Line data Source code
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/irqflags.h>
22 : #include <linux/seccomp.h>
23 : #include <linux/nodemask.h>
24 : #include <linux/rcupdate.h>
25 : #include <linux/refcount.h>
26 : #include <linux/resource.h>
27 : #include <linux/latencytop.h>
28 : #include <linux/sched/prio.h>
29 : #include <linux/sched/types.h>
30 : #include <linux/signal_types.h>
31 : #include <linux/syscall_user_dispatch.h>
32 : #include <linux/mm_types_task.h>
33 : #include <linux/task_io_accounting.h>
34 : #include <linux/posix-timers.h>
35 : #include <linux/rseq.h>
36 : #include <linux/seqlock.h>
37 : #include <linux/kcsan.h>
38 : #include <asm/kmap_size.h>
39 :
40 : /* task_struct member predeclarations (sorted alphabetically): */
41 : struct audit_context;
42 : struct backing_dev_info;
43 : struct bio_list;
44 : struct blk_plug;
45 : struct capture_control;
46 : struct cfs_rq;
47 : struct fs_struct;
48 : struct futex_pi_state;
49 : struct io_context;
50 : struct io_uring_task;
51 : struct mempolicy;
52 : struct nameidata;
53 : struct nsproxy;
54 : struct perf_event_context;
55 : struct pid_namespace;
56 : struct pipe_inode_info;
57 : struct rcu_node;
58 : struct reclaim_state;
59 : struct robust_list_head;
60 : struct root_domain;
61 : struct rq;
62 : struct sched_attr;
63 : struct sched_param;
64 : struct seq_file;
65 : struct sighand_struct;
66 : struct signal_struct;
67 : struct task_delay_info;
68 : struct task_group;
69 :
70 : /*
71 : * Task state bitmask. NOTE! These bits are also
72 : * encoded in fs/proc/array.c: get_task_state().
73 : *
74 : * We have two separate sets of flags: task->state
75 : * is about runnability, while task->exit_state are
76 : * about the task exiting. Confusing, but this way
77 : * modifying one set can't modify the other one by
78 : * mistake.
79 : */
80 :
81 : /* Used in tsk->state: */
82 : #define TASK_RUNNING 0x0000
83 : #define TASK_INTERRUPTIBLE 0x0001
84 : #define TASK_UNINTERRUPTIBLE 0x0002
85 : #define __TASK_STOPPED 0x0004
86 : #define __TASK_TRACED 0x0008
87 : /* Used in tsk->exit_state: */
88 : #define EXIT_DEAD 0x0010
89 : #define EXIT_ZOMBIE 0x0020
90 : #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
91 : /* Used in tsk->state again: */
92 : #define TASK_PARKED 0x0040
93 : #define TASK_DEAD 0x0080
94 : #define TASK_WAKEKILL 0x0100
95 : #define TASK_WAKING 0x0200
96 : #define TASK_NOLOAD 0x0400
97 : #define TASK_NEW 0x0800
98 : #define TASK_STATE_MAX 0x1000
99 :
100 : /* Convenience macros for the sake of set_current_state: */
101 : #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
102 : #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
103 : #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
104 :
105 : #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
106 :
107 : /* Convenience macros for the sake of wake_up(): */
108 : #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
109 :
110 : /* get_task_state(): */
111 : #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
112 : TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
113 : __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
114 : TASK_PARKED)
115 :
116 : #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
117 :
118 : #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
119 :
120 : #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
121 :
122 : #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
123 :
124 : /*
125 : * Special states are those that do not use the normal wait-loop pattern. See
126 : * the comment with set_special_state().
127 : */
128 : #define is_special_task_state(state) \
129 : ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD))
130 :
131 : #define __set_current_state(state_value) \
132 : do { \
133 : WARN_ON_ONCE(is_special_task_state(state_value));\
134 : current->task_state_change = _THIS_IP_; \
135 : current->state = (state_value); \
136 : } while (0)
137 :
138 : #define set_current_state(state_value) \
139 : do { \
140 : WARN_ON_ONCE(is_special_task_state(state_value));\
141 : current->task_state_change = _THIS_IP_; \
142 : smp_store_mb(current->state, (state_value)); \
143 : } while (0)
144 :
145 : #define set_special_state(state_value) \
146 : do { \
147 : unsigned long flags; /* may shadow */ \
148 : WARN_ON_ONCE(!is_special_task_state(state_value)); \
149 : raw_spin_lock_irqsave(¤t->pi_lock, flags); \
150 : current->task_state_change = _THIS_IP_; \
151 : current->state = (state_value); \
152 : raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
153 : } while (0)
154 : #else
155 : /*
156 : * set_current_state() includes a barrier so that the write of current->state
157 : * is correctly serialised wrt the caller's subsequent test of whether to
158 : * actually sleep:
159 : *
160 : * for (;;) {
161 : * set_current_state(TASK_UNINTERRUPTIBLE);
162 : * if (CONDITION)
163 : * break;
164 : *
165 : * schedule();
166 : * }
167 : * __set_current_state(TASK_RUNNING);
168 : *
169 : * If the caller does not need such serialisation (because, for instance, the
170 : * CONDITION test and condition change and wakeup are under the same lock) then
171 : * use __set_current_state().
172 : *
173 : * The above is typically ordered against the wakeup, which does:
174 : *
175 : * CONDITION = 1;
176 : * wake_up_state(p, TASK_UNINTERRUPTIBLE);
177 : *
178 : * where wake_up_state()/try_to_wake_up() executes a full memory barrier before
179 : * accessing p->state.
180 : *
181 : * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
182 : * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
183 : * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
184 : *
185 : * However, with slightly different timing the wakeup TASK_RUNNING store can
186 : * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not
187 : * a problem either because that will result in one extra go around the loop
188 : * and our @cond test will save the day.
189 : *
190 : * Also see the comments of try_to_wake_up().
191 : */
192 : #define __set_current_state(state_value) \
193 : current->state = (state_value)
194 :
195 : #define set_current_state(state_value) \
196 : smp_store_mb(current->state, (state_value))
197 :
198 : /*
199 : * set_special_state() should be used for those states when the blocking task
200 : * can not use the regular condition based wait-loop. In that case we must
201 : * serialize against wakeups such that any possible in-flight TASK_RUNNING stores
202 : * will not collide with our state change.
203 : */
204 : #define set_special_state(state_value) \
205 : do { \
206 : unsigned long flags; /* may shadow */ \
207 : raw_spin_lock_irqsave(¤t->pi_lock, flags); \
208 : current->state = (state_value); \
209 : raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
210 : } while (0)
211 :
212 : #endif
213 :
214 : /* Task command name length: */
215 : #define TASK_COMM_LEN 16
216 :
217 : extern void scheduler_tick(void);
218 :
219 : #define MAX_SCHEDULE_TIMEOUT LONG_MAX
220 :
221 : extern long schedule_timeout(long timeout);
222 : extern long schedule_timeout_interruptible(long timeout);
223 : extern long schedule_timeout_killable(long timeout);
224 : extern long schedule_timeout_uninterruptible(long timeout);
225 : extern long schedule_timeout_idle(long timeout);
226 : asmlinkage void schedule(void);
227 : extern void schedule_preempt_disabled(void);
228 : asmlinkage void preempt_schedule_irq(void);
229 :
230 : extern int __must_check io_schedule_prepare(void);
231 : extern void io_schedule_finish(int token);
232 : extern long io_schedule_timeout(long timeout);
233 : extern void io_schedule(void);
234 :
235 : /**
236 : * struct prev_cputime - snapshot of system and user cputime
237 : * @utime: time spent in user mode
238 : * @stime: time spent in system mode
239 : * @lock: protects the above two fields
240 : *
241 : * Stores previous user/system time values such that we can guarantee
242 : * monotonicity.
243 : */
244 : struct prev_cputime {
245 : #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
246 : u64 utime;
247 : u64 stime;
248 : raw_spinlock_t lock;
249 : #endif
250 : };
251 :
252 : enum vtime_state {
253 : /* Task is sleeping or running in a CPU with VTIME inactive: */
254 : VTIME_INACTIVE = 0,
255 : /* Task is idle */
256 : VTIME_IDLE,
257 : /* Task runs in kernelspace in a CPU with VTIME active: */
258 : VTIME_SYS,
259 : /* Task runs in userspace in a CPU with VTIME active: */
260 : VTIME_USER,
261 : /* Task runs as guests in a CPU with VTIME active: */
262 : VTIME_GUEST,
263 : };
264 :
265 : struct vtime {
266 : seqcount_t seqcount;
267 : unsigned long long starttime;
268 : enum vtime_state state;
269 : unsigned int cpu;
270 : u64 utime;
271 : u64 stime;
272 : u64 gtime;
273 : };
274 :
275 : /*
276 : * Utilization clamp constraints.
