root/include/linux/sched.h

INCLUDED FROM

DEFINITIONS

1. task_pid
2. task_pid_nr
3. task_pid_nr_ns
4. task_pid_vnr
5. task_tgid_nr
6. pid_alive
7. task_pgrp_nr_ns
8. task_pgrp_vnr
9. task_session_nr_ns
10. task_session_vnr
11. task_tgid_nr_ns
12. task_tgid_vnr
13. task_ppid_nr_ns
14. task_ppid_nr
15. task_pgrp_nr
16. task_state_index
17. task_index_to_char
18. task_state_to_char
19. is_global_init
20. is_percpu_thread
21. TASK_PFA_TEST
22. do_set_cpus_allowed
23. set_cpus_allowed_ptr
24. task_nice
25. is_idle_task
26. task_thread_info
27. kick_process
28. set_task_comm
29. scheduler_ipi
30. wait_task_inactive
31. set_tsk_thread_flag
32. clear_tsk_thread_flag
33. update_tsk_thread_flag
34. test_and_set_tsk_thread_flag
35. test_and_clear_tsk_thread_flag
36. test_tsk_thread_flag
37. set_tsk_need_resched
38. clear_tsk_need_resched
39. test_tsk_need_resched
40. _cond_resched
41. cond_resched_rcu
42. spin_needbreak
43. need_resched
44. task_cpu
45. task_cpu
46. set_task_cpu
47. vcpu_is_preempted
48. rseq_set_notify_resume
49. rseq_handle_notify_resume
50. rseq_signal_deliver
51. rseq_preempt
52. rseq_migrate
53. rseq_fork
54. rseq_execve
55. rseq_set_notify_resume
56. rseq_handle_notify_resume
57. rseq_signal_deliver
58. rseq_preempt
59. rseq_migrate
60. rseq_fork
61. rseq_execve
62. exit_umh
63. rseq_syscall

   1 /* SPDX-License-Identifier: GPL-2.0 */
   2 #ifndef _LINUX_SCHED_H
   3 #define _LINUX_SCHED_H
   4 
   5 /*
   6  * Define 'struct task_struct' and provide the main scheduler
   7  * APIs (schedule(), wakeup variants, etc.)
   8  */
   9 
  10 #include <uapi/linux/sched.h>
  11 
  12 #include <asm/current.h>
  13 
  14 #include <linux/pid.h>
  15 #include <linux/sem.h>
  16 #include <linux/shm.h>
  17 #include <linux/kcov.h>
  18 #include <linux/mutex.h>
  19 #include <linux/plist.h>
  20 #include <linux/hrtimer.h>
  21 #include <linux/seccomp.h>
  22 #include <linux/nodemask.h>
  23 #include <linux/rcupdate.h>
  24 #include <linux/refcount.h>
  25 #include <linux/resource.h>
  26 #include <linux/latencytop.h>
  27 #include <linux/sched/prio.h>
  28 #include <linux/sched/types.h>
  29 #include <linux/signal_types.h>
  30 #include <linux/mm_types_task.h>
  31 #include <linux/task_io_accounting.h>
  32 #include <linux/posix-timers.h>
  33 #include <linux/rseq.h>
  34 
  35 /* task_struct member predeclarations (sorted alphabetically): */
  36 struct audit_context;
  37 struct backing_dev_info;
  38 struct bio_list;
  39 struct blk_plug;
  40 struct capture_control;
  41 struct cfs_rq;
  42 struct fs_struct;
  43 struct futex_pi_state;
  44 struct io_context;
  45 struct mempolicy;
  46 struct nameidata;
  47 struct nsproxy;
  48 struct perf_event_context;
  49 struct pid_namespace;
  50 struct pipe_inode_info;
  51 struct rcu_node;
  52 struct reclaim_state;
  53 struct robust_list_head;
  54 struct root_domain;
  55 struct rq;
  56 struct sched_attr;
  57 struct sched_param;
  58 struct seq_file;
  59 struct sighand_struct;
  60 struct signal_struct;
  61 struct task_delay_info;
  62 struct task_group;
  63 
  64 /*
  65  * Task state bitmask. NOTE! These bits are also
  66  * encoded in fs/proc/array.c: get_task_state().
  67  *
  68  * We have two separate sets of flags: task->state
  69  * is about runnability, while task->exit_state are
  70  * about the task exiting. Confusing, but this way
  71  * modifying one set can't modify the other one by
  72  * mistake.
  73  */
  74 
  75 /* Used in tsk->state: */
  76 #define TASK_RUNNING                    0x0000
  77 #define TASK_INTERRUPTIBLE              0x0001
  78 #define TASK_UNINTERRUPTIBLE            0x0002
  79 #define __TASK_STOPPED                  0x0004
  80 #define __TASK_TRACED                   0x0008
  81 /* Used in tsk->exit_state: */
  82 #define EXIT_DEAD                       0x0010
  83 #define EXIT_ZOMBIE                     0x0020
  84 #define EXIT_TRACE                      (EXIT_ZOMBIE | EXIT_DEAD)
  85 /* Used in tsk->state again: */
  86 #define TASK_PARKED                     0x0040
  87 #define TASK_DEAD                       0x0080
  88 #define TASK_WAKEKILL                   0x0100
  89 #define TASK_WAKING                     0x0200
  90 #define TASK_NOLOAD                     0x0400
  91 #define TASK_NEW                        0x0800
  92 #define TASK_STATE_MAX                  0x1000
  93 
  94 /* Convenience macros for the sake of set_current_state: */
  95 #define TASK_KILLABLE                   (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
  96 #define TASK_STOPPED                    (TASK_WAKEKILL | __TASK_STOPPED)
  97 #define TASK_TRACED                     (TASK_WAKEKILL | __TASK_TRACED)
  98 
  99 #define TASK_IDLE                       (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
 100 
 101 /* Convenience macros for the sake of wake_up(): */
 102 #define TASK_NORMAL                     (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 103 
 104 /* get_task_state(): */
 105 #define TASK_REPORT                     (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 106                                          TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 107                                          __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
 108                                          TASK_PARKED)
 109 
 110 #define task_is_traced(task)            ((task->state & __TASK_TRACED) != 0)
 111 
 112 #define task_is_stopped(task)           ((task->state & __TASK_STOPPED) != 0)
 113 
 114 #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 115 
 116 #define task_contributes_to_load(task)  ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
 117                                          (task->flags & PF_FROZEN) == 0 && \
 118                                          (task->state & TASK_NOLOAD) == 0)
 119 
 120 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
 121 
 122 /*
 123  * Special states are those that do not use the normal wait-loop pattern. See
 124  * the comment with set_special_state().
 125  */
 126 #define is_special_task_state(state)                            \
 127         ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD))
 128 
 129 #define __set_current_state(state_value)                        \
 130         do {                                                    \
 131                 WARN_ON_ONCE(is_special_task_state(state_value));\
 132                 current->task_state_change = _THIS_IP_;         \
 133                 current->state = (state_value);                 \
 134         } while (0)
 135 
 136 #define set_current_state(state_value)                          \
 137         do {                                                    \
 138                 WARN_ON_ONCE(is_special_task_state(state_value));\
 139                 current->task_state_change = _THIS_IP_;         \
 140                 smp_store_mb(current->state, (state_value));    \
 141         } while (0)
 142 
 143 #define set_special_state(state_value)                                  \
 144         do {                                                            \
 145                 unsigned long flags; /* may shadow */                   \
 146                 WARN_ON_ONCE(!is_special_task_state(state_value));      \
 147                 raw_spin_lock_irqsave(&current->pi_lock, flags);        \
 148                 current->task_state_change = _THIS_IP_;                 \
 149                 current->state = (state_value);                         \
 150                 raw_spin_unlock_irqrestore(&current->pi_lock, flags);   \
 151         } while (0)
 152 #else
 153 /*
 154  * set_current_state() includes a barrier so that the write of current->state
 155  * is correctly serialised wrt the caller's subsequent test of whether to
 156  * actually sleep:
 157  *
 158  *   for (;;) {
 159  *      set_current_state(TASK_UNINTERRUPTIBLE);
 160  *      if (!need_sleep)
 161  *              break;
 162  *
 163  *      schedule();
 164  *   }
 165  *   __set_current_state(TASK_RUNNING);
 166  *
 167  * If the caller does not need such serialisation (because, for instance, the
 168  * condition test and condition change and wakeup are under the same lock) then
 169  * use __set_current_state().
 170  *
 171  * The above is typically ordered against the wakeup, which does:
 172  *
 173  *   need_sleep = false;
 174  *   wake_up_state(p, TASK_UNINTERRUPTIBLE);
 175  *
 176  * where wake_up_state() executes a full memory barrier before accessing the
 177  * task state.
 178  *
 179  * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
 180  * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
 181  * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
 182  *
 183  * However, with slightly different timing the wakeup TASK_RUNNING store can
 184  * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not
 185  * a problem either because that will result in one extra go around the loop
 186  * and our @cond test will save the day.
 187  *
 188  * Also see the comments of try_to_wake_up().
