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 #include <linux/kcsan.h> 35 36 /* task_struct member predeclarations (sorted alphabetically): */ 37 struct audit_context; 38 struct backing_dev_info; 39 struct bio_list; 40 struct blk_plug; 41 struct capture_control; 42 struct cfs_rq; 43 struct fs_struct; 44 struct futex_pi_state; 45 struct io_context; 46 struct mempolicy; 47 struct nameidata; 48 struct nsproxy; 49 struct perf_event_context; 50 struct pid_namespace; 51 struct pipe_inode_info; 52 struct rcu_node; 53 struct reclaim_state; 54 struct robust_list_head; 55 struct root_domain; 56 struct rq; 57 struct sched_attr; 58 struct sched_param; 59 struct seq_file; 60 struct sighand_struct; 61 struct signal_struct; 62 struct task_delay_info; 63 struct task_group; 64 65 /* 66 * Task state bitmask. NOTE! These bits are also 67 * encoded in fs/proc/array.c: get_task_state(). 68 * 69 * We have two separate sets of flags: task->state 70 * is about runnability, while task->exit_state are 71 * about the task exiting. Confusing, but this way 72 * modifying one set can't modify the other one by 73 * mistake. 74 */ 75 76 /* Used in tsk->state: */ 77 #define TASK_RUNNING 0x0000 78 #define TASK_INTERRUPTIBLE 0x0001 79 #define TASK_UNINTERRUPTIBLE 0x0002 80 #define __TASK_STOPPED 0x0004 81 #define __TASK_TRACED 0x0008 82 /* Used in tsk->exit_state: */ 83 #define EXIT_DEAD 0x0010 84 #define EXIT_ZOMBIE 0x0020 85 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD) 86 /* Used in tsk->state again: */ 87 #define TASK_PARKED 0x0040 88 #define TASK_DEAD 0x0080 89 #define TASK_WAKEKILL 0x0100 90 #define TASK_WAKING 0x0200 91 #define TASK_NOLOAD 0x0400 92 #define TASK_NEW 0x0800 93 #define TASK_STATE_MAX 0x1000 94 95 /* Convenience macros for the sake of set_current_state: */ 96 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 97 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 98 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 99 100 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD) 101 102 /* Convenience macros for the sake of wake_up(): */ 103 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 104 105 /* get_task_state(): */ 106 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 107 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 108 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \ 109 TASK_PARKED) 110 111 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 112 113 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 114 115 #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 116 117 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP 118 119 /* 120 * Special states are those that do not use the normal wait-loop pattern. See 121 * the comment with set_special_state(). 122 */ 123 #define is_special_task_state(state) \ 124 ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD)) 125 126 #define __set_current_state(state_value) \ 127 do { \ 128 WARN_ON_ONCE(is_special_task_state(state_value));\ 129 current->task_state_change = _THIS_IP_; \ 130 current->state = (state_value); \ 131 } while (0) 132 133 #define set_current_state(state_value) \ 134 do { \ 135 WARN_ON_ONCE(is_special_task_state(state_value));\ 136 current->task_state_change = _THIS_IP_; \ 137 smp_store_mb(current->state, (state_value)); \ 138 } while (0) 139 140 #define set_special_state(state_value) \ 141 do { \ 142 unsigned long flags; /* may shadow */ \ 143 WARN_ON_ONCE(!is_special_task_state(state_value)); \ 144 raw_spin_lock_irqsave(¤t->pi_lock, flags); \ 145 current->task_state_change = _THIS_IP_; \ 146 current->state = (state_value); \ 147 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \ 148 } while (0) 149 #else 150 /* 151 * set_current_state() includes a barrier so that the write of current->state 152 * is correctly serialised wrt the caller's subsequent test of whether to 153 * actually sleep: 154 * 155 * for (;;) { 156 * set_current_state(TASK_UNINTERRUPTIBLE); 157 * if (!need_sleep) 158 * break; 159 * 160 * schedule(); 161 * } 162 * __set_current_state(TASK_RUNNING); 163 * 164 * If the caller does not need such serialisation (because, for instance, the 165 * condition test and condition change and wakeup are under the same lock) then 166 * use __set_current_state(). 167 * 168 * The above is typically ordered against the wakeup, which does: 169 * 170 * need_sleep = false; 171 * wake_up_state(p, TASK_UNINTERRUPTIBLE); 172 * 173 * where wake_up_state() executes a full memory barrier before accessing the 174 * task state. 175 * 176 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is, 177 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a 178 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING). 179 * 180 * However, with slightly different timing the wakeup TASK_RUNNING store can 181 * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not 182 * a problem either because that will result in one extra go around the loop 183 * and our @cond test will save the day. 184 * 185 * Also see the comments of try_to_wake_up(). 186 */ 187 #define __set_current_state(state_value) \ 188 current->state = (state_value) 189 190 #define set_current_state(state_value) \ 191 smp_store_mb(current->state, (state_value)) 192 193 /* 194 * set_special_state() should be used for those states when the blocking task 195 * can not use the regular condition based wait-loop. In that case we must 196 * serialize against wakeups such that any possible in-flight TASK_RUNNING stores 197 * will not collide with our state change. 198 */ 199 #define set_special_state(state_value) \ 200 do { \ 201 unsigned long flags; /* may shadow */ \ 202 raw_spin_lock_irqsave(¤t->pi_lock, flags); \ 203 current->state = (state_value); \ 204 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \ 205 } while (0) 206 207 #endif 208 209 /* Task command name length: */ 210 #define TASK_COMM_LEN 16 211 212 extern void scheduler_tick(void); 213 214 #define MAX_SCHEDULE_TIMEOUT LONG_MAX 215 216 extern long schedule_timeout(long timeout); 217 extern long schedule_timeout_interruptible(long timeout); 218 extern long schedule_timeout_killable(long timeout); 219 extern long schedule_timeout_uninterruptible(long timeout); 220 extern long schedule_timeout_idle(long timeout); 221 asmlinkage void schedule(void); 222 extern void schedule_preempt_disabled(void); 223 asmlinkage void preempt_schedule_irq(void); 224 225 extern int __must_check io_schedule_prepare(void); 226 extern void io_schedule_finish(int token); 227 extern long io_schedule_timeout(long timeout); 228 extern void io_schedule(void); 229 230 /** 231 * struct prev_cputime - snapshot of system and user cputime 232 * @utime: time spent in user mode 233 * @stime: time spent in system mode 234 * @lock: protects the above two fields 235 * 236 * Stores previous user/system time values such that we can guarantee 237 * monotonicity. 238 */ 239 struct prev_cputime { 240 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE 241 u64 utime; 242 u64 stime; 243 raw_spinlock_t lock; 244 #endif 245 }; 246 247 enum vtime_state { 248 /* Task is sleeping or running in a CPU with VTIME inactive: */ 249 VTIME_INACTIVE = 0, 250 /* Task is idle */ 251 VTIME_IDLE, 252 /* Task runs in kernelspace in a CPU with VTIME active: */ 253 VTIME_SYS, 254 /* Task runs in userspace in a CPU with VTIME active: */ 255 VTIME_USER, 256 /* Task runs as guests in a CPU with VTIME active: */ 257 VTIME_GUEST, 258 }; 259 260 struct vtime { 261 seqcount_t seqcount; 262 unsigned long long starttime; 263 enum vtime_state state; 264 unsigned int cpu; 265 u64 utime; 266 u64 stime; 267 u64 gtime; 268 }; 269 270 /* 271 * Utilization clamp constraints. 