277 : * @UCLAMP_MIN: Minimum utilization
278 : * @UCLAMP_MAX: Maximum utilization
279 : * @UCLAMP_CNT: Utilization clamp constraints count
280 : */
281 : enum uclamp_id {
282 : UCLAMP_MIN = 0,
283 : UCLAMP_MAX,
284 : UCLAMP_CNT
285 : };
286 :
287 : #ifdef CONFIG_SMP
288 : extern struct root_domain def_root_domain;
289 : extern struct mutex sched_domains_mutex;
290 : #endif
291 :
292 : struct sched_info {
293 : #ifdef CONFIG_SCHED_INFO
294 : /* Cumulative counters: */
295 :
296 : /* # of times we have run on this CPU: */
297 : unsigned long pcount;
298 :
299 : /* Time spent waiting on a runqueue: */
300 : unsigned long long run_delay;
301 :
302 : /* Timestamps: */
303 :
304 : /* When did we last run on a CPU? */
305 : unsigned long long last_arrival;
306 :
307 : /* When were we last queued to run? */
308 : unsigned long long last_queued;
309 :
310 : #endif /* CONFIG_SCHED_INFO */
311 : };
312 :
313 : /*
314 : * Integer metrics need fixed point arithmetic, e.g., sched/fair
315 : * has a few: load, load_avg, util_avg, freq, and capacity.
316 : *
317 : * We define a basic fixed point arithmetic range, and then formalize
318 : * all these metrics based on that basic range.
319 : */
320 : # define SCHED_FIXEDPOINT_SHIFT 10
321 : # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
322 :
323 : /* Increase resolution of cpu_capacity calculations */
324 : # define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
325 : # define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
326 :
327 : struct load_weight {
328 : unsigned long weight;
329 : u32 inv_weight;
330 : };
331 :
332 : /**
333 : * struct util_est - Estimation utilization of FAIR tasks
334 : * @enqueued: instantaneous estimated utilization of a task/cpu
335 : * @ewma: the Exponential Weighted Moving Average (EWMA)
336 : * utilization of a task
337 : *
338 : * Support data structure to track an Exponential Weighted Moving Average
339 : * (EWMA) of a FAIR task's utilization. New samples are added to the moving
340 : * average each time a task completes an activation. Sample's weight is chosen
341 : * so that the EWMA will be relatively insensitive to transient changes to the
342 : * task's workload.
343 : *
344 : * The enqueued attribute has a slightly different meaning for tasks and cpus:
345 : * - task: the task's util_avg at last task dequeue time
346 : * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
347 : * Thus, the util_est.enqueued of a task represents the contribution on the
348 : * estimated utilization of the CPU where that task is currently enqueued.
349 : *
350 : * Only for tasks we track a moving average of the past instantaneous
351 : * estimated utilization. This allows to absorb sporadic drops in utilization
352 : * of an otherwise almost periodic task.
353 : */
354 : struct util_est {
355 : unsigned int enqueued;
356 : unsigned int ewma;
357 : #define UTIL_EST_WEIGHT_SHIFT 2
358 : } __attribute__((__aligned__(sizeof(u64))));
359 :
360 : /*
361 : * The load/runnable/util_avg accumulates an infinite geometric series
362 : * (see __update_load_avg_cfs_rq() in kernel/sched/pelt.c).
363 : *
364 : * [load_avg definition]
365 : *
366 : * load_avg = runnable% * scale_load_down(load)
367 : *
368 : * [runnable_avg definition]
369 : *
370 : * runnable_avg = runnable% * SCHED_CAPACITY_SCALE
371 : *
372 : * [util_avg definition]
373 : *
374 : * util_avg = running% * SCHED_CAPACITY_SCALE
375 : *
376 : * where runnable% is the time ratio that a sched_entity is runnable and
377 : * running% the time ratio that a sched_entity is running.
378 : *
379 : * For cfs_rq, they are the aggregated values of all runnable and blocked
380 : * sched_entities.
381 : *
382 : * The load/runnable/util_avg doesn't directly factor frequency scaling and CPU
383 : * capacity scaling. The scaling is done through the rq_clock_pelt that is used
384 : * for computing those signals (see update_rq_clock_pelt())
385 : *
386 : * N.B., the above ratios (runnable% and running%) themselves are in the
387 : * range of [0, 1]. To do fixed point arithmetics, we therefore scale them
388 : * to as large a range as necessary. This is for example reflected by
389 : * util_avg's SCHED_CAPACITY_SCALE.
390 : *
391 : * [Overflow issue]
392 : *
393 : * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
394 : * with the highest load (=88761), always runnable on a single cfs_rq,
395 : * and should not overflow as the number already hits PID_MAX_LIMIT.
396 : *
397 : * For all other cases (including 32-bit kernels), struct load_weight's
398 : * weight will overflow first before we do, because:
399 : *
400 : * Max(load_avg) <= Max(load.weight)
401 : *
402 : * Then it is the load_weight's responsibility to consider overflow
403 : * issues.
404 : */
405 : struct sched_avg {
406 : u64 last_update_time;
407 : u64 load_sum;
408 : u64 runnable_sum;
409 : u32 util_sum;
410 : u32 period_contrib;
411 : unsigned long load_avg;
412 : unsigned long runnable_avg;
413 : unsigned long util_avg;
414 : struct util_est util_est;
415 : } ____cacheline_aligned;
416 :
417 : struct sched_statistics {
418 : #ifdef CONFIG_SCHEDSTATS
419 : u64 wait_start;
420 : u64 wait_max;
421 : u64 wait_count;
422 : u64 wait_sum;
423 : u64 iowait_count;
424 : u64 iowait_sum;
425 :
426 : u64 sleep_start;
427 : u64 sleep_max;
428 : s64 sum_sleep_runtime;
429 :
430 : u64 block_start;
431 : u64 block_max;
432 : u64 exec_max;
433 : u64 slice_max;
434 :
435 : u64 nr_migrations_cold;
436 : u64 nr_failed_migrations_affine;
437 : u64 nr_failed_migrations_running;
438 : u64 nr_failed_migrations_hot;
439 : u64 nr_forced_migrations;
440 :
441 : u64 nr_wakeups;
442 : u64 nr_wakeups_sync;
443 : u64 nr_wakeups_migrate;
444 : u64 nr_wakeups_local;
445 : u64 nr_wakeups_remote;
446 : u64 nr_wakeups_affine;
447 : u64 nr_wakeups_affine_attempts;
448 : u64 nr_wakeups_passive;
449 : u64 nr_wakeups_idle;
450 : #endif
451 : };
452 :
453 : struct sched_entity {
454 : /* For load-balancing: */
455 : struct load_weight load;
456 : struct rb_node run_node;
457 : struct list_head group_node;
458 : unsigned int on_rq;
459 :
460 : u64 exec_start;
461 : u64 sum_exec_runtime;
462 : u64 vruntime;
463 : u64 prev_sum_exec_runtime;
464 :
465 : u64 nr_migrations;
466 :
467 : struct sched_statistics statistics;
468 :
469 : #ifdef CONFIG_FAIR_GROUP_SCHED
470 : int depth;
471 : struct sched_entity *parent;
472 : /* rq on which this entity is (to be) queued: */
473 : struct cfs_rq *cfs_rq;
474 : /* rq "owned" by this entity/group: */
475 : struct cfs_rq *my_q;
476 : /* cached value of my_q->h_nr_running */
477 : unsigned long runnable_weight;
478 : #endif
479 :
480 : #ifdef CONFIG_SMP
481 : /*
482 : * Per entity load average tracking.
483 : *
484 : * Put into separate cache line so it does not
485 : * collide with read-mostly values above.
486 : */
487 : struct sched_avg avg;
488 : #endif
489 : };
490 :
491 : struct sched_rt_entity {
492 : struct list_head run_list;
493 : unsigned long timeout;
494 : unsigned long watchdog_stamp;
495 : unsigned int time_slice;
496 : unsigned short on_rq;
497 : unsigned short on_list;
498 :
499 : struct sched_rt_entity *back;
500 : #ifdef CONFIG_RT_GROUP_SCHED
501 : struct sched_rt_entity *parent;
502 : /* rq on which this entity is (to be) queued: */
503 : struct rt_rq *rt_rq;
504 : /* rq "owned" by this entity/group: */
505 : struct rt_rq *my_q;
506 : #endif
507 : } __randomize_layout;
508 :
509 : struct sched_dl_entity {
510 : struct rb_node rb_node;
511 :
512 : /*
513 : * Original scheduling parameters. Copied here from sched_attr
514 : * during sched_setattr(), they will remain the same until
515 : * the next sched_setattr().
516 : */
517 : u64 dl_runtime; /* Maximum runtime for each instance */
518 : u64 dl_deadline; /* Relative deadline of each instance */
519 : u64 dl_period; /* Separation of two instances (period) */
520 : u64 dl_bw; /* dl_runtime / dl_period */
521 : u64 dl_density; /* dl_runtime / dl_deadline */
522 :
523 : /*
524 : * Actual scheduling parameters. Initialized with the values above,
525 : * they are continuously updated during task execution. Note that
526 : * the remaining runtime could be < 0 in case we are in overrun.
527 : */
528 : s64 runtime; /* Remaining runtime for this instance */
529 : u64 deadline; /* Absolute deadline for this instance */
530 : unsigned int flags; /* Specifying the scheduler behaviour */
531 :
532 : /*
533 : * Some bool flags:
534 : *
535 : * @dl_throttled tells if we exhausted the runtime. If so, the
536 : * task has to wait for a replenishment to be performed at the
537 : * next firing of dl_timer.
538 : *
539 : * @dl_boosted tells if we are boosted due to DI. If so we are
540 : * outside bandwidth enforcement mechanism (but only until we
541 : * exit the critical section);
542 : *
543 : * @dl_yielded tells if task gave up the CPU before consuming
544 : * all its available runtime during the last job.
