1 /* SPDX-License-Identifier: GPL-2.0 */
6 * Define 'struct task_struct' and provide the main scheduler
7 * APIs (schedule(), wakeup variants, etc.)
10 #include <uapi/linux/sched.h>
12 #include <asm/current.h>
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/signal_types.h>
29 #include <linux/mm_types_task.h>
30 #include <linux/task_io_accounting.h>
31 #include <linux/rseq.h>
33 /* task_struct member predeclarations (sorted alphabetically): */
35 struct backing_dev_info
;
38 struct capture_control
;
41 struct futex_pi_state
;
46 struct perf_event_context
;
48 struct pipe_inode_info
;
51 struct robust_list_head
;
57 struct sighand_struct
;
59 struct task_delay_info
;
63 * Task state bitmask. NOTE! These bits are also
64 * encoded in fs/proc/array.c: get_task_state().
66 * We have two separate sets of flags: task->state
67 * is about runnability, while task->exit_state are
68 * about the task exiting. Confusing, but this way
69 * modifying one set can't modify the other one by
73 /* Used in tsk->state: */
74 #define TASK_RUNNING 0x0000
75 #define TASK_INTERRUPTIBLE 0x0001
76 #define TASK_UNINTERRUPTIBLE 0x0002
77 #define __TASK_STOPPED 0x0004
78 #define __TASK_TRACED 0x0008
79 /* Used in tsk->exit_state: */
80 #define EXIT_DEAD 0x0010
81 #define EXIT_ZOMBIE 0x0020
82 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
83 /* Used in tsk->state again: */
84 #define TASK_PARKED 0x0040
85 #define TASK_DEAD 0x0080
86 #define TASK_WAKEKILL 0x0100
87 #define TASK_WAKING 0x0200
88 #define TASK_NOLOAD 0x0400
89 #define TASK_NEW 0x0800
90 #define TASK_STATE_MAX 0x1000
92 /* Convenience macros for the sake of set_current_state: */
93 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
94 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
95 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
97 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
99 /* Convenience macros for the sake of wake_up(): */
100 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
102 /* get_task_state(): */
103 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
104 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
105 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
108 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
110 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
112 #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
114 #define task_contributes_to_load(task) ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
115 (task->flags & PF_FROZEN) == 0 && \
116 (task->state & TASK_NOLOAD) == 0)
118 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
121 * Special states are those that do not use the normal wait-loop pattern. See
122 * the comment with set_special_state().
124 #define is_special_task_state(state) \
125 ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD))
127 #define __set_current_state(state_value) \
129 WARN_ON_ONCE(is_special_task_state(state_value));\
130 current->task_state_change = _THIS_IP_; \
131 current->state = (state_value); \
134 #define set_current_state(state_value) \
136 WARN_ON_ONCE(is_special_task_state(state_value));\
137 current->task_state_change = _THIS_IP_; \
138 smp_store_mb(current->state, (state_value)); \
141 #define set_special_state(state_value) \
143 unsigned long flags; /* may shadow */ \
144 WARN_ON_ONCE(!is_special_task_state(state_value)); \
145 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
146 current->task_state_change = _THIS_IP_; \
147 current->state = (state_value); \
148 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
152 * set_current_state() includes a barrier so that the write of current->state
153 * is correctly serialised wrt the caller's subsequent test of whether to
157 * set_current_state(TASK_UNINTERRUPTIBLE);
163 * __set_current_state(TASK_RUNNING);
165 * If the caller does not need such serialisation (because, for instance, the
166 * condition test and condition change and wakeup are under the same lock) then
167 * use __set_current_state().
169 * The above is typically ordered against the wakeup, which does:
171 * need_sleep = false;
172 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
174 * where wake_up_state() executes a full memory barrier before accessing the
177 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
178 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
179 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
181 * However, with slightly different timing the wakeup TASK_RUNNING store can
182 * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not
183 * a problem either because that will result in one extra go around the loop
184 * and our @cond test will save the day.
186 * Also see the comments of try_to_wake_up().
188 #define __set_current_state(state_value) \
189 current->state = (state_value)
191 #define set_current_state(state_value) \
192 smp_store_mb(current->state, (state_value))
195 * set_special_state() should be used for those states when the blocking task
196 * can not use the regular condition based wait-loop. In that case we must
197 * serialize against wakeups such that any possible in-flight TASK_RUNNING stores
198 * will not collide with our state change.
200 #define set_special_state(state_value) \
202 unsigned long flags; /* may shadow */ \
203 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
204 current->state = (state_value); \
205 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
210 /* Task command name length: */
211 #define TASK_COMM_LEN 16
213 extern void scheduler_tick(void);
215 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
217 extern long schedule_timeout(long timeout
);
218 extern long schedule_timeout_interruptible(long timeout
);
219 extern long schedule_timeout_killable(long timeout
);
220 extern long schedule_timeout_uninterruptible(long timeout
);
221 extern long schedule_timeout_idle(long timeout
);
222 asmlinkage
void schedule(void);
223 extern void schedule_preempt_disabled(void);
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);
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
236 * Stores previous user/system time values such that we can guarantee
239 struct prev_cputime
{
240 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
248 * struct task_cputime - collected CPU time counts
249 * @utime: time spent in user mode, in nanoseconds
250 * @stime: time spent in kernel mode, in nanoseconds
251 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
253 * This structure groups together three kinds of CPU time that are tracked for
254 * threads and thread groups. Most things considering CPU time want to group
255 * these counts together and treat all three of them in parallel.
257 struct task_cputime
{
260 unsigned long long sum_exec_runtime
;
263 /* Alternate field names when used on cache expirations: */
264 #define virt_exp utime
265 #define prof_exp stime
266 #define sched_exp sum_exec_runtime
269 /* Task is sleeping or running in a CPU with VTIME inactive: */
271 /* Task runs in userspace in a CPU with VTIME active: */
273 /* Task runs in kernelspace in a CPU with VTIME active: */
279 unsigned long long starttime
;
280 enum vtime_state state
;
287 * Utilization clamp constraints.
288 * @UCLAMP_MIN: Minimum utilization
289 * @UCLAMP_MAX: Maximum utilization
290 * @UCLAMP_CNT: Utilization clamp constraints count
299 #ifdef CONFIG_SCHED_INFO
300 /* Cumulative counters: */
302 /* # of times we have run on this CPU: */
303 unsigned long pcount
;
305 /* Time spent waiting on a runqueue: */
306 unsigned long long run_delay
;
310 /* When did we last run on a CPU? */
311 unsigned long long last_arrival
;
313 /* When were we last queued to run? */
314 unsigned long long last_queued
;
316 #endif /* CONFIG_SCHED_INFO */
320 * Integer metrics need fixed point arithmetic, e.g., sched/fair
321 * has a few: load, load_avg, util_avg, freq, and capacity.