 189  */
 190 #define __set_current_state(state_value)                                \
 191         current->state = (state_value)
 192 
 193 #define set_current_state(state_value)                                  \
 194         smp_store_mb(current->state, (state_value))
 195 
 196 /*
 197  * set_special_state() should be used for those states when the blocking task
 198  * can not use the regular condition based wait-loop. In that case we must
 199  * serialize against wakeups such that any possible in-flight TASK_RUNNING stores
 200  * will not collide with our state change.
 201  */
 202 #define set_special_state(state_value)                                  \
 203         do {                                                            \
 204                 unsigned long flags; /* may shadow */                   \
 205                 raw_spin_lock_irqsave(&current->pi_lock, flags);        \
 206                 current->state = (state_value);                         \
 207                 raw_spin_unlock_irqrestore(&current->pi_lock, flags);   \
 208         } while (0)
 209 
 210 #endif
 211 
 212 /* Task command name length: */
 213 #define TASK_COMM_LEN                   16
 214 
 215 extern void scheduler_tick(void);
 216 
 217 #define MAX_SCHEDULE_TIMEOUT            LONG_MAX
 218 
 219 extern long schedule_timeout(long timeout);
 220 extern long schedule_timeout_interruptible(long timeout);
 221 extern long schedule_timeout_killable(long timeout);
 222 extern long schedule_timeout_uninterruptible(long timeout);
 223 extern long schedule_timeout_idle(long timeout);
 224 asmlinkage void schedule(void);
 225 extern void schedule_preempt_disabled(void);
 226 asmlinkage void preempt_schedule_irq(void);
 227 
 228 extern int __must_check io_schedule_prepare(void);
 229 extern void io_schedule_finish(int token);
 230 extern long io_schedule_timeout(long timeout);
 231 extern void io_schedule(void);
 232 
 233 /**
 234  * struct prev_cputime - snapshot of system and user cputime
 235  * @utime: time spent in user mode
 236  * @stime: time spent in system mode
 237  * @lock: protects the above two fields
 238  *
 239  * Stores previous user/system time values such that we can guarantee
 240  * monotonicity.
 241  */
 242 struct prev_cputime {
 243 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
 244         u64                             utime;
 245         u64                             stime;
 246         raw_spinlock_t                  lock;
 247 #endif
 248 };
 249 
 250 enum vtime_state {
 251         /* Task is sleeping or running in a CPU with VTIME inactive: */
 252         VTIME_INACTIVE = 0,
 253         /* Task runs in userspace in a CPU with VTIME active: */
 254         VTIME_USER,
 255         /* Task runs in kernelspace in a CPU with VTIME active: */
 256         VTIME_SYS,
 257 };
 258 
 259 struct vtime {
 260         seqcount_t              seqcount;
 261         unsigned long long      starttime;
 262         enum vtime_state        state;
 263         u64                     utime;
 264         u64                     stime;
 265         u64                     gtime;
 266 };
 267 
 268 /*
 269  * Utilization clamp constraints.
 270  * @UCLAMP_MIN: Minimum utilization
 271  * @UCLAMP_MAX: Maximum utilization
 272  * @UCLAMP_CNT: Utilization clamp constraints count
 273  */
 274 enum uclamp_id {
 275         UCLAMP_MIN = 0,
 276         UCLAMP_MAX,
 277         UCLAMP_CNT
 278 };
 279 
 280 #ifdef CONFIG_SMP
 281 extern struct root_domain def_root_domain;
 282 extern struct mutex sched_domains_mutex;
 283 #endif
 284 
 285 struct sched_info {
 286 #ifdef CONFIG_SCHED_INFO
 287         /* Cumulative counters: */
 288 
 289         /* # of times we have run on this CPU: */
 290         unsigned long                   pcount;
 291 
 292         /* Time spent waiting on a runqueue: */
 293         unsigned long long              run_delay;
 294 
 295         /* Timestamps: */
 296 
 297         /* When did we last run on a CPU? */
 298         unsigned long long              last_arrival;
 299 
 300         /* When were we last queued to run? */
 301         unsigned long long              last_queued;
 302 
 303 #endif /* CONFIG_SCHED_INFO */
 304 };
 305 
 306 /*
 307  * Integer metrics need fixed point arithmetic, e.g., sched/fair
 308  * has a few: load, load_avg, util_avg, freq, and capacity.
 309  *
 310  * We define a basic fixed point arithmetic range, and then formalize
 311  * all these metrics based on that basic range.
 312  */
 313 # define SCHED_FIXEDPOINT_SHIFT         10
 314 # define SCHED_FIXEDPOINT_SCALE         (1L << SCHED_FIXEDPOINT_SHIFT)
 315 
 316 /* Increase resolution of cpu_capacity calculations */
 317 # define SCHED_CAPACITY_SHIFT           SCHED_FIXEDPOINT_SHIFT
 318 # define SCHED_CAPACITY_SCALE           (1L << SCHED_CAPACITY_SHIFT)
 319 
 320 struct load_weight {
 321         unsigned long                   weight;
 322         u32                             inv_weight;
 323 };
 324 
 325 /**
 326  * struct util_est - Estimation utilization of FAIR tasks
 327  * @enqueued: instantaneous estimated utilization of a task/cpu
 328  * @ewma:     the Exponential Weighted Moving Average (EWMA)
 329  *            utilization of a task
 330  *
 331  * Support data structure to track an Exponential Weighted Moving Average
 332  * (EWMA) of a FAIR task's utilization. New samples are added to the moving
 333  * average each time a task completes an activation. Sample's weight is chosen
 334  * so that the EWMA will be relatively insensitive to transient changes to the
 335  * task's workload.
 336  *
 337  * The enqueued attribute has a slightly different meaning for tasks and cpus:
 338  * - task:   the task's util_avg at last task dequeue time
 339  * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
 340  * Thus, the util_est.enqueued of a task represents the contribution on the
 341  * estimated utilization of the CPU where that task is currently enqueued.
 342  *
 343  * Only for tasks we track a moving average of the past instantaneous
 344  * estimated utilization. This allows to absorb sporadic drops in utilization
 345  * of an otherwise almost periodic task.
 346  */
 347 struct util_est {
 348         unsigned int                    enqueued;
 349         unsigned int                    ewma;
 350 #define UTIL_EST_WEIGHT_SHIFT           2
 351 } __attribute__((__aligned__(sizeof(u64))));
 352 
 353 /*
 354  * The load_avg/util_avg accumulates an infinite geometric series
 355  * (see __update_load_avg() in kernel/sched/fair.c).
 356  *
 357  * [load_avg definition]
 358  *
 359  *   load_avg = runnable% * scale_load_down(load)
 360  *
 361  * where runnable% is the time ratio that a sched_entity is runnable.
 362  * For cfs_rq, it is the aggregated load_avg of all runnable and
 363  * blocked sched_entities.
 364  *
 365  * [util_avg definition]
 366  *
 367  *   util_avg = running% * SCHED_CAPACITY_SCALE
 368  *
 369  * where running% is the time ratio that a sched_entity is running on
 370  * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
 371  * and blocked sched_entities.
 372  *
 373  * load_avg and util_avg don't direcly factor frequency scaling and CPU
 374  * capacity scaling. The scaling is done through the rq_clock_pelt that
 375  * is used for computing those signals (see update_rq_clock_pelt())
 376  *
 377  * N.B., the above ratios (runnable% and running%) themselves are in the
 378  * range of [0, 1]. To do fixed point arithmetics, we therefore scale them
 379  * to as large a range as necessary. This is for example reflected by
 380  * util_avg's SCHED_CAPACITY_SCALE.
 381  *
 382  * [Overflow issue]
 383  *
 384  * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
 385  * with the highest load (=88761), always runnable on a single cfs_rq,
 386  * and should not overflow as the number already hits PID_MAX_LIMIT.
 387  *
 388  * For all other cases (including 32-bit kernels), struct load_weight's
 389  * weight will overflow first before we do, because:
 390  *
 391  *    Max(load_avg) <= Max(load.weight)
 392  *
 393  * Then it is the load_weight's responsibility to consider overflow
 394  * issues.