545 : *
546 : * @dl_non_contending tells if the task is inactive while still
547 : * contributing to the active utilization. In other words, it
548 : * indicates if the inactive timer has been armed and its handler
549 : * has not been executed yet. This flag is useful to avoid race
550 : * conditions between the inactive timer handler and the wakeup
551 : * code.
552 : *
553 : * @dl_overrun tells if the task asked to be informed about runtime
554 : * overruns.
555 : */
556 : unsigned int dl_throttled : 1;
557 : unsigned int dl_yielded : 1;
558 : unsigned int dl_non_contending : 1;
559 : unsigned int dl_overrun : 1;
560 :
561 : /*
562 : * Bandwidth enforcement timer. Each -deadline task has its
563 : * own bandwidth to be enforced, thus we need one timer per task.
564 : */
565 : struct hrtimer dl_timer;
566 :
567 : /*
568 : * Inactive timer, responsible for decreasing the active utilization
569 : * at the "0-lag time". When a -deadline task blocks, it contributes
570 : * to GRUB's active utilization until the "0-lag time", hence a
571 : * timer is needed to decrease the active utilization at the correct
572 : * time.
573 : */
574 : struct hrtimer inactive_timer;
575 :
576 : #ifdef CONFIG_RT_MUTEXES
577 : /*
578 : * Priority Inheritance. When a DEADLINE scheduling entity is boosted
579 : * pi_se points to the donor, otherwise points to the dl_se it belongs
580 : * to (the original one/itself).
581 : */
582 : struct sched_dl_entity *pi_se;
583 : #endif
584 : };
585 :
586 : #ifdef CONFIG_UCLAMP_TASK
587 : /* Number of utilization clamp buckets (shorter alias) */
588 : #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT
589 :
590 : /*
591 : * Utilization clamp for a scheduling entity
592 : * @value: clamp value "assigned" to a se
593 : * @bucket_id: bucket index corresponding to the "assigned" value
594 : * @active: the se is currently refcounted in a rq's bucket
595 : * @user_defined: the requested clamp value comes from user-space
596 : *
597 : * The bucket_id is the index of the clamp bucket matching the clamp value
598 : * which is pre-computed and stored to avoid expensive integer divisions from
599 : * the fast path.
600 : *
601 : * The active bit is set whenever a task has got an "effective" value assigned,
602 : * which can be different from the clamp value "requested" from user-space.
603 : * This allows to know a task is refcounted in the rq's bucket corresponding
604 : * to the "effective" bucket_id.
605 : *
606 : * The user_defined bit is set whenever a task has got a task-specific clamp
607 : * value requested from userspace, i.e. the system defaults apply to this task
608 : * just as a restriction. This allows to relax default clamps when a less
609 : * restrictive task-specific value has been requested, thus allowing to
610 : * implement a "nice" semantic. For example, a task running with a 20%
611 : * default boost can still drop its own boosting to 0%.
612 : */
613 : struct uclamp_se {
614 : unsigned int value : bits_per(SCHED_CAPACITY_SCALE);
615 : unsigned int bucket_id : bits_per(UCLAMP_BUCKETS);
616 : unsigned int active : 1;
617 : unsigned int user_defined : 1;
618 : };
619 : #endif /* CONFIG_UCLAMP_TASK */
620 :
621 : union rcu_special {
622 : struct {
623 : u8 blocked;
624 : u8 need_qs;
625 : u8 exp_hint; /* Hint for performance. */
626 : u8 need_mb; /* Readers need smp_mb(). */
627 : } b; /* Bits. */
628 : u32 s; /* Set of bits. */
629 : };
630 :
631 : enum perf_event_task_context {
632 : perf_invalid_context = -1,
633 : perf_hw_context = 0,
634 : perf_sw_context,
635 : perf_nr_task_contexts,
636 : };
637 :
638 : struct wake_q_node {
639 : struct wake_q_node *next;
640 : };
641 :
642 : struct kmap_ctrl {
643 : #ifdef CONFIG_KMAP_LOCAL
644 : int idx;
645 : pte_t pteval[KM_MAX_IDX];
646 : #endif
647 : };
648 :
649 : struct task_struct {
650 : #ifdef CONFIG_THREAD_INFO_IN_TASK
651 : /*
652 : * For reasons of header soup (see current_thread_info()), this
653 : * must be the first element of task_struct.
654 : */
655 : struct thread_info thread_info;
656 : #endif
657 : /* -1 unrunnable, 0 runnable, >0 stopped: */
658 : volatile long state;
659 :
660 : /*
661 : * This begins the randomizable portion of task_struct. Only
662 : * scheduling-critical items should be added above here.
663 : */
664 : randomized_struct_fields_start
665 :
666 : void *stack;
667 : refcount_t usage;
668 : /* Per task flags (PF_*), defined further below: */
669 : unsigned int flags;
670 : unsigned int ptrace;
671 :
672 : #ifdef CONFIG_SMP
673 : int on_cpu;
674 : struct __call_single_node wake_entry;
675 : #ifdef CONFIG_THREAD_INFO_IN_TASK
676 : /* Current CPU: */
677 : unsigned int cpu;
678 : #endif
679 : unsigned int wakee_flips;
680 : unsigned long wakee_flip_decay_ts;
681 : struct task_struct *last_wakee;
682 :
683 : /*
684 : * recent_used_cpu is initially set as the last CPU used by a task
685 : * that wakes affine another task. Waker/wakee relationships can
686 : * push tasks around a CPU where each wakeup moves to the next one.
687 : * Tracking a recently used CPU allows a quick search for a recently
688 : * used CPU that may be idle.
689 : */
690 : int recent_used_cpu;
691 : int wake_cpu;
692 : #endif
693 : int on_rq;
694 :
695 : int prio;
696 : int static_prio;
697 : int normal_prio;
698 : unsigned int rt_priority;
699 :
700 : const struct sched_class *sched_class;
701 : struct sched_entity se;
702 : struct sched_rt_entity rt;
703 : #ifdef CONFIG_CGROUP_SCHED
704 : struct task_group *sched_task_group;
705 : #endif
706 : struct sched_dl_entity dl;
707 :
708 : #ifdef CONFIG_UCLAMP_TASK
709 : /*
710 : * Clamp values requested for a scheduling entity.
711 : * Must be updated with task_rq_lock() held.
712 : */
713 : struct uclamp_se uclamp_req[UCLAMP_CNT];
714 : /*
715 : * Effective clamp values used for a scheduling entity.
716 : * Must be updated with task_rq_lock() held.
717 : */
718 : struct uclamp_se uclamp[UCLAMP_CNT];
719 : #endif
720 :
721 : #ifdef CONFIG_PREEMPT_NOTIFIERS
722 : /* List of struct preempt_notifier: */
723 : struct hlist_head preempt_notifiers;
724 : #endif
725 :
726 : #ifdef CONFIG_BLK_DEV_IO_TRACE
727 : unsigned int btrace_seq;
728 : #endif
729 :
730 : unsigned int policy;
731 : int nr_cpus_allowed;
732 : const cpumask_t *cpus_ptr;
733 : cpumask_t cpus_mask;
734 : void *migration_pending;
735 : #ifdef CONFIG_SMP
736 : unsigned short migration_disabled;
737 : #endif
738 : unsigned short migration_flags;
739 :
740 : #ifdef CONFIG_PREEMPT_RCU
741 : int rcu_read_lock_nesting;
742 : union rcu_special rcu_read_unlock_special;
743 : struct list_head rcu_node_entry;
744 : struct rcu_node *rcu_blocked_node;
745 : #endif /* #ifdef CONFIG_PREEMPT_RCU */
746 :
747 : #ifdef CONFIG_TASKS_RCU
748 : unsigned long rcu_tasks_nvcsw;
749 : u8 rcu_tasks_holdout;
750 : u8 rcu_tasks_idx;
751 : int rcu_tasks_idle_cpu;
752 : struct list_head rcu_tasks_holdout_list;
753 : #endif /* #ifdef CONFIG_TASKS_RCU */
754 :
755 : #ifdef CONFIG_TASKS_TRACE_RCU
756 : int trc_reader_nesting;
757 : int trc_ipi_to_cpu;
758 : union rcu_special trc_reader_special;
759 : bool trc_reader_checked;
760 : struct list_head trc_holdout_list;
761 : #endif /* #ifdef CONFIG_TASKS_TRACE_RCU */
762 :
763 : struct sched_info sched_info;
764 :
765 : struct list_head tasks;
766 : #ifdef CONFIG_SMP
767 : struct plist_node pushable_tasks;
768 : struct rb_node pushable_dl_tasks;
769 : #endif
770 :
771 : struct mm_struct *mm;
772 : struct mm_struct *active_mm;
773 :
774 : /* Per-thread vma caching: */
775 : struct vmacache vmacache;
776 :
777 : #ifdef SPLIT_RSS_COUNTING
778 : struct task_rss_stat rss_stat;
779 : #endif
780 : int exit_state;
781 : int exit_code;
782 : int exit_signal;
783 : /* The signal sent when the parent dies: */
784 : int pdeath_signal;
785 : /* JOBCTL_*, siglock protected: */
786 : unsigned long jobctl;
787 :
788 : /* Used for emulating ABI behavior of previous Linux versions: */
789 : unsigned int personality;
790 :
791 : /* Scheduler bits, serialized by scheduler locks: */
792 : unsigned sched_reset_on_fork:1;
793 : unsigned sched_contributes_to_load:1;
794 : unsigned sched_migrated:1;
795 : #ifdef CONFIG_PSI
796 : unsigned sched_psi_wake_requeue:1;
797 : #endif
798 :
799 : /* Force alignment to the next boundary: */
800 : unsigned :0;
801 :
802 : /* Unserialized, strictly 'current' */
803 :
804 : /*
805 : * This field must not be in the scheduler word above due to wakelist
806 : * queueing no longer being serialized by p->on_cpu. However:
807 : *
808 : * p->XXX = X; ttwu()
809 : * schedule() if (p->on_rq && ..) // false
810 : * smp_mb__after_spinlock(); if (smp_load_acquire(&p->on_cpu) && //true
811 : * deactivate_task() ttwu_queue_wakelist())
812 : * p->on_rq = 0; p->sched_remote_wakeup = Y;
813 : *
814 : * guarantees all stores of 'current' are visible before
815 : * ->sched_remote_wakeup gets used, so it can be in this word.
