447#define MMF_EXE_FILE_CHANGED    18      /* see prctl_set_mm_exe_file() */
 448
 449#define MMF_INIT_MASK           (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
 450
 451struct sighand_struct {
 452        atomic_t                count;
 453        struct k_sigaction      action[_NSIG];
 454        spinlock_t              siglock;
 455        wait_queue_head_t       signalfd_wqh;
 456};
 457
 458struct pacct_struct {
 459        int                     ac_flag;
 460        long                    ac_exitcode;
 461        unsigned long           ac_mem;
 462        cputime_t               ac_utime, ac_stime;
 463        unsigned long           ac_minflt, ac_majflt;
 464};
 465
 466struct cpu_itimer {
 467        cputime_t expires;
 468        cputime_t incr;
 469        u32 error;
 470        u32 incr_error;
 471};
 472
 473/**
 474 * struct task_cputime - collected CPU time counts
 475 * @utime:              time spent in user mode, in &cputime_t units
 476 * @stime:              time spent in kernel mode, in &cputime_t units
 477 * @sum_exec_runtime:   total time spent on the CPU, in nanoseconds
 478 *
 479 * This structure groups together three kinds of CPU time that are
 480 * tracked for threads and thread groups.  Most things considering
 481 * CPU time want to group these counts together and treat all three
 482 * of them in parallel.
 483 */
 484struct task_cputime {
 485        cputime_t utime;
 486        cputime_t stime;
 487        unsigned long long sum_exec_runtime;
 488};
 489/* Alternate field names when used to cache expirations. */
 490#define prof_exp        stime
 491#define virt_exp        utime
 492#define sched_exp       sum_exec_runtime
 493
 494#define INIT_CPUTIME    \
 495        (struct task_cputime) {                                 \
 496                .utime = 0,                                     \
 497                .stime = 0,                                     \
 498                .sum_exec_runtime = 0,                          \
 499        }
 500
 501/*
 502 * Disable preemption until the scheduler is running.
 503 * Reset by start_kernel()->sched_init()->init_idle().
 504 *
 505 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
 506 * before the scheduler is active -- see should_resched().
 507 */
 508#define INIT_PREEMPT_COUNT      (1 + PREEMPT_ACTIVE)
 509
 510/**
 511 * struct thread_group_cputimer - thread group interval timer counts
 512 * @cputime:            thread group interval timers.
 513 * @running:            non-zero when there are timers running and
 514 *                      @cputime receives updates.
 515 * @lock:               lock for fields in this struct.
 516 *
 517 * This structure contains the version of task_cputime, above, that is
 518 * used for thread group CPU timer calculations.
 519 */
 520struct thread_group_cputimer {
 521        struct task_cputime cputime;
 522        int running;
 523        raw_spinlock_t lock;
 524};
 525
 526#include <linux/rwsem.h>
 527struct autogroup;
 528
 529/*
 530 * NOTE! "signal_struct" does not have its own
 531 * locking, because a shared signal_struct always
 532 * implies a shared sighand_struct, so locking
 533 * sighand_struct is always a proper superset of
 534 * the locking of signal_struct.
 535 */
 536struct signal_struct {
 537        atomic_t                sigcnt;
 538        atomic_t                live;
 539        int                     nr_threads;
 540
 541        wait_queue_head_t       wait_chldexit;  /* for wait4() */
 542
 543        /* current thread group signal load-balancing target: */
 544        struct task_struct      *curr_target;
 545
 546        /* shared signal handling: */
 547        struct sigpending       shared_pending;
 548
 549        /* thread group exit support */
 550        int                     group_exit_code;
 551        /* overloaded:
 552         * - notify group_exit_task when ->count is equal to notify_count
 553         * - everyone except group_exit_task is stopped during signal delivery
 554         *   of fatal signals, group_exit_task processes the signal.
 555         */
 556        int                     notify_count;
 557        struct task_struct      *group_exit_task;
 558
 559        /* thread group stop support, overloads group_exit_code too */
 560        int                     group_stop_count;
 561        unsigned int            flags; /* see SIGNAL_* flags below */
 562
 563        /*
 564         * PR_SET_CHILD_SUBREAPER marks a process, like a service
 565         * manager, to re-parent orphan (double-forking) child processes
 566         * to this process instead of 'init'. The service manager is
 567         * able to receive SIGCHLD signals and is able to investigate
 568         * the process until it calls wait(). All children of this
 569         * process will inherit a flag if they should look for a
 570         * child_subreaper process at exit.
 571         */
 572        unsigned int            is_child_subreaper:1;
 573        unsigned int            has_child_subreaper:1;
 574
 575        /* POSIX.1b Interval Timers */
 576        struct list_head posix_timers;
 577
 578        /* ITIMER_REAL timer for the process */
 579        struct hrtimer real_timer;
 580        struct pid *leader_pid;
 581        ktime_t it_real_incr;
 582
 583        /*
 584         * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
 585         * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
 586         * values are defined to 0 and 1 respectively
 587         */
 588        struct cpu_itimer it[2];
 589
 590        /*
 591         * Thread group totals for process CPU timers.
 592         * See thread_group_cputimer(), et al, for details.
 593         */
 594        struct thread_group_cputimer cputimer;
 595
 596        /* Earliest-expiration cache. */
 597        struct task_cputime cputime_expires;
 598
 599        struct list_head cpu_timers[3];
 600
 601        struct pid *tty_old_pgrp;
 602
 603        /* boolean value for session group leader */
 604        int leader;
 605
 606        struct tty_struct *tty; /* NULL if no tty */
 607
 608#ifdef CONFIG_SCHED_AUTOGROUP
 609        struct autogroup *autogroup;
 610#endif
 611        /*
 612         * Cumulative resource counters for dead threads in the group,
 613         * and for reaped dead child processes forked by this group.
 614         * Live threads maintain their own counters and add to these
 615         * in __exit_signal, except for the group leader.
 616         */
 617        cputime_t utime, stime, cutime, cstime;
 618        cputime_t gtime;
 619        cputime_t cgtime;
 620#ifndef CONFIG_VIRT_CPU_ACCOUNTING
 621        cputime_t prev_utime, prev_stime;
 622#endif
 623        unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 624        unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 625        unsigned long inblock, oublock, cinblock, coublock;
 626        unsigned long maxrss, cmaxrss;
 627        struct task_io_accounting ioac;
 628
 629        /*
 630         * Cumulative ns of schedule CPU time fo dead threads in the
 631         * group, not including a zombie group leader, (This only differs
 632         * from jiffies_to_ns(utime + stime) if sched_clock uses something
 633         * other than jiffies.)
 634         */
 635        unsigned long long sum_sched_runtime;
 636
 637        /*
 638         * We don't bother to synchronize most readers of this at all,
 639         * because there is no reader checking a limit that actually needs
 640         * to get both rlim_cur and rlim_max atomically, and either one
 641         * alone is a single word that can safely be read normally.
 642         * getrlimit/setrlimit use task_lock(current->group_leader) to
 643         * protect this instead of the siglock, because they really
 644         * have no need to disable irqs.
 645         */
 646        struct rlimit rlim[RLIM_NLIMITS];
 647
 648#ifdef CONFIG_BSD_PROCESS_ACCT
 649        struct pacct_struct pacct;      /* per-process accounting information */
 650#endif
 651#ifdef CONFIG_TASKSTATS
 652        struct taskstats *stats;
 653#endif
 654#ifdef CONFIG_AUDIT
 655        unsigned audit_tty;
 656        struct tty_audit_buf *tty_audit_buf;
 657#endif
 658#ifdef CONFIG_CGROUPS
 659        /*
 660         * group_rwsem prevents new tasks from entering the threadgroup and
 661         * member tasks from exiting,a more specifically, setting of
 662         * PF_EXITING.  fork and exit paths are protected with this rwsem
 663         * using threadgroup_change_begin/end().  Users which require
 664         * threadgroup to remain stable should use threadgroup_[un]lock()
 665         * which also takes care of exec path.  Currently, cgroup is the
 666         * only user.
 667         */
 668        struct rw_semaphore group_rwsem;
 669#endif
 670
 671        int oom_adj;            /* OOM kill score adjustment (bit shift) */
 672        int oom_score_adj;      /* OOM kill score adjustment */
 673        int oom_score_adj_min;  /* OOM kill score adjustment minimum value.
 674                                 * Only settable by CAP_SYS_RESOURCE. */
 675
 676        struct mutex cred_guard_mutex;  /* guard against foreign influences on
 677                                         * credential calculations
 678                                         * (notably. ptrace) */
 679};
 680
 681/* Context switch must be unlocked if interrupts are to be enabled */
 682#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
 683# define __ARCH_WANT_UNLOCKED_CTXSW
 684#endif
 685
 686/*
 687 * Bits in flags field of signal_struct.
 688 */
 689#define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 690#define SIGNAL_STOP_CONTINUED   0x00000002 /* SIGCONT since WCONTINUED reap */
 691#define SIGNAL_GROUP_EXIT       0x00000004 /* group exit in progress */
 692/*
 693 * Pending notifications to parent.
 694 */
 695#define SIGNAL_CLD_STOPPED      0x00000010
 696#define SIGNAL_CLD_CONTINUED    0x00000020
 697#define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
 698
 699#define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
 700
 701/* If true, all threads except ->group_exit_task have pending SIGKILL */
 702static inline int signal_group_exit(const struct signal_struct *sig)
 703{
 704        return  (sig->flags & SIGNAL_GROUP_EXIT) ||
 705                (sig->group_exit_task != NULL);
 706}
 707
 708/*
 709 * Some day this will be a full-fledged user tracking system..