 395  */
 396 struct sched_avg {
 397         u64                             last_update_time;
 398         u64                             load_sum;
 399         u64                             runnable_load_sum;
 400         u32                             util_sum;
 401         u32                             period_contrib;
 402         unsigned long                   load_avg;
 403         unsigned long                   runnable_load_avg;
 404         unsigned long                   util_avg;
 405         struct util_est                 util_est;
 406 } ____cacheline_aligned;
 407 
 408 struct sched_statistics {
 409 #ifdef CONFIG_SCHEDSTATS
 410         u64                             wait_start;
 411         u64                             wait_max;
 412         u64                             wait_count;
 413         u64                             wait_sum;
 414         u64                             iowait_count;
 415         u64                             iowait_sum;
 416 
 417         u64                             sleep_start;
 418         u64                             sleep_max;
 419         s64                             sum_sleep_runtime;
 420 
 421         u64                             block_start;
 422         u64                             block_max;
 423         u64                             exec_max;
 424         u64                             slice_max;
 425 
 426         u64                             nr_migrations_cold;
 427         u64                             nr_failed_migrations_affine;
 428         u64                             nr_failed_migrations_running;
 429         u64                             nr_failed_migrations_hot;
 430         u64                             nr_forced_migrations;
 431 
 432         u64                             nr_wakeups;
 433         u64                             nr_wakeups_sync;
 434         u64                             nr_wakeups_migrate;
 435         u64                             nr_wakeups_local;
 436         u64                             nr_wakeups_remote;
 437         u64                             nr_wakeups_affine;
 438         u64                             nr_wakeups_affine_attempts;
 439         u64                             nr_wakeups_passive;
 440         u64                             nr_wakeups_idle;
 441 #endif
 442 };
 443 
 444 struct sched_entity {
 445         /* For load-balancing: */
 446         struct load_weight              load;
 447         unsigned long                   runnable_weight;
 448         struct rb_node                  run_node;
 449         struct list_head                group_node;
 450         unsigned int                    on_rq;
 451 
 452         u64                             exec_start;
 453         u64                             sum_exec_runtime;
 454         u64                             vruntime;
 455         u64                             prev_sum_exec_runtime;
 456 
 457         u64                             nr_migrations;
 458 
 459         struct sched_statistics         statistics;
 460 
 461 #ifdef CONFIG_FAIR_GROUP_SCHED
 462         int                             depth;
 463         struct sched_entity             *parent;
 464         /* rq on which this entity is (to be) queued: */
 465         struct cfs_rq                   *cfs_rq;
 466         /* rq "owned" by this entity/group: */
 467         struct cfs_rq                   *my_q;
 468 #endif
 469 
 470 #ifdef CONFIG_SMP
 471         /*
 472          * Per entity load average tracking.
 473          *
 474          * Put into separate cache line so it does not
 475          * collide with read-mostly values above.
 476          */
 477         struct sched_avg                avg;
 478 #endif
 479 };
 480 
 481 struct sched_rt_entity {
 482         struct list_head                run_list;
 483         unsigned long                   timeout;
 484         unsigned long                   watchdog_stamp;
 485         unsigned int                    time_slice;
 486         unsigned short                  on_rq;
 487         unsigned short                  on_list;
 488 
 489         struct sched_rt_entity          *back;
 490 #ifdef CONFIG_RT_GROUP_SCHED
 491         struct sched_rt_entity          *parent;
 492         /* rq on which this entity is (to be) queued: */
 493         struct rt_rq                    *rt_rq;
 494         /* rq "owned" by this entity/group: */
 495         struct rt_rq                    *my_q;
 496 #endif
 497 } __randomize_layout;
 498 
 499 struct sched_dl_entity {
 500         struct rb_node                  rb_node;
 501 
 502         /*
 503          * Original scheduling parameters. Copied here from sched_attr
 504          * during sched_setattr(), they will remain the same until
 505          * the next sched_setattr().
 506          */
 507         u64                             dl_runtime;     /* Maximum runtime for each instance    */
 508         u64                             dl_deadline;    /* Relative deadline of each instance   */
 509         u64                             dl_period;      /* Separation of two instances (period) */
 510         u64                             dl_bw;          /* dl_runtime / dl_period               */
 511         u64                             dl_density;     /* dl_runtime / dl_deadline             */
 512 
 513         /*
 514          * Actual scheduling parameters. Initialized with the values above,
 515          * they are continuously updated during task execution. Note that
 516          * the remaining runtime could be < 0 in case we are in overrun.
 517          */
 518         s64                             runtime;        /* Remaining runtime for this instance  */
 519         u64                             deadline;       /* Absolute deadline for this instance  */
 520         unsigned int                    flags;          /* Specifying the scheduler behaviour   */
 521 
 522         /*
 523          * Some bool flags:
 524          *
 525          * @dl_throttled tells if we exhausted the runtime. If so, the
 526          * task has to wait for a replenishment to be performed at the
 527          * next firing of dl_timer.
 528          *
 529          * @dl_boosted tells if we are boosted due to DI. If so we are
 530          * outside bandwidth enforcement mechanism (but only until we
 531          * exit the critical section);
 532          *
 533          * @dl_yielded tells if task gave up the CPU before consuming
 534          * all its available runtime during the last job.
 535          *
 536          * @dl_non_contending tells if the task is inactive while still
 537          * contributing to the active utilization. In other words, it
 538          * indicates if the inactive timer has been armed and its handler
 539          * has not been executed yet. This flag is useful to avoid race
 540          * conditions between the inactive timer handler and the wakeup
 541          * code.
 542          *
 543          * @dl_overrun tells if the task asked to be informed about runtime
 544          * overruns.
 545          */
 546         unsigned int                    dl_throttled      : 1;
 547         unsigned int                    dl_boosted        : 1;
 548         unsigned int                    dl_yielded        : 1;
 549         unsigned int                    dl_non_contending : 1;
 550         unsigned int                    dl_overrun        : 1;
 551 
 552         /*
 553          * Bandwidth enforcement timer. Each -deadline task has its
 554          * own bandwidth to be enforced, thus we need one timer per task.
 555          */
 556         struct hrtimer                  dl_timer;
 557 
 558         /*
 559          * Inactive timer, responsible for decreasing the active utilization
 560          * at the "0-lag time". When a -deadline task blocks, it contributes
 561          * to GRUB's active utilization until the "0-lag time", hence a
 562          * timer is needed to decrease the active utilization at the correct
 563          * time.
 564          */
 565         struct hrtimer inactive_timer;
 566 };
 567 
 568 #ifdef CONFIG_UCLAMP_TASK
 569 /* Number of utilization clamp buckets (shorter alias) */
 570 #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT
 571 
 572 /*
 573  * Utilization clamp for a scheduling entity
 574  * @value:              clamp value "assigned" to a se
 575  * @bucket_id:          bucket index corresponding to the "assigned" value
 576  * @active:             the se is currently refcounted in a rq's bucket
 577  * @user_defined:       the requested clamp value comes from user-space
 578  *
 579  * The bucket_id is the index of the clamp bucket matching the clamp value
 580  * which is pre-computed and stored to avoid expensive integer divisions from
 581  * the fast path.
 582  *
 583  * The active bit is set whenever a task has got an "effective" value assigned,
 584  * which can be different from the clamp value "requested" from user-space.
 585  * This allows to know a task is refcounted in the rq's bucket corresponding
 586  * to the "effective" bucket_id.
 587  *
 588  * The user_defined bit is set whenever a task has got a task-specific clamp
 589  * value requested from userspace, i.e. the system defaults apply to this task
 590  * just as a restriction. This allows to relax default clamps when a less
 591  * restrictive task-specific value has been requested, thus allowing to
 592  * implement a "nice" semantic. For example, a task running with a 20%
 593  * default boost can still drop its own boosting to 0%.
 594  */
 595 struct uclamp_se {
 596         unsigned int value              : bits_per(SCHED_CAPACITY_SCALE);
 597         unsigned int bucket_id          : bits_per(UCLAMP_BUCKETS);
 598         unsigned int active             : 1;
 599         unsigned int user_defined       : 1;
 600 };
 601 #endif /* CONFIG_UCLAMP_TASK */
 602 
 603 union rcu_special {
 604         struct {
 605                 u8                      blocked;
 606                 u8                      need_qs;
 607                 u8                      exp_hint; /* Hint for performance. */
 608                 u8                      deferred_qs;
 609         } b; /* Bits. */
 610         u32 s; /* Set of bits. */
 611 };
 612 
 613 enum perf_event_task_context {
 614         perf_invalid_context = -1,
 615         perf_hw_context = 0,
 616         perf_sw_context,
 617         perf_nr_task_contexts,
 618 };
 619 
 620 struct wake_q_node {
 621         struct wake_q_node *next;
 622 };
 623 
 624 struct task_struct {
 625 #ifdef CONFIG_THREAD_INFO_IN_TASK
 626         /*
 627          * For reasons of header soup (see current_thread_info()), this
 628          * must be the first element of task_struct.
 629          */
 630         struct thread_info              thread_info;
 631 #endif
 632         /* -1 unrunnable, 0 runnable, >0 stopped: */
 633         volatile long                   state;
 634 
 635         /*
 636          * This begins the randomizable portion of task_struct. Only
 637          * scheduling-critical items should be added above here.
 638          */
 639         randomized_struct_fields_start
 640 
 641         void                            *stack;
 642         refcount_t                      usage;
 643         /* Per task flags (PF_*), defined further below: */
 644         unsigned int                    flags;
 645         unsigned int                    ptrace;
 646 
 647 #ifdef CONFIG_SMP
 648         struct llist_node               wake_entry;
 649         int                             on_cpu;
 650 #ifdef CONFIG_THREAD_INFO_IN_TASK
 651         /* Current CPU: */
 652         unsigned int                    cpu;
 653 #endif
 654         unsigned int                    wakee_flips;
 655         unsigned long                   wakee_flip_decay_ts;
 656         struct task_struct              *last_wakee;
 657 
 658         /*
 659          * recent_used_cpu is initially set as the last CPU used by a task
 660          * that wakes affine another task. Waker/wakee relationships can
 661          * push tasks around a CPU where each wakeup moves to the next one.
 662          * Tracking a recently used CPU allows a quick search for a recently
 663          * used CPU that may be idle.