816 : */
817 : unsigned sched_remote_wakeup:1;
818 :
819 : /* Bit to tell LSMs we're in execve(): */
820 : unsigned in_execve:1;
821 : unsigned in_iowait:1;
822 : #ifndef TIF_RESTORE_SIGMASK
823 : unsigned restore_sigmask:1;
824 : #endif
825 : #ifdef CONFIG_MEMCG
826 : unsigned in_user_fault:1;
827 : #endif
828 : #ifdef CONFIG_COMPAT_BRK
829 : unsigned brk_randomized:1;
830 : #endif
831 : #ifdef CONFIG_CGROUPS
832 : /* disallow userland-initiated cgroup migration */
833 : unsigned no_cgroup_migration:1;
834 : /* task is frozen/stopped (used by the cgroup freezer) */
835 : unsigned frozen:1;
836 : #endif
837 : #ifdef CONFIG_BLK_CGROUP
838 : unsigned use_memdelay:1;
839 : #endif
840 : #ifdef CONFIG_PSI
841 : /* Stalled due to lack of memory */
842 : unsigned in_memstall:1;
843 : #endif
844 :
845 : unsigned long atomic_flags; /* Flags requiring atomic access. */
846 :
847 : struct restart_block restart_block;
848 :
849 : pid_t pid;
850 : pid_t tgid;
851 :
852 : #ifdef CONFIG_STACKPROTECTOR
853 : /* Canary value for the -fstack-protector GCC feature: */
854 : unsigned long stack_canary;
855 : #endif
856 : /*
857 : * Pointers to the (original) parent process, youngest child, younger sibling,
858 : * older sibling, respectively. (p->father can be replaced with
859 : * p->real_parent->pid)
860 : */
861 :
862 : /* Real parent process: */
863 : struct task_struct __rcu *real_parent;
864 :
865 : /* Recipient of SIGCHLD, wait4() reports: */
866 : struct task_struct __rcu *parent;
867 :
868 : /*
869 : * Children/sibling form the list of natural children:
870 : */
871 : struct list_head children;
872 : struct list_head sibling;
873 : struct task_struct *group_leader;
874 :
875 : /*
876 : * 'ptraced' is the list of tasks this task is using ptrace() on.
877 : *
878 : * This includes both natural children and PTRACE_ATTACH targets.
879 : * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
880 : */
881 : struct list_head ptraced;
882 : struct list_head ptrace_entry;
883 :
884 : /* PID/PID hash table linkage. */
885 : struct pid *thread_pid;
886 : struct hlist_node pid_links[PIDTYPE_MAX];
887 : struct list_head thread_group;
888 : struct list_head thread_node;
889 :
890 : struct completion *vfork_done;
891 :
892 : /* CLONE_CHILD_SETTID: */
893 : int __user *set_child_tid;
894 :
895 : /* CLONE_CHILD_CLEARTID: */
896 : int __user *clear_child_tid;
897 :
898 : /* PF_IO_WORKER */
899 : void *pf_io_worker;
900 :
901 : u64 utime;
902 : u64 stime;
903 : #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
904 : u64 utimescaled;
905 : u64 stimescaled;
906 : #endif
907 : u64 gtime;
908 : struct prev_cputime prev_cputime;
909 : #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
910 : struct vtime vtime;
911 : #endif
912 :
913 : #ifdef CONFIG_NO_HZ_FULL
914 : atomic_t tick_dep_mask;
915 : #endif
916 : /* Context switch counts: */
917 : unsigned long nvcsw;
918 : unsigned long nivcsw;
919 :
920 : /* Monotonic time in nsecs: */
921 : u64 start_time;
922 :
923 : /* Boot based time in nsecs: */
924 : u64 start_boottime;
925 :
926 : /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
927 : unsigned long min_flt;
928 : unsigned long maj_flt;
929 :
930 : /* Empty if CONFIG_POSIX_CPUTIMERS=n */
931 : struct posix_cputimers posix_cputimers;
932 :
933 : #ifdef CONFIG_POSIX_CPU_TIMERS_TASK_WORK
934 : struct posix_cputimers_work posix_cputimers_work;
935 : #endif
936 :
937 : /* Process credentials: */
938 :
939 : /* Tracer's credentials at attach: */
940 : const struct cred __rcu *ptracer_cred;
941 :
942 : /* Objective and real subjective task credentials (COW): */
943 : const struct cred __rcu *real_cred;
944 :
945 : /* Effective (overridable) subjective task credentials (COW): */
946 : const struct cred __rcu *cred;
947 :
948 : #ifdef CONFIG_KEYS
949 : /* Cached requested key. */
950 : struct key *cached_requested_key;
951 : #endif
952 :
953 : /*
954 : * executable name, excluding path.
955 : *
956 : * - normally initialized setup_new_exec()
957 : * - access it with [gs]et_task_comm()
958 : * - lock it with task_lock()
959 : */
960 : char comm[TASK_COMM_LEN];
961 :
962 : struct nameidata *nameidata;
963 :
964 : #ifdef CONFIG_SYSVIPC
965 : struct sysv_sem sysvsem;
966 : struct sysv_shm sysvshm;
967 : #endif
968 : #ifdef CONFIG_DETECT_HUNG_TASK
969 : unsigned long last_switch_count;
970 : unsigned long last_switch_time;
971 : #endif
972 : /* Filesystem information: */
973 : struct fs_struct *fs;
974 :
975 : /* Open file information: */
976 : struct files_struct *files;
977 :
978 : #ifdef CONFIG_IO_URING
979 : struct io_uring_task *io_uring;
980 : #endif
981 :
982 : /* Namespaces: */
983 : struct nsproxy *nsproxy;
984 :
985 : /* Signal handlers: */
986 : struct signal_struct *signal;
987 : struct sighand_struct __rcu *sighand;
988 : sigset_t blocked;
989 : sigset_t real_blocked;
990 : /* Restored if set_restore_sigmask() was used: */
991 : sigset_t saved_sigmask;
992 : struct sigpending pending;
993 : unsigned long sas_ss_sp;
994 : size_t sas_ss_size;
995 : unsigned int sas_ss_flags;
996 :
997 : struct callback_head *task_works;
998 :
999 : #ifdef CONFIG_AUDIT
1000 : #ifdef CONFIG_AUDITSYSCALL
1001 : struct audit_context *audit_context;
1002 : #endif
1003 : kuid_t loginuid;
1004 : unsigned int sessionid;
1005 : #endif
1006 : struct seccomp seccomp;
1007 : struct syscall_user_dispatch syscall_dispatch;
1008 :
1009 : /* Thread group tracking: */
1010 : u64 parent_exec_id;
1011 : u64 self_exec_id;
1012 :
1013 : /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
1014 : spinlock_t alloc_lock;
1015 :
1016 : /* Protection of the PI data structures: */
1017 : raw_spinlock_t pi_lock;
1018 :
1019 : struct wake_q_node wake_q;
1020 :
1021 : #ifdef CONFIG_RT_MUTEXES
1022 : /* PI waiters blocked on a rt_mutex held by this task: */
1023 : struct rb_root_cached pi_waiters;
1024 : /* Updated under owner's pi_lock and rq lock */
1025 : struct task_struct *pi_top_task;
1026 : /* Deadlock detection and priority inheritance handling: */
1027 : struct rt_mutex_waiter *pi_blocked_on;
1028 : #endif
1029 :
1030 : #ifdef CONFIG_DEBUG_MUTEXES
1031 : /* Mutex deadlock detection: */
1032 : struct mutex_waiter *blocked_on;
1033 : #endif
1034 :
1035 : #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1036 : int non_block_count;
1037 : #endif
1038 :
1039 : #ifdef CONFIG_TRACE_IRQFLAGS
1040 : struct irqtrace_events irqtrace;
1041 : unsigned int hardirq_threaded;
1042 : u64 hardirq_chain_key;
1043 : int softirqs_enabled;
1044 : int softirq_context;
1045 : int irq_config;
1046 : #endif
1047 :
1048 : #ifdef CONFIG_LOCKDEP
1049 : # define MAX_LOCK_DEPTH 48UL
1050 : u64 curr_chain_key;
1051 : int lockdep_depth;
1052 : unsigned int lockdep_recursion;
1053 : struct held_lock held_locks[MAX_LOCK_DEPTH];
1054 : #endif
1055 :
1056 : #if defined(CONFIG_UBSAN) && !defined(CONFIG_UBSAN_TRAP)
1057 : unsigned int in_ubsan;
1058 : #endif
1059 :
1060 : /* Journalling filesystem info: */
1061 : void *journal_info;
1062 :
1063 : /* Stacked block device info: */
1064 : struct bio_list *bio_list;
1065 :
1066 : #ifdef CONFIG_BLOCK
1067 : /* Stack plugging: */
1068 : struct blk_plug *plug;
1069 : #endif
1070 :
1071 : /* VM state: */
1072 : struct reclaim_state *reclaim_state;
1073 :
1074 : struct backing_dev_info *backing_dev_info;
1075 :
1076 : struct io_context *io_context;
1077 :
1078 : #ifdef CONFIG_COMPACTION
1079 : struct capture_control *capture_control;
1080 : #endif
1081 : /* Ptrace state: */
1082 : unsigned long ptrace_message;
1083 : kernel_siginfo_t *last_siginfo;
1084 :
1085 : struct task_io_accounting ioac;
1086 : #ifdef CONFIG_PSI
1087 : /* Pressure stall state */
1088 : unsigned int psi_flags;
1089 : #endif
1090 : #ifdef CONFIG_TASK_XACCT
1091 : /* Accumulated RSS usage: */
1092 : u64 acct_rss_mem1;
1093 : /* Accumulated virtual memory usage: */
1094 : u64 acct_vm_mem1;
1095 : /* stime + utime since last update: */
1096 : u64 acct_timexpd;
1097 : #endif
1098 : #ifdef CONFIG_CPUSETS
1099 : /* Protected by ->alloc_lock: */
1100 : nodemask_t mems_allowed;
1101 : /* Seqence number to catch updates: */
1102 : seqcount_spinlock_t mems_allowed_seq;
1103 : int cpuset_mem_spread_rotor;
1104 : int cpuset_slab_spread_rotor;
1105 : #endif
1106 : #ifdef CONFIG_CGROUPS
1107 : /* Control Group info protected by css_set_lock: */
1108 : struct css_set __rcu *cgroups;
1109 : /* cg_list protected by css_set_lock and tsk->alloc_lock: */