323 * We define a basic fixed point arithmetic range, and then formalize
324 * all these metrics based on that basic range.
326 # define SCHED_FIXEDPOINT_SHIFT 10
327 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
329 /* Increase resolution of cpu_capacity calculations */
330 # define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
331 # define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
334 unsigned long weight
;
339 * struct util_est - Estimation utilization of FAIR tasks
340 * @enqueued: instantaneous estimated utilization of a task/cpu
341 * @ewma: the Exponential Weighted Moving Average (EWMA)
342 * utilization of a task
344 * Support data structure to track an Exponential Weighted Moving Average
345 * (EWMA) of a FAIR task's utilization. New samples are added to the moving
346 * average each time a task completes an activation. Sample's weight is chosen
347 * so that the EWMA will be relatively insensitive to transient changes to the
350 * The enqueued attribute has a slightly different meaning for tasks and cpus:
351 * - task: the task's util_avg at last task dequeue time
352 * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
353 * Thus, the util_est.enqueued of a task represents the contribution on the
354 * estimated utilization of the CPU where that task is currently enqueued.
356 * Only for tasks we track a moving average of the past instantaneous
357 * estimated utilization. This allows to absorb sporadic drops in utilization
358 * of an otherwise almost periodic task.
361 unsigned int enqueued
;
363 #define UTIL_EST_WEIGHT_SHIFT 2
364 } __attribute__((__aligned__(sizeof(u64
))));
367 * The load_avg/util_avg accumulates an infinite geometric series
368 * (see __update_load_avg() in kernel/sched/fair.c).
370 * [load_avg definition]
372 * load_avg = runnable% * scale_load_down(load)
374 * where runnable% is the time ratio that a sched_entity is runnable.
375 * For cfs_rq, it is the aggregated load_avg of all runnable and
376 * blocked sched_entities.
378 * [util_avg definition]
380 * util_avg = running% * SCHED_CAPACITY_SCALE
382 * where running% is the time ratio that a sched_entity is running on
383 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
384 * and blocked sched_entities.
386 * load_avg and util_avg don't direcly factor frequency scaling and CPU
387 * capacity scaling. The scaling is done through the rq_clock_pelt that
388 * is used for computing those signals (see update_rq_clock_pelt())
390 * N.B., the above ratios (runnable% and running%) themselves are in the
391 * range of [0, 1]. To do fixed point arithmetics, we therefore scale them
392 * to as large a range as necessary. This is for example reflected by
393 * util_avg's SCHED_CAPACITY_SCALE.
397 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
398 * with the highest load (=88761), always runnable on a single cfs_rq,
399 * and should not overflow as the number already hits PID_MAX_LIMIT.
401 * For all other cases (including 32-bit kernels), struct load_weight's
402 * weight will overflow first before we do, because:
404 * Max(load_avg) <= Max(load.weight)
406 * Then it is the load_weight's responsibility to consider overflow
410 u64 last_update_time
;
412 u64 runnable_load_sum
;
415 unsigned long load_avg
;
416 unsigned long runnable_load_avg
;
417 unsigned long util_avg
;
418 struct util_est util_est
;
419 } ____cacheline_aligned
;
421 struct sched_statistics
{
422 #ifdef CONFIG_SCHEDSTATS
432 s64 sum_sleep_runtime
;
439 u64 nr_migrations_cold
;
440 u64 nr_failed_migrations_affine
;
441 u64 nr_failed_migrations_running
;
442 u64 nr_failed_migrations_hot
;
443 u64 nr_forced_migrations
;
447 u64 nr_wakeups_migrate
;
448 u64 nr_wakeups_local
;
449 u64 nr_wakeups_remote
;
450 u64 nr_wakeups_affine
;
451 u64 nr_wakeups_affine_attempts
;
452 u64 nr_wakeups_passive
;
457 struct sched_entity
{
458 /* For load-balancing: */
459 struct load_weight load
;
460 unsigned long runnable_weight
;
461 struct rb_node run_node
;
462 struct list_head group_node
;
466 u64 sum_exec_runtime
;
468 u64 prev_sum_exec_runtime
;
472 struct sched_statistics statistics
;
474 #ifdef CONFIG_FAIR_GROUP_SCHED
476 struct sched_entity
*parent
;
477 /* rq on which this entity is (to be) queued: */
478 struct cfs_rq
*cfs_rq
;
479 /* rq "owned" by this entity/group: */
485 * Per entity load average tracking.
487 * Put into separate cache line so it does not
488 * collide with read-mostly values above.
490 struct sched_avg avg
;
494 struct sched_rt_entity
{
495 struct list_head run_list
;
496 unsigned long timeout
;
497 unsigned long watchdog_stamp
;
498 unsigned int time_slice
;
499 unsigned short on_rq
;
500 unsigned short on_list
;
502 struct sched_rt_entity
*back
;
503 #ifdef CONFIG_RT_GROUP_SCHED
504 struct sched_rt_entity
*parent
;
505 /* rq on which this entity is (to be) queued: */
507 /* rq "owned" by this entity/group: */
510 } __randomize_layout
;
512 struct sched_dl_entity
{
513 struct rb_node rb_node
;
516 * Original scheduling parameters. Copied here from sched_attr
517 * during sched_setattr(), they will remain the same until
518 * the next sched_setattr().
520 u64 dl_runtime
; /* Maximum runtime for each instance */
521 u64 dl_deadline
; /* Relative deadline of each instance */
522 u64 dl_period
; /* Separation of two instances (period) */
523 u64 dl_bw
; /* dl_runtime / dl_period */
524 u64 dl_density
; /* dl_runtime / dl_deadline */
527 * Actual scheduling parameters. Initialized with the values above,
528 * they are continuously updated during task execution. Note that
529 * the remaining runtime could be < 0 in case we are in overrun.
531 s64 runtime
; /* Remaining runtime for this instance */
532 u64 deadline
; /* Absolute deadline for this instance */
533 unsigned int flags
; /* Specifying the scheduler behaviour */
538 * @dl_throttled tells if we exhausted the runtime. If so, the
539 * task has to wait for a replenishment to be performed at the
540 * next firing of dl_timer.