272 * @UCLAMP_MIN: Minimum utilization 273 * @UCLAMP_MAX: Maximum utilization 274 * @UCLAMP_CNT: Utilization clamp constraints count 275 */ 276 enum uclamp_id { 277 UCLAMP_MIN = 0, 278 UCLAMP_MAX, 279 UCLAMP_CNT 280 }; 281 282 #ifdef CONFIG_SMP 283 extern struct root_domain def_root_domain; 284 extern struct mutex sched_domains_mutex; 285 #endif 286 287 struct sched_info { 288 #ifdef CONFIG_SCHED_INFO 289 /* Cumulative counters: */ 290 291 /* # of times we have run on this CPU: */ 292 unsigned long pcount; 293 294 /* Time spent waiting on a runqueue: */ 295 unsigned long long run_delay; 296 297 /* Timestamps: */ 298 299 /* When did we last run on a CPU? */ 300 unsigned long long last_arrival; 301 302 /* When were we last queued to run? */ 303 unsigned long long last_queued; 304 305 #endif /* CONFIG_SCHED_INFO */ 306 }; 307 308 /* 309 * Integer metrics need fixed point arithmetic, e.g., sched/fair 310 * has a few: load, load_avg, util_avg, freq, and capacity. 311 * 312 * We define a basic fixed point arithmetic range, and then formalize 313 * all these metrics based on that basic range. 314 */ 315 # define SCHED_FIXEDPOINT_SHIFT 10 316 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT) 317 318 /* Increase resolution of cpu_capacity calculations */ 319 # define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT 320 # define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT) 321 322 struct load_weight { 323 unsigned long weight; 324 u32 inv_weight; 325 }; 326 327 /** 328 * struct util_est - Estimation utilization of FAIR tasks 329 * @enqueued: instantaneous estimated utilization of a task/cpu 330 * @ewma: the Exponential Weighted Moving Average (EWMA) 331 * utilization of a task 332 * 333 * Support data structure to track an Exponential Weighted Moving Average 334 * (EWMA) of a FAIR task's utilization. New samples are added to the moving 335 * average each time a task completes an activation. Sample's weight is chosen 336 * so that the EWMA will be relatively insensitive to transient changes to the 337 * task's workload. 338 * 339 * The enqueued attribute has a slightly different meaning for tasks and cpus: 340 * - task: the task's util_avg at last task dequeue time 341 * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU 342 * Thus, the util_est.enqueued of a task represents the contribution on the 343 * estimated utilization of the CPU where that task is currently enqueued. 344 * 345 * Only for tasks we track a moving average of the past instantaneous 346 * estimated utilization. This allows to absorb sporadic drops in utilization 347 * of an otherwise almost periodic task. 348 */ 349 struct util_est { 350 unsigned int enqueued; 351 unsigned int ewma; 352 #define UTIL_EST_WEIGHT_SHIFT 2 353 } __attribute__((__aligned__(sizeof(u64)))); 354 355 /* 356 * The load/runnable/util_avg accumulates an infinite geometric series 357 * (see __update_load_avg_cfs_rq() in kernel/sched/pelt.c). 358 * 359 * [load_avg definition] 360 * 361 * load_avg = runnable% * scale_load_down(load) 362 * 363 * [runnable_avg definition] 364 * 365 * runnable_avg = runnable% * SCHED_CAPACITY_SCALE 366 * 367 * [util_avg definition] 368 * 369 * util_avg = running% * SCHED_CAPACITY_SCALE 370 * 371 * where runnable% is the time ratio that a sched_entity is runnable and 372 * running% the time ratio that a sched_entity is running. 373 * 374 * For cfs_rq, they are the aggregated values of all runnable and blocked 375 * sched_entities. 376 * 377 * The load/runnable/util_avg doesn't direcly factor frequency scaling and CPU 378 * capacity scaling. The scaling is done through the rq_clock_pelt that is used 379 * for computing those signals (see update_rq_clock_pelt()) 380 * 381 * N.B., the above ratios (runnable% and running%) themselves are in the 382 * range of [0, 1]. To do fixed point arithmetics, we therefore scale them 383 * to as large a range as necessary. This is for example reflected by 384 * util_avg's SCHED_CAPACITY_SCALE. 385 * 386 * [Overflow issue] 387 * 388 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities 389 * with the highest load (=88761), always runnable on a single cfs_rq, 390 * and should not overflow as the number already hits PID_MAX_LIMIT. 391 * 392 * For all other cases (including 32-bit kernels), struct load_weight's 393 * weight will overflow first before we do, because: 394 * 395 * Max(load_avg) <= Max(load.weight) 396 * 397 * Then it is the load_weight's responsibility to consider overflow 398 * issues. 399 */ 400 struct sched_avg { 401 u64 last_update_time; 402 u64 load_sum; 403 u64 runnable_sum; 404 u32 util_sum; 405 u32 period_contrib; 406 unsigned long load_avg; 407 unsigned long runnable_avg; 408 unsigned long util_avg; 409 struct util_est util_est; 410 } ____cacheline_aligned; 411 412 struct sched_statistics { 413 #ifdef CONFIG_SCHEDSTATS 414 u64 wait_start; 415 u64 wait_max; 416 u64 wait_count; 417 u64 wait_sum; 418 u64 iowait_count; 419 u64 iowait_sum; 420 421 u64 sleep_start; 422 u64 sleep_max; 423 s64 sum_sleep_runtime; 424 425 u64 block_start; 426 u64 block_max; 427 u64 exec_max; 428 u64 slice_max; 429 430 u64 nr_migrations_cold; 431 u64 nr_failed_migrations_affine; 432 u64 nr_failed_migrations_running; 433 u64 nr_failed_migrations_hot; 434 u64 nr_forced_migrations; 435 436 u64 nr_wakeups; 437 u64 nr_wakeups_sync; 438 u64 nr_wakeups_migrate; 439 u64 nr_wakeups_local; 440 u64 nr_wakeups_remote; 441 u64 nr_wakeups_affine; 442 u64 nr_wakeups_affine_attempts; 443 u64 nr_wakeups_passive; 444 u64 nr_wakeups_idle; 445 #endif 446 }; 447 448 struct sched_entity { 449 /* For load-balancing: */ 450 struct load_weight load; 451 struct rb_node run_node; 452 struct list_head group_node; 453 unsigned int on_rq; 454 455 u64 exec_start; 456 u64 sum_exec_runtime; 457 u64 vruntime; 458 u64 prev_sum_exec_runtime; 459 460 u64 nr_migrations; 461 462 struct sched_statistics statistics; 463 464 #ifdef CONFIG_FAIR_GROUP_SCHED 465 int depth; 466 struct sched_entity *parent; 467 /* rq on which this entity is (to be) queued: */ 468 struct cfs_rq *cfs_rq; 469 /* rq "owned" by this entity/group: */ 470 struct cfs_rq *my_q; 471 /* cached value of my_q->h_nr_running */ 472 unsigned long runnable_weight; 473 #endif 474 475 #ifdef CONFIG_SMP 476 /* 477 * Per entity load average tracking. 478 * 479 * Put into separate cache line so it does not 480 * collide with read-mostly values above. 481 */ 482 struct sched_avg avg; 483 #endif 484 }; 485 486 struct sched_rt_entity { 487 struct list_head run_list; 488 unsigned long timeout; 489 unsigned long watchdog_stamp; 490 unsigned int time_slice; 491 unsigned short on_rq; 492 unsigned short on_list; 493 494 struct sched_rt_entity *back; 495 #ifdef CONFIG_RT_GROUP_SCHED 496 struct sched_rt_entity *parent; 497 /* rq on which this entity is (to be) queued: */ 498 struct rt_rq *rt_rq; 499 /* rq "owned" by this entity/group: */ 500 struct rt_rq *my_q; 501 #endif 502 } __randomize_layout; 503 504 struct sched_dl_entity { 505 struct rb_node rb_node; 506 507 /* 508 * Original scheduling parameters. Copied here from sched_attr 509 * during sched_setattr(), they will remain the same until 510 * the next sched_setattr(). 511 */ 512 u64 dl_runtime; /* Maximum runtime for each instance */ 513 u64 dl_deadline; /* Relative deadline of each instance */ 514 u64 dl_period; /* Separation of two instances (period) */ 515 u64 dl_bw; /* dl_runtime / dl_period */ 516 u64 dl_density; /* dl_runtime / dl_deadline */ 517 518 /* 519 * Actual scheduling parameters. Initialized with the values above, 520 * they are continuously updated during task execution. Note that 521 * the remaining runtime could be < 0 in case we are in overrun. 