1110 : struct list_head cg_list;
1111 : #endif
1112 : #ifdef CONFIG_X86_CPU_RESCTRL
1113 : u32 closid;
1114 : u32 rmid;
1115 : #endif
1116 : #ifdef CONFIG_FUTEX
1117 : struct robust_list_head __user *robust_list;
1118 : #ifdef CONFIG_COMPAT
1119 : struct compat_robust_list_head __user *compat_robust_list;
1120 : #endif
1121 : struct list_head pi_state_list;
1122 : struct futex_pi_state *pi_state_cache;
1123 : struct mutex futex_exit_mutex;
1124 : unsigned int futex_state;
1125 : #endif
1126 : #ifdef CONFIG_PERF_EVENTS
1127 : struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1128 : struct mutex perf_event_mutex;
1129 : struct list_head perf_event_list;
1130 : #endif
1131 : #ifdef CONFIG_DEBUG_PREEMPT
1132 : unsigned long preempt_disable_ip;
1133 : #endif
1134 : #ifdef CONFIG_NUMA
1135 : /* Protected by alloc_lock: */
1136 : struct mempolicy *mempolicy;
1137 : short il_prev;
1138 : short pref_node_fork;
1139 : #endif
1140 : #ifdef CONFIG_NUMA_BALANCING
1141 : int numa_scan_seq;
1142 : unsigned int numa_scan_period;
1143 : unsigned int numa_scan_period_max;
1144 : int numa_preferred_nid;
1145 : unsigned long numa_migrate_retry;
1146 : /* Migration stamp: */
1147 : u64 node_stamp;
1148 : u64 last_task_numa_placement;
1149 : u64 last_sum_exec_runtime;
1150 : struct callback_head numa_work;
1151 :
1152 : /*
1153 : * This pointer is only modified for current in syscall and
1154 : * pagefault context (and for tasks being destroyed), so it can be read
1155 : * from any of the following contexts:
1156 : * - RCU read-side critical section
1157 : * - current->numa_group from everywhere
1158 : * - task's runqueue locked, task not running
1159 : */
1160 : struct numa_group __rcu *numa_group;
1161 :
1162 : /*
1163 : * numa_faults is an array split into four regions:
1164 : * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1165 : * in this precise order.
1166 : *
1167 : * faults_memory: Exponential decaying average of faults on a per-node
1168 : * basis. Scheduling placement decisions are made based on these
1169 : * counts. The values remain static for the duration of a PTE scan.
1170 : * faults_cpu: Track the nodes the process was running on when a NUMA
1171 : * hinting fault was incurred.
1172 : * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1173 : * during the current scan window. When the scan completes, the counts
1174 : * in faults_memory and faults_cpu decay and these values are copied.
1175 : */
1176 : unsigned long *numa_faults;
1177 : unsigned long total_numa_faults;
1178 :
1179 : /*
1180 : * numa_faults_locality tracks if faults recorded during the last
1181 : * scan window were remote/local or failed to migrate. The task scan
1182 : * period is adapted based on the locality of the faults with different
1183 : * weights depending on whether they were shared or private faults
1184 : */
1185 : unsigned long numa_faults_locality[3];
1186 :
1187 : unsigned long numa_pages_migrated;
1188 : #endif /* CONFIG_NUMA_BALANCING */
1189 :
1190 : #ifdef CONFIG_RSEQ
1191 : struct rseq __user *rseq;
1192 : u32 rseq_sig;
1193 : /*
1194 : * RmW on rseq_event_mask must be performed atomically
1195 : * with respect to preemption.
1196 : */
1197 : unsigned long rseq_event_mask;
1198 : #endif
1199 :
1200 : struct tlbflush_unmap_batch tlb_ubc;
1201 :
1202 : union {
1203 : refcount_t rcu_users;
1204 : struct rcu_head rcu;
1205 : };
1206 :
1207 : /* Cache last used pipe for splice(): */
1208 : struct pipe_inode_info *splice_pipe;
1209 :
1210 : struct page_frag task_frag;
1211 :
1212 : #ifdef CONFIG_TASK_DELAY_ACCT
1213 : struct task_delay_info *delays;
1214 : #endif
1215 :
1216 : #ifdef CONFIG_FAULT_INJECTION
1217 : int make_it_fail;
1218 : unsigned int fail_nth;
1219 : #endif
1220 : /*
1221 : * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1222 : * balance_dirty_pages() for a dirty throttling pause:
1223 : */
1224 : int nr_dirtied;
1225 : int nr_dirtied_pause;
1226 : /* Start of a write-and-pause period: */
1227 : unsigned long dirty_paused_when;
1228 :
1229 : #ifdef CONFIG_LATENCYTOP
1230 : int latency_record_count;
1231 : struct latency_record latency_record[LT_SAVECOUNT];
1232 : #endif
1233 : /*
1234 : * Time slack values; these are used to round up poll() and
1235 : * select() etc timeout values. These are in nanoseconds.
1236 : */
1237 : u64 timer_slack_ns;
1238 : u64 default_timer_slack_ns;
1239 :
1240 : #if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
1241 : unsigned int kasan_depth;
1242 : #endif
1243 :
1244 : #ifdef CONFIG_KCSAN
1245 : struct kcsan_ctx kcsan_ctx;
1246 : #ifdef CONFIG_TRACE_IRQFLAGS
1247 : struct irqtrace_events kcsan_save_irqtrace;
1248 : #endif
1249 : #endif
1250 :
1251 : #if IS_ENABLED(CONFIG_KUNIT)
1252 : struct kunit *kunit_test;
1253 : #endif
1254 :
1255 : #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1256 : /* Index of current stored address in ret_stack: */
1257 : int curr_ret_stack;
1258 : int curr_ret_depth;
1259 :
1260 : /* Stack of return addresses for return function tracing: */
1261 : struct ftrace_ret_stack *ret_stack;
1262 :
1263 : /* Timestamp for last schedule: */
1264 : unsigned long long ftrace_timestamp;
1265 :
1266 : /*
1267 : * Number of functions that haven't been traced
1268 : * because of depth overrun:
1269 : */
1270 : atomic_t trace_overrun;
1271 :
1272 : /* Pause tracing: */
1273 : atomic_t tracing_graph_pause;
1274 : #endif
1275 :
1276 : #ifdef CONFIG_TRACING
1277 : /* State flags for use by tracers: */
1278 : unsigned long trace;
1279 :
1280 : /* Bitmask and counter of trace recursion: */
1281 : unsigned long trace_recursion;
1282 : #endif /* CONFIG_TRACING */
1283 :
1284 : #ifdef CONFIG_KCOV
1285 : /* See kernel/kcov.c for more details. */
1286 :
1287 : /* Coverage collection mode enabled for this task (0 if disabled): */
1288 : unsigned int kcov_mode;
1289 :
1290 : /* Size of the kcov_area: */
1291 : unsigned int kcov_size;
1292 :
1293 : /* Buffer for coverage collection: */
1294 : void *kcov_area;
1295 :
1296 : /* KCOV descriptor wired with this task or NULL: */
1297 : struct kcov *kcov;
1298 :
1299 : /* KCOV common handle for remote coverage collection: */
1300 : u64 kcov_handle;
1301 :
1302 : /* KCOV sequence number: */
1303 : int kcov_sequence;
1304 :
1305 : /* Collect coverage from softirq context: */
1306 : unsigned int kcov_softirq;
1307 : #endif
1308 :
1309 : #ifdef CONFIG_MEMCG
1310 : struct mem_cgroup *memcg_in_oom;
1311 : gfp_t memcg_oom_gfp_mask;
1312 : int memcg_oom_order;
1313 :
1314 : /* Number of pages to reclaim on returning to userland: */
1315 : unsigned int memcg_nr_pages_over_high;
1316 :
1317 : /* Used by memcontrol for targeted memcg charge: */
1318 : struct mem_cgroup *active_memcg;
1319 : #endif
1320 :
1321 : #ifdef CONFIG_BLK_CGROUP
1322 : struct request_queue *throttle_queue;
1323 : #endif
1324 :
1325 : #ifdef CONFIG_UPROBES
1326 : struct uprobe_task *utask;
1327 : #endif
1328 : #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1329 : unsigned int sequential_io;
1330 : unsigned int sequential_io_avg;
1331 : #endif
1332 : struct kmap_ctrl kmap_ctrl;
1333 : #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1334 : unsigned long task_state_change;
1335 : #endif
1336 : int pagefault_disabled;
1337 : #ifdef CONFIG_MMU
1338 : struct task_struct *oom_reaper_list;
1339 : #endif
1340 : #ifdef CONFIG_VMAP_STACK
1341 : struct vm_struct *stack_vm_area;
1342 : #endif
1343 : #ifdef CONFIG_THREAD_INFO_IN_TASK
1344 : /* A live task holds one reference: */
1345 : refcount_t stack_refcount;
1346 : #endif
1347 : #ifdef CONFIG_LIVEPATCH
1348 : int patch_state;
1349 : #endif
1350 : #ifdef CONFIG_SECURITY
1351 : /* Used by LSM modules for access restriction: */
1352 : void *security;
1353 : #endif
1354 :
1355 : #ifdef CONFIG_GCC_PLUGIN_STACKLEAK
1356 : unsigned long lowest_stack;
1357 : unsigned long prev_lowest_stack;
1358 : #endif
1359 :
1360 : #ifdef CONFIG_X86_MCE
1361 : void __user *mce_vaddr;
1362 : __u64 mce_kflags;
1363 : u64 mce_addr;
1364 : __u64 mce_ripv : 1,
1365 : mce_whole_page : 1,
1366 : __mce_reserved : 62;
1367 : struct callback_head mce_kill_me;
1368 : #endif
1369 :
1370 : #ifdef CONFIG_KRETPROBES
1371 : struct llist_head kretprobe_instances;
1372 : #endif
1373 :
1374 : /*
1375 : * New fields for task_struct should be added above here, so that
1376 : * they are included in the randomized portion of task_struct.