542 * @dl_boosted tells if we are boosted due to DI. If so we are
543 * outside bandwidth enforcement mechanism (but only until we
544 * exit the critical section);
546 * @dl_yielded tells if task gave up the CPU before consuming
547 * all its available runtime during the last job.
549 * @dl_non_contending tells if the task is inactive while still
550 * contributing to the active utilization. In other words, it
551 * indicates if the inactive timer has been armed and its handler
552 * has not been executed yet. This flag is useful to avoid race
553 * conditions between the inactive timer handler and the wakeup
556 * @dl_overrun tells if the task asked to be informed about runtime
559 unsigned int dl_throttled
: 1;
560 unsigned int dl_boosted
: 1;
561 unsigned int dl_yielded
: 1;
562 unsigned int dl_non_contending
: 1;
563 unsigned int dl_overrun
: 1;
566 * Bandwidth enforcement timer. Each -deadline task has its
567 * own bandwidth to be enforced, thus we need one timer per task.
569 struct hrtimer dl_timer
;
572 * Inactive timer, responsible for decreasing the active utilization
573 * at the "0-lag time". When a -deadline task blocks, it contributes
574 * to GRUB's active utilization until the "0-lag time", hence a
575 * timer is needed to decrease the active utilization at the correct
578 struct hrtimer inactive_timer
;
581 #ifdef CONFIG_UCLAMP_TASK
582 /* Number of utilization clamp buckets (shorter alias) */
583 #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT
586 * Utilization clamp for a scheduling entity
587 * @value: clamp value "assigned" to a se
588 * @bucket_id: bucket index corresponding to the "assigned" value
589 * @active: the se is currently refcounted in a rq's bucket
590 * @user_defined: the requested clamp value comes from user-space
592 * The bucket_id is the index of the clamp bucket matching the clamp value
593 * which is pre-computed and stored to avoid expensive integer divisions from
596 * The active bit is set whenever a task has got an "effective" value assigned,
597 * which can be different from the clamp value "requested" from user-space.
598 * This allows to know a task is refcounted in the rq's bucket corresponding
599 * to the "effective" bucket_id.
601 * The user_defined bit is set whenever a task has got a task-specific clamp
602 * value requested from userspace, i.e. the system defaults apply to this task
603 * just as a restriction. This allows to relax default clamps when a less
604 * restrictive task-specific value has been requested, thus allowing to
605 * implement a "nice" semantic. For example, a task running with a 20%
606 * default boost can still drop its own boosting to 0%.
609 unsigned int value
: bits_per(SCHED_CAPACITY_SCALE
);
610 unsigned int bucket_id
: bits_per(UCLAMP_BUCKETS
);
611 unsigned int active
: 1;
612 unsigned int user_defined
: 1;
614 #endif /* CONFIG_UCLAMP_TASK */
620 u8 exp_hint
; /* Hint for performance. */
623 u32 s
; /* Set of bits. */
626 enum perf_event_task_context
{
627 perf_invalid_context
= -1,
630 perf_nr_task_contexts
,
634 struct wake_q_node
*next
;
638 #ifdef CONFIG_THREAD_INFO_IN_TASK
640 * For reasons of header soup (see current_thread_info()), this
641 * must be the first element of task_struct.
643 struct thread_info thread_info
;
645 /* -1 unrunnable, 0 runnable, >0 stopped: */
649 * This begins the randomizable portion of task_struct. Only
650 * scheduling-critical items should be added above here.
652 randomized_struct_fields_start
656 /* Per task flags (PF_*), defined further below: */
661 struct llist_node wake_entry
;
663 #ifdef CONFIG_THREAD_INFO_IN_TASK
667 unsigned int wakee_flips
;
668 unsigned long wakee_flip_decay_ts
;
669 struct task_struct
*last_wakee
;
672 * recent_used_cpu is initially set as the last CPU used by a task
673 * that wakes affine another task. Waker/wakee relationships can
674 * push tasks around a CPU where each wakeup moves to the next one.
675 * Tracking a recently used CPU allows a quick search for a recently
676 * used CPU that may be idle.
686 unsigned int rt_priority
;
688 const struct sched_class
*sched_class
;
689 struct sched_entity se
;
690 struct sched_rt_entity rt
;
691 #ifdef CONFIG_CGROUP_SCHED
692 struct task_group
*sched_task_group
;
694 struct sched_dl_entity dl
;
696 #ifdef CONFIG_UCLAMP_TASK
697 /* Clamp values requested for a scheduling entity */
698 struct uclamp_se uclamp_req
[UCLAMP_CNT
];
699 /* Effective clamp values used for a scheduling entity */
700 struct uclamp_se uclamp
[UCLAMP_CNT
];
703 #ifdef CONFIG_PREEMPT_NOTIFIERS
704 /* List of struct preempt_notifier: */
705 struct hlist_head preempt_notifiers
;
708 #ifdef CONFIG_BLK_DEV_IO_TRACE
709 unsigned int btrace_seq
;
714 const cpumask_t
*cpus_ptr
;
717 #ifdef CONFIG_PREEMPT_RCU
718 int rcu_read_lock_nesting
;
719 union rcu_special rcu_read_unlock_special
;
720 struct list_head rcu_node_entry
;
721 struct rcu_node
*rcu_blocked_node
;
722 #endif /* #ifdef CONFIG_PREEMPT_RCU */
724 #ifdef CONFIG_TASKS_RCU
725 unsigned long rcu_tasks_nvcsw
;
726 u8 rcu_tasks_holdout
;
728 int rcu_tasks_idle_cpu
;
729 struct list_head rcu_tasks_holdout_list
;
730 #endif /* #ifdef CONFIG_TASKS_RCU */
732 struct sched_info sched_info
;
734 struct list_head tasks
;
736 struct plist_node pushable_tasks
;
737 struct rb_node pushable_dl_tasks
;
740 struct mm_struct
*mm
;
741 struct mm_struct
*active_mm
;
743 /* Per-thread vma caching: */
744 struct vmacache vmacache
;
746 #ifdef SPLIT_RSS_COUNTING
747 struct task_rss_stat rss_stat
;
752 /* The signal sent when the parent dies: */
754 /* JOBCTL_*, siglock protected: */
755 unsigned long jobctl
;
757 /* Used for emulating ABI behavior of previous Linux versions: */
758 unsigned int personality
;
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;
766 unsigned sched_psi_wake_requeue
:1;
769 /* Force alignment to the next boundary: */
772 /* Unserialized, strictly 'current' */
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;
781 unsigned in_user_fault
:1;
783 #ifdef CONFIG_COMPAT_BRK
784 unsigned brk_randomized
:1;
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) */
792 #ifdef CONFIG_BLK_CGROUP
793 /* to be used once the psi infrastructure lands upstream. */
794 unsigned use_memdelay
:1;
797 unsigned long atomic_flags
; /* Flags requiring atomic access. */
799 struct restart_block restart_block
;
804 #ifdef CONFIG_STACKPROTECTOR
805 /* Canary value for the -fstack-protector GCC feature: */
806 unsigned long stack_canary
;
809 * Pointers to the (original) parent process, youngest child, younger sibling,
810 * older sibling, respectively. (p->father can be replaced with
811 * p->real_parent->pid)
814 /* Real parent process: */
815 struct task_struct __rcu
*real_parent
;
817 /* Recipient of SIGCHLD, wait4() reports: */
818 struct task_struct __rcu
*parent
;
821 * Children/sibling form the list of natural children:
823 struct list_head children
;
824 struct list_head sibling
;
825 struct task_struct
*group_leader
;
828 * 'ptraced' is the list of tasks this task is using ptrace() on.