522 */ 523 s64 runtime; /* Remaining runtime for this instance */ 524 u64 deadline; /* Absolute deadline for this instance */ 525 unsigned int flags; /* Specifying the scheduler behaviour */ 526 527 /* 528 * Some bool flags: 529 * 530 * @dl_throttled tells if we exhausted the runtime. If so, the 531 * task has to wait for a replenishment to be performed at the 532 * next firing of dl_timer. 533 * 534 * @dl_boosted tells if we are boosted due to DI. If so we are 535 * outside bandwidth enforcement mechanism (but only until we 536 * exit the critical section); 537 * 538 * @dl_yielded tells if task gave up the CPU before consuming 539 * all its available runtime during the last job. 540 * 541 * @dl_non_contending tells if the task is inactive while still 542 * contributing to the active utilization. In other words, it 543 * indicates if the inactive timer has been armed and its handler 544 * has not been executed yet. This flag is useful to avoid race 545 * conditions between the inactive timer handler and the wakeup 546 * code. 547 * 548 * @dl_overrun tells if the task asked to be informed about runtime 549 * overruns. 550 */ 551 unsigned int dl_throttled : 1; 552 unsigned int dl_boosted : 1; 553 unsigned int dl_yielded : 1; 554 unsigned int dl_non_contending : 1; 555 unsigned int dl_overrun : 1; 556 557 /* 558 * Bandwidth enforcement timer. Each -deadline task has its 559 * own bandwidth to be enforced, thus we need one timer per task. 560 */ 561 struct hrtimer dl_timer; 562 563 /* 564 * Inactive timer, responsible for decreasing the active utilization 565 * at the "0-lag time". When a -deadline task blocks, it contributes 566 * to GRUB's active utilization until the "0-lag time", hence a 567 * timer is needed to decrease the active utilization at the correct 568 * time. 569 */ 570 struct hrtimer inactive_timer; 571 }; 572 573 #ifdef CONFIG_UCLAMP_TASK 574 /* Number of utilization clamp buckets (shorter alias) */ 575 #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT 576 577 /* 578 * Utilization clamp for a scheduling entity 579 * @value: clamp value "assigned" to a se 580 * @bucket_id: bucket index corresponding to the "assigned" value 581 * @active: the se is currently refcounted in a rq's bucket 582 * @user_defined: the requested clamp value comes from user-space 583 * 584 * The bucket_id is the index of the clamp bucket matching the clamp value 585 * which is pre-computed and stored to avoid expensive integer divisions from 586 * the fast path. 587 * 588 * The active bit is set whenever a task has got an "effective" value assigned, 589 * which can be different from the clamp value "requested" from user-space. 590 * This allows to know a task is refcounted in the rq's bucket corresponding 591 * to the "effective" bucket_id. 592 * 593 * The user_defined bit is set whenever a task has got a task-specific clamp 594 * value requested from userspace, i.e. the system defaults apply to this task 595 * just as a restriction. This allows to relax default clamps when a less 596 * restrictive task-specific value has been requested, thus allowing to 597 * implement a "nice" semantic. For example, a task running with a 20% 598 * default boost can still drop its own boosting to 0%. 599 */ 600 struct uclamp_se { 601 unsigned int value : bits_per(SCHED_CAPACITY_SCALE); 602 unsigned int bucket_id : bits_per(UCLAMP_BUCKETS); 603 unsigned int active : 1; 604 unsigned int user_defined : 1; 605 }; 606 #endif /* CONFIG_UCLAMP_TASK */ 607 608 union rcu_special { 609 struct { 610 u8 blocked; 611 u8 need_qs; 612 u8 exp_hint; /* Hint for performance. */ 613 u8 need_mb; /* Readers need smp_mb(). */ 614 } b; /* Bits. */ 615 u32 s; /* Set of bits. */ 616 }; 617 618 enum perf_event_task_context { 619 perf_invalid_context = -1, 620 perf_hw_context = 0, 621 perf_sw_context, 622 perf_nr_task_contexts, 623 }; 624 625 struct wake_q_node { 626 struct wake_q_node *next; 627 }; 628 629 struct task_struct { 630 #ifdef CONFIG_THREAD_INFO_IN_TASK 631 /* 632 * For reasons of header soup (see current_thread_info()), this 633 * must be the first element of task_struct. 634 */ 635 struct thread_info thread_info; 636 #endif 637 /* -1 unrunnable, 0 runnable, >0 stopped: */ 638 volatile long state; 639 640 /* 641 * This begins the randomizable portion of task_struct. Only 642 * scheduling-critical items should be added above here. 643 */ 644 randomized_struct_fields_start 645 646 void *stack; 647 refcount_t usage; 648 /* Per task flags (PF_*), defined further below: */ 649 unsigned int flags; 650 unsigned int ptrace; 651 652 #ifdef CONFIG_SMP 653 int on_cpu; 654 struct __call_single_node wake_entry; 655 #ifdef CONFIG_THREAD_INFO_IN_TASK 656 /* Current CPU: */ 657 unsigned int cpu; 658 #endif 659 unsigned int wakee_flips; 660 unsigned long wakee_flip_decay_ts; 661 struct task_struct *last_wakee; 662 663 /* 664 * recent_used_cpu is initially set as the last CPU used by a task 665 * that wakes affine another task. Waker/wakee relationships can 666 * push tasks around a CPU where each wakeup moves to the next one. 667 * Tracking a recently used CPU allows a quick search for a recently 668 * used CPU that may be idle. 669 */ 670 int recent_used_cpu; 671 int wake_cpu; 672 #endif 673 int on_rq; 674 675 int prio; 676 int static_prio; 677 int normal_prio; 678 unsigned int rt_priority; 679 680 const struct sched_class *sched_class; 681 struct sched_entity se; 682 struct sched_rt_entity rt; 683 #ifdef CONFIG_CGROUP_SCHED 684 struct task_group *sched_task_group; 685 #endif 686 struct sched_dl_entity dl; 687 688 #ifdef CONFIG_UCLAMP_TASK 689 /* Clamp values requested for a scheduling entity */ 690 struct uclamp_se uclamp_req[UCLAMP_CNT]; 691 /* Effective clamp values used for a scheduling entity */ 692 struct uclamp_se uclamp[UCLAMP_CNT]; 693 #endif 694 695 #ifdef CONFIG_PREEMPT_NOTIFIERS 696 /* List of struct preempt_notifier: */ 697 struct hlist_head preempt_notifiers; 698 #endif 699 700 #ifdef CONFIG_BLK_DEV_IO_TRACE 701 unsigned int btrace_seq; 702 #endif 703 704 unsigned int policy; 705 int nr_cpus_allowed; 706 const cpumask_t *cpus_ptr; 707 cpumask_t cpus_mask; 708 709 #ifdef CONFIG_PREEMPT_RCU 710 int rcu_read_lock_nesting; 711 union rcu_special rcu_read_unlock_special; 712 struct list_head rcu_node_entry; 713 struct rcu_node *rcu_blocked_node; 714 #endif /* #ifdef CONFIG_PREEMPT_RCU */ 715 716 #ifdef CONFIG_TASKS_RCU 717 unsigned long rcu_tasks_nvcsw; 718 u8 rcu_tasks_holdout; 719 u8 rcu_tasks_idx; 720 int rcu_tasks_idle_cpu; 721 struct list_head rcu_tasks_holdout_list; 722 #endif /* #ifdef CONFIG_TASKS_RCU */ 723 724 #ifdef CONFIG_TASKS_TRACE_RCU 725 int trc_reader_nesting; 726 int trc_ipi_to_cpu; 727 union rcu_special trc_reader_special; 728 bool trc_reader_checked; 729 struct list_head trc_holdout_list; 730 #endif /* #ifdef CONFIG_TASKS_TRACE_RCU */ 731 732 struct sched_info sched_info; 733 734 struct list_head tasks; 735 #ifdef CONFIG_SMP 736 struct plist_node pushable_tasks; 737 struct rb_node pushable_dl_tasks; 738 #endif 739 740 struct mm_struct *mm; 741 struct mm_struct *active_mm; 742 743 /* Per-thread vma caching: */ 744 struct vmacache vmacache; 745 746 #ifdef SPLIT_RSS_COUNTING 747 struct task_rss_stat rss_stat; 748 #endif 749 int exit_state; 750 int exit_code; 751 int exit_signal; 752 /* The signal sent when the parent dies: */ 753 int pdeath_signal; 754 /* JOBCTL_*, siglock protected: */ 755 unsigned long jobctl; 756 757 /* Used for emulating ABI behavior of previous Linux versions: */ 758 unsigned int personality; 759 760 /* Scheduler bits, serialized by scheduler locks: */ 761 unsigned sched_reset_on_fork:1; 762 unsigned sched_contributes_to_load:1; 763 unsigned sched_migrated:1; 764 unsigned sched_remote_wakeup:1; 765 #ifdef CONFIG_PSI 766 unsigned sched_psi_wake_requeue:1; 767 #endif 768 769 /* Force alignment to the next boundary: */ 770 unsigned :0; 771 772 /* Unserialized, strictly 