1377 : */
1378 : randomized_struct_fields_end
1379 :
1380 : /* CPU-specific state of this task: */
1381 : struct thread_struct thread;
1382 :
1383 : /*
1384 : * WARNING: on x86, 'thread_struct' contains a variable-sized
1385 : * structure. It *MUST* be at the end of 'task_struct'.
1386 : *
1387 : * Do not put anything below here!
1388 : */
1389 : };
1390 :
1391 10706 : static inline struct pid *task_pid(struct task_struct *task)
1392 : {
1393 10706 : return task->thread_pid;
1394 : }
1395 :
1396 : /*
1397 : * the helpers to get the task's different pids as they are seen
1398 : * from various namespaces
1399 : *
1400 : * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1401 : * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1402 : * current.
1403 : * task_xid_nr_ns() : id seen from the ns specified;
1404 : *
1405 : * see also pid_nr() etc in include/linux/pid.h
1406 : */
1407 : pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1408 :
1409 307 : static inline pid_t task_pid_nr(struct task_struct *tsk)
1410 : {
1411 307 : return tsk->pid;
1412 : }
1413 :
1414 1922 : static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1415 : {
1416 1922 : return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1417 : }
1418 :
1419 1352 : static inline pid_t task_pid_vnr(struct task_struct *tsk)
1420 : {
1421 1352 : return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1422 : }
1423 :
1424 :
1425 793 : static inline pid_t task_tgid_nr(struct task_struct *tsk)
1426 : {
1427 793 : return tsk->tgid;
1428 : }
1429 :
1430 : /**
1431 : * pid_alive - check that a task structure is not stale
1432 : * @p: Task structure to be checked.
1433 : *
1434 : * Test if a process is not yet dead (at most zombie state)
1435 : * If pid_alive fails, then pointers within the task structure
1436 : * can be stale and must not be dereferenced.
1437 : *
1438 : * Return: 1 if the process is alive. 0 otherwise.
1439 : */
1440 56 : static inline int pid_alive(const struct task_struct *p)
1441 : {
1442 56 : return p->thread_pid != NULL;
1443 : }
1444 :
1445 141 : static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1446 : {
1447 141 : return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1448 : }
1449 :
1450 : static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1451 : {
1452 : return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1453 : }
1454 :
1455 :
1456 141 : static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1457 : {
1458 141 : return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1459 : }
1460 :
1461 : static inline pid_t task_session_vnr(struct task_struct *tsk)
1462 : {
1463 : return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1464 : }
1465 :
1466 1064 : static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1467 : {
1468 1064 : return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns);
1469 : }
1470 :
1471 1604 : static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1472 : {
1473 1604 : return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL);
1474 : }
1475 :
1476 : static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1477 : {
1478 : pid_t pid = 0;
1479 :
1480 : rcu_read_lock();
1481 : if (pid_alive(tsk))
1482 : pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1483 : rcu_read_unlock();
1484 :
1485 : return pid;
1486 : }
1487 :
1488 : static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1489 : {
1490 : return task_ppid_nr_ns(tsk, &init_pid_ns);
1491 : }
1492 :
1493 : /* Obsolete, do not use: */
1494 : static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1495 : {
1496 : return task_pgrp_nr_ns(tsk, &init_pid_ns);
1497 : }
1498 :
1499 : #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1500 : #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1501 :
1502 197 : static inline unsigned int task_state_index(struct task_struct *tsk)
1503 : {
1504 197 : unsigned int tsk_state = READ_ONCE(tsk->state);
1505 197 : unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT;
1506 :
1507 197 : BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX);
1508 :
1509 197 : if (tsk_state == TASK_IDLE)
1510 25 : state = TASK_REPORT_IDLE;
1511 :
1512 197 : return fls(state);
1513 : }
1514 :
1515 0 : static inline char task_index_to_char(unsigned int state)
1516 : {
1517 0 : static const char state_char[] = "RSDTtXZPI";
1518 :
1519 0 : BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1);
1520 :
1521 0 : return state_char[state];
1522 : }
1523 :
1524 0 : static inline char task_state_to_char(struct task_struct *tsk)
1525 : {
1526 0 : return task_index_to_char(task_state_index(tsk));
1527 : }
1528 :
1529 : /**
1530 : * is_global_init - check if a task structure is init. Since init
1531 : * is free to have sub-threads we need to check tgid.
1532 : * @tsk: Task structure to be checked.
1533 : *
1534 : * Check if a task structure is the first user space task the kernel created.
1535 : *
1536 : * Return: 1 if the task structure is init. 0 otherwise.
1537 : */
1538 793 : static inline int is_global_init(struct task_struct *tsk)
1539 : {
1540 793 : return task_tgid_nr(tsk) == 1;
1541 : }
1542 :
1543 : extern struct pid *cad_pid;
1544 :
1545 : /*
1546 : * Per process flags
1547 : */
1548 : #define PF_VCPU 0x00000001 /* I'm a virtual CPU */
1549 : #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1550 : #define PF_EXITING 0x00000004 /* Getting shut down */
1551 : #define PF_IO_WORKER 0x00000010 /* Task is an IO worker */
1552 : #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1553 : #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1554 : #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1555 : #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1556 : #define PF_DUMPCORE 0x00000200 /* Dumped core */
1557 : #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1558 : #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1559 : #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1560 : #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1561 : #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1562 : #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1563 : #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1564 : #define PF_KSWAPD 0x00020000 /* I am kswapd */
1565 : #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1566 : #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1567 : #define PF_LOCAL_THROTTLE 0x00100000 /* Throttle writes only against the bdi I write to,
1568 : * I am cleaning dirty pages from some other bdi. */
1569 : #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1570 : #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1571 : #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1572 : #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_mask */
1573 : #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1574 : #define PF_MEMALLOC_NOCMA 0x10000000 /* All allocation request will have _GFP_MOVABLE cleared */
1575 : #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1576 : #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1577 :
1578 : /*
1579 : * Only the _current_ task can read/write to tsk->flags, but other
1580 : * tasks can access tsk->flags in readonly mode for example
1581 : * with tsk_used_math (like during threaded core dumping).
1582 : * There is however an exception to this rule during ptrace
1583 : * or during fork: the ptracer task is allowed to write to the
1584 : * child->flags of its traced child (same goes for fork, the parent
1585 : * can write to the child->flags), because we're guaranteed the
1586 : * child is not running and in turn not changing child->flags
1587 : * at the same time the parent does it.