830 * This includes both natural children and PTRACE_ATTACH targets.
831 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
833 struct list_head ptraced
;
834 struct list_head ptrace_entry
;
836 /* PID/PID hash table linkage. */
837 struct pid
*thread_pid
;
838 struct hlist_node pid_links
[PIDTYPE_MAX
];
839 struct list_head thread_group
;
840 struct list_head thread_node
;
842 struct completion
*vfork_done
;
844 /* CLONE_CHILD_SETTID: */
845 int __user
*set_child_tid
;
847 /* CLONE_CHILD_CLEARTID: */
848 int __user
*clear_child_tid
;
852 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
857 struct prev_cputime prev_cputime
;
858 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
862 #ifdef CONFIG_NO_HZ_FULL
863 atomic_t tick_dep_mask
;
865 /* Context switch counts: */
867 unsigned long nivcsw
;
869 /* Monotonic time in nsecs: */
872 /* Boot based time in nsecs: */
875 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
876 unsigned long min_flt
;
877 unsigned long maj_flt
;
879 #ifdef CONFIG_POSIX_TIMERS
880 struct task_cputime cputime_expires
;
881 struct list_head cpu_timers
[3];
884 /* Process credentials: */
886 /* Tracer's credentials at attach: */
887 const struct cred __rcu
*ptracer_cred
;
889 /* Objective and real subjective task credentials (COW): */
890 const struct cred __rcu
*real_cred
;
892 /* Effective (overridable) subjective task credentials (COW): */
893 const struct cred __rcu
*cred
;
896 /* Cached requested key. */
897 struct key
*cached_requested_key
;
901 * executable name, excluding path.
903 * - normally initialized setup_new_exec()
904 * - access it with [gs]et_task_comm()
905 * - lock it with task_lock()
907 char comm
[TASK_COMM_LEN
];
909 struct nameidata
*nameidata
;
911 #ifdef CONFIG_SYSVIPC
912 struct sysv_sem sysvsem
;
913 struct sysv_shm sysvshm
;
915 #ifdef CONFIG_DETECT_HUNG_TASK
916 unsigned long last_switch_count
;
917 unsigned long last_switch_time
;
919 /* Filesystem information: */
920 struct fs_struct
*fs
;
922 /* Open file information: */
923 struct files_struct
*files
;
926 struct nsproxy
*nsproxy
;
928 /* Signal handlers: */
929 struct signal_struct
*signal
;
930 struct sighand_struct
*sighand
;
932 sigset_t real_blocked
;
933 /* Restored if set_restore_sigmask() was used: */
934 sigset_t saved_sigmask
;
935 struct sigpending pending
;
936 unsigned long sas_ss_sp
;
938 unsigned int sas_ss_flags
;
940 struct callback_head
*task_works
;
943 #ifdef CONFIG_AUDITSYSCALL
944 struct audit_context
*audit_context
;
947 unsigned int sessionid
;
949 struct seccomp seccomp
;
951 /* Thread group tracking: */
955 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
956 spinlock_t alloc_lock
;
958 /* Protection of the PI data structures: */
959 raw_spinlock_t pi_lock
;
961 struct wake_q_node wake_q
;
963 #ifdef CONFIG_RT_MUTEXES
964 /* PI waiters blocked on a rt_mutex held by this task: */
965 struct rb_root_cached pi_waiters
;
966 /* Updated under owner's pi_lock and rq lock */
967 struct task_struct
*pi_top_task
;
968 /* Deadlock detection and priority inheritance handling: */
969 struct rt_mutex_waiter
*pi_blocked_on
;
972 #ifdef CONFIG_DEBUG_MUTEXES
973 /* Mutex deadlock detection: */
974 struct mutex_waiter
*blocked_on
;
977 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
981 #ifdef CONFIG_TRACE_IRQFLAGS
982 unsigned int irq_events
;
983 unsigned long hardirq_enable_ip
;
984 unsigned long hardirq_disable_ip
;
985 unsigned int hardirq_enable_event
;
986 unsigned int hardirq_disable_event
;
987 int hardirqs_enabled
;
989 unsigned long softirq_disable_ip
;
990 unsigned long softirq_enable_ip
;
991 unsigned int softirq_disable_event
;
992 unsigned int softirq_enable_event
;
993 int softirqs_enabled
;
997 #ifdef CONFIG_LOCKDEP
998 # define MAX_LOCK_DEPTH 48UL
1001 unsigned int lockdep_recursion
;
1002 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1006 unsigned int in_ubsan
;
1009 /* Journalling filesystem info: */
1012 /* Stacked block device info: */
1013 struct bio_list
*bio_list
;
1016 /* Stack plugging: */
1017 struct blk_plug
*plug
;
1021 struct reclaim_state
*reclaim_state
;
1023 struct backing_dev_info
*backing_dev_info
;
1025 struct io_context
*io_context
;
1027 #ifdef CONFIG_COMPACTION
1028 struct capture_control
*capture_control
;
1031 unsigned long ptrace_message
;
1032 kernel_siginfo_t
*last_siginfo
;
1034 struct task_io_accounting ioac
;
1036 /* Pressure stall state */
1037 unsigned int psi_flags
;
1039 #ifdef CONFIG_TASK_XACCT
1040 /* Accumulated RSS usage: */
1042 /* Accumulated virtual memory usage: */
1044 /* stime + utime since last update: */
1047 #ifdef CONFIG_CPUSETS
1048 /* Protected by ->alloc_lock: */
1049 nodemask_t mems_allowed
;
1050 /* Seqence number to catch updates: */
1051 seqcount_t mems_allowed_seq
;
1052 int cpuset_mem_spread_rotor
;
1053 int cpuset_slab_spread_rotor
;
1055 #ifdef CONFIG_CGROUPS
1056 /* Control Group info protected by css_set_lock: */
1057 struct css_set __rcu
*cgroups
;
1058 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
1059 struct list_head cg_list
;
1061 #ifdef CONFIG_X86_CPU_RESCTRL
1066 struct robust_list_head __user
*robust_list
;
1067 #ifdef CONFIG_COMPAT
1068 struct compat_robust_list_head __user
*compat_robust_list
;
1070 struct list_head pi_state_list
;
1071 struct futex_pi_state
*pi_state_cache
;
1073 #ifdef CONFIG_PERF_EVENTS