'current' */ 773 774 /* Bit to tell LSMs we're in execve(): */ 775 unsigned in_execve:1; 776 unsigned in_iowait:1; 777 #ifndef TIF_RESTORE_SIGMASK 778 unsigned restore_sigmask:1; 779 #endif 780 #ifdef CONFIG_MEMCG 781 unsigned in_user_fault:1; 782 #endif 783 #ifdef CONFIG_COMPAT_BRK 784 unsigned brk_randomized:1; 785 #endif 786 #ifdef CONFIG_CGROUPS 787 /* disallow userland-initiated cgroup migration */ 788 unsigned no_cgroup_migration:1; 789 /* task is frozen/stopped (used by the cgroup freezer) */ 790 unsigned frozen:1; 791 #endif 792 #ifdef CONFIG_BLK_CGROUP 793 unsigned use_memdelay:1; 794 #endif 795 #ifdef CONFIG_PSI 796 /* Stalled due to lack of memory */ 797 unsigned in_memstall:1; 798 #endif 799 800 unsigned long atomic_flags; /* Flags requiring atomic access. */ 801 802 struct restart_block restart_block; 803 804 pid_t pid; 805 pid_t tgid; 806 807 #ifdef CONFIG_STACKPROTECTOR 808 /* Canary value for the -fstack-protector GCC feature: */ 809 unsigned long stack_canary; 810 #endif 811 /* 812 * Pointers to the (original) parent process, youngest child, younger sibling, 813 * older sibling, respectively. (p->father can be replaced with 814 * p->real_parent->pid) 815 */ 816 817 /* Real parent process: */ 818 struct task_struct __rcu *real_parent; 819 820 /* Recipient of SIGCHLD, wait4() reports: */ 821 struct task_struct __rcu *parent; 822 823 /* 824 * Children/sibling form the list of natural children: 825 */ 826 struct list_head children; 827 struct list_head sibling; 828 struct task_struct *group_leader; 829 830 /* 831 * 'ptraced' is the list of tasks this task is using ptrace() on. 832 * 833 * This includes both natural children and PTRACE_ATTACH targets. 834 * 'ptrace_entry' is this task's link on the p->parent->ptraced list. 835 */ 836 struct list_head ptraced; 837 struct list_head ptrace_entry; 838 839 /* PID/PID hash table linkage. */ 840 struct pid *thread_pid; 841 struct hlist_node pid_links[PIDTYPE_MAX]; 842 struct list_head thread_group; 843 struct list_head thread_node; 844 845 struct completion *vfork_done; 846 847 /* CLONE_CHILD_SETTID: */ 848 int __user *set_child_tid; 849 850 /* CLONE_CHILD_CLEARTID: */ 851 int __user *clear_child_tid; 852 853 u64 utime; 854 u64 stime; 855 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME 856 u64 utimescaled; 857 u64 stimescaled; 858 #endif 859 u64 gtime; 860 struct prev_cputime prev_cputime; 861 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN 862 struct vtime vtime; 863 #endif 864 865 #ifdef CONFIG_NO_HZ_FULL 866 atomic_t tick_dep_mask; 867 #endif 868 /* Context switch counts: */ 869 unsigned long nvcsw; 870 unsigned long nivcsw; 871 872 /* Monotonic time in nsecs: */ 873 u64 start_time; 874 875 /* Boot based time in nsecs: */ 876 u64 start_boottime; 877 878 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */ 879 unsigned long min_flt; 880 unsigned long maj_flt; 881 882 /* Empty if CONFIG_POSIX_CPUTIMERS=n */ 883 struct posix_cputimers posix_cputimers; 884 885 /* Process credentials: */ 886 887 /* Tracer's credentials at attach: */ 888 const struct cred __rcu *ptracer_cred; 889 890 /* Objective and real subjective task credentials (COW): */ 891 const struct cred __rcu *real_cred; 892 893 /* Effective (overridable) subjective task credentials (COW): */ 894 const struct cred __rcu *cred; 895 896 #ifdef CONFIG_KEYS 897 /* Cached requested key. */ 898 struct key *cached_requested_key; 899 #endif 900 901 /* 902 * executable name, excluding path. 903 * 904 * - normally initialized setup_new_exec() 905 * - access it with [gs]et_task_comm() 906 * - lock it with task_lock() 907 */ 908 char comm[TASK_COMM_LEN]; 909 910 struct nameidata *nameidata; 911 912 #ifdef CONFIG_SYSVIPC 913 struct sysv_sem sysvsem; 914 struct sysv_shm sysvshm; 915 #endif 916 #ifdef CONFIG_DETECT_HUNG_TASK 917 unsigned long last_switch_count; 918 unsigned long last_switch_time; 919 #endif 920 /* Filesystem information: */ 921 struct fs_struct *fs; 922 923 /* Open file information: */ 924 struct files_struct *files; 925 926 /* Namespaces: */ 927 struct nsproxy *nsproxy; 928 929 /* Signal handlers: */ 930 struct signal_struct *signal; 931 struct sighand_struct __rcu *sighand; 932 sigset_t blocked; 933 sigset_t real_blocked; 934 /* Restored if set_restore_sigmask() was used: */ 935 sigset_t saved_sigmask; 936 struct sigpending pending; 937 unsigned long sas_ss_sp; 938 size_t sas_ss_size; 939 unsigned int sas_ss_flags; 940 941 struct callback_head *task_works; 942 943 #ifdef CONFIG_AUDIT 944 #ifdef CONFIG_AUDITSYSCALL 945 struct audit_context *audit_context; 946 #endif 947 kuid_t loginuid; 948 unsigned int sessionid; 949 #endif 950 struct seccomp seccomp; 951 952 /* Thread group tracking: */ 953 u64 parent_exec_id; 954 u64 self_exec_id; 955 956 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */ 957 spinlock_t alloc_lock; 958 959 /* Protection of the PI data structures: */ 960 raw_spinlock_t pi_lock; 961 962 struct wake_q_node wake_q; 963 964 #ifdef CONFIG_RT_MUTEXES 965 /* PI waiters blocked on a rt_mutex held by this task: */ 966 struct rb_root_cached pi_waiters; 967 /* Updated under owner's pi_lock and rq lock */ 968 struct task_struct *pi_top_task; 969 /* Deadlock detection and priority inheritance handling: */ 970 struct rt_mutex_waiter *pi_blocked_on; 971 #endif 972 973 #ifdef CONFIG_DEBUG_MUTEXES 974 /* Mutex deadlock detection: */ 975 struct mutex_waiter *blocked_on; 976 #endif 977 978 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP 979 int non_block_count; 980 #endif 981 982 #ifdef CONFIG_TRACE_IRQFLAGS 983 unsigned int irq_events; 984 unsigned int hardirq_threaded; 985 unsigned long hardirq_enable_ip; 986 unsigned long hardirq_disable_ip; 987 unsigned int hardirq_enable_event; 988 unsigned int hardirq_disable_event; 989 int hardirqs_enabled; 990 int hardirq_context; 991 u64 hardirq_chain_key; 992 unsigned long softirq_disable_ip; 993 unsigned long softirq_enable_ip; 994 unsigned int softirq_disable_event; 995 unsigned int softirq_enable_event; 996 int softirqs_enabled; 997 int softirq_context; 998 int irq_config; 999 #endif 1000 1001 #ifdef CONFIG_LOCKDEP 1002 # define MAX_LOCK_DEPTH 48UL 1003 u64 curr_chain_key; 1004 int lockdep_depth; 1005 unsigned int lockdep_recursion; 1006 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1007 #endif 1008 1009 #ifdef CONFIG_UBSAN 1010 unsigned int in_ubsan; 1011 #endif 1012 1013 /* Journalling filesystem info: */ 1014 void *journal_info; 1015 1016 /* Stacked block device info: */ 1017 struct bio_list *bio_list; 1018 1019 #ifdef CONFIG_BLOCK 1020 /* Stack plugging: */ 1021 struct blk_plug *plug; 1022 #endif 1023 1024 /* VM state: */ 1025 struct reclaim_state *reclaim_state; 1026 1027 struct backing_dev_info *backing_dev_info; 1028 1029 struct io_context *io_context; 1030 1031 #ifdef CONFIG_COMPACTION 1032 struct capture_control *capture_control; 1033 #endif 1034 /* Ptrace state: */ 1035 unsigned long ptrace_message; 1036 kernel_siginfo_t *last_siginfo; 1037 1038 struct task_io_accounting ioac; 1039 #ifdef CONFIG_PSI 1040 /* Pressure stall state */ 1041 unsigned int psi_flags; 1042 #endif 1043 #ifdef CONFIG_TASK_XACCT 1044 /* Accumulated RSS usage: */ 1045 u64 acct_rss_mem1; 1046 /* Accumulated virtual memory usage: */ 1047 u64 acct_vm_mem1; 1048 /* stime + utime since last update: */ 1049 u64 acct_timexpd; 1050 #endif 1051 #ifdef CONFIG_CPUSETS 1052 /* Protected by ->alloc_lock: */ 1053 nodemask_t mems_allowed; 1054 /* Seqence number to catch updates: */ 1055 seqcount_t mems_allowed_seq; 1056 int cpuset_mem_spread_rotor; 1057 int cpuset_slab_spread_rotor; 1058 #endif 1059 #ifdef CONFIG_CGROUPS 1060 /* Control Group info protected by css_set_lock: */ 1061 struct css_set __rcu *cgroups; 1062 /* cg_list protected by