1588 : */
1589 : #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1590 : #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1591 : #define clear_used_math() clear_stopped_child_used_math(current)
1592 : #define set_used_math() set_stopped_child_used_math(current)
1593 :
1594 : #define conditional_stopped_child_used_math(condition, child) \
1595 : do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1596 :
1597 : #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1598 :
1599 : #define copy_to_stopped_child_used_math(child) \
1600 : do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1601 :
1602 : /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1603 : #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1604 : #define used_math() tsk_used_math(current)
1605 :
1606 : static inline bool is_percpu_thread(void)
1607 : {
1608 : #ifdef CONFIG_SMP
1609 : return (current->flags & PF_NO_SETAFFINITY) &&
1610 : (current->nr_cpus_allowed == 1);
1611 : #else
1612 : return true;
1613 : #endif
1614 : }
1615 :
1616 : /* Per-process atomic flags. */
1617 : #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1618 : #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1619 : #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1620 : #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */
1621 : #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/
1622 : #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */
1623 : #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */
1624 : #define PFA_SPEC_SSB_NOEXEC 7 /* Speculative Store Bypass clear on execve() */
1625 :
1626 : #define TASK_PFA_TEST(name, func) \
1627 : static inline bool task_##func(struct task_struct *p) \
1628 : { return test_bit(PFA_##name, &p->atomic_flags); }
1629 :
1630 : #define TASK_PFA_SET(name, func) \
1631 : static inline void task_set_##func(struct task_struct *p) \
1632 : { set_bit(PFA_##name, &p->atomic_flags); }
1633 :
1634 : #define TASK_PFA_CLEAR(name, func) \
1635 : static inline void task_clear_##func(struct task_struct *p) \
1636 : { clear_bit(PFA_##name, &p->atomic_flags); }
1637 :
1638 4021 : TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1639 151 : TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1640 :
1641 : TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1642 : TASK_PFA_SET(SPREAD_PAGE, spread_page)
1643 : TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1644 :
1645 : TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1646 : TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1647 : TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1648 :
1649 56 : TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
1650 0 : TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
1651 0 : TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
1652 :
1653 56 : TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1654 0 : TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1655 0 : TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1656 :
1657 56 : TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1658 0 : TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1659 :
1660 0 : TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable)
1661 0 : TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable)
1662 0 : TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable)
1663 :
1664 0 : TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1665 0 : TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1666 :
1667 : static inline void
1668 : current_restore_flags(unsigned long orig_flags, unsigned long flags)
1669 : {
1670 : current->flags &= ~flags;
1671 : current->flags |= orig_flags & flags;
1672 : }
1673 :
1674 : extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1675 : extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1676 : #ifdef CONFIG_SMP
1677 : extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1678 : extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1679 : #else
1680 : static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1681 : {
1682 : }
1683 : static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1684 : {
1685 : if (!cpumask_test_cpu(0, new_mask))
1686 : return -EINVAL;
1687 : return 0;
1688 : }
1689 : #endif
1690 :
1691 : extern int yield_to(struct task_struct *p, bool preempt);
1692 : extern void set_user_nice(struct task_struct *p, long nice);
1693 : extern int task_prio(const struct task_struct *p);
1694 :
1695 : /**
1696 : * task_nice - return the nice value of a given task.
1697 : * @p: the task in question.
1698 : *
1699 : * Return: The nice value [ -20 ... 0 ... 19 ].
1700 : */
1701 1154 : static inline int task_nice(const struct task_struct *p)
1702 : {
1703 1154 : return PRIO_TO_NICE((p)->static_prio);
1704 : }
1705 :
1706 : extern int can_nice(const struct task_struct *p, const int nice);
1707 : extern int task_curr(const struct task_struct *p);
1708 : extern int idle_cpu(int cpu);
1709 : extern int available_idle_cpu(int cpu);
1710 : extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1711 : extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1712 : extern void sched_set_fifo(struct task_struct *p);
1713 : extern void sched_set_fifo_low(struct task_struct *p);
1714 : extern void sched_set_normal(struct task_struct *p, int nice);
1715 : extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1716 : extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *);
1717 : extern struct task_struct *idle_task(int cpu);
1718 :
1719 : /**
1720 : * is_idle_task - is the specified task an idle task?
1721 : * @p: the task in question.
1722 : *
1723 : * Return: 1 if @p is an idle task. 0 otherwise.
1724 : */
1725 3480518 : static __always_inline bool is_idle_task(const struct task_struct *p)
1726 : {
1727 3431754 : return !!(p->flags & PF_IDLE);
1728 : }
1729 :
1730 : extern struct task_struct *curr_task(int cpu);
1731 : extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1732 :
1733 : void yield(void);
1734 :
1735 : union thread_union {
1736 : #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1737 : struct task_struct task;
1738 : #endif
1739 : #ifndef CONFIG_THREAD_INFO_IN_TASK
1740 : struct thread_info thread_info;
1741 : #endif
1742 : unsigned long stack[THREAD_SIZE/sizeof(long)];
1743 : };
1744 :
1745 : #ifndef CONFIG_THREAD_INFO_IN_TASK
1746 : extern struct thread_info init_thread_info;
1747 : #endif
1748 :
1749 : extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
1750 :
1751 : #ifdef CONFIG_THREAD_INFO_IN_TASK
1752 453124 : static inline struct thread_info *task_thread_info(struct task_struct *task)
1753 : {
1754 195851 : return &task->thread_info;
1755 : }
1756 : #elif !defined(__HAVE_THREAD_FUNCTIONS)
1757 : # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1758 : #endif
1759 :
1760 : /*
1761 : * find a task by one of its numerical ids
1762 : *
1763 : * find_task_by_pid_ns():
1764 : * finds a task by its pid in the specified namespace
1765 : * find_task_by_vpid():
1766 : * finds a task by its virtual pid
1767 : *
1768 : * see also find_vpid() etc in include/linux/pid.h
1769 : */
1770 :
1771 : extern struct task_struct *find_task_by_vpid(pid_t nr);
1772 : extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1773 :
1774 : /*
1775 : * find a task by its virtual pid and get the task struct
1776 : */
1777 : extern struct task_struct *find_get_task_by_vpid(pid_t nr);
1778 :
1779 : extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1780 : extern int wake_up_process(struct task_struct *tsk);
1781 : extern void wake_up_new_task(struct task_struct *tsk);
1782 :
1783 : #ifdef CONFIG_SMP
1784 : extern void kick_process(struct task_struct *tsk);
1785 : #else
1786 : static inline void kick_process(struct task_struct *tsk) { }
1787 : #endif
1788 :
1789 : extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1790 :
1791 125 : static inline void set_task_comm(struct task_struct *tsk, const char *from)
1792 : {
1793 125 : __set_task_comm(tsk, from, false);
1794 76 : }
1795 :
1796 : extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
1797 : #define get_task_comm(buf, tsk) ({ \
1798 : BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1799 : __get_task_comm(buf, sizeof(buf), tsk); \
1800 : })
1801 :
1802 : #ifdef CONFIG_SMP
1803 2525 : static __always_inline void scheduler_ipi(void)
1804 : {
1805 : /*
1806 : * Fold TIF_NEED_RESCHED into the preempt_count; anybody setting
1807 : * TIF_NEED_RESCHED remotely (for the first time) will also send
1808 : * this IPI.
1809 : */
1810 2525 : preempt_fold_need_resched();
1811 : }
1812 : extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1813 : #else
1814 : static inline void scheduler_ipi(void) { }
1815 : static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1816 : {
1817 : return 1;
1818 : }
1819 : #endif
1820 :
1821 : /*
1822 : * Set thread flags in other task's structures.
1823 : * See asm/thread_info.h for TIF_xxxx flags available:
1824 : */
1825 16239 : static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1826 : {
1827 28435 : set_ti_thread_flag(task_thread_info(tsk), flag);
1828 0 : }
1829 :
1830 31523 : static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1831 : {
1832 60406 : clear_ti_thread_flag(task_thread_info(tsk), flag);
1833 0 : }
1834 :
1835 : static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag,
1836 : bool value)
1837 : {
1838 : update_ti_thread_flag(task_thread_info(tsk), flag, value);
1839 : }
1840 :
1841 56851 : static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1842 : {
1843 56851 : return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1844 : }
1845 :
1846 2646 : static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1847 : {
1848 2646 : return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1849 : }
1850 :
1851 277188 : static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1852 : {
1853 228833 : return test_ti_thread_flag(task_thread_info(tsk), flag);
1854 : }
1855 :
1856 12190 : static inline void set_tsk_need_resched(struct task_struct *tsk)
1857 : {
1858 12190 : set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1859 : }
1860 :
1861 28876 : static inline void clear_tsk_need_resched(struct task_struct *tsk)
1862 : {
1863 28876 : clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1864 : }
1865 :
1866 37201 : static inline int test_tsk_need_resched(struct task_struct *tsk)
1867 : {
1868 37201 : return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1869 : }
1870 :
1871 : /*
1872 : * cond_resched() and cond_resched_lock(): latency reduction via
1873 : * explicit rescheduling in places that are safe. The return
1874 : * value indicates whether a reschedule was done in fact.