 710 */
 711struct user_struct {
 712        atomic_t __count;       /* reference count */
 713        atomic_t processes;     /* How many processes does this user have? */
 714        atomic_t files;         /* How many open files does this user have? */
 715        atomic_t sigpending;    /* How many pending signals does this user have? */
 716#ifdef CONFIG_INOTIFY_USER
 717        atomic_t inotify_watches; /* How many inotify watches does this user have? */
 718        atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 719#endif
 720#ifdef CONFIG_FANOTIFY
 721        atomic_t fanotify_listeners;
 722#endif
 723#ifdef CONFIG_EPOLL
 724        atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
 725#endif
 726#ifdef CONFIG_POSIX_MQUEUE
 727        /* protected by mq_lock */
 728        unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 729#endif
 730        unsigned long locked_shm; /* How many pages of mlocked shm ? */
 731
 732#ifdef CONFIG_KEYS
 733        struct key *uid_keyring;        /* UID specific keyring */
 734        struct key *session_keyring;    /* UID's default session keyring */
 735#endif
 736
 737        /* Hash table maintenance information */
 738        struct hlist_node uidhash_node;
 739        kuid_t uid;
 740
 741#ifdef CONFIG_PERF_EVENTS
 742        atomic_long_t locked_vm;
 743#endif
 744};
 745
 746extern int uids_sysfs_init(void);
 747
 748extern struct user_struct *find_user(kuid_t);
 749
 750extern struct user_struct root_user;
 751#define INIT_USER (&root_user)
 752
 753
 754struct backing_dev_info;
 755struct reclaim_state;
 756
 757#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 758struct sched_info {
 759        /* cumulative counters */
 760        unsigned long pcount;         /* # of times run on this cpu */
 761        unsigned long long run_delay; /* time spent waiting on a runqueue */
 762
 763        /* timestamps */
 764        unsigned long long last_arrival,/* when we last ran on a cpu */
 765                           last_queued; /* when we were last queued to run */
 766};
 767#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
 768
 769#ifdef CONFIG_TASK_DELAY_ACCT
 770struct task_delay_info {
 771        spinlock_t      lock;
 772        unsigned int    flags;  /* Private per-task flags */
 773
 774        /* For each stat XXX, add following, aligned appropriately
 775         *
 776         * struct timespec XXX_start, XXX_end;
 777         * u64 XXX_delay;
 778         * u32 XXX_count;
 779         *
 780         * Atomicity of updates to XXX_delay, XXX_count protected by
 781         * single lock above (split into XXX_lock if contention is an issue).
 782         */
 783
 784        /*
 785         * XXX_count is incremented on every XXX operation, the delay
 786         * associated with the operation is added to XXX_delay.
 787         * XXX_delay contains the accumulated delay time in nanoseconds.
 788         */
 789        struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
 790        u64 blkio_delay;        /* wait for sync block io completion */
 791        u64 swapin_delay;       /* wait for swapin block io completion */
 792        u32 blkio_count;        /* total count of the number of sync block */
 793                                /* io operations performed */
 794        u32 swapin_count;       /* total count of the number of swapin block */
 795                                /* io operations performed */
 796
 797        struct timespec freepages_start, freepages_end;
 798        u64 freepages_delay;    /* wait for memory reclaim */
 799        u32 freepages_count;    /* total count of memory reclaim */
 800};
 801#endif  /* CONFIG_TASK_DELAY_ACCT */
 802
 803static inline int sched_info_on(void)
 804{
 805#ifdef CONFIG_SCHEDSTATS
 806        return 1;
 807#elif defined(CONFIG_TASK_DELAY_ACCT)
 808        extern int delayacct_on;
 809        return delayacct_on;
 810#else
 811        return 0;
 812#endif
 813}
 814
 815enum cpu_idle_type {
 816        CPU_IDLE,
 817        CPU_NOT_IDLE,
 818        CPU_NEWLY_IDLE,
 819        CPU_MAX_IDLE_TYPES
 820};
 821
 822/*
 823 * Increase resolution of nice-level calculations for 64-bit architectures.
 824 * The extra resolution improves shares distribution and load balancing of
 825 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
 826 * hierarchies, especially on larger systems. This is not a user-visible change
 827 * and does not change the user-interface for setting shares/weights.
 828 *
 829 * We increase resolution only if we have enough bits to allow this increased
 830 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
 831 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
 832 * increased costs.
 833 */
 834#if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load  */
 835# define SCHED_LOAD_RESOLUTION  10
 836# define scale_load(w)          ((w) << SCHED_LOAD_RESOLUTION)
 837# define scale_load_down(w)     ((w) >> SCHED_LOAD_RESOLUTION)
 838#else
 839# define SCHED_LOAD_RESOLUTION  0
 840# define scale_load(w)          (w)
 841# define scale_load_down(w)     (w)
 842#endif
 843
 844#define SCHED_LOAD_SHIFT        (10 + SCHED_LOAD_RESOLUTION)
 845#define SCHED_LOAD_SCALE        (1L << SCHED_LOAD_SHIFT)
 846
 847/*
 848 * Increase resolution of cpu_power calculations
 849 */
 850#define SCHED_POWER_SHIFT       10
 851#define SCHED_POWER_SCALE       (1L << SCHED_POWER_SHIFT)
 852
 853/*
 854 * sched-domains (multiprocessor balancing) declarations:
 855 */
 856#ifdef CONFIG_SMP
 857#define SD_LOAD_BALANCE         0x0001  /* Do load balancing on this domain. */
 858#define SD_BALANCE_NEWIDLE      0x0002  /* Balance when about to become idle */
 859#define SD_BALANCE_EXEC         0x0004  /* Balance on exec */
 860#define SD_BALANCE_FORK         0x0008  /* Balance on fork, clone */
 861#define SD_BALANCE_WAKE         0x0010  /* Balance on wakeup */
 862#define SD_WAKE_AFFINE          0x0020  /* Wake task to waking CPU */
 863#define SD_PREFER_LOCAL         0x0040  /* Prefer to keep tasks local to this domain */
 864#define SD_SHARE_CPUPOWER       0x0080  /* Domain members share cpu power */
 865#define SD_SHARE_PKG_RESOURCES  0x0200  /* Domain members share cpu pkg resources */
 866#define SD_SERIALIZE            0x0400  /* Only a single load balancing instance */
 867#define SD_ASYM_PACKING         0x0800  /* Place busy groups earlier in the domain */
 868#define SD_PREFER_SIBLING       0x1000  /* Prefer to place tasks in a sibling domain */
 869#define SD_OVERLAP              0x2000  /* sched_domains of this level overlap */
 870
 871extern int __weak arch_sd_sibiling_asym_packing(void);
 872
 873struct sched_group_power {
 874        atomic_t ref;
 875        /*
 876         * CPU power of this group, SCHED_LOAD_SCALE being max power for a
 877         * single CPU.
 878         */
 879        unsigned int power, power_orig;
 880        unsigned long next_update;
 881        /*
 882         * Number of busy cpus in this group.
 883         */
 884        atomic_t nr_busy_cpus;
 885
 886        unsigned long cpumask[0]; /* iteration mask */
 887};
 888
 889struct sched_group {
 890        struct sched_group *next;       /* Must be a circular list */
 891        atomic_t ref;
 892
 893        unsigned int group_weight;
 894        struct sched_group_power *sgp;
 895
 896        /*
 897         * The CPUs this group covers.
 898         *
 899         * NOTE: this field is variable length. (Allocated dynamically
 900         * by attaching extra space to the end of the structure,
 901         * depending on how many CPUs the kernel has booted up with)
 902         */
 903        unsigned long cpumask[0];
 904};
 905
 906static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
 907{
 908        return to_cpumask(sg->cpumask);
 909}
 910
 911/*
 912 * cpumask masking which cpus in the group are allowed to iterate up the domain
 913 * tree.
 914 */
 915static inline struct cpumask *sched_group_mask(struct sched_group *sg)
 916{
 917        return to_cpumask(sg->sgp->cpumask);
 918}
 919
 920/**
 921 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
 922 * @group: The group whose first cpu is to be returned.
 923 */
 924static inline unsigned int group_first_cpu(struct sched_group *group)
 925{
 926        return cpumask_first(sched_group_cpus(group));
 927}
 928
 929struct sched_domain_attr {
 930        int relax_domain_level;
 931};
 932
 933#define SD_ATTR_INIT    (struct sched_domain_attr) {    \
 934        .relax_domain_level = -1,                       \
 935}
 936
 937extern int sched_domain_level_max;
 938
 939struct sched_domain {
 940        /* These fields must be setup */
 941        struct sched_domain *parent;    /* top domain must be null terminated */
 942        struct sched_domain *child;     /* bottom domain must be null terminated */
 943        struct sched_group *groups;     /* the balancing groups of the domain */
 944        unsigned long min_interval;     /* Minimum balance interval ms */
 945        unsigned long max_interval;     /* Maximum balance interval ms */
 946        unsigned int busy_factor;       /* less balancing by factor if busy */
 947        unsigned int imbalance_pct;     /* No balance until over watermark */
 948        unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
 949        unsigned int busy_idx;
 950        unsigned int idle_idx;
 951        unsigned int newidle_idx;
 952        unsigned int wake_idx;
 953        unsigned int forkexec_idx;
 954        unsigned int smt_gain;
 955        int flags;                      /* See SD_* */
 956        int level;
 957
 958        /* Runtime fields. */
 959        unsigned long last_balance;     /* init to jiffies. units in jiffies */
 960        unsigned int balance_interval;  /* initialise to 1. units in ms. */
 961        unsigned int nr_balance_failed; /* initialise to 0 */
 962
 963        u64 last_update;
 964
 965#ifdef CONFIG_SCHEDSTATS
 966        /* load_balance() stats */
 967        unsigned int lb_count[CPU_MAX_IDLE_TYPES];
 968        unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
 969        unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
 970        unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
 971        unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
 972        unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
 973        unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
 974        unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
 975
 976        /* Active load balancing */
 977        unsigned int alb_count;
 978        unsigned int alb_failed;
 979        unsigned int alb_pushed;
 980
 981        /* SD_BALANCE_EXEC stats */
 982        unsigned int sbe_count;
 983        unsigned int sbe_balanced;
 984        unsigned int sbe_pushed;
 985
 986        /* SD_BALANCE_FORK stats */
 987        unsigned int sbf_count;
 988        unsigned int sbf_balanced;
 989        unsigned int sbf_pushed;
 990
 991        /* try_to_wake_up() stats */
 992        unsigned int ttwu_wake_remote;
 993        unsigned int ttwu_move_affine;
 994        unsigned int ttwu_move_balance;
 995#endif
 996#ifdef CONFIG_SCHED_DEBUG
 997        char *name;
 998#endif
 999        union {
1000                void *private;          /* used during construction */
1001                struct rcu_head rcu;    /* used during destruction */
1002        };
1003
1004        unsigned int span_weight;
1005        /*
1006         * Span of all CPUs in this domain.