css_set_lock and tsk->alloc_lock: */ 1063 struct list_head cg_list; 1064 #endif 1065 #ifdef CONFIG_X86_CPU_RESCTRL 1066 u32 closid; 1067 u32 rmid; 1068 #endif 1069 #ifdef CONFIG_FUTEX 1070 struct robust_list_head __user *robust_list; 1071 #ifdef CONFIG_COMPAT 1072 struct compat_robust_list_head __user *compat_robust_list; 1073 #endif 1074 struct list_head pi_state_list; 1075 struct futex_pi_state *pi_state_cache; 1076 struct mutex futex_exit_mutex; 1077 unsigned int futex_state; 1078 #endif 1079 #ifdef CONFIG_PERF_EVENTS 1080 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts]; 1081 struct mutex perf_event_mutex; 1082 struct list_head perf_event_list; 1083 #endif 1084 #ifdef CONFIG_DEBUG_PREEMPT 1085 unsigned long preempt_disable_ip; 1086 #endif 1087 #ifdef CONFIG_NUMA 1088 /* Protected by alloc_lock: */ 1089 struct mempolicy *mempolicy; 1090 short il_prev; 1091 short pref_node_fork; 1092 #endif 1093 #ifdef CONFIG_NUMA_BALANCING 1094 int numa_scan_seq; 1095 unsigned int numa_scan_period; 1096 unsigned int numa_scan_period_max; 1097 int numa_preferred_nid; 1098 unsigned long numa_migrate_retry; 1099 /* Migration stamp: */ 1100 u64 node_stamp; 1101 u64 last_task_numa_placement; 1102 u64 last_sum_exec_runtime; 1103 struct callback_head numa_work; 1104 1105 /* 1106 * This pointer is only modified for current in syscall and 1107 * pagefault context (and for tasks being destroyed), so it can be read 1108 * from any of the following contexts: 1109 * - RCU read-side critical section 1110 * - current->numa_group from everywhere 1111 * - task's runqueue locked, task not running 1112 */ 1113 struct numa_group __rcu *numa_group; 1114 1115 /* 1116 * numa_faults is an array split into four regions: 1117 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer 1118 * in this precise order. 1119 * 1120 * faults_memory: Exponential decaying average of faults on a per-node 1121 * basis. Scheduling placement decisions are made based on these 1122 * counts. The values remain static for the duration of a PTE scan. 1123 * faults_cpu: Track the nodes the process was running on when a NUMA 1124 * hinting fault was incurred. 1125 * faults_memory_buffer and faults_cpu_buffer: Record faults per node 1126 * during the current scan window. When the scan completes, the counts 1127 * in faults_memory and faults_cpu decay and these values are copied. 1128 */ 1129 unsigned long *numa_faults; 1130 unsigned long total_numa_faults; 1131 1132 /* 1133 * numa_faults_locality tracks if faults recorded during the last 1134 * scan window were remote/local or failed to migrate. The task scan 1135 * period is adapted based on the locality of the faults with different 1136 * weights depending on whether they were shared or private faults 1137 */ 1138 unsigned long numa_faults_locality[3]; 1139 1140 unsigned long numa_pages_migrated; 1141 #endif /* CONFIG_NUMA_BALANCING */ 1142 1143 #ifdef CONFIG_RSEQ 1144 struct rseq __user *rseq; 1145 u32 rseq_sig; 1146 /* 1147 * RmW on rseq_event_mask must be performed atomically 1148 * with respect to preemption. 1149 */ 1150 unsigned long rseq_event_mask; 1151 #endif 1152 1153 struct tlbflush_unmap_batch tlb_ubc; 1154 1155 union { 1156 refcount_t rcu_users; 1157 struct rcu_head rcu; 1158 }; 1159 1160 /* Cache last used pipe for splice(): */ 1161 struct pipe_inode_info *splice_pipe; 1162 1163 struct page_frag task_frag; 1164 1165 #ifdef CONFIG_TASK_DELAY_ACCT 1166 struct task_delay_info *delays; 1167 #endif 1168 1169 #ifdef CONFIG_FAULT_INJECTION 1170 int make_it_fail; 1171 unsigned int fail_nth; 1172 #endif 1173 /* 1174 * When (nr_dirtied >= nr_dirtied_pause), it's time to call 1175 * balance_dirty_pages() for a dirty throttling pause: 1176 */ 1177 int nr_dirtied; 1178 int nr_dirtied_pause; 1179 /* Start of a write-and-pause period: */ 1180 unsigned long dirty_paused_when; 1181 1182 #ifdef CONFIG_LATENCYTOP 1183 int latency_record_count; 1184 struct latency_record latency_record[LT_SAVECOUNT]; 1185 #endif 1186 /* 1187 * Time slack values; these are used to round up poll() and 1188 * select() etc timeout values. These are in nanoseconds. 1189 */ 1190 u64 timer_slack_ns; 1191 u64 default_timer_slack_ns; 1192 1193 #ifdef CONFIG_KASAN 1194 unsigned int kasan_depth; 1195 #endif 1196 #ifdef CONFIG_KCSAN 1197 struct kcsan_ctx kcsan_ctx; 1198 #endif 1199 1200 #ifdef CONFIG_FUNCTION_GRAPH_TRACER 1201 /* Index of current stored address in ret_stack: */ 1202 int curr_ret_stack; 1203 int curr_ret_depth; 1204 1205 /* Stack of return addresses for return function tracing: */ 1206 struct ftrace_ret_stack *ret_stack; 1207 1208 /* Timestamp for last schedule: */ 1209 unsigned long long ftrace_timestamp; 1210 1211 /* 1212 * Number of functions that haven't been traced 1213 * because of depth overrun: 1214 */ 1215 atomic_t trace_overrun; 1216 1217 /* Pause tracing: */ 1218 atomic_t tracing_graph_pause; 1219 #endif 1220 1221 #ifdef CONFIG_TRACING 1222 /* State flags for use by tracers: */ 1223 unsigned long trace; 1224 1225 /* Bitmask and counter of trace recursion: */ 1226 unsigned long trace_recursion; 1227 #endif /* CONFIG_TRACING */ 1228 1229 #ifdef CONFIG_KCOV 1230 /* See kernel/kcov.c for more details. */ 1231 1232 /* Coverage collection mode enabled for this task (0 if disabled): */ 1233 unsigned int kcov_mode; 1234 1235 /* Size of the kcov_area: */ 1236 unsigned int kcov_size; 1237 1238 /* Buffer for coverage collection: */ 1239 void *kcov_area; 1240 1241 /* KCOV descriptor wired with this task or NULL: */ 1242 struct kcov *kcov; 1243 1244 /* KCOV common handle for remote coverage collection: */ 1245 u64 kcov_handle; 1246 1247 /* KCOV sequence number: */ 1248 int kcov_sequence; 1249 1250 /* Collect coverage from softirq context: */ 1251 unsigned int kcov_softirq; 1252 #endif 1253 1254 #ifdef CONFIG_MEMCG 1255 struct mem_cgroup *memcg_in_oom; 1256 gfp_t memcg_oom_gfp_mask; 1257 int memcg_oom_order; 1258 1259 /* Number of pages to reclaim on returning to userland: */ 1260 unsigned int memcg_nr_pages_over_high; 1261 1262 /* Used by memcontrol for targeted memcg charge: */ 1263 struct mem_cgroup *active_memcg; 1264 #endif 1265 1266 #ifdef CONFIG_BLK_CGROUP 1267 struct request_queue *throttle_queue; 1268 #endif 1269 1270 #ifdef CONFIG_UPROBES 1271 struct uprobe_task *utask; 1272 #endif 1273 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE) 1274 unsigned int sequential_io; 1275 unsigned int sequential_io_avg; 1276 #endif 1277 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP 1278 unsigned long task_state_change; 1279 #endif 1280 int pagefault_disabled; 1281 #ifdef CONFIG_MMU 1282 struct task_struct *oom_reaper_list; 1283 #endif 1284 #ifdef CONFIG_VMAP_STACK 1285 struct vm_struct *stack_vm_area; 1286 #endif 1287 #ifdef CONFIG_THREAD_INFO_IN_TASK 1288 /* A live task holds one reference: */ 1289 refcount_t stack_refcount; 1290 #endif 1291 #ifdef CONFIG_LIVEPATCH 1292 int patch_state; 1293 #endif 1294 #ifdef CONFIG_SECURITY 1295 /* Used by LSM modules for access restriction: */ 1296 void *security; 1297 #endif 1298 1299 #ifdef CONFIG_GCC_PLUGIN_STACKLEAK 1300 unsigned long lowest_stack; 1301 unsigned long prev_lowest_stack; 1302 #endif 1303 1304 #ifdef CONFIG_X86_MCE 1305 u64 mce_addr; 1306 __u64 mce_ripv : 1, 1307 mce_whole_page : 1, 1308 __mce_reserved : 62; 1309 struct callback_head mce_kill_me; 1310 #endif 1311 1312 /* 1313 * New fields for task_struct should be added above here, so that 1314 * they are included in the randomized portion of task_struct. 1315 */ 1316 randomized_struct_fields_end 1317 1318 /* CPU-specific state of this task: */ 1319 struct thread_struct thread; 1320 1321 /* 1322 * WARNING: on x86, 'thread_struct' contains a variable-sized 1323 * structure. It *MUST* be at the end of 'task_struct'. 