1074 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1075 struct mutex perf_event_mutex
;
1076 struct list_head perf_event_list
;
1078 #ifdef CONFIG_DEBUG_PREEMPT
1079 unsigned long preempt_disable_ip
;
1082 /* Protected by alloc_lock: */
1083 struct mempolicy
*mempolicy
;
1085 short pref_node_fork
;
1087 #ifdef CONFIG_NUMA_BALANCING
1089 unsigned int numa_scan_period
;
1090 unsigned int numa_scan_period_max
;
1091 int numa_preferred_nid
;
1092 unsigned long numa_migrate_retry
;
1093 /* Migration stamp: */
1095 u64 last_task_numa_placement
;
1096 u64 last_sum_exec_runtime
;
1097 struct callback_head numa_work
;
1100 * This pointer is only modified for current in syscall and
1101 * pagefault context (and for tasks being destroyed), so it can be read
1102 * from any of the following contexts:
1103 * - RCU read-side critical section
1104 * - current->numa_group from everywhere
1105 * - task's runqueue locked, task not running
1107 struct numa_group __rcu
*numa_group
;
1110 * numa_faults is an array split into four regions:
1111 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1112 * in this precise order.
1114 * faults_memory: Exponential decaying average of faults on a per-node
1115 * basis. Scheduling placement decisions are made based on these
1116 * counts. The values remain static for the duration of a PTE scan.
1117 * faults_cpu: Track the nodes the process was running on when a NUMA
1118 * hinting fault was incurred.
1119 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1120 * during the current scan window. When the scan completes, the counts
1121 * in faults_memory and faults_cpu decay and these values are copied.
1123 unsigned long *numa_faults
;
1124 unsigned long total_numa_faults
;
1127 * numa_faults_locality tracks if faults recorded during the last
1128 * scan window were remote/local or failed to migrate. The task scan
1129 * period is adapted based on the locality of the faults with different
1130 * weights depending on whether they were shared or private faults
1132 unsigned long numa_faults_locality
[3];
1134 unsigned long numa_pages_migrated
;
1135 #endif /* CONFIG_NUMA_BALANCING */
1138 struct rseq __user
*rseq
;
1141 * RmW on rseq_event_mask must be performed atomically
1142 * with respect to preemption.
1144 unsigned long rseq_event_mask
;
1147 struct tlbflush_unmap_batch tlb_ubc
;
1149 struct rcu_head rcu
;
1151 /* Cache last used pipe for splice(): */
1152 struct pipe_inode_info
*splice_pipe
;
1154 struct page_frag task_frag
;
1156 #ifdef CONFIG_TASK_DELAY_ACCT
1157 struct task_delay_info
*delays
;
1160 #ifdef CONFIG_FAULT_INJECTION
1162 unsigned int fail_nth
;
1165 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1166 * balance_dirty_pages() for a dirty throttling pause:
1169 int nr_dirtied_pause
;
1170 /* Start of a write-and-pause period: */
1171 unsigned long dirty_paused_when
;
1173 #ifdef CONFIG_LATENCYTOP
1174 int latency_record_count
;
1175 struct latency_record latency_record
[LT_SAVECOUNT
];
1178 * Time slack values; these are used to round up poll() and
1179 * select() etc timeout values. These are in nanoseconds.
1182 u64 default_timer_slack_ns
;
1185 unsigned int kasan_depth
;
1188 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1189 /* Index of current stored address in ret_stack: */
1193 /* Stack of return addresses for return function tracing: */
1194 struct ftrace_ret_stack
*ret_stack
;
1196 /* Timestamp for last schedule: */
1197 unsigned long long ftrace_timestamp
;
1200 * Number of functions that haven't been traced
1201 * because of depth overrun:
1203 atomic_t trace_overrun
;
1205 /* Pause tracing: */
1206 atomic_t tracing_graph_pause
;
1209 #ifdef CONFIG_TRACING
1210 /* State flags for use by tracers: */
1211 unsigned long trace
;
1213 /* Bitmask and counter of trace recursion: */
1214 unsigned long trace_recursion
;
1215 #endif /* CONFIG_TRACING */
1218 /* Coverage collection mode enabled for this task (0 if disabled): */
1219 unsigned int kcov_mode
;
1221 /* Size of the kcov_area: */
1222 unsigned int kcov_size
;
1224 /* Buffer for coverage collection: */
1227 /* KCOV descriptor wired with this task or NULL: */
1232 struct mem_cgroup
*memcg_in_oom
;
1233 gfp_t memcg_oom_gfp_mask
;
1234 int memcg_oom_order
;
1236 /* Number of pages to reclaim on returning to userland: */
1237 unsigned int memcg_nr_pages_over_high
;
1239 /* Used by memcontrol for targeted memcg charge: */
1240 struct mem_cgroup
*active_memcg
;
1243 #ifdef CONFIG_BLK_CGROUP
1244 struct request_queue
*throttle_queue
;
1247 #ifdef CONFIG_UPROBES
1248 struct uprobe_task
*utask
;
1250 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1251 unsigned int sequential_io
;
1252 unsigned int sequential_io_avg
;
1254 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1255 unsigned long task_state_change
;
1257 int pagefault_disabled
;
1259 struct task_struct
*oom_reaper_list
;
1261 #ifdef CONFIG_VMAP_STACK
1262 struct vm_struct
*stack_vm_area
;
1264 #ifdef CONFIG_THREAD_INFO_IN_TASK
1265 /* A live task holds one reference: */
1266 refcount_t stack_refcount
;
1268 #ifdef CONFIG_LIVEPATCH
1271 #ifdef CONFIG_SECURITY
1272 /* Used by LSM modules for access restriction: */
1276 #ifdef CONFIG_GCC_PLUGIN_STACKLEAK
1277 unsigned long lowest_stack
;
1278 unsigned long prev_lowest_stack
;
1282 * New fields for task_struct should be added above here, so that
1283 * they are included in the randomized portion of task_struct.