1007         *
1008         * NOTE: this field is variable length. (Allocated dynamically
1009         * by attaching extra space to the end of the structure,
1010         * depending on how many CPUs the kernel has booted up with)
1011         */
1012        unsigned long span[0];
1013};
1014
1015static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1016{
1017        return to_cpumask(sd->span);
1018}
1019
1020extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1021                                    struct sched_domain_attr *dattr_new);
1022
1023/* Allocate an array of sched domains, for partition_sched_domains(). */
1024cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1025void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1026
1027/* Test a flag in parent sched domain */
1028static inline int test_sd_parent(struct sched_domain *sd, int flag)
1029{
1030        if (sd->parent && (sd->parent->flags & flag))
1031                return 1;
1032
1033        return 0;
1034}
1035
1036unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1037unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1038
1039bool cpus_share_cache(int this_cpu, int that_cpu);
1040
1041#else /* CONFIG_SMP */
1042
1043struct sched_domain_attr;
1044
1045static inline void
1046partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1047                        struct sched_domain_attr *dattr_new)
1048{
1049}
1050
1051static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1052{
1053        return true;
1054}
1055
1056#endif  /* !CONFIG_SMP */
1057
1058
1059struct io_context;                      /* See blkdev.h */
1060
1061
1062#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1063extern void prefetch_stack(struct task_struct *t);
1064#else
1065static inline void prefetch_stack(struct task_struct *t) { }
1066#endif
1067
1068struct audit_context;           /* See audit.c */
1069struct mempolicy;
1070struct pipe_inode_info;
1071struct uts_namespace;
1072
1073struct rq;
1074struct sched_domain;
1075
1076/*
1077 * wake flags
1078 */
1079#define WF_SYNC         0x01            /* waker goes to sleep after wakup */
1080#define WF_FORK         0x02            /* child wakeup after fork */
1081#define WF_MIGRATED     0x04            /* internal use, task got migrated */
1082
1083#define ENQUEUE_WAKEUP          1
1084#define ENQUEUE_HEAD            2
1085#ifdef CONFIG_SMP
1086#define ENQUEUE_WAKING          4       /* sched_class::task_waking was called */
1087#else
1088#define ENQUEUE_WAKING          0
1089#endif
1090
1091#define DEQUEUE_SLEEP           1
1092
1093struct sched_class {
1094        const struct sched_class *next;
1095
1096        void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1097        void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1098        void (*yield_task) (struct rq *rq);
1099        bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1100
1101        void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1102
1103        struct task_struct * (*pick_next_task) (struct rq *rq);
1104        void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1105
1106#ifdef CONFIG_SMP
1107        int  (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1108
1109        void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1110        void (*post_schedule) (struct rq *this_rq);
1111        void (*task_waking) (struct task_struct *task);
1112        void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1113
1114        void (*set_cpus_allowed)(struct task_struct *p,
1115                                 const struct cpumask *newmask);
1116
1117        void (*rq_online)(struct rq *rq);
1118        void (*rq_offline)(struct rq *rq);
1119#endif
1120
1121        void (*set_curr_task) (struct rq *rq);
1122        void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1123        void (*task_fork) (struct task_struct *p);
1124
1125        void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1126        void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1127        void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1128                             int oldprio);
1129
1130        unsigned int (*get_rr_interval) (struct rq *rq,
1131                                         struct task_struct *task);
1132
1133#ifdef CONFIG_FAIR_GROUP_SCHED
1134        void (*task_move_group) (struct task_struct *p, int on_rq);
1135#endif
1136};
1137
1138struct load_weight {
1139        unsigned long weight, inv_weight;
1140};
1141
1142#ifdef CONFIG_SCHEDSTATS
1143struct sched_statistics {
1144        u64                     wait_start;
1145        u64                     wait_max;
1146        u64                     wait_count;
1147        u64                     wait_sum;
1148        u64                     iowait_count;
1149        u64                     iowait_sum;
1150
1151        u64                     sleep_start;
1152        u64                     sleep_max;
1153        s64                     sum_sleep_runtime;
1154
1155        u64                     block_start;
1156        u64                     block_max;
1157        u64                     exec_max;
1158        u64                     slice_max;
1159
1160        u64                     nr_migrations_cold;
1161        u64                     nr_failed_migrations_affine;
1162        u64                     nr_failed_migrations_running;
1163        u64                     nr_failed_migrations_hot;
1164        u64                     nr_forced_migrations;
1165
1166        u64                     nr_wakeups;
1167        u64                     nr_wakeups_sync;
1168        u64                     nr_wakeups_migrate;
1169        u64                     nr_wakeups_local;
1170        u64                     nr_wakeups_remote;
1171        u64                     nr_wakeups_affine;
1172        u64                     nr_wakeups_affine_attempts;
1173        u64                     nr_wakeups_passive;
1174        u64                     nr_wakeups_idle;
1175};
1176#endif
1177
1178struct sched_entity {
1179        struct load_weight      load;           /* for load-balancing */
1180        struct rb_node          run_node;
1181        struct list_head        group_node;
1182        unsigned int            on_rq;
1183
1184        u64                     exec_start;
1185        u64                     sum_exec_runtime;
1186        u64                     vruntime;
1187        u64                     prev_sum_exec_runtime;
1188
1189        u64                     nr_migrations;
1190
1191#ifdef CONFIG_SCHEDSTATS
1192        struct sched_statistics statistics;
1193#endif
1194
1195#ifdef CONFIG_FAIR_GROUP_SCHED
1196        struct sched_entity     *parent;
1197        /* rq on which this entity is (to be) queued: */
1198        struct cfs_rq           *cfs_rq;
1199        /* rq "owned" by this entity/group: */
1200        struct cfs_rq           *my_q;
1201#endif
1202};
1203
1204struct sched_rt_entity {
1205        struct list_head run_list;
1206        unsigned long timeout;
1207        unsigned int time_slice;
1208
1209        struct sched_rt_entity *back;
1210#ifdef CONFIG_RT_GROUP_SCHED
1211        struct sched_rt_entity  *parent;
1212        /* rq on which this entity is (to be) queued: */
1213        struct rt_rq            *rt_rq;
1214        /* rq "owned" by this entity/group: */
1215        struct rt_rq            *my_q;
1216#endif
1217};
1218
1219/*
1220 * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1221 * Timeslices get refilled after they expire.
1222 */
1223#define RR_TIMESLICE            (100 * HZ / 1000)
1224
1225struct rcu_node;
1226
1227enum perf_event_task_context {
1228        perf_invalid_context = -1,
1229        perf_hw_context = 0,
1230        perf_sw_context,
1231        perf_nr_task_contexts,
1232};
1233
1234struct task_struct {
1235        volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1236        void *stack;
1237        atomic_t usage;
1238        unsigned int flags;     /* per process flags, defined below */
1239        unsigned int ptrace;
1240
1241#ifdef CONFIG_SMP
1242        struct llist_node wake_entry;
1243        int on_cpu;
1244#endif
1245        int on_rq;
1246
1247        int prio, static_prio, normal_prio;
1248        unsigned int rt_priority;
1249        const struct sched_class *sched_class;
1250        struct sched_entity se;
1251        struct sched_rt_entity rt;
1252#ifdef CONFIG_CGROUP_SCHED
1253        struct task_group *sched_task_group;
1254#endif
1255
1256#ifdef CONFIG_PREEMPT_NOTIFIERS
1257        /* list of struct preempt_notifier: */
1258        struct hlist_head preempt_notifiers;
1259#endif
1260
1261        /*
1262         * fpu_counter contains the number of consecutive context switches
1263         * that the FPU is used. If this is over a threshold, the lazy fpu
1264         * saving becomes unlazy to save the trap. This is an unsigned char
1265         * so that after 256 times the counter wraps and the behavior turns
1266         * lazy again; this to deal with bursty apps that only use FPU for
1267         * a short time
1268         */
1269        unsigned char fpu_counter;
1270#ifdef CONFIG_BLK_DEV_IO_TRACE
1271        unsigned int btrace_seq;
1272#endif
1273
1274        unsigned int policy;
1275        int nr_cpus_allowed;
1276        cpumask_t cpus_allowed;
1277
1278#ifdef CONFIG_PREEMPT_RCU
1279        int rcu_read_lock_nesting;
1280        char rcu_read_unlock_special;
1281        struct list_head rcu_node_entry;
1282#endif /* #ifdef CONFIG_PREEMPT_RCU */
1283#ifdef CONFIG_TREE_PREEMPT_RCU
1284        struct rcu_node *rcu_blocked_node;
1285#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1286#ifdef CONFIG_RCU_BOOST
1287        struct rt_mutex *rcu_boost_mutex;
1288#endif /* #ifdef CONFIG_RCU_BOOST */
1289
1290#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1291        struct sched_info sched_info;
1292#endif
1293
1294        struct list_head tasks;
1295#ifdef CONFIG_SMP
1296        struct plist_node pushable_tasks;
1297#endif
1298
1299        struct mm_struct *mm, *active_mm;
1300#ifdef CONFIG_COMPAT_BRK
1301        unsigned brk_randomized:1;
1302#endif
1303#if defined(SPLIT_RSS_COUNTING)
1304        struct task_rss_stat    rss_stat;
1305#endif
1306/* task state */
1307        int exit_state;
1308        int exit_code, exit_signal;
1309        int pdeath_signal;  /*  The signal sent when the parent dies  */
1310        unsigned int jobctl;    /* JOBCTL_*, siglock protected */
1311        /* ??? */
1312        unsigned int personality;
1313        unsigned did_exec:1;
1314        unsigned in_execve:1;   /* Tell the LSMs that the process is doing an
1315                                 * execve */
1316        unsigned in_iowait:1;
1317
1318        /* task may not gain privileges */
1319        unsigned no_new_privs:1;
1320
1321        /* Revert to default priority/policy when forking */
1322        unsigned sched_reset_on_fork:1;
1323        unsigned sched_contributes_to_load:1;
1324
1325        pid_t pid;
1326        pid_t tgid;
1327
1328#ifdef CONFIG_CC_STACKPROTECTOR
1329        /* Canary value for the -fstack-protector gcc feature */
1330        unsigned long stack_canary;
1331#endif
1332        /*
1333         * pointers to (original) parent process, youngest child, younger sibling,
1334         * older sibling, respectively.  (p->father can be replaced with
1335         * p->real_parent->pid)
1336         */
1337        struct task_struct __rcu *real_parent; /* real parent process */
1338        struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1339        /*
1340         * children/sibling forms the list of my natural children
1341         */
1342        struct list_head children;      /* list of my children */
1343        struct list_head sibling;       /* linkage in my parent's children list */
1344        struct task_struct *group_leader;       /* threadgroup leader */
1345
1346        /*
1347         * ptraced is the list of tasks this task is using ptrace on.