1324 * 1325 * Do not put anything below here! 1326 */ 1327 }; 1328 1329 static inline struct pid *task_pid(struct task_struct *task) 1330 { 1331 return task->thread_pid; 1332 } 1333 1334 /* 1335 * the helpers to get the task's different pids as they are seen 1336 * from various namespaces 1337 * 1338 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1339 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1340 * current. 1341 * task_xid_nr_ns() : id seen from the ns specified; 1342 * 1343 * see also pid_nr() etc in include/linux/pid.h 1344 */ 1345 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns); 1346 1347 static inline pid_t task_pid_nr(struct task_struct *tsk) 1348 { 1349 return tsk->pid; 1350 } 1351 1352 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) 1353 { 1354 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); 1355 } 1356 1357 static inline pid_t task_pid_vnr(struct task_struct *tsk) 1358 { 1359 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); 1360 } 1361 1362 1363 static inline pid_t task_tgid_nr(struct task_struct *tsk) 1364 { 1365 return tsk->tgid; 1366 } 1367 1368 /** 1369 * pid_alive - check that a task structure is not stale 1370 * @p: Task structure to be checked. 1371 * 1372 * Test if a process is not yet dead (at most zombie state) 1373 * If pid_alive fails, then pointers within the task structure 1374 * can be stale and must not be dereferenced. 1375 * 1376 * Return: 1 if the process is alive. 0 otherwise. 1377 */ 1378 static inline int pid_alive(const struct task_struct *p) 1379 { 1380 return p->thread_pid != NULL; 1381 } 1382 1383 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) 1384 { 1385 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); 1386 } 1387 1388 static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1389 { 1390 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); 1391 } 1392 1393 1394 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) 1395 { 1396 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); 1397 } 1398 1399 static inline pid_t task_session_vnr(struct task_struct *tsk) 1400 { 1401 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); 1402 } 1403 1404 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) 1405 { 1406 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns); 1407 } 1408 1409 static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1410 { 1411 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL); 1412 } 1413 1414 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns) 1415 { 1416 pid_t pid = 0; 1417 1418 rcu_read_lock(); 1419 if (pid_alive(tsk)) 1420 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns); 1421 rcu_read_unlock(); 1422 1423 return pid; 1424 } 1425 1426 static inline pid_t task_ppid_nr(const struct task_struct *tsk) 1427 { 1428 return task_ppid_nr_ns(tsk, &init_pid_ns); 1429 } 1430 1431 /* Obsolete, do not use: */ 1432 static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1433 { 1434 return task_pgrp_nr_ns(tsk, &init_pid_ns); 1435 } 1436 1437 #define TASK_REPORT_IDLE (TASK_REPORT + 1) 1438 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1) 1439 1440 static inline unsigned int task_state_index(struct task_struct *tsk) 1441 { 1442 unsigned int tsk_state = READ_ONCE(tsk->state); 1443 unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT; 1444 1445 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX); 1446 1447 if (tsk_state == TASK_IDLE) 1448 state = TASK_REPORT_IDLE; 1449 1450 return fls(state); 1451 } 1452 1453 static inline char task_index_to_char(unsigned int state) 1454 { 1455 static const char state_char[] = "RSDTtXZPI"; 1456 1457 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1); 1458 1459 return state_char[state]; 1460 } 1461 1462 static inline char task_state_to_char(struct task_struct *tsk) 1463 { 1464 return task_index_to_char(task_state_index(tsk)); 1465 } 1466 1467 /** 1468 * is_global_init - check if a task structure is init. Since init 1469 * is free to have sub-threads we need to check tgid. 1470 * @tsk: Task structure to be checked. 1471 * 1472 * Check if a task structure is the first user space task the kernel created. 1473 * 1474 * Return: 1 if the task structure is init. 0 otherwise. 1475 */ 1476 static inline int is_global_init(struct task_struct *tsk) 1477 { 1478 return task_tgid_nr(tsk) == 1; 1479 } 1480 1481 extern struct pid *cad_pid; 1482 1483 /* 1484 * Per process flags 1485 */ 1486 #define PF_IDLE 0x00000002 /* I am an IDLE thread */ 1487 #define PF_EXITING 0x00000004 /* Getting shut down */ 1488 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1489 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ 1490 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */ 1491 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */ 1492 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */ 1493 #define PF_DUMPCORE 0x00000200 /* Dumped core */ 1494 #define PF_SIGNALED 0x00000400 /* Killed by a signal */ 1495 #define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1496 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */ 1497 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */ 1498 #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */ 1499 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */ 1500 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */ 1501 #define PF_KSWAPD 0x00020000 /* I am kswapd */ 1502 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */ 1503 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */ 1504 #define PF_LOCAL_THROTTLE 0x00100000 /* Throttle writes only against the bdi I write to, 1505 * I am cleaning dirty pages from some other bdi. */ 1506 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1507 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */ 1508 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1509 #define PF_UMH 0x02000000 /* I'm an Usermodehelper process */ 1510 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_mask */ 1511 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ 1512 #define PF_MEMALLOC_NOCMA 0x10000000 /* All allocation request will have _GFP_MOVABLE cleared */ 1513 #define PF_IO_WORKER 0x20000000 /* Task is an IO worker */ 1514 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */ 1515 #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */ 1516 1517 /* 1518 * Only the _current_ task can read/write to tsk->flags, but other 1519 * tasks can access tsk->flags in readonly mode for example 1520 * with tsk_used_math (like during threaded core dumping). 1521 * There is however an exception to this rule during ptrace 1522 * or during fork: the ptracer task is allowed to write to the 1523 * child->flags of its traced child (same goes for fork, the parent 1524 * can write to the child->flags), because we're guaranteed the 1525 * child is not running and in turn not changing child->flags 1526 * at the same time the parent does it. 1527 */ 1528 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1529 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1530 #define clear_used_math() clear_stopped_child_used_math(current) 1531 #define set_used_math() set_stopped_child_used_math(current) 1532 1533 #define conditional_stopped_child_used_math(condition, child) \ 1534 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1535 1536 #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current) 1537 1538 #define copy_to_stopped_child_used_math(child) \ 1539 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1540 1541 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1542 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1543 #define used_math() tsk_used_math(current) 1544 1545 static inline bool is_percpu_thread(void) 1546 { 1547 #ifdef CONFIG_SMP 1548 return (current->flags & PF_NO_SETAFFINITY) && 1549 (current->nr_cpus_allowed == 1); 1550 #else 1551 return true; 1552 #endif 1553 } 1554 1555 /* Per-process atomic flags. */ 1556 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */ 1557 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */ 1558 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */ 1559 #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */ 1560 #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/ 1561 #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */ 1562 #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */ 1563 #define PFA_SPEC_SSB_NOEXEC 7 /* Speculative Store Bypass clear on execve() */ 1564 1565 #define TASK_PFA_TEST(name, func) \ 1566 static inline bool task_##func(struct task_struct *p) \ 1567 { return test_bit(PFA_##name, &p->atomic_flags); } 1568 1569 #define TASK_PFA_SET(name, func) \ 1570 static inline void task_set_##func(struct task_struct *p) \ 1571 { set_bit(PFA_##name, &p->atomic_flags); } 1572 1573 #define TASK_PFA_CLEAR(name, func) \ 1574 static inline void task_clear_##func(struct task_struct *p) \ 1575 { clear_bit(PFA_##name, &p->atomic_flags); } 1576 1577 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs) 1578 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs) 1579 1580 TASK_PFA_TEST(SPREAD_PAGE, spread_page) 1581 TASK_PFA_SET(SPREAD_PAGE, spread_page) 1582 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page) 1583 1584 TASK_PFA_TEST(SPREAD_SLAB, spread_slab) 1585 TASK_PFA_SET(SPREAD_SLAB, spread_slab) 1586 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) 1587 1588 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable) 1589 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable) 1590 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable) 1591 1592 TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec) 1593 TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec) 1594 TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec) 1595 1596 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) 1597 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) 1598 1599 TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable) 1600 TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable) 1601 TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable) 1602 1603 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable) 1604 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable) 1605 1606 static inline void 1607 current_restore_flags(unsigned long orig_flags, unsigned long flags) 1608 { 1609 current->flags &= ~flags; 1610 current->flags |= orig_flags & flags; 1611 } 1612 1613 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial); 1614 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed); 1615 #ifdef CONFIG_SMP 1616 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask); 1617 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask); 1618 #else 1619 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask) 1620 { 1621 } 1622 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask) 1623 { 1624 if (!cpumask_test_cpu(0, new_mask)) 1625 return -EINVAL; 1626 return 0; 1627 } 1628 #endif 1629 1630 extern int yield_to(struct task_struct *p, bool preempt); 1631 extern void set_user_nice(struct task_struct *p, long nice); 1632 extern int task_prio(const struct task_struct *p); 1633 1634 /** 1635 * task_nice - return the nice value of a given task. 1636 * @p: the task in question. 1637 * 1638 * Return: The nice value [ -20 ... 0 ... 19 ]. 1639 */ 1640 static inline int task_nice(const struct task_struct *p) 1641 { 1642 return PRIO_TO_NICE((p)->static_prio); 1643 } 1644 1645 extern int can_nice(const struct task_struct *p, const int nice); 1646 extern int task_curr(const struct task_struct *p); 1647 extern int idle_cpu(int cpu); 1648 extern int available_idle_cpu(int cpu); 1649 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *); 1650 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *); 1651 extern int sched_setattr(struct task_struct *, const struct sched_attr *); 1652 extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *); 1653 extern struct task_struct *idle_task(int cpu); 1654 1655 /** 1656 * is_idle_task - is the specified task an idle task? 1657 * @p: the task in question. 1658 * 1659 * Return: 1 if @p is an idle task. 0 otherwise. 1660 */ 1661 static inline bool is_idle_task(const struct task_struct *p) 1662 { 1663 return !!(p->flags & PF_IDLE); 1664 } 1665 1666 extern struct task_struct *curr_task(int cpu); 1667 extern void ia64_set_curr_task(int cpu, struct task_struct *p); 1668 1669 void yield(void); 1670 1671 union thread_union { 1672 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK 1673 struct task_struct task; 1674 #endif 1675 #ifndef CONFIG_THREAD_INFO_IN_TASK 1676 struct thread_info thread_info; 1677 #endif 1678 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1679 }; 1680 1681 #ifndef CONFIG_THREAD_INFO_IN_TASK 1682 extern struct thread_info init_thread_info; 1683 #endif 1684 1685 extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)]; 1686 1687 #ifdef CONFIG_THREAD_INFO_IN_TASK 1688 static inline struct thread_info *task_thread_info(struct task_struct *task) 1689 { 1690 return &task->thread_info; 1691 } 1692 #elif !defined(__HAVE_THREAD_FUNCTIONS) 1693 # define task_thread_info(task) ((struct thread_info *)(task)->stack) 1694 #endif 1695 1696 /* 1697 * find a task by one of its numerical ids 1698 * 1699 * find_task_by_pid_ns(): 1700 * finds a task by its pid in the specified namespace 1701 * find_task_by_vpid(): 1702 * finds a task by its virtual pid 1703 * 1704 * see also find_vpid() etc in include/linux/pid.h 1705 */ 1706 1707 extern struct task_struct *find_task_by_vpid(pid_t nr); 1708 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns); 1709 1710 /* 1711 * find a task by its virtual pid and get the task struct 1712 */ 1713 extern struct task_struct *find_get_task_by_vpid(pid_t nr); 1714 1715 extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1716 extern int wake_up_process(struct task_struct *tsk); 1717 extern void wake_up_new_task(struct task_struct *tsk); 1718 1719 #ifdef CONFIG_SMP 1720 extern void kick_process(struct task_struct *tsk); 1721 #else 1722 static inline void kick_process(struct task_struct *tsk) { } 1723 #endif 1724 1725 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec); 1726 1727 static inline void set_task_comm(struct task_struct *tsk, const char *from) 1728 { 1729 __set_task_comm(tsk, from, false); 1730 } 1731 1732 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk); 1733 #define get_task_comm(buf, tsk) ({ \ 1734 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \ 1735 __get_task_comm(buf, sizeof(buf), tsk); \ 1736 }) 1737 1738 #ifdef CONFIG_SMP 1739 static __always_inline void scheduler_ipi(void) 1740 { 1741 /* 1742 * Fold TIF_NEED_RESCHED into the preempt_count; anybody setting 1743 * TIF_NEED_RESCHED remotely (for the first time) will also send 1744 * this IPI. 1745 */ 1746 preempt_fold_need_resched(); 1747 } 1748 extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 1749 #else 1750 static inline void scheduler_ipi(void) { } 1751 static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state) 1752 { 1753 return 1; 1754 } 1755 #endif 1756 1757 /* 1758 * Set thread flags in other task's structures. 