1875 : * cond_resched_lock() will drop the spinlock before scheduling,
1876 : */
1877 : #if !defined(CONFIG_PREEMPTION) || defined(CONFIG_PREEMPT_DYNAMIC)
1878 : extern int __cond_resched(void);
1879 :
1880 : #ifdef CONFIG_PREEMPT_DYNAMIC
1881 :
1882 : DECLARE_STATIC_CALL(cond_resched, __cond_resched);
1883 :
1884 : static __always_inline int _cond_resched(void)
1885 : {
1886 : return static_call_mod(cond_resched)();
1887 : }
1888 :
1889 : #else
1890 :
1891 630990 : static inline int _cond_resched(void)
1892 : {
1893 630990 : return __cond_resched();
1894 : }
1895 :
1896 : #endif /* CONFIG_PREEMPT_DYNAMIC */
1897 :
1898 : #else
1899 :
1900 : static inline int _cond_resched(void) { return 0; }
1901 :
1902 : #endif /* !defined(CONFIG_PREEMPTION) || defined(CONFIG_PREEMPT_DYNAMIC) */
1903 :
1904 : #define cond_resched() ({ \
1905 : ___might_sleep(__FILE__, __LINE__, 0); \
1906 : _cond_resched(); \
1907 : })
1908 :
1909 : extern int __cond_resched_lock(spinlock_t *lock);
1910 : extern int __cond_resched_rwlock_read(rwlock_t *lock);
1911 : extern int __cond_resched_rwlock_write(rwlock_t *lock);
1912 :
1913 : #define cond_resched_lock(lock) ({ \
1914 : ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1915 : __cond_resched_lock(lock); \
1916 : })
1917 :
1918 : #define cond_resched_rwlock_read(lock) ({ \
1919 : __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
1920 : __cond_resched_rwlock_read(lock); \
1921 : })
1922 :
1923 : #define cond_resched_rwlock_write(lock) ({ \
1924 : __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
1925 : __cond_resched_rwlock_write(lock); \
1926 : })
1927 :
1928 0 : static inline void cond_resched_rcu(void)
1929 : {
1930 : #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1931 0 : rcu_read_unlock();
1932 0 : cond_resched();
1933 0 : rcu_read_lock();
1934 : #endif
1935 0 : }
1936 :
1937 : /*
1938 : * Does a critical section need to be broken due to another
1939 : * task waiting?: (technically does not depend on CONFIG_PREEMPTION,
1940 : * but a general need for low latency)
1941 : */
1942 1656 : static inline int spin_needbreak(spinlock_t *lock)
1943 : {
1944 : #ifdef CONFIG_PREEMPTION
1945 : return spin_is_contended(lock);
1946 : #else
1947 1656 : return 0;
1948 : #endif
1949 : }
1950 :
1951 : /*
1952 : * Check if a rwlock is contended.
1953 : * Returns non-zero if there is another task waiting on the rwlock.
1954 : * Returns zero if the lock is not contended or the system / underlying
1955 : * rwlock implementation does not support contention detection.
1956 : * Technically does not depend on CONFIG_PREEMPTION, but a general need
1957 : * for low latency.
1958 : */
1959 0 : static inline int rwlock_needbreak(rwlock_t *lock)
1960 : {
1961 : #ifdef CONFIG_PREEMPTION
1962 : return rwlock_is_contended(lock);
1963 : #else
1964 0 : return 0;
1965 : #endif
1966 : }
1967 :
1968 1573180 : static __always_inline bool need_resched(void)
1969 : {
1970 1573180 : return unlikely(tif_need_resched());
1971 : }
1972 :
1973 : /*
1974 : * Wrappers for p->thread_info->cpu access. No-op on UP.
1975 : */
1976 : #ifdef CONFIG_SMP
1977 :
1978 584061 : static inline unsigned int task_cpu(const struct task_struct *p)
1979 : {
1980 : #ifdef CONFIG_THREAD_INFO_IN_TASK
1981 573612 : return READ_ONCE(p->cpu);
1982 : #else
1983 : return READ_ONCE(task_thread_info(p)->cpu);
1984 : #endif
1985 : }
1986 :
1987 : extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1988 :
1989 : #else
1990 :
1991 : static inline unsigned int task_cpu(const struct task_struct *p)
1992 : {
1993 : return 0;
1994 : }
1995 :
1996 : static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1997 : {
1998 : }
1999 :
2000 : #endif /* CONFIG_SMP */
2001 :
2002 : /*
2003 : * In order to reduce various lock holder preemption latencies provide an
2004 : * interface to see if a vCPU is currently running or not.
2005 : *
2006 : * This allows us to terminate optimistic spin loops and block, analogous to
2007 : * the native optimistic spin heuristic of testing if the lock owner task is
2008 : * running or not.
2009 : */
2010 : #ifndef vcpu_is_preempted
2011 : static inline bool vcpu_is_preempted(int cpu)
2012 : {
2013 : return false;
2014 : }
2015 : #endif
2016 :
2017 : extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2018 : extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2019 :
2020 : #ifndef TASK_SIZE_OF
2021 : #define TASK_SIZE_OF(tsk) TASK_SIZE
2022 : #endif
2023 :
2024 : #ifdef CONFIG_SMP
2025 : /* Returns effective CPU energy utilization, as seen by the scheduler */
2026 : unsigned long sched_cpu_util(int cpu, unsigned long max);
2027 : #endif /* CONFIG_SMP */
2028 :
2029 : #ifdef CONFIG_RSEQ
2030 :
2031 : /*
2032 : * Map the event mask on the user-space ABI enum rseq_cs_flags
2033 : * for direct mask checks.
2034 : */
2035 : enum rseq_event_mask_bits {
2036 : RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT,
2037 : RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT,
2038 : RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT,
2039 : };
2040 :
2041 : enum rseq_event_mask {
2042 : RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT),
2043 : RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT),
2044 : RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT),
2045 : };
2046 :
2047 : static inline void rseq_set_notify_resume(struct task_struct *t)
2048 : {
2049 : if (t->rseq)
2050 : set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
2051 : }
2052 :
2053 : void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs);
2054 :
2055 : static inline void rseq_handle_notify_resume(struct ksignal *ksig,
2056 : struct pt_regs *regs)
2057 : {
2058 : if (current->rseq)
2059 : __rseq_handle_notify_resume(ksig, regs);
2060 : }
2061 :
2062 : static inline void rseq_signal_deliver(struct ksignal *ksig,
2063 : struct pt_regs *regs)
2064 : {
2065 : preempt_disable();
2066 : __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask);
2067 : preempt_enable();
2068 : rseq_handle_notify_resume(ksig, regs);
2069 : }
2070 :
2071 : /* rseq_preempt() requires preemption to be disabled. */
2072 : static inline void rseq_preempt(struct task_struct *t)
2073 : {
2074 : __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask);
2075 : rseq_set_notify_resume(t);
2076 : }
2077 :
2078 : /* rseq_migrate() requires preemption to be disabled. */
2079 : static inline void rseq_migrate(struct task_struct *t)
2080 : {
2081 : __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask);
2082 : rseq_set_notify_resume(t);
2083 : }
2084 :
2085 : /*
2086 : * If parent process has a registered restartable sequences area, the
2087 : * child inherits. Unregister rseq for a clone with CLONE_VM set.
2088 : */
2089 : static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
2090 : {
2091 : if (clone_flags & CLONE_VM) {
2092 : t->rseq = NULL;
2093 : t->rseq_sig = 0;
2094 : t->rseq_event_mask = 0;
2095 : } else {
2096 : t->rseq = current->rseq;
2097 : t->rseq_sig = current->rseq_sig;
2098 : t->rseq_event_mask = current->rseq_event_mask;
2099 : }
2100 : }
2101 :
2102 : static inline void rseq_execve(struct task_struct *t)
2103 : {
2104 : t->rseq = NULL;
2105 : t->rseq_sig = 0;
2106 : t->rseq_event_mask = 0;
2107 : }
2108 :
2109 : #else
2110 :
2111 : static inline void rseq_set_notify_resume(struct task_struct *t)
2112 : {
2113 : }
2114 51835 : static inline void rseq_handle_notify_resume(struct ksignal *ksig,
2115 : struct pt_regs *regs)
2116 : {
2117 51835 : }
2118 1265 : static inline void rseq_signal_deliver(struct ksignal *ksig,
2119 : struct pt_regs *regs)
2120 : {
2121 1265 : }
2122 26512 : static inline void rseq_preempt(struct task_struct *t)
2123 : {
2124 26512 : }
2125 2952 : static inline void rseq_migrate(struct task_struct *t)
2126 : {
2127 2952 : }
2128 916 : static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
2129 : {
2130 916 : }
2131 1021 : static inline void rseq_execve(struct task_struct *t)
2132 : {
2133 1021 : }
2134 :
2135 : #endif
2136 :
2137 : #ifdef CONFIG_DEBUG_RSEQ
2138 :
2139 : void rseq_syscall(struct pt_regs *regs);
2140 :
2141 : #else
2142 :
2143 296002 : static inline void rseq_syscall(struct pt_regs *regs)
2144 : {
2145 296002 : }
2146 :
2147 : #endif
2148 :
2149 : const struct sched_avg *sched_trace_cfs_rq_avg(struct cfs_rq *cfs_rq);
2150 : char *sched_trace_cfs_rq_path(struct cfs_rq *cfs_rq, char *str, int len);
2151 : int sched_trace_cfs_rq_cpu(struct cfs_rq *cfs_rq);
2152 :
2153 : const struct sched_avg *sched_trace_rq_avg_rt(struct rq *rq);
2154 : const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq);
2155 : const struct sched_avg *sched_trace_rq_avg_irq(struct rq *rq);
2156 :
2157 : int sched_trace_rq_cpu(struct rq *rq);
2158 : int sched_trace_rq_cpu_capacity(struct rq *rq);
2159 : int sched_trace_rq_nr_running(struct rq *rq);
2160 :
2161 : const struct cpumask *sched_trace_rd_span(struct root_domain *rd);
2162 :
2163 : #endif
|