1348         * This includes both natural children and PTRACE_ATTACH targets.
1349         * p->ptrace_entry is p's link on the p->parent->ptraced list.
1350         */
1351        struct list_head ptraced;
1352        struct list_head ptrace_entry;
1353
1354        /* PID/PID hash table linkage. */
1355        struct pid_link pids[PIDTYPE_MAX];
1356        struct list_head thread_group;
1357
1358        struct completion *vfork_done;          /* for vfork() */
1359        int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1360        int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1361
1362        cputime_t utime, stime, utimescaled, stimescaled;
1363        cputime_t gtime;
1364#ifndef CONFIG_VIRT_CPU_ACCOUNTING
1365        cputime_t prev_utime, prev_stime;
1366#endif
1367        unsigned long nvcsw, nivcsw; /* context switch counts */
1368        struct timespec start_time;             /* monotonic time */
1369        struct timespec real_start_time;        /* boot based time */
1370/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1371        unsigned long min_flt, maj_flt;
1372
1373        struct task_cputime cputime_expires;
1374        struct list_head cpu_timers[3];
1375
1376/* process credentials */
1377        const struct cred __rcu *real_cred; /* objective and real subjective task
1378                                         * credentials (COW) */
1379        const struct cred __rcu *cred;  /* effective (overridable) subjective task
1380                                         * credentials (COW) */
1381        char comm[TASK_COMM_LEN]; /* executable name excluding path
1382                                     - access with [gs]et_task_comm (which lock
1383                                       it with task_lock())
1384                                     - initialized normally by setup_new_exec */
1385/* file system info */
1386        int link_count, total_link_count;
1387#ifdef CONFIG_SYSVIPC
1388/* ipc stuff */
1389        struct sysv_sem sysvsem;
1390#endif
1391#ifdef CONFIG_DETECT_HUNG_TASK
1392/* hung task detection */
1393        unsigned long last_switch_count;
1394#endif
1395/* CPU-specific state of this task */
1396        struct thread_struct thread;
1397/* filesystem information */
1398        struct fs_struct *fs;
1399/* open file information */
1400        struct files_struct *files;
1401/* namespaces */
1402        struct nsproxy *nsproxy;
1403/* signal handlers */
1404        struct signal_struct *signal;
1405        struct sighand_struct *sighand;
1406
1407        sigset_t blocked, real_blocked;
1408        sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1409        struct sigpending pending;
1410
1411        unsigned long sas_ss_sp;
1412        size_t sas_ss_size;
1413        int (*notifier)(void *priv);
1414        void *notifier_data;
1415        sigset_t *notifier_mask;
1416        struct callback_head *task_works;
1417
1418        struct audit_context *audit_context;
1419#ifdef CONFIG_AUDITSYSCALL
1420        uid_t loginuid;
1421        unsigned int sessionid;
1422#endif
1423        struct seccomp seccomp;
1424
1425/* Thread group tracking */
1426        u32 parent_exec_id;
1427        u32 self_exec_id;
1428/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1429 * mempolicy */
1430        spinlock_t alloc_lock;
1431
1432        /* Protection of the PI data structures: */
1433        raw_spinlock_t pi_lock;
1434
1435#ifdef CONFIG_RT_MUTEXES
1436        /* PI waiters blocked on a rt_mutex held by this task */
1437        struct plist_head pi_waiters;
1438        /* Deadlock detection and priority inheritance handling */
1439        struct rt_mutex_waiter *pi_blocked_on;
1440#endif
1441
1442#ifdef CONFIG_DEBUG_MUTEXES
1443        /* mutex deadlock detection */
1444        struct mutex_waiter *blocked_on;
1445#endif
1446#ifdef CONFIG_TRACE_IRQFLAGS
1447        unsigned int irq_events;
1448        unsigned long hardirq_enable_ip;
1449        unsigned long hardirq_disable_ip;
1450        unsigned int hardirq_enable_event;
1451        unsigned int hardirq_disable_event;
1452        int hardirqs_enabled;
1453        int hardirq_context;
1454        unsigned long softirq_disable_ip;
1455        unsigned long softirq_enable_ip;
1456        unsigned int softirq_disable_event;
1457        unsigned int softirq_enable_event;
1458        int softirqs_enabled;
1459        int softirq_context;
1460#endif
1461#ifdef CONFIG_LOCKDEP
1462# define MAX_LOCK_DEPTH 48UL
1463        u64 curr_chain_key;
1464        int lockdep_depth;
1465        unsigned int lockdep_recursion;
1466        struct held_lock held_locks[MAX_LOCK_DEPTH];
1467        gfp_t lockdep_reclaim_gfp;
1468#endif
1469
1470/* journalling filesystem info */
1471        void *journal_info;
1472
1473/* stacked block device info */
1474        struct bio_list *bio_list;
1475
1476#ifdef CONFIG_BLOCK
1477/* stack plugging */
1478        struct blk_plug *plug;
1479#endif
1480
1481/* VM state */
1482        struct reclaim_state *reclaim_state;
1483
1484        struct backing_dev_info *backing_dev_info;
1485
1486        struct io_context *io_context;
1487
1488        unsigned long ptrace_message;
1489        siginfo_t *last_siginfo; /* For ptrace use.  */
1490        struct task_io_accounting ioac;
1491#if defined(CONFIG_TASK_XACCT)
1492        u64 acct_rss_mem1;      /* accumulated rss usage */
1493        u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1494        cputime_t acct_timexpd; /* stime + utime since last update */
1495#endif
1496#ifdef CONFIG_CPUSETS
1497        nodemask_t mems_allowed;        /* Protected by alloc_lock */
1498        seqcount_t mems_allowed_seq;    /* Seqence no to catch updates */
1499        int cpuset_mem_spread_rotor;
1500        int cpuset_slab_spread_rotor;
1501#endif
1502#ifdef CONFIG_CGROUPS
1503        /* Control Group info protected by css_set_lock */
1504        struct css_set __rcu *cgroups;
1505        /* cg_list protected by css_set_lock and tsk->alloc_lock */
1506        struct list_head cg_list;
1507#endif
1508#ifdef CONFIG_FUTEX
1509        struct robust_list_head __user *robust_list;
1510#ifdef CONFIG_COMPAT
1511        struct compat_robust_list_head __user *compat_robust_list;
1512#endif
1513        struct list_head pi_state_list;
1514        struct futex_pi_state *pi_state_cache;
1515#endif
1516#ifdef CONFIG_PERF_EVENTS
1517        struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1518        struct mutex perf_event_mutex;
1519        struct list_head perf_event_list;
1520#endif
1521#ifdef CONFIG_NUMA
1522        struct mempolicy *mempolicy;    /* Protected by alloc_lock */
1523        short il_next;
1524        short pref_node_fork;
1525#endif
1526        struct rcu_head rcu;
1527
1528        /*
1529         * cache last used pipe for splice
1530         */
1531        struct pipe_inode_info *splice_pipe;
1532#ifdef  CONFIG_TASK_DELAY_ACCT
1533        struct task_delay_info *delays;
1534#endif
1535#ifdef CONFIG_FAULT_INJECTION
1536        int make_it_fail;
1537#endif
1538        /*
1539         * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1540         * balance_dirty_pages() for some dirty throttling pause
1541         */
1542        int nr_dirtied;
1543        int nr_dirtied_pause;
1544        unsigned long dirty_paused_when; /* start of a write-and-pause period */
1545
1546#ifdef CONFIG_LATENCYTOP
1547        int latency_record_count;
1548        struct latency_record latency_record[LT_SAVECOUNT];
1549#endif
1550        /*
1551         * time slack values; these are used to round up poll() and
1552         * select() etc timeout values. These are in nanoseconds.
1553         */
1554        unsigned long timer_slack_ns;
1555        unsigned long default_timer_slack_ns;
1556
1557#ifdef CONFIG_FUNCTION_GRAPH_TRACER
1558        /* Index of current stored address in ret_stack */
1559        int curr_ret_stack;
1560        /* Stack of return addresses for return function tracing */
1561        struct ftrace_ret_stack *ret_stack;
1562        /* time stamp for last schedule */
1563        unsigned long long ftrace_timestamp;
1564        /*
1565         * Number of functions that haven't been traced
1566         * because of depth overrun.