1285 randomized_struct_fields_end
1287 /* CPU-specific state of this task: */
1288 struct thread_struct thread
;
1291 * WARNING: on x86, 'thread_struct' contains a variable-sized
1292 * structure. It *MUST* be at the end of 'task_struct'.
1294 * Do not put anything below here!
1298 static inline struct pid
*task_pid(struct task_struct
*task
)
1300 return task
->thread_pid
;
1304 * the helpers to get the task's different pids as they are seen
1305 * from various namespaces
1307 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1308 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1310 * task_xid_nr_ns() : id seen from the ns specified;
1312 * see also pid_nr() etc in include/linux/pid.h
1314 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
, struct pid_namespace
*ns
);
1316 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1321 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1323 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1326 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1328 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1332 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1338 * pid_alive - check that a task structure is not stale
1339 * @p: Task structure to be checked.
1341 * Test if a process is not yet dead (at most zombie state)
1342 * If pid_alive fails, then pointers within the task structure
1343 * can be stale and must not be dereferenced.
1345 * Return: 1 if the process is alive. 0 otherwise.
1347 static inline int pid_alive(const struct task_struct
*p
)
1349 return p
->thread_pid
!= NULL
;
1352 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1354 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1357 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1359 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1363 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1365 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1368 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1370 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1373 static inline pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1375 return __task_pid_nr_ns(tsk
, PIDTYPE_TGID
, ns
);
1378 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1380 return __task_pid_nr_ns(tsk
, PIDTYPE_TGID
, NULL
);
1383 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
1389 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
1395 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
1397 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
1400 /* Obsolete, do not use: */
1401 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1403 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1406 #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1407 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1409 static inline unsigned int task_state_index(struct task_struct
*tsk
)
1411 unsigned int tsk_state
= READ_ONCE(tsk
->state
);
1412 unsigned int state
= (tsk_state
| tsk
->exit_state
) & TASK_REPORT
;
1414 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX
);
1416 if (tsk_state
== TASK_IDLE
)
1417 state
= TASK_REPORT_IDLE
;
1422 static inline char task_index_to_char(unsigned int state
)
1424 static const char state_char
[] = "RSDTtXZPI";
1426 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX
) != sizeof(state_char
) - 1);
1428 return state_char
[state
];
1431 static inline char task_state_to_char(struct task_struct
*tsk
)
1433 return task_index_to_char(task_state_index(tsk
));
1437 * is_global_init - check if a task structure is init. Since init
1438 * is free to have sub-threads we need to check tgid.
1439 * @tsk: Task structure to be checked.
1441 * Check if a task structure is the first user space task the kernel created.
1443 * Return: 1 if the task structure is init. 0 otherwise.
1445 static inline int is_global_init(struct task_struct
*tsk
)
1447 return task_tgid_nr(tsk
) == 1;
1450 extern struct pid
*cad_pid
;
1455 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1456 #define PF_EXITING 0x00000004 /* Getting shut down */
1457 #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */
1458 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1459 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1460 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1461 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1462 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1463 #define PF_DUMPCORE 0x00000200 /* Dumped core */
1464 #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1465 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1466 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1467 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1468 #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1469 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1470 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1471 #define PF_KSWAPD 0x00020000 /* I am kswapd */
1472 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1473 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1474 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1475 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1476 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1477 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1478 #define PF_MEMSTALL 0x01000000 /* Stalled due to lack of memory */
1479 #define PF_UMH 0x02000000 /* I'm an Usermodehelper process */
1480 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_mask */
1481 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1482 #define PF_MEMALLOC_NOCMA 0x10000000 /* All allocation request will have _GFP_MOVABLE cleared */
1483 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1484 #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1487 * Only the _current_ task can read/write to tsk->flags, but other
1488 * tasks can access tsk->flags in readonly mode for example
1489 * with tsk_used_math (like during threaded core dumping).
1490 * There is however an exception to this rule during ptrace
1491 * or during fork: the ptracer task is allowed to write to the
1492 * child->flags of its traced child (same goes for fork, the parent
1493 * can write to the child->flags), because we're guaranteed the
1494 * child is not running and in turn not changing child->flags
1495 * at the same time the parent does it.