1759 * See asm/thread_info.h for TIF_xxxx flags available: 1760 */ 1761 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 1762 { 1763 set_ti_thread_flag(task_thread_info(tsk), flag); 1764 } 1765 1766 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1767 { 1768 clear_ti_thread_flag(task_thread_info(tsk), flag); 1769 } 1770 1771 static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag, 1772 bool value) 1773 { 1774 update_ti_thread_flag(task_thread_info(tsk), flag, value); 1775 } 1776 1777 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 1778 { 1779 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 1780 } 1781 1782 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1783 { 1784 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 1785 } 1786 1787 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 1788 { 1789 return test_ti_thread_flag(task_thread_info(tsk), flag); 1790 } 1791 1792 static inline void set_tsk_need_resched(struct task_struct *tsk) 1793 { 1794 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1795 } 1796 1797 static inline void clear_tsk_need_resched(struct task_struct *tsk) 1798 { 1799 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1800 } 1801 1802 static inline int test_tsk_need_resched(struct task_struct *tsk) 1803 { 1804 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 1805 } 1806 1807 /* 1808 * cond_resched() and cond_resched_lock(): latency reduction via 1809 * explicit rescheduling in places that are safe. The return 1810 * value indicates whether a reschedule was done in fact. 1811 * cond_resched_lock() will drop the spinlock before scheduling, 1812 */ 1813 #ifndef CONFIG_PREEMPTION 1814 extern int _cond_resched(void); 1815 #else 1816 static inline int _cond_resched(void) { return 0; } 1817 #endif 1818 1819 #define cond_resched() ({ \ 1820 ___might_sleep(__FILE__, __LINE__, 0); \ 1821 _cond_resched(); \ 1822 }) 1823 1824 extern int __cond_resched_lock(spinlock_t *lock); 1825 1826 #define cond_resched_lock(lock) ({ \ 1827 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\ 1828 __cond_resched_lock(lock); \ 1829 }) 1830 1831 static inline void cond_resched_rcu(void) 1832 { 1833 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU) 1834 rcu_read_unlock(); 1835 cond_resched(); 1836 rcu_read_lock(); 1837 #endif 1838 } 1839 1840 /* 1841 * Does a critical section need to be broken due to another 1842 * task waiting?: (technically does not depend on CONFIG_PREEMPTION, 1843 * but a general need for low latency) 1844 */ 1845 static inline int spin_needbreak(spinlock_t *lock) 1846 { 1847 #ifdef CONFIG_PREEMPTION 1848 return spin_is_contended(lock); 1849 #else 1850 return 0; 1851 #endif 1852 } 1853 1854 static __always_inline bool need_resched(void) 1855 { 1856 return unlikely(tif_need_resched()); 1857 } 1858 1859 /* 1860 * Wrappers for p->thread_info->cpu access. No-op on UP. 1861 */ 1862 #ifdef CONFIG_SMP 1863 1864 static inline unsigned int task_cpu(const struct task_struct *p) 1865 { 1866 #ifdef CONFIG_THREAD_INFO_IN_TASK 1867 return READ_ONCE(p->cpu); 1868 #else 1869 return READ_ONCE(task_thread_info(p)->cpu); 1870 #endif 1871 } 1872 1873 extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 1874 1875 #else 1876 1877 static inline unsigned int task_cpu(const struct task_struct *p) 1878 { 1879 return 0; 1880 } 1881 1882 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 1883 { 1884 } 1885 1886 #endif /* CONFIG_SMP */ 1887 1888 /* 1889 * In order to reduce various lock holder preemption latencies provide an 1890 * interface to see if a vCPU is currently running or not. 1891 * 1892 * This allows us to terminate optimistic spin loops and block, analogous to 1893 * the native optimistic spin heuristic of testing if the lock owner task is 1894 * running or not. 1895 */ 1896 #ifndef vcpu_is_preempted 1897 static inline bool vcpu_is_preempted(int cpu) 1898 { 1899 return false; 1900 } 1901 #endif 1902 1903 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); 1904 extern long sched_getaffinity(pid_t pid, struct cpumask *mask); 1905 1906 #ifndef TASK_SIZE_OF 1907 #define TASK_SIZE_OF(tsk) TASK_SIZE 1908 #endif 1909 1910 #ifdef CONFIG_RSEQ 1911 1912 /* 1913 * Map the event mask on the user-space ABI enum rseq_cs_flags 1914 * for direct mask checks. 1915 */ 1916 enum rseq_event_mask_bits { 1917 RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT, 1918 RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT, 1919 RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT, 1920 }; 1921 1922 enum rseq_event_mask { 1923 RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT), 1924 RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT), 1925 RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT), 1926 }; 1927 1928 static inline void rseq_set_notify_resume(struct task_struct *t) 1929 { 1930 if (t->rseq) 1931 set_tsk_thread_flag(t, TIF_NOTIFY_RESUME); 1932 } 1933 1934 void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs); 1935 1936 static inline void rseq_handle_notify_resume(struct ksignal *ksig, 1937 struct pt_regs *regs) 1938 { 1939 if (current->rseq) 1940 __rseq_handle_notify_resume(ksig, regs); 1941 } 1942 1943 static inline void rseq_signal_deliver(struct ksignal *ksig, 1944 struct pt_regs *regs) 1945 { 1946 preempt_disable(); 1947 __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask); 1948 preempt_enable(); 1949 rseq_handle_notify_resume(ksig, regs); 1950 } 1951 1952 /* rseq_preempt() requires preemption to be disabled. */ 1953 static inline void rseq_preempt(struct task_struct *t) 1954 { 1955 __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask); 1956 rseq_set_notify_resume(t); 1957 } 1958 1959 /* rseq_migrate() requires preemption to be disabled. */ 1960 static inline void rseq_migrate(struct task_struct *t) 1961 { 1962 __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask); 1963 rseq_set_notify_resume(t); 1964 } 1965 1966 /* 1967 * If parent process has a registered restartable sequences area, the 1968 * child inherits. Unregister rseq for a clone with CLONE_VM set. 1969 */ 1970 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) 1971 { 1972 if (clone_flags & CLONE_VM) { 1973 t->rseq = NULL; 1974 t->rseq_sig = 0; 1975 t->rseq_event_mask = 0; 1976 } else { 1977 t->rseq = current->rseq; 1978 t->rseq_sig = current->rseq_sig; 1979 t->rseq_event_mask = current->rseq_event_mask; 1980 } 1981 } 1982 1983 static inline void rseq_execve(struct task_struct *t) 1984 { 1985 t->rseq = NULL; 1986 t->rseq_sig = 0; 1987 t->rseq_event_mask = 0; 1988 } 1989 1990 #else 1991 1992 static inline void rseq_set_notify_resume(struct task_struct *t) 1993 { 1994 } 1995 static inline void rseq_handle_notify_resume(struct ksignal *ksig, 1996 struct pt_regs *regs) 1997 { 1998 } 1999 static inline void rseq_signal_deliver(struct ksignal *ksig, 2000 struct pt_regs *regs) 2001 { 2002 } 2003 static inline void rseq_preempt(struct task_struct *t) 2004 { 2005 } 2006 static inline void rseq_migrate(struct task_struct *t) 2007 { 2008 } 2009 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) 2010 { 2011 } 2012 static inline void rseq_execve(struct task_struct *t) 2013 { 2014 } 2015 2016 #endif 2017 2018 void __exit_umh(struct task_struct *tsk); 2019 2020 static inline void exit_umh(struct task_struct *tsk) 2021 { 2022 if (unlikely(tsk->flags & PF_UMH)) 2023 __exit_umh(tsk); 2024 } 2025 2026 #ifdef CONFIG_DEBUG_RSEQ 2027 2028 void rseq_syscall(struct pt_regs *regs); 2029 2030 #else 2031 2032 static inline void rseq_syscall(struct pt_regs *regs) 2033 { 2034 } 2035 2036 #endif 2037 2038 const struct sched_avg *sched_trace_cfs_rq_avg(struct cfs_rq *cfs_rq); 2039 char *sched_trace_cfs_rq_path(struct cfs_rq *cfs_rq, char *str, int len); 2040 int sched_trace_cfs_rq_cpu(struct cfs_rq *cfs_rq); 2041 2042 const struct sched_avg *sched_trace_rq_avg_rt(struct rq *rq); 2043 const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq); 2044 const struct sched_avg *sched_trace_rq_avg_irq(struct rq *rq); 2045 2046 int sched_trace_rq_cpu(struct rq *rq); 2047 2048 const struct cpumask *sched_trace_rd_span(struct root_domain *rd); 2049 2050 #endif 2051