1567         */
1568        atomic_t trace_overrun;
1569        /* Pause for the tracing */
1570        atomic_t tracing_graph_pause;
1571#endif
1572#ifdef CONFIG_TRACING
1573        /* state flags for use by tracers */
1574        unsigned long trace;
1575        /* bitmask and counter of trace recursion */
1576        unsigned long trace_recursion;
1577#endif /* CONFIG_TRACING */
1578#ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1579        struct memcg_batch_info {
1580                int do_batch;   /* incremented when batch uncharge started */
1581                struct mem_cgroup *memcg; /* target memcg of uncharge */
1582                unsigned long nr_pages; /* uncharged usage */
1583                unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1584        } memcg_batch;
1585#endif
1586#ifdef CONFIG_HAVE_HW_BREAKPOINT
1587        atomic_t ptrace_bp_refcnt;
1588#endif
1589#ifdef CONFIG_UPROBES
1590        struct uprobe_task *utask;
1591#endif
1592};
1593
1594/* Future-safe accessor for struct task_struct's cpus_allowed. */
1595#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1596
1597/*
1598 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1599 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1600 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1601 * values are inverted: lower p->prio value means higher priority.
1602 *
1603 * The MAX_USER_RT_PRIO value allows the actual maximum
1604 * RT priority to be separate from the value exported to
1605 * user-space.  This allows kernel threads to set their
1606 * priority to a value higher than any user task. Note:
1607 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1608 */
1609
1610#define MAX_USER_RT_PRIO        100
1611#define MAX_RT_PRIO             MAX_USER_RT_PRIO
1612
1613#define MAX_PRIO                (MAX_RT_PRIO + 40)
1614#define DEFAULT_PRIO            (MAX_RT_PRIO + 20)
1615
1616static inline int rt_prio(int prio)
1617{
1618        if (unlikely(prio < MAX_RT_PRIO))
1619                return 1;
1620        return 0;
1621}
1622
1623static inline int rt_task(struct task_struct *p)
1624{
1625        return rt_prio(p->prio);
1626}
1627
1628static inline struct pid *task_pid(struct task_struct *task)
1629{
1630        return task->pids[PIDTYPE_PID].pid;
1631}
1632
1633static inline struct pid *task_tgid(struct task_struct *task)
1634{
1635        return task->group_leader->pids[PIDTYPE_PID].pid;
1636}
1637
1638/*
1639 * Without tasklist or rcu lock it is not safe to dereference
1640 * the result of task_pgrp/task_session even if task == current,
1641 * we can race with another thread doing sys_setsid/sys_setpgid.
1642 */
1643static inline struct pid *task_pgrp(struct task_struct *task)
1644{
1645        return task->group_leader->pids[PIDTYPE_PGID].pid;
1646}
1647
1648static inline struct pid *task_session(struct task_struct *task)
1649{
1650        return task->group_leader->pids[PIDTYPE_SID].pid;
1651}
1652
1653struct pid_namespace;
1654
1655/*
1656 * the helpers to get the task's different pids as they are seen
1657 * from various namespaces
1658 *
1659 * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1660 * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1661 *                     current.
1662 * task_xid_nr_ns()  : id seen from the ns specified;
1663 *
1664 * set_task_vxid()   : assigns a virtual id to a task;
1665 *
1666 * see also pid_nr() etc in include/linux/pid.h
1667 */
1668pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1669                        struct pid_namespace *ns);
1670
1671static inline pid_t task_pid_nr(struct task_struct *tsk)
1672{
1673        return tsk->pid;
1674}
1675
1676static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1677                                        struct pid_namespace *ns)
1678{
1679        return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1680}
1681
1682static inline pid_t task_pid_vnr(struct task_struct *tsk)
1683{
1684        return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1685}
1686
1687
1688static inline pid_t task_tgid_nr(struct task_struct *tsk)
1689{
1690        return tsk->tgid;
1691}
1692
1693pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1694
1695static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1696{
1697        return pid_vnr(task_tgid(tsk));
1698}
1699
1700
1701static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1702                                        struct pid_namespace *ns)
1703{
1704        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1705}
1706
1707static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1708{
1709        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1710}
1711
1712
1713static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1714                                        struct pid_namespace *ns)
1715{
1716        return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1717}
1718
1719static inline pid_t task_session_vnr(struct task_struct *tsk)
1720{
1721        return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1722}
1723
1724/* obsolete, do not use */
1725static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1726{
1727        return task_pgrp_nr_ns(tsk, &init_pid_ns);
1728}
1729
1730/**
1731 * pid_alive - check that a task structure is not stale
1732 * @p: Task structure to be checked.
1733 *
1734 * Test if a process is not yet dead (at most zombie state)
1735 * If pid_alive fails, then pointers within the task structure
1736 * can be stale and must not be dereferenced.
1737 */
1738static inline int pid_alive(struct task_struct *p)
1739{
1740        return p->pids[PIDTYPE_PID].pid != NULL;
1741}
1742
1743/**
1744 * is_global_init - check if a task structure is init
1745 * @tsk: Task structure to be checked.
1746 *
1747 * Check if a task structure is the first user space task the kernel created.
1748 */
1749static inline int is_global_init(struct task_struct *tsk)
1750{
1751        return tsk->pid == 1;
1752}
1753
1754/*
1755 * is_container_init:
1756 * check whether in the task is init in its own pid namespace.
1757 */
1758extern int is_container_init(struct task_struct *tsk);
1759
1760extern struct pid *cad_pid;
1761
1762extern void free_task(struct task_struct *tsk);
1763#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1764
1765extern void __put_task_struct(struct task_struct *t);
1766
1767static inline void put_task_struct(struct task_struct *t)
1768{
1769        if (atomic_dec_and_test(&t->usage))
1770                __put_task_struct(t);
1771}
1772
1773extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1774extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1775
1776/*
1777 * Per process flags
1778 */
1779#define PF_EXITING      0x00000004      /* getting shut down */
1780#define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
1781#define PF_VCPU         0x00000010      /* I'm a virtual CPU */
1782#define PF_WQ_WORKER    0x00000020      /* I'm a workqueue worker */
1783#define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
1784#define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1785#define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
1786#define PF_DUMPCORE     0x00000200      /* dumped core */
1787#define PF_SIGNALED     0x00000400      /* killed by a signal */
1788#define PF_MEMALLOC     0x00000800      /* Allocating memory */
1789#define PF_NPROC_EXCEEDED 0x00001000    /* set_user noticed that RLIMIT_NPROC was exceeded */
1790#define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
1791#define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
1792#define PF_FROZEN       0x00010000      /* frozen for system suspend */
1793#define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
1794#define PF_KSWAPD       0x00040000      /* I am kswapd */
1795#define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
1796#define PF_KTHREAD      0x00200000      /* I am a kernel thread */
1797#define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
1798#define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
1799#define PF_SPREAD_PAGE  0x01000000      /* Spread page cache over cpuset */
1800#define PF_SPREAD_SLAB  0x02000000      /* Spread some slab caches over cpuset */
1801#define PF_THREAD_BOUND 0x04000000      /* Thread bound to specific cpu */
1802#define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1803#define PF_MEMPOLICY    0x10000000      /* Non-default NUMA mempolicy */
1804#define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
1805#define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezable */
1806
1807/*
1808 * Only the _current_ task can read/write to tsk->flags, but other
1809 * tasks can access tsk->flags in readonly mode for example
1810 * with tsk_used_math (like during threaded core dumping).
1811 * There is however an exception to this rule during ptrace
1812 * or during fork: the ptracer task is allowed to write to the
1813 * child->flags of its traced child (same goes for fork, the parent
1814 * can write to the child->flags), because we're guaranteed the
1815 * child is not running and in turn not changing child->flags
1816 * at the same time the parent does it.