1497 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1498 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1499 #define clear_used_math() clear_stopped_child_used_math(current)
1500 #define set_used_math() set_stopped_child_used_math(current)
1502 #define conditional_stopped_child_used_math(condition, child) \
1503 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1505 #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1507 #define copy_to_stopped_child_used_math(child) \
1508 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1510 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1511 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1512 #define used_math() tsk_used_math(current)
1514 static inline bool is_percpu_thread(void)
1517 return (current
->flags
& PF_NO_SETAFFINITY
) &&
1518 (current
->nr_cpus_allowed
== 1);
1524 /* Per-process atomic flags. */
1525 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1526 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1527 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1528 #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */
1529 #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/
1530 #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */
1531 #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */
1532 #define PFA_SPEC_SSB_NOEXEC 7 /* Speculative Store Bypass clear on execve() */
1534 #define TASK_PFA_TEST(name, func) \
1535 static inline bool task_##func(struct task_struct *p) \
1536 { return test_bit(PFA_##name, &p->atomic_flags); }
1538 #define TASK_PFA_SET(name, func) \
1539 static inline void task_set_##func(struct task_struct *p) \
1540 { set_bit(PFA_##name, &p->atomic_flags); }
1542 #define TASK_PFA_CLEAR(name, func) \
1543 static inline void task_clear_##func(struct task_struct *p) \
1544 { clear_bit(PFA_##name, &p->atomic_flags); }
1546 TASK_PFA_TEST(NO_NEW_PRIVS
, no_new_privs
)
1547 TASK_PFA_SET(NO_NEW_PRIVS
, no_new_privs
)
1549 TASK_PFA_TEST(SPREAD_PAGE
, spread_page
)
1550 TASK_PFA_SET(SPREAD_PAGE
, spread_page
)
1551 TASK_PFA_CLEAR(SPREAD_PAGE
, spread_page
)
1553 TASK_PFA_TEST(SPREAD_SLAB
, spread_slab
)
1554 TASK_PFA_SET(SPREAD_SLAB
, spread_slab
)
1555 TASK_PFA_CLEAR(SPREAD_SLAB
, spread_slab
)
1557 TASK_PFA_TEST(SPEC_SSB_DISABLE
, spec_ssb_disable
)
1558 TASK_PFA_SET(SPEC_SSB_DISABLE
, spec_ssb_disable
)
1559 TASK_PFA_CLEAR(SPEC_SSB_DISABLE
, spec_ssb_disable
)
1561 TASK_PFA_TEST(SPEC_SSB_NOEXEC
, spec_ssb_noexec
)
1562 TASK_PFA_SET(SPEC_SSB_NOEXEC
, spec_ssb_noexec
)
1563 TASK_PFA_CLEAR(SPEC_SSB_NOEXEC
, spec_ssb_noexec
)
1565 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE
, spec_ssb_force_disable
)
1566 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE
, spec_ssb_force_disable
)
1568 TASK_PFA_TEST(SPEC_IB_DISABLE
, spec_ib_disable
)
1569 TASK_PFA_SET(SPEC_IB_DISABLE
, spec_ib_disable
)
1570 TASK_PFA_CLEAR(SPEC_IB_DISABLE
, spec_ib_disable
)
1572 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE
, spec_ib_force_disable
)
1573 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE
, spec_ib_force_disable
)
1576 current_restore_flags(unsigned long orig_flags
, unsigned long flags
)
1578 current
->flags
&= ~flags
;
1579 current
->flags
|= orig_flags
& flags
;
1582 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
, const struct cpumask
*trial
);
1583 extern int task_can_attach(struct task_struct
*p
, const struct cpumask
*cs_cpus_allowed
);
1585 extern void do_set_cpus_allowed(struct task_struct
*p
, const struct cpumask
*new_mask
);
1586 extern int set_cpus_allowed_ptr(struct task_struct
*p
, const struct cpumask
*new_mask
);
1588 static inline void do_set_cpus_allowed(struct task_struct
*p
, const struct cpumask
*new_mask
)
1591 static inline int set_cpus_allowed_ptr(struct task_struct
*p
, const struct cpumask
*new_mask
)
1593 if (!cpumask_test_cpu(0, new_mask
))
1599 extern int yield_to(struct task_struct
*p
, bool preempt
);
1600 extern void set_user_nice(struct task_struct
*p
, long nice
);
1601 extern int task_prio(const struct task_struct
*p
);
1604 * task_nice - return the nice value of a given task.
1605 * @p: the task in question.
1607 * Return: The nice value [ -20 ... 0 ... 19 ].
1609 static inline int task_nice(const struct task_struct
*p
)
1611 return PRIO_TO_NICE((p
)->static_prio
);
1614 extern int can_nice(const struct task_struct
*p
, const int nice
);
1615 extern int task_curr(const struct task_struct
*p
);
1616 extern int idle_cpu(int cpu
);
1617 extern int available_idle_cpu(int cpu
);
1618 extern int sched_setscheduler(struct task_struct
*, int, const struct sched_param
*);
1619 extern int sched_setscheduler_nocheck(struct task_struct
*, int, const struct sched_param
*);
1620 extern int sched_setattr(struct task_struct
*, const struct sched_attr
*);
1621 extern int sched_setattr_nocheck(struct task_struct
*, const struct sched_attr
*);
1622 extern struct task_struct
*idle_task(int cpu
);
1625 * is_idle_task - is the specified task an idle task?
1626 * @p: the task in question.
1628 * Return: 1 if @p is an idle task. 0 otherwise.
1630 static inline bool is_idle_task(const struct task_struct
*p
)
1632 return !!(p
->flags
& PF_IDLE
);
1635 extern struct task_struct
*curr_task(int cpu
);
1636 extern void ia64_set_curr_task(int cpu
, struct task_struct
*p
);
1640 union thread_union
{
1641 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1642 struct task_struct task
;
1644 #ifndef CONFIG_THREAD_INFO_IN_TASK
1645 struct thread_info thread_info
;
1647 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1650 #ifndef CONFIG_THREAD_INFO_IN_TASK
1651 extern struct thread_info init_thread_info
;
1654 extern unsigned long init_stack
[THREAD_SIZE
/ sizeof(unsigned long)];
1656 #ifdef CONFIG_THREAD_INFO_IN_TASK
1657 static inline struct thread_info
*task_thread_info(struct task_struct
*task
)
1659 return &task
->thread_info
;
1661 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1662 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1666 * find a task by one of its numerical ids
1668 * find_task_by_pid_ns():
1669 * finds a task by its pid in the specified namespace
1670 * find_task_by_vpid():
1671 * finds a task by its virtual pid
1673 * see also find_vpid() etc in include/linux/pid.h
1676 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1677 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
, struct pid_namespace
*ns
);
1680 * find a task by its virtual pid and get the task struct
1682 extern struct task_struct
*find_get_task_by_vpid(pid_t nr
);
1684 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
1685 extern int wake_up_process(struct task_struct
*tsk
);
1686 extern void wake_up_new_task(struct task_struct
*tsk
);
1689 extern void kick_process(struct task_struct
*tsk
);
1691 static inline void kick_process(struct task_struct
*tsk
) { }
1694 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
1696 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
1698 __set_task_comm(tsk
, from
, false);
1701 extern char *__get_task_comm(char *to
, size_t len
, struct task_struct
*tsk
);
1702 #define get_task_comm(buf, tsk) ({ \
1703 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1704 __get_task_comm(buf, sizeof(buf), tsk); \
1708 void scheduler_ipi(void);
1709 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
1711 static inline void scheduler_ipi(void) { }
1712 static inline unsigned long wait_task_inactive(struct task_struct
*p
, long match_state
)
1719 * Set thread flags in other task's structures.
1720 * See asm/thread_info.h for TIF_xxxx flags available:
1722 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1724 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1727 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1729 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1732 static inline void update_tsk_thread_flag(struct task_struct
*tsk
, int flag
,
1735 update_ti_thread_flag(task_thread_info(tsk
), flag
, value
);
1738 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1740 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1743 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1745 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1748 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1750 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1753 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1755 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1758 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1760 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1763 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
1765 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
1769 * cond_resched() and cond_resched_lock(): latency reduction via
1770 * explicit rescheduling in places that are safe. The return
1771 * value indicates whether a reschedule was done in fact.