 664          */
 665         int                             recent_used_cpu;
 666         int                             wake_cpu;
 667 #endif
 668         int                             on_rq;
 669 
 670         int                             prio;
 671         int                             static_prio;
 672         int                             normal_prio;
 673         unsigned int                    rt_priority;
 674 
 675         const struct sched_class        *sched_class;
 676         struct sched_entity             se;
 677         struct sched_rt_entity          rt;
 678 #ifdef CONFIG_CGROUP_SCHED
 679         struct task_group               *sched_task_group;
 680 #endif
 681         struct sched_dl_entity          dl;
 682 
 683 #ifdef CONFIG_UCLAMP_TASK
 684         /* Clamp values requested for a scheduling entity */
 685         struct uclamp_se                uclamp_req[UCLAMP_CNT];
 686         /* Effective clamp values used for a scheduling entity */
 687         struct uclamp_se                uclamp[UCLAMP_CNT];
 688 #endif
 689 
 690 #ifdef CONFIG_PREEMPT_NOTIFIERS
 691         /* List of struct preempt_notifier: */
 692         struct hlist_head               preempt_notifiers;
 693 #endif
 694 
 695 #ifdef CONFIG_BLK_DEV_IO_TRACE
 696         unsigned int                    btrace_seq;
 697 #endif
 698 
 699         unsigned int                    policy;
 700         int                             nr_cpus_allowed;
 701         const cpumask_t                 *cpus_ptr;
 702         cpumask_t                       cpus_mask;
 703 
 704 #ifdef CONFIG_PREEMPT_RCU
 705         int                             rcu_read_lock_nesting;
 706         union rcu_special               rcu_read_unlock_special;
 707         struct list_head                rcu_node_entry;
 708         struct rcu_node                 *rcu_blocked_node;
 709 #endif /* #ifdef CONFIG_PREEMPT_RCU */
 710 
 711 #ifdef CONFIG_TASKS_RCU
 712         unsigned long                   rcu_tasks_nvcsw;
 713         u8                              rcu_tasks_holdout;
 714         u8                              rcu_tasks_idx;
 715         int                             rcu_tasks_idle_cpu;
 716         struct list_head                rcu_tasks_holdout_list;
 717 #endif /* #ifdef CONFIG_TASKS_RCU */
 718 
 719         struct sched_info               sched_info;
 720 
 721         struct list_head                tasks;
 722 #ifdef CONFIG_SMP
 723         struct plist_node               pushable_tasks;
 724         struct rb_node                  pushable_dl_tasks;
 725 #endif
 726 
 727         struct mm_struct                *mm;
 728         struct mm_struct                *active_mm;
 729 
 730         /* Per-thread vma caching: */
 731         struct vmacache                 vmacache;
 732 
 733 #ifdef SPLIT_RSS_COUNTING
 734         struct task_rss_stat            rss_stat;
 735 #endif
 736         int                             exit_state;
 737         int                             exit_code;
 738         int                             exit_signal;
 739         /* The signal sent when the parent dies: */
 740         int                             pdeath_signal;
 741         /* JOBCTL_*, siglock protected: */
 742         unsigned long                   jobctl;
 743 
 744         /* Used for emulating ABI behavior of previous Linux versions: */
 745         unsigned int                    personality;
 746 
 747         /* Scheduler bits, serialized by scheduler locks: */
 748         unsigned                        sched_reset_on_fork:1;
 749         unsigned                        sched_contributes_to_load:1;
 750         unsigned                        sched_migrated:1;
 751         unsigned                        sched_remote_wakeup:1;
 752 #ifdef CONFIG_PSI
 753         unsigned                        sched_psi_wake_requeue:1;
 754 #endif
 755 
 756         /* Force alignment to the next boundary: */
 757         unsigned                        :0;
 758 
 759         /* Unserialized, strictly 'current' */
 760 
 761         /* Bit to tell LSMs we're in execve(): */
 762         unsigned                        in_execve:1;
 763         unsigned                        in_iowait:1;
 764 #ifndef TIF_RESTORE_SIGMASK
 765         unsigned                        restore_sigmask:1;
 766 #endif
 767 #ifdef CONFIG_MEMCG
 768         unsigned                        in_user_fault:1;
 769 #endif
 770 #ifdef CONFIG_COMPAT_BRK
 771         unsigned                        brk_randomized:1;
 772 #endif
 773 #ifdef CONFIG_CGROUPS
 774         /* disallow userland-initiated cgroup migration */
 775         unsigned                        no_cgroup_migration:1;
 776         /* task is frozen/stopped (used by the cgroup freezer) */
 777         unsigned                        frozen:1;
 778 #endif
 779 #ifdef CONFIG_BLK_CGROUP
 780         /* to be used once the psi infrastructure lands upstream. */
 781         unsigned                        use_memdelay:1;
 782 #endif
 783 
 784         unsigned long                   atomic_flags; /* Flags requiring atomic access. */
 785 
 786         struct restart_block            restart_block;
 787 
 788         pid_t                           pid;
 789         pid_t                           tgid;
 790 
 791 #ifdef CONFIG_STACKPROTECTOR
 792         /* Canary value for the -fstack-protector GCC feature: */
 793         unsigned long                   stack_canary;
 794 #endif
 795         /*
 796          * Pointers to the (original) parent process, youngest child, younger sibling,
 797          * older sibling, respectively.  (p->father can be replaced with
 798          * p->real_parent->pid)
 799          */
 800 
 801         /* Real parent process: */
 802         struct task_struct __rcu        *real_parent;
 803 
 804         /* Recipient of SIGCHLD, wait4() reports: */
 805         struct task_struct __rcu        *parent;
 806 
 807         /*
 808          * Children/sibling form the list of natural children:
 809          */
 810         struct list_head                children;
 811         struct list_head                sibling;
 812         struct task_struct              *group_leader;
 813 
 814         /*
 815          * 'ptraced' is the list of tasks this task is using ptrace() on.
 816          *
 817          * This includes both natural children and PTRACE_ATTACH targets.
 818          * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
 819          */
 820         struct list_head                ptraced;
 821         struct list_head                ptrace_entry;
 822 
 823         /* PID/PID hash table linkage. */
 824         struct pid                      *thread_pid;
 825         struct hlist_node               pid_links[PIDTYPE_MAX];
 826         struct list_head                thread_group;
 827         struct list_head                thread_node;
 828 
 829         struct completion               *vfork_done;
 830 
 831         /* CLONE_CHILD_SETTID: */
 832         int __user                      *set_child_tid;
 833 
 834         /* CLONE_CHILD_CLEARTID: */
 835         int __user                      *clear_child_tid;
 836 
 837         u64                             utime;
 838         u64                             stime;
 839 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
 840         u64                             utimescaled;
 841         u64                             stimescaled;
 842 #endif
 843         u64                             gtime;
 844         struct prev_cputime             prev_cputime;
 845 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
 846         struct vtime                    vtime;
 847 #endif
 848 
 849 #ifdef CONFIG_NO_HZ_FULL
 850         atomic_t                        tick_dep_mask;
 851 #endif
 852         /* Context switch counts: */
 853         unsigned long                   nvcsw;
 854         unsigned long                   nivcsw;
 855 
 856         /* Monotonic time in nsecs: */
 857         u64                             start_time;
 858 
 859         /* Boot based time in nsecs: */
 860         u64                             real_start_time;
 861 
 862         /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
 863         unsigned long                   min_flt;
 864         unsigned long                   maj_flt;
 865 
 866         /* Empty if CONFIG_POSIX_CPUTIMERS=n */
 867         struct posix_cputimers          posix_cputimers;
 868 
 869         /* Process credentials: */
 870 
 871         /* Tracer's credentials at attach: */
 872         const struct cred __rcu         *ptracer_cred;
 873 
 874         /* Objective and real subjective task credentials (COW): */
 875         const struct cred __rcu         *real_cred;
 876 
 877         /* Effective (overridable) subjective task credentials (COW): */
 878         const struct cred __rcu         *cred;
 879 
 880 #ifdef CONFIG_KEYS
 881         /* Cached requested key. */
 882         struct key                      *cached_requested_key;
 883 #endif
 884 
 885         /*
 886          * executable name, excluding path.