1817 */
1818#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1819#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1820#define clear_used_math() clear_stopped_child_used_math(current)
1821#define set_used_math() set_stopped_child_used_math(current)
1822#define conditional_stopped_child_used_math(condition, child) \
1823        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1824#define conditional_used_math(condition) \
1825        conditional_stopped_child_used_math(condition, current)
1826#define copy_to_stopped_child_used_math(child) \
1827        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1828/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1829#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1830#define used_math() tsk_used_math(current)
1831
1832/*
1833 * task->jobctl flags
1834 */
1835#define JOBCTL_STOP_SIGMASK     0xffff  /* signr of the last group stop */
1836
1837#define JOBCTL_STOP_DEQUEUED_BIT 16     /* stop signal dequeued */
1838#define JOBCTL_STOP_PENDING_BIT 17      /* task should stop for group stop */
1839#define JOBCTL_STOP_CONSUME_BIT 18      /* consume group stop count */
1840#define JOBCTL_TRAP_STOP_BIT    19      /* trap for STOP */
1841#define JOBCTL_TRAP_NOTIFY_BIT  20      /* trap for NOTIFY */
1842#define JOBCTL_TRAPPING_BIT     21      /* switching to TRACED */
1843#define JOBCTL_LISTENING_BIT    22      /* ptracer is listening for events */
1844
1845#define JOBCTL_STOP_DEQUEUED    (1 << JOBCTL_STOP_DEQUEUED_BIT)
1846#define JOBCTL_STOP_PENDING     (1 << JOBCTL_STOP_PENDING_BIT)
1847#define JOBCTL_STOP_CONSUME     (1 << JOBCTL_STOP_CONSUME_BIT)
1848#define JOBCTL_TRAP_STOP        (1 << JOBCTL_TRAP_STOP_BIT)
1849#define JOBCTL_TRAP_NOTIFY      (1 << JOBCTL_TRAP_NOTIFY_BIT)
1850#define JOBCTL_TRAPPING         (1 << JOBCTL_TRAPPING_BIT)
1851#define JOBCTL_LISTENING        (1 << JOBCTL_LISTENING_BIT)
1852
1853#define JOBCTL_TRAP_MASK        (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1854#define JOBCTL_PENDING_MASK     (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1855
1856extern bool task_set_jobctl_pending(struct task_struct *task,
1857                                    unsigned int mask);
1858extern void task_clear_jobctl_trapping(struct task_struct *task);
1859extern void task_clear_jobctl_pending(struct task_struct *task,
1860                                      unsigned int mask);
1861
1862#ifdef CONFIG_PREEMPT_RCU
1863
1864#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1865#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1866
1867static inline void rcu_copy_process(struct task_struct *p)
1868{
1869        p->rcu_read_lock_nesting = 0;
1870        p->rcu_read_unlock_special = 0;
1871#ifdef CONFIG_TREE_PREEMPT_RCU
1872        p->rcu_blocked_node = NULL;
1873#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1874#ifdef CONFIG_RCU_BOOST
1875        p->rcu_boost_mutex = NULL;
1876#endif /* #ifdef CONFIG_RCU_BOOST */
1877        INIT_LIST_HEAD(&p->rcu_node_entry);
1878}
1879
1880#else
1881
1882static inline void rcu_copy_process(struct task_struct *p)
1883{
1884}
1885
1886#endif
1887
1888static inline void tsk_restore_flags(struct task_struct *task,
1889                                unsigned long orig_flags, unsigned long flags)
1890{
1891        task->flags &= ~flags;
1892        task->flags |= orig_flags & flags;
1893}
1894
1895#ifdef CONFIG_SMP
1896extern void do_set_cpus_allowed(struct task_struct *p,
1897                               const struct cpumask *new_mask);
1898
1899extern int set_cpus_allowed_ptr(struct task_struct *p,
1900                                const struct cpumask *new_mask);
1901#else
1902static inline void do_set_cpus_allowed(struct task_struct *p,
1903                                      const struct cpumask *new_mask)
1904{
1905}
1906static inline int set_cpus_allowed_ptr(struct task_struct *p,
1907                                       const struct cpumask *new_mask)
1908{
1909        if (!cpumask_test_cpu(0, new_mask))
1910                return -EINVAL;
1911        return 0;
1912}
1913#endif
1914
1915#ifdef CONFIG_NO_HZ
1916void calc_load_enter_idle(void);
1917void calc_load_exit_idle(void);
1918#else
1919static inline void calc_load_enter_idle(void) { }
1920static inline void calc_load_exit_idle(void) { }
1921#endif /* CONFIG_NO_HZ */
1922
1923#ifndef CONFIG_CPUMASK_OFFSTACK
1924static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1925{
1926        return set_cpus_allowed_ptr(p, &new_mask);
1927}
1928#endif
1929
1930/*
1931 * Do not use outside of architecture code which knows its limitations.
1932 *
1933 * sched_clock() has no promise of monotonicity or bounded drift between
1934 * CPUs, use (which you should not) requires disabling IRQs.
1935 *
1936 * Please use one of the three interfaces below.
1937 */
1938extern unsigned long long notrace sched_clock(void);
1939/*
1940 * See the comment in kernel/sched/clock.c
1941 */
1942extern u64 cpu_clock(int cpu);
1943extern u64 local_clock(void);
1944extern u64 sched_clock_cpu(int cpu);
1945
1946
1947extern void sched_clock_init(void);
1948
1949#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1950static inline void sched_clock_tick(void)
1951{
1952}
1953
1954static inline void sched_clock_idle_sleep_event(void)
1955{
1956}
1957
1958static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1959{
1960}
1961#else
1962/*
1963 * Architectures can set this to 1 if they have specified
1964 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1965 * but then during bootup it turns out that sched_clock()
1966 * is reliable after all:
1967 */
1968extern int sched_clock_stable;
1969
1970extern void sched_clock_tick(void);
1971extern void sched_clock_idle_sleep_event(void);
1972extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1973#endif
1974
1975#ifdef CONFIG_IRQ_TIME_ACCOUNTING
1976/*
1977 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1978 * The reason for this explicit opt-in is not to have perf penalty with
1979 * slow sched_clocks.
1980 */
1981extern void enable_sched_clock_irqtime(void);
1982extern void disable_sched_clock_irqtime(void);
1983#else
1984static inline void enable_sched_clock_irqtime(void) {}
1985static inline void disable_sched_clock_irqtime(void) {}
1986#endif
1987
1988extern unsigned long long
1989task_sched_runtime(struct task_struct *task);
1990
1991/* sched_exec is called by processes performing an exec */
1992#ifdef CONFIG_SMP
1993extern void sched_exec(void);
1994#else
1995#define sched_exec()   {}
1996#endif
1997
1998extern void sched_clock_idle_sleep_event(void);
1999extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2000
2001#ifdef CONFIG_HOTPLUG_CPU
2002extern void idle_task_exit(void);
2003#else
2004static inline void idle_task_exit(void) {}
2005#endif
2006
2007#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2008extern void wake_up_idle_cpu(int cpu);
2009#else
2010static inline void wake_up_idle_cpu(int cpu) { }
2011#endif
2012
2013extern unsigned int sysctl_sched_latency;
2014extern unsigned int sysctl_sched_min_granularity;
2015extern unsigned int sysctl_sched_wakeup_granularity;
2016extern unsigned int sysctl_sched_child_runs_first;
2017
2018enum sched_tunable_scaling {
2019        SCHED_TUNABLESCALING_NONE,
2020        SCHED_TUNABLESCALING_LOG,
2021        SCHED_TUNABLESCALING_LINEAR,
2022        SCHED_TUNABLESCALING_END,
2023};
2024extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2025
2026#ifdef CONFIG_SCHED_DEBUG
2027extern unsigned int sysctl_sched_migration_cost;
2028extern unsigned int sysctl_sched_nr_migrate;
2029extern unsigned int sysctl_sched_time_avg;
2030extern unsigned int sysctl_timer_migration;
2031extern unsigned int sysctl_sched_shares_window;
2032
2033int sched_proc_update_handler(struct ctl_table *table, int write,
2034                void __user *buffer, size_t *length,
2035                loff_t *ppos);
2036#endif
2037#ifdef CONFIG_SCHED_DEBUG
2038static inline unsigned int get_sysctl_timer_migration(void)
2039{
2040        return sysctl_timer_migration;
2041}
2042#else
2043static inline unsigned int get_sysctl_timer_migration(void)
2044{
2045        return 1;
2046}
2047#endif
2048extern unsigned int sysctl_sched_rt_period;
2049extern int sysctl_sched_rt_runtime;
2050
2051int sched_rt_handler(struct ctl_table *table, int write,
2052                void __user *buffer, size_t *lenp,
2053                loff_t *ppos);
2054
2055#ifdef CONFIG_SCHED_AUTOGROUP
2056extern unsigned int sysctl_sched_autogroup_enabled;
2057
2058extern void sched_autogroup_create_attach(struct task_struct *p);
2059extern void sched_autogroup_detach(struct task_struct *p);
2060extern void sched_autogroup_fork(struct signal_struct *sig);
2061extern void sched_autogroup_exit(struct signal_struct *sig);
2062#ifdef CONFIG_PROC_FS
2063extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2064extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2065#endif
2066#else
2067static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2068static inline void sched_autogroup_detach(struct task_struct *p) { }
2069static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2070static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2071#endif
2072
2073#ifdef CONFIG_CFS_BANDWIDTH
2074extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2075#endif
2076
2077#ifdef CONFIG_RT_MUTEXES
2078extern int rt_mutex_getprio(struct task_struct *p);
2079extern void rt_mutex_setprio(struct task_struct *p, int prio);
2080extern void rt_mutex_adjust_pi(struct task_struct *p);
2081static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2082{
2083        return tsk->pi_blocked_on != NULL;
2084}
2085#else
2086static inline int rt_mutex_getprio(struct task_struct *p)
2087{
2088        return p->normal_prio;
2089}
2090# define rt_mutex_adjust_pi(p)          do { } while (0)
2091static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2092{
2093        return false;
2094}
2095#endif
2096
2097extern bool yield_to(struct task_struct *p, bool preempt);
2098extern void set_user_nice(struct task_struct *p, long nice);
2099extern int task_prio(const struct task_struct *p);
2100extern int task_nice(const struct task_struct *p);
2101extern int can_nice(const struct task_struct *p, const int nice);
2102extern int task_curr(const struct task_struct *p);
2103extern int idle_cpu(int cpu);
2104extern int sched_setscheduler(struct task_struct *, int,
2105                              const struct sched_param *);
2106extern int sched_setscheduler_nocheck(struct task_struct *, int,
2107                                      const struct sched_param *);
2108extern struct task_struct *idle_task(int cpu);
2109/**
2110 * is_idle_task - is the specified task an idle task?
2111 * @p: the task in question.
2112 */
2113static inline bool is_idle_task(const struct task_struct *p)
2114{
2115        return p->pid == 0;
2116}
2117extern struct task_struct *curr_task(int cpu);
2118extern void set_curr_task(int cpu, struct task_struct *p);
2119
2120void yield(void);
2121
2122/*
2123 * The default (Linux) execution domain.