1772 * cond_resched_lock() will drop the spinlock before scheduling,
1774 #ifndef CONFIG_PREEMPT
1775 extern int _cond_resched(void);
1777 static inline int _cond_resched(void) { return 0; }
1780 #define cond_resched() ({ \
1781 ___might_sleep(__FILE__, __LINE__, 0); \
1785 extern int __cond_resched_lock(spinlock_t
*lock
);
1787 #define cond_resched_lock(lock) ({ \
1788 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1789 __cond_resched_lock(lock); \
1792 static inline void cond_resched_rcu(void)
1794 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1802 * Does a critical section need to be broken due to another
1803 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1804 * but a general need for low latency)
1806 static inline int spin_needbreak(spinlock_t
*lock
)
1808 #ifdef CONFIG_PREEMPT
1809 return spin_is_contended(lock
);
1815 static __always_inline
bool need_resched(void)
1817 return unlikely(tif_need_resched());
1821 * Wrappers for p->thread_info->cpu access. No-op on UP.
1825 static inline unsigned int task_cpu(const struct task_struct
*p
)
1827 #ifdef CONFIG_THREAD_INFO_IN_TASK
1828 return READ_ONCE(p
->cpu
);
1830 return READ_ONCE(task_thread_info(p
)->cpu
);
1834 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
1838 static inline unsigned int task_cpu(const struct task_struct
*p
)
1843 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
1847 #endif /* CONFIG_SMP */
1850 * In order to reduce various lock holder preemption latencies provide an
1851 * interface to see if a vCPU is currently running or not.
1853 * This allows us to terminate optimistic spin loops and block, analogous to
1854 * the native optimistic spin heuristic of testing if the lock owner task is
1857 #ifndef vcpu_is_preempted
1858 # define vcpu_is_preempted(cpu) false
1861 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
1862 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
1864 #ifndef TASK_SIZE_OF
1865 #define TASK_SIZE_OF(tsk) TASK_SIZE
1871 * Map the event mask on the user-space ABI enum rseq_cs_flags
1872 * for direct mask checks.
1874 enum rseq_event_mask_bits
{
1875 RSEQ_EVENT_PREEMPT_BIT
= RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT
,
1876 RSEQ_EVENT_SIGNAL_BIT
= RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT
,
1877 RSEQ_EVENT_MIGRATE_BIT
= RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT
,
1880 enum rseq_event_mask
{
1881 RSEQ_EVENT_PREEMPT
= (1U << RSEQ_EVENT_PREEMPT_BIT
),
1882 RSEQ_EVENT_SIGNAL
= (1U << RSEQ_EVENT_SIGNAL_BIT
),
1883 RSEQ_EVENT_MIGRATE
= (1U << RSEQ_EVENT_MIGRATE_BIT
),
1886 static inline void rseq_set_notify_resume(struct task_struct
*t
)
1889 set_tsk_thread_flag(t
, TIF_NOTIFY_RESUME
);
1892 void __rseq_handle_notify_resume(struct ksignal
*sig
, struct pt_regs
*regs
);
1894 static inline void rseq_handle_notify_resume(struct ksignal
*ksig
,
1895 struct pt_regs
*regs
)
1898 __rseq_handle_notify_resume(ksig
, regs
);
1901 static inline void rseq_signal_deliver(struct ksignal
*ksig
,
1902 struct pt_regs
*regs
)
1905 __set_bit(RSEQ_EVENT_SIGNAL_BIT
, ¤t
->rseq_event_mask
);
1907 rseq_handle_notify_resume(ksig
, regs
);
1910 /* rseq_preempt() requires preemption to be disabled. */
1911 static inline void rseq_preempt(struct task_struct
*t
)
1913 __set_bit(RSEQ_EVENT_PREEMPT_BIT
, &t
->rseq_event_mask
);
1914 rseq_set_notify_resume(t
);
1917 /* rseq_migrate() requires preemption to be disabled. */
1918 static inline void rseq_migrate(struct task_struct
*t
)
1920 __set_bit(RSEQ_EVENT_MIGRATE_BIT
, &t
->rseq_event_mask
);
1921 rseq_set_notify_resume(t
);
1925 * If parent process has a registered restartable sequences area, the
1926 * child inherits. Only applies when forking a process, not a thread.
1928 static inline void rseq_fork(struct task_struct
*t
, unsigned long clone_flags
)
1930 if (clone_flags
& CLONE_THREAD
) {
1933 t
->rseq_event_mask
= 0;
1935 t
->rseq
= current
->rseq
;
1936 t
->rseq_sig
= current
->rseq_sig
;
1937 t
->rseq_event_mask
= current
->rseq_event_mask
;
1941 static inline void rseq_execve(struct task_struct
*t
)
1945 t
->rseq_event_mask
= 0;
1950 static inline void rseq_set_notify_resume(struct task_struct
*t
)
1953 static inline void rseq_handle_notify_resume(struct ksignal
*ksig
,
1954 struct pt_regs
*regs
)
1957 static inline void rseq_signal_deliver(struct ksignal
*ksig
,
1958 struct pt_regs
*regs
)
1961 static inline void rseq_preempt(struct task_struct
*t
)
1964 static inline void rseq_migrate(struct task_struct
*t
)
1967 static inline void rseq_fork(struct task_struct
*t
, unsigned long clone_flags
)
1970 static inline void rseq_execve(struct task_struct
*t
)
1976 void __exit_umh(struct task_struct
*tsk
);
1978 static inline void exit_umh(struct task_struct
*tsk
)
1980 if (unlikely(tsk
->flags
& PF_UMH
))
1984 #ifdef CONFIG_DEBUG_RSEQ
1986 void rseq_syscall(struct pt_regs
*regs
);
1990 static inline void rseq_syscall(struct pt_regs
*regs
)
1996 const struct sched_avg
*sched_trace_cfs_rq_avg(struct cfs_rq
*cfs_rq
);
1997 char *sched_trace_cfs_rq_path(struct cfs_rq
*cfs_rq
, char *str
, int len
);
1998 int sched_trace_cfs_rq_cpu(struct cfs_rq
*cfs_rq
);
2000 const struct sched_avg
*sched_trace_rq_avg_rt(struct rq
*rq
);
2001 const struct sched_avg
*sched_trace_rq_avg_dl(struct rq
*rq
);
2002 const struct sched_avg
*sched_trace_rq_avg_irq(struct rq
*rq
);
2004 int sched_trace_rq_cpu(struct rq
*rq
);
2006 const struct cpumask
*sched_trace_rd_span(struct root_domain
*rd
);