 887          *
 888          * - normally initialized setup_new_exec()
 889          * - access it with [gs]et_task_comm()
 890          * - lock it with task_lock()
 891          */
 892         char                            comm[TASK_COMM_LEN];
 893 
 894         struct nameidata                *nameidata;
 895 
 896 #ifdef CONFIG_SYSVIPC
 897         struct sysv_sem                 sysvsem;
 898         struct sysv_shm                 sysvshm;
 899 #endif
 900 #ifdef CONFIG_DETECT_HUNG_TASK
 901         unsigned long                   last_switch_count;
 902         unsigned long                   last_switch_time;
 903 #endif
 904         /* Filesystem information: */
 905         struct fs_struct                *fs;
 906 
 907         /* Open file information: */
 908         struct files_struct             *files;
 909 
 910         /* Namespaces: */
 911         struct nsproxy                  *nsproxy;
 912 
 913         /* Signal handlers: */
 914         struct signal_struct            *signal;
 915         struct sighand_struct           *sighand;
 916         sigset_t                        blocked;
 917         sigset_t                        real_blocked;
 918         /* Restored if set_restore_sigmask() was used: */
 919         sigset_t                        saved_sigmask;
 920         struct sigpending               pending;
 921         unsigned long                   sas_ss_sp;
 922         size_t                          sas_ss_size;
 923         unsigned int                    sas_ss_flags;
 924 
 925         struct callback_head            *task_works;
 926 
 927 #ifdef CONFIG_AUDIT
 928 #ifdef CONFIG_AUDITSYSCALL
 929         struct audit_context            *audit_context;
 930 #endif
 931         kuid_t                          loginuid;
 932         unsigned int                    sessionid;
 933 #endif
 934         struct seccomp                  seccomp;
 935 
 936         /* Thread group tracking: */
 937         u64                             parent_exec_id;
 938         u64                             self_exec_id;
 939 
 940         /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
 941         spinlock_t                      alloc_lock;
 942 
 943         /* Protection of the PI data structures: */
 944         raw_spinlock_t                  pi_lock;
 945 
 946         struct wake_q_node              wake_q;
 947 
 948 #ifdef CONFIG_RT_MUTEXES
 949         /* PI waiters blocked on a rt_mutex held by this task: */
 950         struct rb_root_cached           pi_waiters;
 951         /* Updated under owner's pi_lock and rq lock */
 952         struct task_struct              *pi_top_task;
 953         /* Deadlock detection and priority inheritance handling: */
 954         struct rt_mutex_waiter          *pi_blocked_on;
 955 #endif
 956 
 957 #ifdef CONFIG_DEBUG_MUTEXES
 958         /* Mutex deadlock detection: */
 959         struct mutex_waiter             *blocked_on;
 960 #endif
 961 
 962 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
 963         int                             non_block_count;
 964 #endif
 965 
 966 #ifdef CONFIG_TRACE_IRQFLAGS
 967         unsigned int                    irq_events;
 968         unsigned long                   hardirq_enable_ip;
 969         unsigned long                   hardirq_disable_ip;
 970         unsigned int                    hardirq_enable_event;
 971         unsigned int                    hardirq_disable_event;
 972         int                             hardirqs_enabled;
 973         int                             hardirq_context;
 974         unsigned long                   softirq_disable_ip;
 975         unsigned long                   softirq_enable_ip;
 976         unsigned int                    softirq_disable_event;
 977         unsigned int                    softirq_enable_event;
 978         int                             softirqs_enabled;
 979         int                             softirq_context;
 980 #endif
 981 
 982 #ifdef CONFIG_LOCKDEP
 983 # define MAX_LOCK_DEPTH                 48UL
 984         u64                             curr_chain_key;
 985         int                             lockdep_depth;
 986         unsigned int                    lockdep_recursion;
 987         struct held_lock                held_locks[MAX_LOCK_DEPTH];
 988 #endif
 989 
 990 #ifdef CONFIG_UBSAN
 991         unsigned int                    in_ubsan;
 992 #endif
 993 
 994         /* Journalling filesystem info: */
 995         void                            *journal_info;
 996 
 997         /* Stacked block device info: */
 998         struct bio_list                 *bio_list;
 999 
1000 #ifdef CONFIG_BLOCK
1001         /* Stack plugging: */
1002         struct blk_plug                 *plug;
1003 #endif
1004 
1005         /* VM state: */
1006         struct reclaim_state            *reclaim_state;
1007 
1008         struct backing_dev_info         *backing_dev_info;
1009 
1010         struct io_context               *io_context;
1011 
1012 #ifdef CONFIG_COMPACTION
1013         struct capture_control          *capture_control;
1014 #endif
1015         /* Ptrace state: */
1016         unsigned long                   ptrace_message;
1017         kernel_siginfo_t                *last_siginfo;
1018 
1019         struct task_io_accounting       ioac;
1020 #ifdef CONFIG_PSI
1021         /* Pressure stall state */
1022         unsigned int                    psi_flags;
1023 #endif
1024 #ifdef CONFIG_TASK_XACCT
1025         /* Accumulated RSS usage: */
1026         u64                             acct_rss_mem1;
1027         /* Accumulated virtual memory usage: */
1028         u64                             acct_vm_mem1;
1029         /* stime + utime since last update: */
1030         u64                             acct_timexpd;
1031 #endif
1032 #ifdef CONFIG_CPUSETS
1033         /* Protected by ->alloc_lock: */
1034         nodemask_t                      mems_allowed;
1035         /* Seqence number to catch updates: */
1036         seqcount_t                      mems_allowed_seq;
1037         int                             cpuset_mem_spread_rotor;
1038         int                             cpuset_slab_spread_rotor;
1039 #endif
1040 #ifdef CONFIG_CGROUPS
1041         /* Control Group info protected by css_set_lock: */
1042         struct css_set __rcu            *cgroups;
1043         /* cg_list protected by css_set_lock and tsk->alloc_lock: */
1044         struct list_head                cg_list;
1045 #endif
1046 #ifdef CONFIG_X86_CPU_RESCTRL
1047         u32                             closid;
1048         u32                             rmid;
1049 #endif
1050 #ifdef CONFIG_FUTEX
1051         struct robust_list_head __user  *robust_list;
1052 #ifdef CONFIG_COMPAT
1053         struct compat_robust_list_head __user *compat_robust_list;
1054 #endif
1055         struct list_head                pi_state_list;
1056         struct futex_pi_state           *pi_state_cache;
1057         struct mutex                    futex_exit_mutex;
1058         unsigned int                    futex_state;
1059 #endif
1060 #ifdef CONFIG_PERF_EVENTS
1061         struct perf_event_context       *perf_event_ctxp[perf_nr_task_contexts];
1062         struct mutex                    perf_event_mutex;
1063         struct list_head                perf_event_list;
1064 #endif
1065 #ifdef CONFIG_DEBUG_PREEMPT
1066         unsigned long                   preempt_disable_ip;
1067 #endif
1068 #ifdef CONFIG_NUMA
1069         /* Protected by alloc_lock: */
1070         struct mempolicy                *mempolicy;
1071         short                           il_prev;
1072         short                           pref_node_fork;
1073 #endif
1074 #ifdef CONFIG_NUMA_BALANCING
1075         int                             numa_scan_seq;
1076         unsigned int                    numa_scan_period;
1077         unsigned int                    numa_scan_period_max;
1078         int                             numa_preferred_nid;
1079         unsigned long                   numa_migrate_retry;
1080         /* Migration stamp: */
1081         u64                             node_stamp;
1082         u64                             last_task_numa_placement;
1083         u64                             last_sum_exec_runtime;
1084         struct callback_head            numa_work;
1085 
1086         /*
1087          * This pointer is only modified for current in syscall and
1088          * pagefault context (and for tasks being destroyed), so it can be read
1089          * from any of the following contexts:
1090          *  - RCU read-side critical section
1091          *  - current->numa_group from everywhere
1092          *  - task's runqueue locked, task not running
1093          */
1094         struct numa_group __rcu         *numa_group;
1095 
1096         /*
1097          * numa_faults is an array split into four regions:
1098          * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1099          * in this precise order.
1100          *
1101          * faults_memory: Exponential decaying average of faults on a per-node
1102          * basis. Scheduling placement decisions are made based on these
1103          * counts. The values remain static for the duration of a PTE scan.
1104          * faults_cpu: Track the nodes the process was running on when a NUMA
1105          * hinting fault was incurred.
1106          * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1107          * during the current scan window. When the scan completes, the counts
1108          * in faults_memory and faults_cpu decay and these values are copied.
1109          */
1110         unsigned long                   *numa_faults;
1111         unsigned long                   total_numa_faults;
1112 
1113         /*
1114          * numa_faults_locality tracks if faults recorded during the last
1115          * scan window were remote/local or failed to migrate. The task scan
1116          * period is adapted based on the locality of the faults with different
1117          * weights depending on whether they were shared or private faults
1118          */
1119         unsigned long                   numa_faults_locality[3];
1120 
1121         unsigned long                   numa_pages_migrated;
1122 #endif /* CONFIG_NUMA_BALANCING */
1123 
1124 #ifdef CONFIG_RSEQ
1125         struct rseq __user *rseq;
1126         u32 rseq_sig;
1127         /*
1128          * RmW on rseq_event_mask must be performed atomically
1129          * with respect to preemption.
1130          */
1131         unsigned long rseq_event_mask;
1132 #endif
1133 
1134         struct tlbflush_unmap_batch     tlb_ubc;
1135 
1136         union {
1137                 refcount_t              rcu_users;
1138                 struct rcu_head         rcu;
1139         };
1140 
1141         /* Cache last used pipe for splice(): */
1142         struct pipe_inode_info          *splice_pipe;
1143 
1144         struct page_frag                task_frag;
1145 
1146 #ifdef CONFIG_TASK_DELAY_ACCT
1147         struct task_delay_info          *delays;
1148 #endif
1149 
1150 #ifdef CONFIG_FAULT_INJECTION
1151         int                             make_it_fail;
1152         unsigned int                    fail_nth;
1153 #endif
1154         /*
1155          * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1156          * balance_dirty_pages() for a dirty throttling pause:
1157          */
1158         int                             nr_dirtied;
1159         int                             nr_dirtied_pause;
1160         /* Start of a write-and-pause period: */
1161         unsigned long                   dirty_paused_when;
1162 
1163 #ifdef CONFIG_LATENCYTOP
1164         int                             latency_record_count;
1165         struct latency_record           latency_record[LT_SAVECOUNT];
1166 #endif
1167         /*
1168          * Time slack values; these are used to round up poll() and
1169          * select() etc timeout values. These are in nanoseconds.