2124 */
2125extern struct exec_domain       default_exec_domain;
2126
2127union thread_union {
2128        struct thread_info thread_info;
2129        unsigned long stack[THREAD_SIZE/sizeof(long)];
2130};
2131
2132#ifndef __HAVE_ARCH_KSTACK_END
2133static inline int kstack_end(void *addr)
2134{
2135        /* Reliable end of stack detection:
2136         * Some APM bios versions misalign the stack
2137         */
2138        return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2139}
2140#endif
2141
2142extern union thread_union init_thread_union;
2143extern struct task_struct init_task;
2144
2145extern struct   mm_struct init_mm;
2146
2147extern struct pid_namespace init_pid_ns;
2148
2149/*
2150 * find a task by one of its numerical ids
2151 *
2152 * find_task_by_pid_ns():
2153 *      finds a task by its pid in the specified namespace
2154 * find_task_by_vpid():
2155 *      finds a task by its virtual pid
2156 *
2157 * see also find_vpid() etc in include/linux/pid.h
2158 */
2159
2160extern struct task_struct *find_task_by_vpid(pid_t nr);
2161extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2162                struct pid_namespace *ns);
2163
2164extern void __set_special_pids(struct pid *pid);
2165
2166/* per-UID process charging. */
2167extern struct user_struct * alloc_uid(kuid_t);
2168static inline struct user_struct *get_uid(struct user_struct *u)
2169{
2170        atomic_inc(&u->__count);
2171        return u;
2172}
2173extern void free_uid(struct user_struct *);
2174
2175#include <asm/current.h>
2176
2177extern void xtime_update(unsigned long ticks);
2178
2179extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2180extern int wake_up_process(struct task_struct *tsk);
2181extern void wake_up_new_task(struct task_struct *tsk);
2182#ifdef CONFIG_SMP
2183 extern void kick_process(struct task_struct *tsk);
2184#else
2185 static inline void kick_process(struct task_struct *tsk) { }
2186#endif
2187extern void sched_fork(struct task_struct *p);
2188extern void sched_dead(struct task_struct *p);
2189
2190extern void proc_caches_init(void);
2191extern void flush_signals(struct task_struct *);
2192extern void __flush_signals(struct task_struct *);
2193extern void ignore_signals(struct task_struct *);
2194extern void flush_signal_handlers(struct task_struct *, int force_default);
2195extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2196
2197static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2198{
2199        unsigned long flags;
2200        int ret;
2201
2202        spin_lock_irqsave(&tsk->sighand->siglock, flags);
2203        ret = dequeue_signal(tsk, mask, info);
2204        spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2205
2206        return ret;
2207}
2208
2209extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2210                              sigset_t *mask);
2211extern void unblock_all_signals(void);
2212extern void release_task(struct task_struct * p);
2213extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2214extern int force_sigsegv(int, struct task_struct *);
2215extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2216extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2217extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2218extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2219                                const struct cred *, u32);
2220extern int kill_pgrp(struct pid *pid, int sig, int priv);
2221extern int kill_pid(struct pid *pid, int sig, int priv);
2222extern int kill_proc_info(int, struct siginfo *, pid_t);
2223extern __must_check bool do_notify_parent(struct task_struct *, int);
2224extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2225extern void force_sig(int, struct task_struct *);
2226extern int send_sig(int, struct task_struct *, int);
2227extern int zap_other_threads(struct task_struct *p);
2228extern struct sigqueue *sigqueue_alloc(void);
2229extern void sigqueue_free(struct sigqueue *);
2230extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2231extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2232extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2233
2234static inline void restore_saved_sigmask(void)
2235{
2236        if (test_and_clear_restore_sigmask())
2237                __set_current_blocked(&current->saved_sigmask);
2238}
2239
2240static inline sigset_t *sigmask_to_save(void)
2241{
2242        sigset_t *res = &current->blocked;
2243        if (unlikely(test_restore_sigmask()))
2244                res = &current->saved_sigmask;
2245        return res;
2246}
2247
2248static inline int kill_cad_pid(int sig, int priv)
2249{
2250        return kill_pid(cad_pid, sig, priv);
2251}
2252
2253/* These can be the second arg to send_sig_info/send_group_sig_info.  */
2254#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2255#define SEND_SIG_PRIV   ((struct siginfo *) 1)
2256#define SEND_SIG_FORCED ((struct siginfo *) 2)
2257
2258/*
2259 * True if we are on the alternate signal stack.
2260 */
2261static inline int on_sig_stack(unsigned long sp)
2262{
2263#ifdef CONFIG_STACK_GROWSUP
2264        return sp >= current->sas_ss_sp &&
2265                sp - current->sas_ss_sp < current->sas_ss_size;
2266#else
2267        return sp > current->sas_ss_sp &&
2268                sp - current->sas_ss_sp <= current->sas_ss_size;
2269#endif
2270}
2271
2272static inline int sas_ss_flags(unsigned long sp)
2273{
2274        return (current->sas_ss_size == 0 ? SS_DISABLE
2275                : on_sig_stack(sp) ? SS_ONSTACK : 0);
2276}
2277
2278/*
2279 * Routines for handling mm_structs
2280 */
2281extern struct mm_struct * mm_alloc(void);
2282
2283/* mmdrop drops the mm and the page tables */
2284extern void __mmdrop(struct mm_struct *);
2285static inline void mmdrop(struct mm_struct * mm)
2286{
2287        if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2288                __mmdrop(mm);
2289}
2290
2291/* mmput gets rid of the mappings and all user-space */
2292extern void mmput(struct mm_struct *);
2293/* Grab a reference to a task's mm, if it is not already going away */
2294extern struct mm_struct *get_task_mm(struct task_struct *task);
2295/*
2296 * Grab a reference to a task's mm, if it is not already going away
2297 * and ptrace_may_access with the mode parameter passed to it
2298 * succeeds.
2299 */
2300extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2301/* Remove the current tasks stale references to the old mm_struct */
2302extern void mm_release(struct task_struct *, struct mm_struct *);
2303/* Allocate a new mm structure and copy contents from tsk->mm */
2304extern struct mm_struct *dup_mm(struct task_struct *tsk);
2305
2306extern int copy_thread(unsigned long, unsigned long, unsigned long,
2307                        struct task_struct *, struct pt_regs *);
2308extern void flush_thread(void);
2309extern void exit_thread(void);
2310
2311extern void exit_files(struct task_struct *);
2312extern void __cleanup_sighand(struct sighand_struct *);
2313
2314extern void exit_itimers(struct signal_struct *);
2315extern void flush_itimer_signals(void);
2316
2317extern void do_group_exit(int);
2318
2319extern void daemonize(const char *, ...);
2320extern int allow_signal(int);
2321extern int disallow_signal(int);
2322
2323extern int do_execve(const char *,
2324                     const char __user * const __user *,
2325                     const char __user * const __user *, struct pt_regs *);
2326extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2327struct task_struct *fork_idle(int);
2328
2329extern void set_task_comm(struct task_struct *tsk, char *from);
2330extern char *get_task_comm(char *to, struct task_struct *tsk);
2331
2332#ifdef CONFIG_SMP
2333void scheduler_ipi(void);
2334extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2335#else
2336static inline void scheduler_ipi(void) { }
2337static inline unsigned long wait_task_inactive(struct task_struct *p,
2338                                               long match_state)
2339{
2340        return 1;
2341}
2342#endif
2343
2344#define next_task(p) \
2345        list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2346
2347#define for_each_process(p) \
2348        for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2349
2350extern bool current_is_single_threaded(void);
2351
2352/*
2353 * Careful: do_each_thread/while_each_thread is a double loop so
2354 *          'break' will not work as expected - use goto instead.
2355 */
2356#define do_each_thread(g, t) \
2357        for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2358
2359#define while_each_thread(g, t) \
2360        while ((t = next_thread(t)) != g)
2361
2362static inline int get_nr_threads(struct task_struct *tsk)
2363{
2364        return tsk->signal->nr_threads;
2365}
2366
2367static inline bool thread_group_leader(struct task_struct *p)
2368{
2369        return p->exit_signal >= 0;
2370}
2371
2372/* Do to the insanities of de_thread it is possible for a process
2373 * to have the pid of the thread group leader without actually being
2374 * the thread group leader.  For iteration through the pids in proc
2375 * all we care about is that we have a task with the appropriate
2376 * pid, we don't actually care if we have the right task.
2377 */
2378static inline int has_group_leader_pid(struct task_struct *p)
2379{
2380        return p->pid == p->tgid;
2381}
2382
2383static inline
2384int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2385{
2386        return p1->tgid == p2->tgid;
2387}
2388
2389static inline struct task_struct *next_thread(const struct task_struct *p)
2390{
2391        return list_entry_rcu(p->thread_group.next,
2392                              struct task_struct, thread_group);
2393}
2394
2395static inline int thread_group_empty(struct task_struct *p)
2396{
2397        return list_empty(&p->thread_group);
2398}
2399
2400#define delay_group_leader(p) \
2401                (thread_group_leader(p) && !thread_group_empty(p))
2402
2403/*
2404 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2405 * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2406 * pins the final release of task.io_context.  Also protects ->cpuset and
2407 * ->cgroup.subsys[]. And ->vfork_done.
2408 *
2409 * Nests both inside and outside of read_lock(&tasklist_lock).
2410 * It must not be nested with write_lock_irq(&tasklist_lock),
2411 * neither inside nor outside.
2412 */
2413static inline void task_lock(struct task_struct *p)
2414{
2415        spin_lock(&p->alloc_lock);
2416}
2417
2418static inline void task_unlock(struct task_struct *p)
2419{
2420        spin_unlock(&p->alloc_lock);
2421}
2422
2423extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2424                                                        unsigned long *flags);
2425
2426static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2427                                                       unsigned long *flags)
2428{
2429        struct sighand_struct *ret;
2430
2431        ret = __lock_task_sighand(tsk, flags);
2432        (void)__cond_lock(&tsk->sighand->siglock, ret);
2433        return ret;
2434}
2435
2436static inline void unlock_task_sighand(struct task_struct *tsk,
2437                                                unsigned long *flags)
2438{
2439        spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2440}
2441
2442#ifdef CONFIG_CGROUPS
2443static inline void threadgroup_change_begin(struct task_struct *tsk)
2444{
2445        down_read(&tsk->signal->group_rwsem);
2446}
2447static inline void threadgroup_change_end(struct task_struct *tsk)
2448{
2449        up_read(&tsk->signal->group_rwsem);
2450}
2451
2452/**
2453 * threadgroup_lock - lock threadgroup
2454 * @tsk: member task of the threadgroup to lock
2455 *
2456 * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2457 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2458 * perform exec.  This is useful for cases where the threadgroup needs to
2459 * stay stable across blockable operations.
2460 *
2461 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2462 * synchronization.  While held, no new task will be added to threadgroup
2463 * and no existing live task will have its PF_EXITING set.