1170          */
1171         u64                             timer_slack_ns;
1172         u64                             default_timer_slack_ns;
1173 
1174 #ifdef CONFIG_KASAN
1175         unsigned int                    kasan_depth;
1176 #endif
1177 
1178 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1179         /* Index of current stored address in ret_stack: */
1180         int                             curr_ret_stack;
1181         int                             curr_ret_depth;
1182 
1183         /* Stack of return addresses for return function tracing: */
1184         struct ftrace_ret_stack         *ret_stack;
1185 
1186         /* Timestamp for last schedule: */
1187         unsigned long long              ftrace_timestamp;
1188 
1189         /*
1190          * Number of functions that haven't been traced
1191          * because of depth overrun:
1192          */
1193         atomic_t                        trace_overrun;
1194 
1195         /* Pause tracing: */
1196         atomic_t                        tracing_graph_pause;
1197 #endif
1198 
1199 #ifdef CONFIG_TRACING
1200         /* State flags for use by tracers: */
1201         unsigned long                   trace;
1202 
1203         /* Bitmask and counter of trace recursion: */
1204         unsigned long                   trace_recursion;
1205 #endif /* CONFIG_TRACING */
1206 
1207 #ifdef CONFIG_KCOV
1208         /* Coverage collection mode enabled for this task (0 if disabled): */
1209         unsigned int                    kcov_mode;
1210 
1211         /* Size of the kcov_area: */
1212         unsigned int                    kcov_size;
1213 
1214         /* Buffer for coverage collection: */
1215         void                            *kcov_area;
1216 
1217         /* KCOV descriptor wired with this task or NULL: */
1218         struct kcov                     *kcov;
1219 #endif
1220 
1221 #ifdef CONFIG_MEMCG
1222         struct mem_cgroup               *memcg_in_oom;
1223         gfp_t                           memcg_oom_gfp_mask;
1224         int                             memcg_oom_order;
1225 
1226         /* Number of pages to reclaim on returning to userland: */
1227         unsigned int                    memcg_nr_pages_over_high;
1228 
1229         /* Used by memcontrol for targeted memcg charge: */
1230         struct mem_cgroup               *active_memcg;
1231 #endif
1232 
1233 #ifdef CONFIG_BLK_CGROUP
1234         struct request_queue            *throttle_queue;
1235 #endif
1236 
1237 #ifdef CONFIG_UPROBES
1238         struct uprobe_task              *utask;
1239 #endif
1240 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1241         unsigned int                    sequential_io;
1242         unsigned int                    sequential_io_avg;
1243 #endif
1244 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1245         unsigned long                   task_state_change;
1246 #endif
1247         int                             pagefault_disabled;
1248 #ifdef CONFIG_MMU
1249         struct task_struct              *oom_reaper_list;
1250 #endif
1251 #ifdef CONFIG_VMAP_STACK
1252         struct vm_struct                *stack_vm_area;
1253 #endif
1254 #ifdef CONFIG_THREAD_INFO_IN_TASK
1255         /* A live task holds one reference: */
1256         refcount_t                      stack_refcount;
1257 #endif
1258 #ifdef CONFIG_LIVEPATCH
1259         int patch_state;
1260 #endif
1261 #ifdef CONFIG_SECURITY
1262         /* Used by LSM modules for access restriction: */
1263         void                            *security;
1264 #endif
1265 
1266 #ifdef CONFIG_GCC_PLUGIN_STACKLEAK
1267         unsigned long                   lowest_stack;
1268         unsigned long                   prev_lowest_stack;
1269 #endif
1270 
1271         /*
1272          * New fields for task_struct should be added above here, so that
1273          * they are included in the randomized portion of task_struct.
1274          */
1275         randomized_struct_fields_end
1276 
1277         /* CPU-specific state of this task: */
1278         struct thread_struct            thread;
1279 
1280         /*
1281          * WARNING: on x86, 'thread_struct' contains a variable-sized
1282          * structure.  It *MUST* be at the end of 'task_struct'.
1283          *
1284          * Do not put anything below here!
1285          */
1286 };
1287 
1288 static inline struct pid *task_pid(struct task_struct *task)
1289 {
1290         return task->thread_pid;
1291 }
1292 
1293 /*
1294  * the helpers to get the task's different pids as they are seen
1295  * from various namespaces
1296  *
1297  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1298  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1299  *                     current.
1300  * task_xid_nr_ns()  : id seen from the ns specified;
1301  *
1302  * see also pid_nr() etc in include/linux/pid.h
1303  */
1304 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1305 
1306 static inline pid_t task_pid_nr(struct task_struct *tsk)
1307 {
1308         return tsk->pid;
1309 }
1310 
1311 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1312 {
1313         return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1314 }
1315 
1316 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1317 {
1318         return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1319 }
1320 
1321 
1322 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1323 {
1324         return tsk->tgid;
1325 }
1326 
1327 /**
1328  * pid_alive - check that a task structure is not stale
1329  * @p: Task structure to be checked.
1330  *
1331  * Test if a process is not yet dead (at most zombie state)
1332  * If pid_alive fails, then pointers within the task structure
1333  * can be stale and must not be dereferenced.
1334  *
1335  * Return: 1 if the process is alive. 0 otherwise.
1336  */
1337 static inline int pid_alive(const struct task_struct *p)
1338 {
1339         return p->thread_pid != NULL;
1340 }
1341 
1342 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1343 {
1344         return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1345 }
1346 
1347 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1348 {
1349         return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1350 }
1351 
1352 
1353 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1354 {
1355         return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1356 }
1357 
1358 static inline pid_t task_session_vnr(struct task_struct *tsk)
1359 {
1360         return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1361 }
1362 
1363 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1364 {
1365         return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns);
1366 }
1367 
1368 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1369 {
1370         return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL);
1371 }
1372 
1373 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1374 {
1375         pid_t pid = 0;
1376 
1377         rcu_read_lock();
1378         if (pid_alive(tsk))
1379                 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1380         rcu_read_unlock();
1381 
1382         return pid;
1383 }
1384 
1385 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1386 {
1387         return task_ppid_nr_ns(tsk, &init_pid_ns);
1388 }
1389 
1390 /* Obsolete, do not use: */
1391 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1392 {
1393         return task_pgrp_nr_ns(tsk, &init_pid_ns);
1394 }
1395 
1396 #define TASK_REPORT_IDLE        (TASK_REPORT + 1)
1397 #define TASK_REPORT_MAX         (TASK_REPORT_IDLE << 1)
1398 
1399 static inline unsigned int task_state_index(struct task_struct *tsk)
1400 {
1401         unsigned int tsk_state = READ_ONCE(tsk->state);
1402         unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT;
1403 
1404         BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX);
1405 
1406         if (tsk_state == TASK_IDLE)
1407                 state = TASK_REPORT_IDLE;
1408 
1409         return fls(state);
1410 }
1411 
1412 static inline char task_index_to_char(unsigned int state)
1413 {
1414         static const char state_char[] = "RSDTtXZPI";
1415 
1416         BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1);
1417 
1418         return state_char[state];
1419 }
1420 
1421 static inline char task_state_to_char(struct task_struct *tsk)
1422 {
1423         return task_index_to_char(task_state_index(tsk));
1424 }
1425 
1426 /**
1427  * is_global_init - check if a task structure is init. Since init
1428  * is free to have sub-threads we need to check tgid.
1429  * @tsk: Task structure to be checked.
1430  *
1431  * Check if a task structure is the first user space task the kernel created.
1432  *
1433  * Return: 1 if the task structure is init. 0 otherwise.
1434  */
1435 static inline int is_global_init(struct task_struct *tsk)
1436 {
1437         return task_tgid_nr(tsk) == 1;
1438 }
1439 
1440 extern struct pid *cad_pid;
1441 
1442 /*
1443  * Per process flags
1444  */
1445 #define PF_IDLE                 0x00000002      /* I am an IDLE thread */
1446 #define PF_EXITING              0x00000004      /* Getting shut down */
1447 #define PF_VCPU                 0x00000010      /* I'm a virtual CPU */
1448 #define PF_WQ_WORKER            0x00000020      /* I'm a workqueue worker */
1449 #define PF_FORKNOEXEC           0x00000040      /* Forked but didn't exec */
1450 #define PF_MCE_PROCESS          0x00000080      /* Process policy on mce errors */
1451 #define PF_SUPERPRIV            0x00000100      /* Used super-user privileges */
1452 #define PF_DUMPCORE             0x00000200      /* Dumped core */
1453 #define PF_SIGNALED             0x00000400      /* Killed by a signal */
1454 #define PF_MEMALLOC             0x00000800      /* Allocating memory */
1455 #define PF_NPROC_EXCEEDED       0x00001000      /* set_user() noticed that RLIMIT_NPROC was exceeded */
1456 #define PF_USED_MATH            0x00002000      /* If unset the fpu must be initialized before use */
1457 #define PF_USED_ASYNC           0x00004000      /* Used async_schedule*(), used by module init */
1458 #define PF_NOFREEZE             0x00008000      /* This thread should not be frozen */
1459 #define PF_FROZEN               0x00010000      /* Frozen for system suspend */
1460 #define PF_KSWAPD               0x00020000      /* I am kswapd */
1461 #define PF_MEMALLOC_NOFS        0x00040000      /* All allocation requests will inherit GFP_NOFS */
1462 #define PF_MEMALLOC_NOIO        0x00080000      /* All allocation requests will inherit GFP_NOIO */
1463 #define PF_LESS_THROTTLE        0x00100000      /* Throttle me less: I clean memory */
1464 #define PF_KTHREAD              0x00200000      /* I am a kernel thread */
1465 #define PF_RANDOMIZE            0x00400000      /* Randomize virtual address space */
1466 #define PF_SWAPWRITE            0x00800000      /* Allowed to write to swap */
1467 #define PF_MEMSTALL             0x01000000      /* Stalled due to lack of memory */
1468 #define PF_UMH                  0x02000000      /* I'm an Usermodehelper process */
1469 #define PF_NO_SETAFFINITY       0x04000000      /* Userland is not allowed to meddle with cpus_mask */
1470 #define PF_MCE_EARLY            0x08000000      /* Early kill for mce process policy */
1471 #define PF_MEMALLOC_NOCMA       0x10000000      /* All allocation request will have _GFP_MOVABLE cleared */
1472 #define PF_FREEZER_SKIP         0x40000000      /* Freezer should not count it as freezable */
1473 #define PF_SUSPEND_TASK         0x80000000      /* This thread called freeze_processes() and should not be frozen */
1474 
1475 /*
1476  * Only the _current_ task can read/write to tsk->flags, but other
1477  * tasks can access tsk->flags in readonly mode for example
1478  * with tsk_used_math (like during threaded core dumping).
1479  * There is however an exception to this rule during ptrace
1480  * or during fork: the ptracer task is allowed to write to the
1481  * child->flags of its traced child (same goes for fork, the parent
1482  * can write to the child->flags), because we're guaranteed the
1483  * child is not running and in turn not changing child->flags
1484  * at the same time the parent does it.
1485  */
1486 #define clear_stopped_child_used_math(child)    do { (child)->flags &= ~PF_USED_MATH; } while (0)
1487 #define set_stopped_child_used_math(child)      do { (child)->flags |= PF_USED_MATH; } while (0)
1488 #define clear_used_math()                       clear_stopped_child_used_math(current)
1489 #define set_used_math()                         set_stopped_child_used_math(current)
1490 
1491 #define conditional_stopped_child_used_math(condition, child) \
1492         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1493 
1494 #define conditional_used_math(condition)        conditional_stopped_child_used_math(condition, current)
1495 
1496 #define copy_to_stopped_child_used_math(child) \
1497         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1498 
1499 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1500 #define tsk_used_math(p)                        ((p)->flags & PF_USED_MATH)
1501 #define used_math()                             tsk_used_math(current)
1502 
1503 static inline bool is_percpu_thread(void)
1504 {
1505 #ifdef CONFIG_SMP
1506         return (current->flags & PF_NO_SETAFFINITY) &&
1507                 (current->nr_cpus_allowed  == 1);
1508 #else
1509         return true;
1510 #endif
1511 }
1512 
1513 /* Per-process atomic flags. */
1514 #define PFA_NO_NEW_PRIVS                0       /* May not gain new privileges. */
1515 #define PFA_SPREAD_PAGE                 1       /* Spread page cache over cpuset */
1516 #define PFA_SPREAD_SLAB                 2       /* Spread some slab caches over cpuset */
1517 #define PFA_SPEC_SSB_DISABLE            3       /* Speculative Store Bypass disabled */
1518 #define PFA_SPEC_SSB_FORCE_DISABLE      4       /* Speculative Store Bypass force disabled*/
1519 #define PFA_SPEC_IB_DISABLE             5       /* Indirect branch speculation restricted */
1520 #define PFA_SPEC_IB_FORCE_DISABLE       6       /* Indirect branch speculation permanently restricted */
1521 #define PFA_SPEC_SSB_NOEXEC             7       /* Speculative Store Bypass clear on execve() */
1522 
1523 #define TASK_PFA_TEST(name, func)                                       \
1524         static inline bool task_##func(struct task_struct *p)           \
1525         { return test_bit(PFA_##name, &p->atomic_flags); }
1526 
1527 #define TASK_PFA_SET(name, func)                                        \
1528         static inline void task_set_##func(struct task_struct *p)       \
1529         { set_bit(PFA_##name, &p->atomic_flags); }
1530 
1531 #define TASK_PFA_CLEAR(name, func)                                      \
1532         static inline void task_clear_##func(struct task_struct *p)     \
1533         { clear_bit(PFA_##name, &p->atomic_flags); }
1534 
1535 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1536 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1537 
1538 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1539 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1540 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1541 
1542 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1543 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1544 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1545 
1546 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
1547 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
1548 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
1549 
1550 TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1551 TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1552 TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1553 
1554 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1555 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1556 
1557 TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable)
1558 TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable)
1559 TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable)
1560 
1561 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1562 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1563 
1564 static inline void
1565 current_restore_flags(unsigned long orig_flags, unsigned long flags)
1566 {
1567         current->flags &= ~flags;
1568         current->flags |= orig_flags & flags;
1569 }
1570 
1571 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1572 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1573 #ifdef CONFIG_SMP
1574 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1575 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1576 #else
1577 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1578 {
1579 }
1580 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1581 {
1582         if (!cpumask_test_cpu(0, new_mask))
1583                 return -EINVAL;
1584         return 0;
1585 }
1586 #endif
1587 
1588 extern int yield_to(struct task_struct *p, bool preempt);
1589 extern void set_user_nice(struct task_struct *p, long nice);
1590 extern int task_prio(const struct task_struct *p);
1591 
1592 /**
1593  * task_nice - return the nice value of a given task.
1594  * @p: the task in question.
1595  *
1596  * Return: The nice value [ -20 ... 0 ... 19 ].
1597  */
1598 static inline int task_nice(const struct task_struct *p)
1599 {
1600         return PRIO_TO_NICE((p)->static_prio);
1601 }
1602 
1603 extern int can_nice(const struct task_struct *p, const int nice);
1604 extern int task_curr(const struct task_struct *p);
1605 extern int idle_cpu(int cpu);
1606 extern int available_idle_cpu(int cpu);
1607 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1608 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1609 extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1610 extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *);
1611 extern struct task_struct *idle_task(int cpu);
1612 
1613 /**
1614  * is_idle_task - is the specified task an idle task?
1615  * @p: the task in question.
1616  *
1617  * Return: 1 if @p is an idle task. 0 otherwise.
1618  */
1619 static inline bool is_idle_task(const struct task_struct *p)
1620 {
1621         return !!(p->flags & PF_IDLE);
1622 }
1623 
1624 extern struct task_struct *curr_task(int cpu);
1625 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1626 
1627 void yield(void);
1628 
1629 union thread_union {
1630 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1631         struct task_struct task;
1632 #endif
1633 #ifndef CONFIG_THREAD_INFO_IN_TASK
1634         struct thread_info thread_info;
1635 #endif
1636         unsigned long stack[THREAD_SIZE/sizeof(long)];
1637 };
1638 
1639 #ifndef CONFIG_THREAD_INFO_IN_TASK
1640 extern struct thread_info init_thread_info;
1641 #endif
1642 
1643 extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
1644 
1645 #ifdef CONFIG_THREAD_INFO_IN_TASK
1646 static inline struct thread_info *task_thread_info(struct task_struct *task)
1647 {
1648         return &task->thread_info;
1649 }
1650 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1651 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1652 #endif
1653 
1654 /*
1655  * find a task by one of its numerical ids
1656  *
1657  * find_task_by_pid_ns():
1658  *      finds a task by its pid in the specified namespace
1659  * find_task_by_vpid():
1660  *      finds a task by its virtual pid
1661  *
1662  * see also find_vpid() etc in include/linux/pid.h
1663  */
1664 
1665 extern struct task_struct *find_task_by_vpid(pid_t nr);
1666 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1667 
1668 /*
1669  * find a task by its virtual pid and get the task struct
1670  */
1671 extern struct task_struct *find_get_task_by_vpid(pid_t nr);
1672 
1673 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1674 extern int wake_up_process(struct task_struct *tsk);
1675 extern void wake_up_new_task(struct task_struct *tsk);
1676 
1677 #ifdef CONFIG_SMP
1678 extern void kick_process(struct task_struct *tsk);
1679 #else
1680 static inline void kick_process(struct task_struct *tsk) { }
1681 #endif
1682 
1683 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1684 
1685 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1686 {
1687         __set_task_comm(tsk, from, false);
1688 }
1689 
1690 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
1691 #define get_task_comm(buf, tsk) ({                      \
1692         BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN);     \
1693         __get_task_comm(buf, sizeof(buf), tsk);         \
1694 })
1695 
1696 #ifdef CONFIG_SMP
1697 void scheduler_ipi(void);
1698 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1699 #else
1700 static inline void scheduler_ipi(void) { }
1701 static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1702 {
1703         return 1;
1704 }
1705 #endif
1706 
1707 /*
1708  * Set thread flags in other task's structures.