2464 *
2465 * During exec, a task goes and puts its thread group through unusual
2466 * changes.  After de-threading, exclusive access is assumed to resources
2467 * which are usually shared by tasks in the same group - e.g. sighand may
2468 * be replaced with a new one.  Also, the exec'ing task takes over group
2469 * leader role including its pid.  Exclude these changes while locked by
2470 * grabbing cred_guard_mutex which is used to synchronize exec path.
2471 */
2472static inline void threadgroup_lock(struct task_struct *tsk)
2473{
2474        /*
2475         * exec uses exit for de-threading nesting group_rwsem inside
2476         * cred_guard_mutex. Grab cred_guard_mutex first.
2477         */
2478        mutex_lock(&tsk->signal->cred_guard_mutex);
2479        down_write(&tsk->signal->group_rwsem);
2480}
2481
2482/**
2483 * threadgroup_unlock - unlock threadgroup
2484 * @tsk: member task of the threadgroup to unlock
2485 *
2486 * Reverse threadgroup_lock().
2487 */
2488static inline void threadgroup_unlock(struct task_struct *tsk)
2489{
2490        up_write(&tsk->signal->group_rwsem);
2491        mutex_unlock(&tsk->signal->cred_guard_mutex);
2492}
2493#else
2494static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2495static inline void threadgroup_change_end(struct task_struct *tsk) {}
2496static inline void threadgroup_lock(struct task_struct *tsk) {}
2497static inline void threadgroup_unlock(struct task_struct *tsk) {}
2498#endif
2499
2500#ifndef __HAVE_THREAD_FUNCTIONS
2501
2502#define task_thread_info(task)  ((struct thread_info *)(task)->stack)
2503#define task_stack_page(task)   ((task)->stack)
2504
2505static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2506{
2507        *task_thread_info(p) = *task_thread_info(org);
2508        task_thread_info(p)->task = p;
2509}
2510
2511static inline unsigned long *end_of_stack(struct task_struct *p)
2512{
2513        return (unsigned long *)(task_thread_info(p) + 1);
2514}
2515
2516#endif
2517
2518static inline int object_is_on_stack(void *obj)
2519{
2520        void *stack = task_stack_page(current);
2521
2522        return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2523}
2524
2525extern void thread_info_cache_init(void);
2526
2527#ifdef CONFIG_DEBUG_STACK_USAGE
2528static inline unsigned long stack_not_used(struct task_struct *p)
2529{
2530        unsigned long *n = end_of_stack(p);
2531
2532        do {    /* Skip over canary */
2533                n++;
2534        } while (!*n);
2535
2536        return (unsigned long)n - (unsigned long)end_of_stack(p);
2537}
2538#endif
2539
2540/* set thread flags in other task's structures
2541 * - see asm/thread_info.h for TIF_xxxx flags available
2542 */
2543static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2544{
2545        set_ti_thread_flag(task_thread_info(tsk), flag);
2546}
2547
2548static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2549{
2550        clear_ti_thread_flag(task_thread_info(tsk), flag);
2551}
2552
2553static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2554{
2555        return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2556}
2557
2558static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2559{
2560        return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2561}
2562
2563static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2564{
2565        return test_ti_thread_flag(task_thread_info(tsk), flag);
2566}
2567
2568static inline void set_tsk_need_resched(struct task_struct *tsk)
2569{
2570        set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2571}
2572
2573static inline void clear_tsk_need_resched(struct task_struct *tsk)
2574{
2575        clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2576}
2577
2578static inline int test_tsk_need_resched(struct task_struct *tsk)
2579{
2580        return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2581}
2582
2583static inline int restart_syscall(void)
2584{
2585        set_tsk_thread_flag(current, TIF_SIGPENDING);
2586        return -ERESTARTNOINTR;
2587}
2588
2589static inline int signal_pending(struct task_struct *p)
2590{
2591        return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2592}
2593
2594static inline int __fatal_signal_pending(struct task_struct *p)
2595{
2596        return unlikely(sigismember(&p->pending.signal, SIGKILL));
2597}
2598
2599static inline int fatal_signal_pending(struct task_struct *p)
2600{
2601        return signal_pending(p) && __fatal_signal_pending(p);
2602}
2603
2604static inline int signal_pending_state(long state, struct task_struct *p)
2605{
2606        if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2607                return 0;
2608        if (!signal_pending(p))
2609                return 0;
2610
2611        return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2612}
2613
2614static inline int need_resched(void)
2615{
2616        return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2617}
2618
2619/*
2620 * cond_resched() and cond_resched_lock(): latency reduction via
2621 * explicit rescheduling in places that are safe. The return
2622 * value indicates whether a reschedule was done in fact.
2623 * cond_resched_lock() will drop the spinlock before scheduling,
2624 * cond_resched_softirq() will enable bhs before scheduling.
2625 */
2626extern int _cond_resched(void);
2627
2628#define cond_resched() ({                       \
2629        __might_sleep(__FILE__, __LINE__, 0);   \
2630        _cond_resched();                        \
2631})
2632
2633extern int __cond_resched_lock(spinlock_t *lock);
2634
2635#ifdef CONFIG_PREEMPT_COUNT
2636#define PREEMPT_LOCK_OFFSET     PREEMPT_OFFSET
2637#else
2638#define PREEMPT_LOCK_OFFSET     0
2639#endif
2640
2641#define cond_resched_lock(lock) ({                              \
2642        __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2643        __cond_resched_lock(lock);                              \
2644})
2645
2646extern int __cond_resched_softirq(void);
2647
2648#define cond_resched_softirq() ({                                       \
2649        __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);      \
2650        __cond_resched_softirq();                                       \
2651})
2652
2653/*
2654 * Does a critical section need to be broken due to another
2655 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2656 * but a general need for low latency)
2657 */
2658static inline int spin_needbreak(spinlock_t *lock)
2659{
2660#ifdef CONFIG_PREEMPT
2661        return spin_is_contended(lock);
2662#else
2663        return 0;
2664#endif
2665}
2666
2667/*
2668 * Thread group CPU time accounting.
2669 */
2670void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2671void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2672
2673static inline void thread_group_cputime_init(struct signal_struct *sig)
2674{
2675        raw_spin_lock_init(&sig->cputimer.lock);
2676}
2677
2678/*
2679 * Reevaluate whether the task has signals pending delivery.
2680 * Wake the task if so.
2681 * This is required every time the blocked sigset_t changes.
2682 * callers must hold sighand->siglock.
2683 */
2684extern void recalc_sigpending_and_wake(struct task_struct *t);
2685extern void recalc_sigpending(void);
2686
2687extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2688
2689/*
2690 * Wrappers for p->thread_info->cpu access. No-op on UP.
2691 */
2692#ifdef CONFIG_SMP
2693
2694static inline unsigned int task_cpu(const struct task_struct *p)
2695{
2696        return task_thread_info(p)->cpu;
2697}
2698
2699extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2700
2701#else
2702
2703static inline unsigned int task_cpu(const struct task_struct *p)
2704{
2705        return 0;
2706}
2707
2708static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2709{
2710}
2711
2712#endif /* CONFIG_SMP */
2713
2714extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2715extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2716
2717extern void normalize_rt_tasks(void);
2718
2719#ifdef CONFIG_CGROUP_SCHED
2720
2721extern struct task_group root_task_group;
2722
2723extern struct task_group *sched_create_group(struct task_group *parent);
2724extern void sched_destroy_group(struct task_group *tg);
2725extern void sched_move_task(struct task_struct *tsk);
2726#ifdef CONFIG_FAIR_GROUP_SCHED
2727extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2728extern unsigned long sched_group_shares(struct task_group *tg);
2729#endif
2730#ifdef CONFIG_RT_GROUP_SCHED
2731extern int sched_group_set_rt_runtime(struct task_group *tg,
2732                                      long rt_runtime_us);
2733extern long sched_group_rt_runtime(struct task_group *tg);
2734extern int sched_group_set_rt_period(struct task_group *tg,
2735                                      long rt_period_us);
2736extern long sched_group_rt_period(struct task_group *tg);
2737extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2738#endif
2739#endif /* CONFIG_CGROUP_SCHED */
2740
2741extern int task_can_switch_user(struct user_struct *up,
2742                                        struct task_struct *tsk);
2743
2744#ifdef CONFIG_TASK_XACCT
2745static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2746{
2747        tsk->ioac.rchar += amt;
2748}
2749
2750static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2751{
2752        tsk->ioac.wchar += amt;
2753}
2754
2755static inline void inc_syscr(struct task_struct *tsk)
2756{
2757        tsk->ioac.syscr++;
2758}
2759
2760static inline void inc_syscw(struct task_struct *tsk)
2761{
2762        tsk->ioac.syscw++;
2763}
2764#else
2765static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2766{
2767}
2768
2769static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2770{
2771}
2772
2773static inline void inc_syscr(struct task_struct *tsk)
2774{
2775}
2776
2777static inline void inc_syscw(struct task_struct *tsk)
2778{
2779}
2780#endif
2781
2782#ifndef TASK_SIZE_OF
2783#define TASK_SIZE_OF(tsk)       TASK_SIZE
2784#endif
2785
2786#ifdef CONFIG_MM_OWNER
2787extern void mm_update_next_owner(struct mm_struct *mm);
2788extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2789#else
2790static inline void mm_update_next_owner(struct mm_struct *mm)
2791{
2792}
2793
2794static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2795{
2796}
2797#endif /* CONFIG_MM_OWNER */
2798
2799static inline unsigned long task_rlimit(const struct task_struct *tsk,
2800                unsigned int limit)
2801{
2802        return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2803}
2804
2805static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2806                unsigned int limit)
2807{
2808        return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2809}
2810
2811static inline unsigned long rlimit(unsigned int limit)
2812{
2813        return task_rlimit(current, limit);
2814}
2815
2816static inline unsigned long rlimit_max(unsigned int limit)
2817{
2818        return task_rlimit_max(current, limit);
2819}
2820
2821#endif /* __KERNEL__ */
2822
2823#endif
2824
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.