1709  * See asm/thread_info.h for TIF_xxxx flags available:
1710  */
1711 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1712 {
1713         set_ti_thread_flag(task_thread_info(tsk), flag);
1714 }
1715 
1716 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1717 {
1718         clear_ti_thread_flag(task_thread_info(tsk), flag);
1719 }
1720 
1721 static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag,
1722                                           bool value)
1723 {
1724         update_ti_thread_flag(task_thread_info(tsk), flag, value);
1725 }
1726 
1727 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1728 {
1729         return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1730 }
1731 
1732 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1733 {
1734         return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1735 }
1736 
1737 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1738 {
1739         return test_ti_thread_flag(task_thread_info(tsk), flag);
1740 }
1741 
1742 static inline void set_tsk_need_resched(struct task_struct *tsk)
1743 {
1744         set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1745 }
1746 
1747 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1748 {
1749         clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1750 }
1751 
1752 static inline int test_tsk_need_resched(struct task_struct *tsk)
1753 {
1754         return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1755 }
1756 
1757 /*
1758  * cond_resched() and cond_resched_lock(): latency reduction via
1759  * explicit rescheduling in places that are safe. The return
1760  * value indicates whether a reschedule was done in fact.
1761  * cond_resched_lock() will drop the spinlock before scheduling,
1762  */
1763 #ifndef CONFIG_PREEMPTION
1764 extern int _cond_resched(void);
1765 #else
1766 static inline int _cond_resched(void) { return 0; }
1767 #endif
1768 
1769 #define cond_resched() ({                       \
1770         ___might_sleep(__FILE__, __LINE__, 0);  \
1771         _cond_resched();                        \
1772 })
1773 
1774 extern int __cond_resched_lock(spinlock_t *lock);
1775 
1776 #define cond_resched_lock(lock) ({                              \
1777         ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1778         __cond_resched_lock(lock);                              \
1779 })
1780 
1781 static inline void cond_resched_rcu(void)
1782 {
1783 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1784         rcu_read_unlock();
1785         cond_resched();
1786         rcu_read_lock();
1787 #endif
1788 }
1789 
1790 /*
1791  * Does a critical section need to be broken due to another
1792  * task waiting?: (technically does not depend on CONFIG_PREEMPTION,
1793  * but a general need for low latency)
1794  */
1795 static inline int spin_needbreak(spinlock_t *lock)
1796 {
1797 #ifdef CONFIG_PREEMPTION
1798         return spin_is_contended(lock);
1799 #else
1800         return 0;
1801 #endif
1802 }
1803 
1804 static __always_inline bool need_resched(void)
1805 {
1806         return unlikely(tif_need_resched());
1807 }
1808 
1809 /*
1810  * Wrappers for p->thread_info->cpu access. No-op on UP.
1811  */
1812 #ifdef CONFIG_SMP
1813 
1814 static inline unsigned int task_cpu(const struct task_struct *p)
1815 {
1816 #ifdef CONFIG_THREAD_INFO_IN_TASK
1817         return READ_ONCE(p->cpu);
1818 #else
1819         return READ_ONCE(task_thread_info(p)->cpu);
1820 #endif
1821 }
1822 
1823 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1824 
1825 #else
1826 
1827 static inline unsigned int task_cpu(const struct task_struct *p)
1828 {
1829         return 0;
1830 }
1831 
1832 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1833 {
1834 }
1835 
1836 #endif /* CONFIG_SMP */
1837 
1838 /*
1839  * In order to reduce various lock holder preemption latencies provide an
1840  * interface to see if a vCPU is currently running or not.
1841  *
1842  * This allows us to terminate optimistic spin loops and block, analogous to
1843  * the native optimistic spin heuristic of testing if the lock owner task is
1844  * running or not.
1845  */
1846 #ifndef vcpu_is_preempted
1847 static inline bool vcpu_is_preempted(int cpu)
1848 {
1849         return false;
1850 }
1851 #endif
1852 
1853 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
1854 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
1855 
1856 #ifndef TASK_SIZE_OF
1857 #define TASK_SIZE_OF(tsk)       TASK_SIZE
1858 #endif
1859 
1860 #ifdef CONFIG_RSEQ
1861 
1862 /*
1863  * Map the event mask on the user-space ABI enum rseq_cs_flags
1864  * for direct mask checks.
1865  */
1866 enum rseq_event_mask_bits {
1867         RSEQ_EVENT_PREEMPT_BIT  = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT,
1868         RSEQ_EVENT_SIGNAL_BIT   = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT,
1869         RSEQ_EVENT_MIGRATE_BIT  = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT,
1870 };
1871 
1872 enum rseq_event_mask {
1873         RSEQ_EVENT_PREEMPT      = (1U << RSEQ_EVENT_PREEMPT_BIT),
1874         RSEQ_EVENT_SIGNAL       = (1U << RSEQ_EVENT_SIGNAL_BIT),
1875         RSEQ_EVENT_MIGRATE      = (1U << RSEQ_EVENT_MIGRATE_BIT),
1876 };
1877 
1878 static inline void rseq_set_notify_resume(struct task_struct *t)
1879 {
1880         if (t->rseq)
1881                 set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
1882 }
1883 
1884 void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs);
1885 
1886 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1887                                              struct pt_regs *regs)
1888 {
1889         if (current->rseq)
1890                 __rseq_handle_notify_resume(ksig, regs);
1891 }
1892 
1893 static inline void rseq_signal_deliver(struct ksignal *ksig,
1894                                        struct pt_regs *regs)
1895 {
1896         preempt_disable();
1897         __set_bit(RSEQ_EVENT_SIGNAL_BIT, &current->rseq_event_mask);
1898         preempt_enable();
1899         rseq_handle_notify_resume(ksig, regs);
1900 }
1901 
1902 /* rseq_preempt() requires preemption to be disabled. */
1903 static inline void rseq_preempt(struct task_struct *t)
1904 {
1905         __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask);
1906         rseq_set_notify_resume(t);
1907 }
1908 
1909 /* rseq_migrate() requires preemption to be disabled. */
1910 static inline void rseq_migrate(struct task_struct *t)
1911 {
1912         __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask);
1913         rseq_set_notify_resume(t);
1914 }
1915 
1916 /*
1917  * If parent process has a registered restartable sequences area, the
1918  * child inherits. Unregister rseq for a clone with CLONE_VM set.
1919  */
1920 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1921 {
1922         if (clone_flags & CLONE_VM) {
1923                 t->rseq = NULL;
1924                 t->rseq_sig = 0;
1925                 t->rseq_event_mask = 0;
1926         } else {
1927                 t->rseq = current->rseq;
1928                 t->rseq_sig = current->rseq_sig;
1929                 t->rseq_event_mask = current->rseq_event_mask;
1930         }
1931 }
1932 
1933 static inline void rseq_execve(struct task_struct *t)
1934 {
1935         t->rseq = NULL;
1936         t->rseq_sig = 0;
1937         t->rseq_event_mask = 0;
1938 }
1939 
1940 #else
1941 
1942 static inline void rseq_set_notify_resume(struct task_struct *t)
1943 {
1944 }
1945 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1946                                              struct pt_regs *regs)
1947 {
1948 }
1949 static inline void rseq_signal_deliver(struct ksignal *ksig,
1950                                        struct pt_regs *regs)
1951 {
1952 }
1953 static inline void rseq_preempt(struct task_struct *t)
1954 {
1955 }
1956 static inline void rseq_migrate(struct task_struct *t)
1957 {
1958 }
1959 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1960 {
1961 }
1962 static inline void rseq_execve(struct task_struct *t)
1963 {
1964 }
1965 
1966 #endif
1967 
1968 void __exit_umh(struct task_struct *tsk);
1969 
1970 static inline void exit_umh(struct task_struct *tsk)
1971 {
1972         if (unlikely(tsk->flags & PF_UMH))
1973                 __exit_umh(tsk);
1974 }
1975 
1976 #ifdef CONFIG_DEBUG_RSEQ
1977 
1978 void rseq_syscall(struct pt_regs *regs);
1979 
1980 #else
1981 
1982 static inline void rseq_syscall(struct pt_regs *regs)
1983 {
1984 }
1985 
1986 #endif
1987 
1988 const struct sched_avg *sched_trace_cfs_rq_avg(struct cfs_rq *cfs_rq);
1989 char *sched_trace_cfs_rq_path(struct cfs_rq *cfs_rq, char *str, int len);
1990 int sched_trace_cfs_rq_cpu(struct cfs_rq *cfs_rq);
1991 
1992 const struct sched_avg *sched_trace_rq_avg_rt(struct rq *rq);
1993 const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq);
1994 const struct sched_avg *sched_trace_rq_avg_irq(struct rq *rq);
1995 
1996 int sched_trace_rq_cpu(struct rq *rq);
1997 
1998 const struct cpumask *sched_trace_rd_span(struct root_domain *rd);
1999 
2000 #endif